Datasets:

ccdv

/

arxiv-summarization

like 70

diffusive acceleration at newtonian shock fronts is an extensively studied phenomenon . detailed discussions of the current status of the investigations can be found in some recent excellent reviews @xcite . while much is by now well understood , some issues are still subjects of much debate , for the theoretical and phenomenological implications that they may have . the most important of these is the backreaction of the accelerated particles on the shock : the violation of the _ test particle approximation _ occurs when the acceleration process becomes sufficiently efficient to generate pressures of the accelerated particles which are comparable with the incoming gas kinetic pressure . both the spectrum of the particles and the structure of the shock are changed by this phenomenon , which is therefore intrinsically nonlinear . at present there are three viable approaches to account for the backreaction of the particles upon the shock : one is based on the ever - improving numerical simulations @xcite that allow a self - consistent treatment of several effects . the second approach is a _ fluid _ approach , and treats cosmic rays as a relativistic second fluid . this class of models was proposed and discussed in @xcite . these models allow one to obtain the thermodynamics of the modified shocks , but do not provide information about the spectrum of accelerated particles . the third approach is analytical and may be very helpful to understand the physics of the nonlinear effects in a way that sometimes is difficult to achieve through simulations , due to their intrinsic complexity . in ref . @xcite a perturbative approach was adopted , in which the pressure of accelerated particles was treated as a small perturbation . by construction this method provides an answer only for weakly modified shocks . an alternative approach was proposed in @xcite , based on the assumption that the diffusion of the particles is sufficiently energy dependent that different parts of the fluid are affected by particles with different average energies . the way the calculations are carried out implies a sort of separate solution of the transport equation for subrelativistic and relativistic particles , so that the two spectra must be somehow connected at @xmath0 _ a posteriori_. recently , in @xcite , the effects of the non - linear backreaction of accelerated particles on the maximum achievable energy were investigated , together with the effects of geometry . the maximum energy of the particles accelerated in supernova remnants in the presence of large acceleration efficiencies was also studied in @xcite . the need for a _ practical _ solution of the acceleration problem in the non - linear regime was recognized in @xcite , where a simple analytical broken - power - law approximation of the non - linear spectra was presented . recently , some promising analytical solutions of the problem of non - linear shock acceleration have appeared in the literature @xcite . these solutions seem to avoid many of the limitations of previous approaches . numerical simulations have been instrumental to identify the dramatic effects of the particles backreaction : they showed that even when the fraction of particles injected from the thermal gas is relatively small , the energy channelled into these few particles can be an appreciable part of the kinetic energy of the unshocked fluid , making the test particle approach unsuitable . the most visible effects of the backreaction of the accelerated particles on the shock appear in the spectrum of the accelerated particles , which shows a peculiar flattening at the highest energies . the analytical approaches reproduce well the basic features arising from nonlinear effects in shock acceleration . while several calculations exist of the nonlinear effects in the shock acceleration of quasi - monochromatic particles injected at a shock surface , there is no description at present of how these effects appear , if they do , when the shock propagates in a medium where ( pre)accelerated particles already exist . the linear theory of this phenomenon was developed by bell @xcite , but has never been generalized to its nonlinear extension . in fact , the backreaction can severely affect the process of re - energization of preaccelerated particles : bell already showed that for strong shocks the energy content of a region where cosmic rays were present could be easily enhanced by a factor @xmath1 at each shock passage , so that equipartition could be readily reached . in these conditions the backreaction of the accelerated particles should be expected . we report here on the first analytical treatment of the shock acceleration in the presence of seed nonthermal particles , with the inclusion of their nonlinear backreaction on the shock . our approach is a generalization of the analytical method introduced in @xcite to describe the nonlinear shock acceleration with monochromatic injection of quasi - thermal particles . in fact , we present here also a general calculation that accounts for both thermal particles and seed nonthermal particles . this case may be of interest for the study of supernova shocks propagating through the interstellar medium ( ism ) where pressure balance exists between gas and cosmic rays . nonlinear effects in shock acceleration of thermal particles result in the appearance of multiple solutions in certain regions of the parameter space . this behaviour resembles that of critical systems , with a bifurcation occurring when some threshold is reached . in the case of shock acceleration , it is not easy to find a way of discriminating among the multiple solutions when they appear . neverthless , in @xcite , a two fluid approach has been used to demonstrate that when three solutions appear , the one with intermediate efficiency for particle acceleration is unstable to corrugations in the shock structure and emission of acustic waves . plausibility arguments may be put forward to justify that the system made of the shock plus the accelerated particles may sit at the critical point , but the author is not aware of any real proof that this is what happens . the physical parameters that play a role in this approach to criticality are the maximum momentum achievable by the particles in the acceleration process , the mach number of the shock , and the injection efficiency , namely the fraction of thermal particles crossing the shock that are accelerated to nonthermal energies . the last of them , the injection efficiency , hides a crucial physics problem by itself , and may play an important role in establishing the level of shock modification . this efficiency parameter in reality is defined by the microphysics of the shock and should not be a free parameter of the problem . unfortunately , our poor knowledge of such microphysics , in particular for collisionless shocks , does not allow us to establish a clear and universal connection between the injection efficiency and the macroscopic shock properties . the paper is structured as follows : in [ sec : nonlin ] we describe the effect of non linearity on shock acceleration and our mathematical approach to describe it . in particular we generalize previous calculations to the case in which seed particles exist in the region where the shock is propagating ; in [ sec : gasdyn ] we describe the gas dynamics in the presence of a non - negligible pressure of accelerated particles . in [ sec : results ] we describe our results , with particular attention for the onset of the particle backreaction and for the appearance of multiple solutions . we conclude in [ sec : conclusions ] . in this section we solve the diffusion - convection equation for the cosmic rays in the most general case in which particles are injected according to some function @xmath2 and the shock propagates within a region where some cosmic ray distribution exists , that we denote as @xmath3 , spatially homogeneous before the shock crossing ( upstream ) . for simplicity we limit ourselves to the case of one - dimensional shocks , but the introduction of different geometrical effects is relatively simple , and in fact many of our conclusions should not be affected by geometry . the equation that describes the diffusive transport of particles in one dimension is @xmath4 - u \frac{\partial f ( x , p)}{\partial x } + \frac{1}{3 } \frac{d u}{d x}~p~\frac{\partial f(x , p)}{\partial p } + q(x , p ) = 0 , \label{eq : trans}\ ] ] where we assumed stationarity ( @xmath5 ) . the @xmath6 axis is oriented from upstream to downstream , as in fig . the presence of pre - existing cosmic rays is introduced as a boundary condition at upstream infinity , by requiring that @xmath7 . one should keep in mind that the common picture of a fluid whose speed is constant ( @xmath8 ) until it hits the shock surface is not appropriate for modified shocks . in fact , in this case , the pressure of the accelerated particles may become large enough to slow down the fluid before it crosses the shock surface . therefore in general at upstream infinity the gas flows at speed @xmath9 , different from @xmath8 ( fluid speed immediately upstream of the shock ) . the two quantities are approximately equal only when the accelerated particles do not have dynamical relevance . the injection term is taken in the form @xmath10 . as a first step , we integrate eq . [ eq : trans ] around @xmath11 , from @xmath12 to @xmath13 , denoted in fig . 1 as points `` 1 '' and `` 2 '' respectively , so that the following equation can be written : @xmath14_2 - \left [ d \frac{\partial f}{\partial x}\right]_1 + \frac{1}{3 } p \frac{d f_0}{d p } ( u_2 - u_1 ) + q_0(p)= 0,\ ] ] where @xmath8 ( @xmath15 ) is the fluid speed immediately upstream ( downstream ) of the shock and @xmath16 is the particle distribution function at the shock location . by requiring that the distribution function downstream is independent of the spatial coordinate ( homogeneity ) , we obtain @xmath17_2=0 $ ] , so that the boundary condition at the shock can be rewritten as @xmath14_1 = \frac{1}{3 } p \frac{d f_0}{d p } ( u_2 - u_1 ) + q_0(p ) . \label{eq : boundaryshock}\ ] ] we can now perform the integration of eq . ( [ eq : trans ] ) from @xmath18 to @xmath12 ( point `` 1 '' ) , in order to take into account the boundary condition at upstream infinity . ( [ eq : boundaryshock ] ) we obtain @xmath19 we can now introduce a quantity @xmath20 defined as @xmath21 whose physical meaning is instrumental to understand the nonlinear backreaction of particles . the function @xmath20 is the average fluid velocity experienced by particles with momentum @xmath22 while diffusing upstream away from the shock surface . in other words , the effect of the average is that , instead of a constant speed @xmath8 upstream , a particle with momentum @xmath22 experiences a spatially variable speed , due to the pressure of the accelerated particles that is slowing down the fluid . since the diffusion coefficient is in general @xmath22-dependent , particles with different energies _ feel _ a different compression coefficient , higher at higher energies if , as expected , the diffusion coefficient is an increasing function of momentum . the role of @xmath20 can also be explained as follows : the distribution function @xmath23 at a distance @xmath6 from the shock surface can be written as @xcite @xmath24,\ ] ] where @xmath25 is the local slope of @xmath26 and the diffusion coefficient @xmath27 has been assumed independent of the location @xmath6 . in first approximation , we can assume that the exponential factor remains important when it is of order unity , namely when its argument is much less than unity . we can therefore introduce a distance @xmath28 , which is the distance at which the exponential equals one . this means that @xmath29\approx u(x_p),\ ] ] so that the speed @xmath20 can be interpreted as the fluid speed at the point @xmath28 where the particles with momentum @xmath22 reverse their motion in the upstream fluid and return to the shock . note that the diffusion coefficient enters the calculation of the distance @xmath28 but does not enter directly the calculation of @xmath20 . in other words , different diffusion coefficients may move the point where the fluid speed is @xmath20 closer to or farther from the shock surface , but do not affect the value of @xmath20 . this approach is similar to that introduced in @xcite . with the introduction of @xmath20 , eq . ( [ eq : step ] ) becomes @xmath30 + u_0 f_\infty + q_0(p ) = 0 , \label{eq : step1}\ ] ] where we used the fact that @xmath31.\ ] ] eq . ( [ eq : step1 ] ) can be written in a way that resembles more the equation for shocks with no particles backreaction but in the presence of seed particles with distribution @xmath3 and injection function @xmath32 : @xmath33 - u_0 f_\infty -q_0(p ) \right\}. \label{eq : transport}\ ] ] the solution of this equation can be written in the following implicit form : @xmath34 @xmath35}{u_{\bar p } - u_2 } \exp\left\{-\int_{\bar p}^p \frac{dp'}{p ' } \frac{3}{u_{p ' } - u_2}\left[u_{p'}+\frac{1}{3}p ' \frac{du_{p'}}{d p'}\right ] \right\}. \label{eq : solut}\ ] ] in the case of monochromatic injection with momentum @xmath36 at the shock surface , we can write @xmath37 where @xmath38 is the gas density immediately upstream ( @xmath12 ) and @xmath39 parametrizes the fraction of the particles crossing the shock which are going to take part in the acceleration process . the injection term in eq . ( [ eq : solut ] ) becomes @xmath40\right\}=\ ] ] @xmath41\right\}. \label{eq : inje}\ ] ] here we introduced the two quantities @xmath42 and @xmath43 , which are respectively the compression factor at the gas subshock and the total compression factor between upstream infinity and downstream . the two compression factors would be equal in the test particle approximation . for a modified shock , @xmath44 can attain values much larger than @xmath45 and more in general , much larger than @xmath46 , which is the maximum value achievable for an ordinary strong non - relativistic shock . the increase of the total compression factor compared with the prediction for an ordinary shock is responsible for the peculiar flattening of the spectra of accelerated particles that represents a feature of nonlinear effects in shock acceleration . in terms of @xmath45 and @xmath44 the density immediately upstream is @xmath47 . in eq . ( [ eq : inje ] ) we can introduce a dimensionless quantity @xmath48 so that @xmath49 introducing the same formalism also for the reacceleration term in eq . ( [ eq : solut ] ) , we obtain the general expression @xmath50 @xmath51 the solution of the problem is known if the velocity field @xmath48 is known . the nonlinearity of the problem reflects in the fact that @xmath52 is in turn a function of @xmath16 as it is clear from the definition of @xmath20 . in order to solve the problem we need to write the equations for the thermodynamics of the system including the gas , the reaccelerated cosmic rays , the cosmic rays accelerated from the thermal pool and the shock itself . we write and solve these equations in the next section . the velocity , density and thermodynamic properties of the fluid can be determined by the mass and momentum conservation equations , with the inclusion of the pressure of the accelerated particles and of the preexisting cosmic rays . we write these equations between a point far upstream ( @xmath18 ) , where the fluid velocity is @xmath9 and the density is @xmath53 , and the point where the fluid upstream velocity is @xmath20 ( density @xmath54 ) . the index @xmath22 denotes quantities measured at the point where the fluid velocity is @xmath20 , namely at the point @xmath28 that can be reached only by particles with momentum @xmath55 . the mass conservation implies : @xmath56 conservation of momentum reads : @xmath57 where @xmath58 and @xmath59 are the gas pressures at the points @xmath18 and @xmath60 respectively , and @xmath61 is the pressure in accelerated particles at the point @xmath28 ( we used the symbol @xmath62 to mean _ cosmic rays _ , in the sense of _ accelerated particles _ ) . the mass flow in accelerated particles has reasonably been neglected . our basic assumption , similar to that used in @xcite , is that the diffusion is @xmath22-dependent and more specifically that the diffusion coefficient @xmath27 is an increasing function of @xmath22 . therefore the typical distance that a particle with momentum @xmath22 moves away from the shock is approximately @xmath63 , larger for high energy particles than for lower energy particles increases with @xmath22 faster than @xmath20 does , therefore @xmath64 is a monotonically increasing function of @xmath22 . ] . as a consequence , at each given point @xmath28 only particles with momentum larger than @xmath22 are able to affect appreciably the fluid . strictly speaking the validity of the assumption depends on how strongly the diffusion coefficient depends on the momentum @xmath22 . the cosmic ray pressure at @xmath28 is the sum of two terms : one is the pressure contributed by the adiabatic compression of the cosmic rays at upstream infinity , and the second is the pressure of the particles accelerated or reaccelerated at the shock ( @xmath65 ) and able to reach the position @xmath28 . since only particles with momentum @xmath66 can reach the point @xmath60 , we can write @xmath67 @xmath68 where @xmath69 is the velocity of particles with momentum @xmath22 , @xmath70 is the maximum momentum achievable in the specific situation under investigation , and @xmath71 is the adiabatic index for the accelerated particles . let us consider separately the case of a strongly modified and weakly modified shock , in order to determine the best choice for @xmath71 . in the case of strongly modified shocks , we will show that most energy is piled up in the region @xmath72 , therefore in this case we can safely adopt @xmath73 , appropriate for a relativistic gas . for weakly modified shocks , the accelerated particles have an approximately power law spectrum with a slope @xmath74 . it can be shown that in this case @xmath75 , so that the relativistic result @xmath73 still applies for @xmath76 ( strong shocks ) . for steeper spectra ( @xmath77 ) a larger adiabatic index should be adopted , but in those cases the solution is basically independent of the choice of @xmath71 because of the weak backreaction of the particles . for the purpose of carrying out our numerical calculation we will therefore always take @xmath73 . the pressure of cosmic rays at upstream infinity is simply @xmath78 where @xmath79 is some minimum momentum in the spectrum of seed particles . for simplicity , we assume that @xmath80 , namely the minimum momentum of the seed particles coincides with the momentum at which particles are injected in the shock and are accelerated . from eq . ( [ eq : pressure ] ) we can see that there is a maximum distance , corresponding to the propagation of particles with momentum @xmath70 such that at larger distances the fluid is unaffected by the accelerated particles and @xmath81 . we will show later that for strongly modified shocks the integral in eq . ( [ eq : cr ] ) is dominated by the region @xmath82 . this improves even more the validity of our approximation @xmath83 . this also suggests that different choices for the diffusion coefficient @xmath27 may affect the value of @xmath70 , but at fixed @xmath70 the spectra of the accelerated particles should not change in a significant way . assuming an adiabatic compression of the gas in the upstream region , we can write @xmath84 where we used mass conservation , eq . ( [ eq : mass ] ) . the gas pressure far upstream is @xmath85 , where @xmath86 is the ratio of specific heats for the gas ( @xmath87 for an ideal gas ) and @xmath88 is the mach number of the fluid far upstream . we introduce now a parameter @xmath89 that quantifies the relative weight of the cosmic ray pressure at upstream infinity compared with the pressure of the gas at the same location , @xmath90 . using this parameter and the definition of the function @xmath52 , the equation for momentum conservation becomes @xmath91 + \frac{1}{\rho_0 u_0 ^ 2 } \frac{d{\tilde p}_{cr}}{dp } = 0.\ ] ] using the definition of @xmath92 and multiplying by @xmath22 , this equation becomes @xmath93 = \frac{4\pi}{3 \rho_0 u_0 ^ 2 } p^4 v(p ) f_0(p ) , \label{eq : eqtosolve}\ ] ] where @xmath16 depends on @xmath52 as written in eq . ( [ eq : laeffe ] ) . ( [ eq : eqtosolve ] ) is therefore an integral - differential nonlinear equation for @xmath52 . the solution of this equation also provides the spectrum of the accelerated particles . the last missing piece is the connection between @xmath45 and @xmath44 , the two compression factors appearing in eq . ( [ eq : solut ] ) . the compression factor at the gas shock around @xmath11 can be written in terms of the mach number @xmath94 of the gas immediately upstream through the well known expression @xmath95 on the other hand , if the upstream gas evolution is adiabatic , then the mach number at @xmath12 can be written in terms of the mach number of the fluid at upstream infinity @xmath88 as @xmath96 so that from the expression for @xmath45 we obtain @xmath97^{\frac{1}{\gamma_g+1}}. \label{eq : rsub_rtot}\ ] ] now that an expression between @xmath45 and @xmath44 has been found , eq . ( [ eq : eqtosolve ] ) basically is an equation for @xmath45 , with the boundary condition that @xmath98 . finding the value of @xmath45 ( and the corresponding value for @xmath44 ) such that @xmath98 also provides the whole function @xmath52 and , through eq . ( [ eq : solut ] ) , the distribution function @xmath26 for the particles resulting from acceleration and reacceleration in the nonlinear regime . if the backreaction of the accelerated particles is small , the _ test particle _ solution must be recovered . in this section we investigate the shock modification due to the backreaction of the accelerated particles . we split the discussion in two parts : in [ subsec : seed ] we consider the case of pre - existing seed particles populating the region where the shock is propagating , and re - energized by the shock . we find the conditions for the onset of the nonlinear regime in which the shock gets modified by the re - accelerated particles . in particular we show that the critical behaviour found for shock acceleration and manifesting itself through the appearance of three solutions , takes place also in this case . in [ subsec : thermal ] we explore the more complicated situation in which a shock accelerates a new population of particles while possibly reaccelerating a pre - existing population of seed particles . here we assume that a fluid is moving with speed @xmath9 in a region where the temperature is @xmath99 . the mach number is some pre - defined value @xmath88 , which can be easily related to @xmath9 and @xmath99 . in the region of interest we assume that a population of seed particles is present with energy per particle already higher than some injection energy necessary for the particles to _ feel _ the shock as a discontinuity . for simplicity we assume that these seed particles have a spectrum which is a power law in momentum in the form : @xmath100 in this section we explore the situation in which the shock reacceleration of a pre - existing population of cosmic rays can modify the shock structure , but there is no acceleration of particles different from those that are already present as seed particles . this is equivalent to turn off the injection term ( @xmath101 in eq . ( [ eq : laeffe ] ) ) . in the next section we discuss what happens when both components are present . the crucial difference between the two components is that the seed particles are by definition already above the threshold for _ injection _ , so that there is no injection efficiency that instead represents such a crucial parameter for the case of acceleration of particles extracted from the thermal pool . having in mind the case of shocks propagating in the interstellar medium of our galaxy , we consider here the case in which the inflowing gas and the seed particles are in pressure equilibrium , namely @xmath102 . in terms of the parameter @xmath103 this implies @xmath104 . the results can then be easily repeated for a generic value of @xmath89 . ( [ eq : eqtosolve ] ) gives the quantity @xmath52 as a function of the momentum @xmath22 for any choice of @xmath45 and @xmath44 , while the acceptable solutions are those with the right matching conditions at @xmath70 . for simplicity , let us assume that the @xmath70 in eq . ( [ eq : f_infty ] ) is the same maximum momentum that particles injected at the shock would achieve : this value only depends on the environmental conditions ( energy losses of the particles ) and/or on the geometry of the shock , which may allow the escape of the particles . the momentum @xmath70 is the same as in the distribution of the pre - exisiting seed particles if , for instance , the seed particles have been accelerated by a shock identical to the one we are considering . in any case , this assumption is not needed for the validity of our conclusions and may be easily relaxed , it simply serves to avoid parameter proliferation . within this assumption , particles re - energized at the shock are simply redistributed in the momentum range between @xmath36 and @xmath70 . the solution , namely the right pair of values for the compression parameters @xmath45 and @xmath44 [ related through eq . ( [ eq : rsub_rtot ] ) ] , is obtained when the solution corresponding to @xmath98 is selected . in fig . [ fig : m150 ] , we plot @xmath105 as a function of @xmath44 for a shock having mach number at infinity @xmath106 . the different lines are labelled by a number representing the @xmath107 of @xmath70 in units of @xmath108 . in other words the parameter @xmath70 changes between @xmath109 and @xmath110 . the physical solutions are found by determining the intersections of each curve with the horizontal line corresponding to @xmath98 . the minimum momentum of the seed particles is taken as @xmath111 . the intersection at @xmath112 is close to the well known linear solution , namely the solution that one would obtain in the test particle approximation . the energy channelled into the nonthermal particles for this solution is very small , and the shock remains approximately unmodified . increasing the value of @xmath70 , the solution moves toward slightly larger values of @xmath44 ( namely the shock becomes more modified ) . for some values of @xmath70 , multiple solutions appear . in particular three regions of @xmath70 can be identified : \1 ) @xmath113 : in this region ( low values of @xmath70 ) the only solution is very close to the one obtained in the test particle approximation . for the values of the parameters in fig . [ fig : m150 ] , @xmath114 . \2 ) @xmath115 : in this region ( intermediate values of @xmath70 three solutions appear , two of which imply a strong modification of the shock , namely an appreciable part of the energy flowing through the shock is converted into energy of the accelerated particles . for the values of the parameters in fig . [ fig : m150 ] , @xmath116 . \3 ) @xmath117 : in this region ( high values of @xmath70 ) the solution becomes one again , and the shock is always strongly modified ( @xmath118 ) . this critical behaviour appears also when one fixes the maximum momentum and uses the mach number of the shock as the order parameter . the results are shown in fig . [ fig : pmax10_5 ] , where mach numbers between 10 and 500 have been considered , at fixed @xmath119 . one can see that for mach numbers below @xmath120 there is only one solution . for mach numbers between @xmath120 and @xmath121 three solutions appear , one of which roughly corresponds to an unmodified shock . for larger mach numbers , only this linear solution remains , and the shock is always only weakly modified . the same situation is plotted in fig . [ fig : pmax10_5rsub ] , where instead of @xmath44 on the x - axis there is @xmath122 , and @xmath45 is the compression coefficient at the gas subshock . for unmodified shocks one expects @xmath123 . again three regions can be identified , in the parameter @xmath88 . \1 ) for @xmath124 only one solution exists and does not necessarily correspond to unmodified shocks . in fact one can see that for @xmath125 , the compression coefficient at the subshock is @xmath126 , and @xmath127 , for the situation plotted in figs . [ fig : pmax10_5 ] and [ fig : pmax10_5rsub ] . even smaller values of @xmath88 do not give a perfectly unmodified shock . \2 ) for @xmath128 three solutions appear , two of which imply a strong modification of the shock . for the parameters used in figs . [ fig : pmax10_5 ] and [ fig : pmax10_5rsub ] , @xmath129 . \3 ) for @xmath130 only the nearly unmodified solution exists . in order to emphasize the fact that the multiple solutions actually correspond to physical solutions with very different spectral characteristics for the accelerated particles we plot in fig . [ fig : spectra ] the spectra for @xmath131 ( for which there is only one solution ) and @xmath106 ( for which there are three solutions ) . in both cases we used @xmath132 . the solution corresponding to mach number @xmath131 has @xmath133 and @xmath134 , resulting in a total pressure of the accelerated particles @xmath135 ( solid line in fig . [ fig : spectra ] ) . the three dashed lines in fig . [ fig : spectra ] are for @xmath106 and represent the spectra for the three solutions . in numbers , these solutions ( from top to bottom in fig . [ fig : spectra ] ) are summarized as follows : _ first solution : _ @xmath136 , @xmath137 , @xmath138 _ second solution : _ @xmath139 , @xmath140 , @xmath141 _ third solution : _ @xmath142 , @xmath143 , @xmath144 in the cases of strong shock modification , the asymptotic spectra for @xmath145 are @xmath146 , while the linear theory , in case of strong shocks ( @xmath147 ) , would predict @xmath148 . in the region of very low energies , the spectra of the reaccelerated particles tend to zero , as known from the linear theory as well . nonlinear effects in shock acceleration were first investigated in the case when a fraction of the thermal gas crossing the shock surface is energized to nonthermal energies . in this section we wish to apply the analytical method discussed in [ sec : nonlin ] and [ sec : gasdyn ] to the most general case in which thermal particles are accelerated but seed particles may already be present in the environment . we start our discussion with the case of acceleration of particles from the thermal distribution , in order to show that the multiple solutions already found in other analytical approaches @xcite are also obtained by adopting the approach illustrated here . in fig . [ fig : noseed ] , we plot the @xmath105 as a function of @xmath44 for the case in which a fraction @xmath39 of the particles ( as indicated in the plots ) crossing the shock is actually accelerated to suprathermal energies . the mach number is chosen to be @xmath106 and the maximum momentum is taken as @xmath149 . one can easily see that @xmath150 , the value for unmodified shocks , for small values of @xmath39 ( low efficiencies ) , while @xmath44 increases above 4 for @xmath151 . three solutions appear for intermediate efficiencies , while the solution for the accelerated particles always predicts a strongly modified shock for high efficiencies , @xmath152 . an asymptotic value of @xmath153 is achieved for the parameters used here . this example demonstrates that the critical behaviour shown to appear for high values of @xmath39 in several analytical or semi - analytical calculations is in fact also predicted by the approach proposed here . next question to ask is however whether the presence of seed particles can change the critical behavior of shocks . in order to answer this question we consider a situation in which particles are accelerated from the thermal pool with efficiency @xmath39 , and at the same time the shock propagates in a medium where the preshock pressure in seed particles equals the gas pressure . this situation is considered to resemble that of a supernova explosion in our galaxy , where cosmic rays fill the volume remaining in quasi - equipartition with the gas . in order to test the critical structure of the shock we calculate @xmath105 as a function of the compression factor @xmath44 between upstream infinity and downstream . our results are plotted in fig . [ fig : crseed ] . the lines refer to the cases @xmath154 from bottom to top , as in fig . [ fig : noseed ] . it is clear from this plot that the shock may be modified by the backreaction of the accelerated particles even for very low values of @xmath39 , because of the presence of seed particles . in other words , the nonlinearity of the shock can well be dominated by the presence of preaccelerated particles rather than by the acceleration of particles from the thermal pool . clearly this depends however on the values of the parameters ( injection momentum , @xmath39 , maximum momentum achievable , mach number ) . the injection momentum , in principle , could be related to @xmath39 in order to reduce the number of free parameters of the problem . this is possible in the assumption that the particles downstream keep a thermal distribution . simulations suggest that the particles injected in the accelerator are the ones with momentum a few times the thermal momentum of the particles downstream , so that @xmath39 can be simply calculated by integration of the maxwell - boltzmann ( mb ) distribution above this minimum momentum . unfortunately , we do not know whether the particle distribution downstream is in fact maxwell - boltzmann - like . moreover , we do not know exactly what is the threshold to impose on the injection momentum , which is a delicate issue because the number of particles taking part to the acceleration process would result from the integral of the mb distribution over its exponentially decreasing part . therefore we preferred here to keep @xmath39 and the injection momentum as separate free parameters . we proposed a semi - analytical approach to show that the backreaction of particles accelerated at a shock is able to affect the shock itself , in such a way that the shock and the accelerated particles become parts of a nonlinear system . in particular , for the first time we included in this kind of calculations the seed particles that may be present in the region where the shock is propagating and that can be re - energized by the shock . while a test - particle approach to this problem was first presented in @xcite , a nonlinear treatment was never investigated . in the pioneering work of ref . @xcite , it was recognized that the energy of the seed particles could be enhanced by about one order of magnitude at each shock passage , and that after an infinite number of strong shocks passing through the region , the spectrum of the particles would tend to the asymptotic spectrum @xmath155 . two comments are in order . first , the continuous increase of the cosmic ray energy due to the re - energization of seed particles leads unavoidably the shock to be modified by the nonthermal pressure , unless one starts from an uninterestingly small pressure of seed particles at the beginning . second , the fact that the spectrum becomes flatter than @xmath156 , which is the result for a strong non - relativistic shock , implies that most of the nonthermal energy is pushed to the highest energies , therefore the shock again can be more easily modified . both these points suggest that a nonlinear treatment of shock re - acceleration is required . in our galaxy , cosmic rays are observed to be in rough equipartition with the gas pressure and with magnetic fields , therefore supernova shocks or shocks generated in other environments propagate in a medium in which the seed particles ( cosmic rays ) are non - negligible . in these circumstances the non - linear effects may be very important . we showed here that in fact for some regions of the parameter space , the shock is modified mainly by the backreaction of the seed particles rather than by the cosmic rays accelerated at the shock from the thermal pool . the spectra of the re - accelerated particles have also been calculated . the interesting phenomenon of the appearance of multiple shock solutions , already known for the case of shock acceleration , appears also for the case of reacceleration . this puzzling phenomenon may suggest that the shock behaves as a self - regulating system settling on the critical point , as proposed in @xcite . on the other hand , it is possible that the multiple solutions may be the artifact of some of the assumptions used in the analytical approach , in particular the request for time independent ( stationary ) solutions and the fact that the role of the self - generated waves on the diffusion coefficient is not taken into account . further investigation , in particular in the direction of a detailed comparison of our results with numerical simulations of shock acceleration is required in order to unveil the physical meaning of the multiple solutions for modified shocks . from the phenomenological point of view , it would be certainly worth to study the implications of nonlinear shock reacceleration on the nonthermal activity in astrophysical environments where the effect is expected to play an important role , in particular in the case of supernova remnants . in particular , as pointed out by the referee , the suggested dominance of reaccelerated ambient seed particles over freshly injected particles may have serious implications for the spectra of secondary nuclei resulting from spallation processes . the author gratefully acknowledges useful discussions with m. baring , d. ellison and m. vietri . the author is also grateful to the referee , l.oc . drury for his very useful remarks . this work was partially supported through grant cofin 2002 at the arcetri astrophysical observatory .

particles crossing repeatedly the surface of a shock wave can be energized by first order fermi acceleration . the linear theory is successful in describing the acceleration process as long as the pressure of the accelerated particles remains negligible compared to the kinetic pressure of the incoming gas ( the so - called test particle approximation ) . when this condition is no longer fulfilled , the shock is modified by the pressure of the accelerated particles in a nonlinear way , namely the spectrum of accelerated particles and the shock structure determine each other . in this paper we present the first description of the nonlinear regime of shock acceleration when the shock propagates in a medium where seed particles are already present . this case may apply for instance to supernova shocks propagating into the interstellar medium , where cosmic rays are in equipartition with the gas pressure . we find that the appearance of multiple solutions , previously found in alternative descriptions of the nonlinear regime , occurs also for the case of reacceleration of seed particles . moreover , for parameters of concern for supernova shocks , the shock is likely to turn nonlinear mainly due to the presence of the pre - existing cosmic rays , rather than due to the acceleration of new particles from the thermal pool . we investigate here the onset of the nonlinear regime for the three following cases : 1 ) seed particles in equipartition with the gas pressure ; 2 ) particles accelerated from the thermal pool ; 3 ) combination of 1 ) and 2 ) . cosmic rays , high energy , acceleration

a plethora of observations concur that the universe at present enters a phase of accelerated expansion . in fact , most cosmologists accept that over 70% of the universe content at present corresponds to the elusive dark energy ; a substance with pressure negative enough to cause the observed acceleration @xcite . the simplest form of dark energy is a positive cosmological constant @xmath0 , which however , needs to be incredibly fine - tuned to explain the observations @xcite . this is why theorists have looked for alternatives , which could explain the observations while setting , as was originally assumed . a promising idea is to consider that the universe at present is entering a late - time inflationary period @xcite . the credibility of this option is supported also by the fact that the generic predictions of inflation in the early universe are in excellent agreement with the observations . the scalar field responsible for this late - inflation period is called quintessence because it is the fifth element after baryons , photons , cdm and neutrinos @xcite . since they are based on the same idea , it is natural to attempt to unify early universe inflation with quintessence . quintessential inflation was thus born @xcite . this attempt has many advantages . firstly , quintessential inflation models allow the treatment of both inflation and quintessence within a single theoretical framework . also , quintessential inflation dispenses with the tuning problem of the initial conditions for quintessence . finally , unified models for inflation and quintessence are more economic because they avoid introducing yet another unobserved scalar field . for quintessential inflation to work one needs a scalar field with a runaway potential , such that the minimum has not been reached until today and , therefore , there is residual potential density , which can cause the observed accelerated expansion . string moduli fields are suitable because they are typically characterised by such runaway potentials . the problem with such fields , however , is how to stabilise them temporarily , in order to use them as inflatons in the early universe . in this work ( see also ref . @xcite ) we achieve this by considering that , during its early evolution our modulus crosses an enhanced symmetry point ( esp ) in field space . when this occurs the modulus is trapped temporarily at the esp @xcite , which leads to a period of inflation . after inflation the modulus picks up speed again in field space resulting into a period of kinetic density domination ( kination ) @xcite . kination ends when the thermal bath of the hot big bang ( hbb ) takes over . during the hbb , due to cosmological friction @xcite , the modulus freezes at some large value and remains there until the present , when its potential density dominates and drives the late - time accelerated expansion @xcite . is is evident that , in order for the modulus to become quintessence , it should not decay after the end of inflation . reheating , therefore should be achieved by other means . we assume that the thermal bath of the hbb is due to the decay of some curvaton field @xcite as suggested in refs . @xcite . by considering a curvaton we do not add an _ ad hoc _ degree of freedom , because the curvaton can be a realistic field , already present in simple extensions of the standard model ( e.g. a right - handed sneutrino @xcite , a flat direction of the ( n)mssm @xcite or a pseudo nambu - goldstone boson @xcite possibly associated with the peccei - quinn symmetry @xcite ) . apart from reheating , the curvaton can provide the correct amplitude of curvature perturbations in the universe . consequently , the energy scale of inflation can be much lower than the grand unified scale @xcite . in fact , in certain curvaton models , the hubble scale during inflation can be as low as the electroweak scale @xcite . string theories contain a number of flat directions which are parametrised by the so - called moduli fields , which correspond to the size and shape of the compactified extra dimensions . many such flat directions are lifted by non - perturbative effects , such as gaugino condensation or d - brane instantons @xcite . the superpotential , then , is of the form @xmath1 where @xmath2 const . is the tree level contribution from fluxes , @xmath3 and @xmath4 are constants and @xmath5 is a khler modulus in units of @xmath6 . hence , the non - perturbative superpotential @xmath7 results in a runaway scalar potential characteristic of string compactifications . for example , in type iib compactifications with a single khler modulus , @xmath8 re(@xmath5 ) is the so - called volume modulus , which parametrises the volume of the compactified space . in this case , the runaway behaviour leads to decompactification of the internal manifold . the tree level khler potential for a modulus , in units of @xmath9 , is @xmath10 and the corresponding supergravity potential is to secure the validity of the supergravity approximation and we have assumed that the esp lies at a minimum in the direction of im(@xmath5 ) . ] @xmath11 to study the cosmology , we turn to the canonically normalised modulus @xmath12 which , due to eq . ( [ tree ] ) , is associated with @xmath13 as @xmath14 suppose that the universe is initially dominated by the above modulus . the non - perturbative scalar potential in eq . ( [ vnp ] ) is very steep ( exponential of an exponential ) , which means that the field soon becomes dominated by its kinetic density . once this is so , the particular form of the potential ceases to be of importance . to achieve inflation we assume that , while rolling , the modulus crosses an esp and becomes temporarily trapped at it . in string compactifications there are distinguished points in moduli space at which there is enhancement of the gauge symmetries @xcite . this results in some massive states of the theory becoming massless at these points . even though from the classical point of view an esp is not a special point , as the modulus approaches it certain states in the string spectrum become massless @xcite . in turn , these massless modes create an interaction potential that may drive the field back to the symmetry point . in that way a modulus can become trapped at an esp @xcite . the strength of the symmetry point depends on the degree of enhancement of the symmetry . such modulus trapping can lead to a period of so - called ` trapped inflation ' @xcite , when the trapping is strong enough to make the kinetic density of the modulus fall below the potential density at the esp . however , it turns out that the number of e - foldings of trapped inflation can not be very large . therefore , with respect to cosmology , the main virtue of the esps relies on their ability to trap the field and hold it there , at least temporarily . because esps are fixed points of the symmetries we have @xmath15 where the prime denotes derivative with respect to @xmath12 and @xmath16 is the value of the modulus at the esp . the above means that the esp is located either at a local extremum ( maximum or minimum ) or at a flat inflection point of the scalar potential , where . this means that the presence of an esp deforms the non - perturbative scalar potential ( see fig . [ esp ] ) . this deformation may be enough so that , after trapped inflation , the field undergoes slow - roll inflation over the flat region of the scalar potential at the vicinity of the esp . the total duration of inflation may , thus , be enough to solve the flatness and horizon problems of the hbb . let us briefly study the trapping of the modulus at the esp . we assume that around the esp there is a contribution to the scalar potential due to the enhanced interaction between the modulus @xmath12 and another field @xmath17 , which we take to be also a scalar field @xcite . the interaction potential is @xmath18 where @xmath19 with @xmath20 being a dimensionless coupling constant . thus , at the esp the @xmath17 particles are massless . the time dependence of the effective ( mass)@xmath21 of the @xmath17 field results in the creation of @xmath17-particles . this takes place when the field is within the production window @xmath22 , where @xmath23 is the kinetic density of the modulus when crossing the esp and the dot denotes derivative with respect to the cosmic time @xmath24 . the effective scalar potential near the esp is @xmath25 where @xmath26 with @xmath27 being the ` background ' scalar potential . following ref . @xcite we have @xmath28 , where @xmath29 denotes the number density of @xmath17 particles produced after the crossing of the esp . this means that @xmath30 and the field climbs a _ linear _ potential since @xmath29 is constant outside the production window . after the first crossing , the field reaches the amplitude @xmath31 , determined by its initial kinetic density . to avoid overshooting the esp we require @xmath32 since for larger values the coupling softens @xcite . after reaching @xmath31 , the field reverses direction and crosses the production window again , generating more @xmath17 particles and , therefore , increasing @xmath33 . thus , it now has to climb a steeper potential reaching an amplitude @xmath34 . the process continues until the ever decreasing amplitude becomes comparable to the production window ( see fig . [ trap1 ] ) . at this moment particle production stops . after the end of particle production , @xmath35 remains roughly constant during an oscillation and the modulus continues oscillating in the quadratic interaction potential . studying this oscillation , we found that , due to the universe expansion , the amplitude and frequency decrease as @xmath36 and @xmath37 @xcite , where the scale factor @xmath38 is normalised to unity at the end of particle production . hence , the quadratic potential becomes gradually `` diluted '' due to the universe expansion ( see fig . [ trap2 ] ) . the above mean that the kinetic density of the oscillating modulus scales as . when @xmath39 becomes redshifted below @xmath40 , trapped inflation begins . the above process assimilates a multitude of initial conditions ( provided overshooting the esp is avoided ) because any kinetic density in excess of @xmath40 is depleted before the onset of trapped inflation . trapped inflation dilutes exponentially the density of the @xmath17particles , which quickly redshifts @xmath41 . therefore , after a rather limited number of e - foldings of trapped inflation , the modulus is released from the esp . since the esp is located at a locally flat region of the potential there is a chance that , after @xmath41 becomes negligible , the modulus drives a period of slow - roll inflation while sliding away from the esp . to study this period we need to quantify the deformation of the scalar potential due to an esp . the appearance of an esp generates either a local extremum or a flat inflection point at @xmath16 . in all cases , in the vicinity of the esp , the scalar potential can be approximated by a cubic polynomial@xcite . hence , the characteristics of the potential depend only on @xmath42 and @xmath43 . in fact , we can parametrise the deformation of the scalar potential using @xcite @xmath44 where @xmath45 and @xmath46 is the hubble parameter during inflation : @xmath47 . the @xmath48 parameter accounts for the strength of the symmetry point ; the smaller the @xmath48 , the stronger the deformation and the wider the inflationary plateau . the requirement that the deformation becomes negligible at distances larger than @xmath6 results in the lower bound @xmath49 , which also guarantees that the modulus does not overshoot the esp @xcite . by studying inflation after the modulus escapes trapping , we have obtained the following results , depending on the esp morphology @xcite . in each case , one has to achieve enough inflationary e - folds to solve the horizon and flatness problems , while also taking care that the curvature perturbations due to the modulus are not excessive compared to observations . consider first the case of a flat inflection point . in this case , we can have enough e - foldings of slow roll inflation if @xmath50 . the case of a local minimum is indistinguishable from the above if @xmath51 . if the opposite is true then the modulus becomes trapped in the local minimum and must escape through tunnelling . afterwards the modulus can drive a period of slow - roll inflation with total number of e - foldings given by @xmath52 . hence , to solve the horizon and flatness problems we need @xmath53 . finally , in the case of a local maximum , after the end of trapping , one can have a phase of fast / slow roll inflation provided @xmath54 . thus , we have found that , in all cases , _ enough slow - roll inflation to solve the horizon and flatness problems of the hbb is attainable provided _ @xmath55 . choosing for illustrative purposes an intermediate value for the hubble scale : @xmath56 tev we have found that one can achieve enough inflationary e - foldings ( up to @xmath57 ) without producing excessive curvature perturbation if @xmath58 . thus , _ there is ample parameter space for slow - roll inflation to occur after the modulus escapes trapping at the esp_. note also , that , while @xmath46 is determined by the location of the esp in field space ( by the vacuum density @xmath40 ) , the values of @xmath59 and @xmath60 are due to the deformation of the scalar potential at the vicinity of the esp which is not directly related to @xmath40 . hence , the requirement that the latter are smaller than @xmath46 does not necessarily imply fine - tunning . after inflation , the field rolls away from the esp . soon the influence of the esp on the scalar potential diminishes and @xmath61 . the steepness of @xmath62 results in the kinetic domination of the modulus density . as a result a period of kination occurs , during which the field equation is : @xmath63 . hence , the density of the universe scales as @xmath64 @xcite . during kination , the scalar field is oblivious of the particular form of the potential . kination is terminated when the density of the decay products of a curvaton field dominates the kinetic density of the modulus @xcite . thus , the end of kination corresponds to reheating , with reheating temperature @xmath65 , where @xmath66 is the hubble parameter at reheating . after the onset of the hbb , the rolling scalar field is subject to cosmological friction @xcite , which asymptotically freezes the field at the value @xmath67 . note that this value depends on @xmath68 which , in turn , is determined by curvaton physics . the modulus remains frozen until the present when it plays the role of quintessence . this guarantees that there is no dangerous variation of fundamental constants during the hbb . the evolution of the modulus until today is depicted in fig . since @xmath69 , the modulus rolls to large values before freezing . at such values we can assume that the scalar potential is @xmath70 where @xmath71 is a density scale and @xmath72 . the above is a typical uplift potential introduced by flux compactifications as discussed below . if the modulus is to account for the required dark energy , it must satisfy the coincidence requirement : @xmath73 , where @xmath74 is the dark energy density parameter and @xmath75 is the critical density at present . hence , @xmath76 thus , the density scale @xmath71 is determined by @xmath68 which , in turn , is determined by curvaton physics . an upper bound on @xmath71 is obtained by demanding that reheating occurs before big bang nucleosynthesis ( bbn ) : @xmath77 where @xmath78 mev is the temperature at bbn . the scalar potential in eq . ( [ exps ] ) may have a multitude of origins . for example , using the volume modulus , we may consider a stack of @xmath79branes located at the tip of a klebanov - strassler throat . the uplift potential is @xcite @xmath80 where @xmath81 and @xmath82 , in the warp factor , are the units of rr and ns three - form fluxes . to satisfy eq . ( [ cbound ] ) we must have @xmath83 . this can be attained by choosing the ratio of fluxes as @xmath84 . taking @xmath85 , only twice as many units of @xmath82 flux as those of @xmath81 flux are needed . it is also possible to consider fluxes of gauge fields on @xmath86branes @xcite . in this case , the scalar potential obtains a contribution @xmath87 where @xmath88 depends on the strength of the gauge fields considered . the constraint in eq . ( [ cbound ] ) requires now @xmath89 tev . the future of the modulus after unfreezing depends on the steepness of the scalar potential , or equivalently the value of @xmath90 in eq . ( [ exps ] ) . @xmath91 for @xmath92 , the modulus dominates the universe for ever , leading to eternal acceleration . this results in future horizons , which pose a problem for the formulation of the s - matrix in string theory @xcite . @xmath91 for @xmath93 , the modulus dominates the universe but results only in a brief accelerated expansion period . such is the fate of the @xmath94 case . @xmath91 for @xmath95 , the modulus does not dominate the universe , albeit causing a brief period of accelerated expansion . afterwards the modulus density remains at a constant ratio with the background matter density . this is the fate of the @xmath96 case . @xmath91 for @xmath97 , the modulus does not cause any accelerated expansion and so can not be used as quintessence . after unfreezing , the modulus rolls fast down the quintessential tail of the scalar potential with its density approaching asymptotically kinetic domination ( and subsequently freezing at a value larger than @xmath98 @xcite ) . this case corresponds to @xmath99 . the brief acceleration period caused by the unfreezing modulus is due to the fact that the modulus oscillates around an attractor solution @xcite , which in itself does not result to acceleration @xcite . in ref . @xcite it was claimed that brief acceleration occurs if @xmath100 , which corresponds to the range @xmath101 . more recent studies , however , have reduced this range . brief acceleration in the range @xmath102 has been confirmed by ref . this includes the @xmath94 case which corresponds to the most popular uplift potential . the range for @xmath90 was expanded further in ref . @xcite , where it is shown that brief acceleration can explain the observations at least up to . since the data are interpreted using a number of priors , we believe that the @xmath96 case is still marginally acceptable . quintessential inflation is possible to achieve in string theory with flux compactifications using as inflaton a string modulus which rolls down its runaway potential . inflation is due to the presence of an enhanced symmetry point ( esp ) , which traps the modulus and creates a locally flat region over which the modulus can slow - roll . there is ample parameter space for successful inflation provided : @xmath103 . trapping assimilates a multitude of initial conditions provided overshooting the esp is avoided . after inflation , the modulus becomes ( again ) kinetically dominated causing a period of kination . reheating is due to the universe domination by the decay products of a curvaton field , which also accounts for the correct amplitude of the curvature perturbations in the universe . during the hot big bang , the modulus freezes and remains constant until the present . at the frozen value , the potential is dominated by an uplift term of exponential form . this residual potential density begins to dominate today , when the modulus unfreezes , leading to a brief period of late inflation . curvaton physics fixes the reheating temperature and determines the value of the frozen modulus @xmath98 and the density scale @xmath71 in the uplift potential . coincidence and bbn constrains on @xmath71 allow realistic values much less fine - tunned than the cosmological constant @xmath0 in the @xmath0cdm model . this work was done with juan c. bueno snchez and was supported by the e.u . marie curie research and training network _ universenet " _ ( mrtn - ct-2006 - 035863 ) and by pparc ( pp / d000394/1 ) . i am grateful to the royal society for supporting my participation to the idm2006 conference . 99 p. j. e. peebles and b. ratra , rev . * 75 * ( 2003 ) 559 ; e. j. copeland , m. sami and s. tsujikawa , int . j. mod . d * 15 * ( 2006 ) 1753 . s. weinberg , rev . * 61 * ( 1989 ) 1 . b. ratra and p. j. e. peebles , phys . d * 37 * ( 1988 ) 3406 . r. caldwell , r. dave and p. j. steinhardt , phys . * 80 * ( 1998 ) 1582 . p. j. e. peebles and a. vilenkin , phys . d * 59 * ( 1999 ) 063505 . r. a. frewin and j. e. lidsey , int . j. mod . d * 2 * ( 1993 ) 323 ; m. peloso and f. rosati , jhep * 9912 * ( 1999 ) 026 ; k. dimopoulos , nucl . . proc . suppl . * 95 * ( 2001 ) 70 ; a. s. majumdar , phys . d * 64 * ( 2001 ) 083503 ; n. j. nunes and e. j. copeland , phys . d * 66 * ( 2002 ) 043524 ; a. gonzalez , t. matos and i. quiros , phys . d * 71 * ( 2005 ) 084029 . k. dimopoulos and j. w. f. valle , astropart . phys . * 18 * ( 2002 ) 287 . k. dimopoulos , phys . d * 68 * ( 2003 ) 123506 . j. c. bueno sanchez and k. dimopoulos , phys . lett . b * 642 * ( 2006 ) 294 ; hep - th/0606223 . l. kofman , a. linde , x. liu , a. maloney , l. mcallister and e. silverstein , jhep * 0405 * ( 2004 ) 030 . m. joyce and t. prokopec , phys . d * 57 * ( 1998 ) 6022 . r. brustein , s. p. de alwis and p. martens , phys . d * 70 * ( 2004 ) 126012 ; t. barreiro , b. de carlos , e. copeland and n. j. nunes , phys . d * 72 * ( 2005 ) 106004 . d. h. lyth and d. wands , phys . b * 524 * ( 2002 ) 5 ; k. enqvist and m. s. sloth , nucl . b * 626 * ( 2002 ) 395 . b. feng and m. z. li , phys . b * 564 * ( 2003 ) 169 ; a. r. liddle and l. a. urena - lopez , phys . d * 68 * ( 2003 ) 043517 . j. mcdonald , phys . d * 68 * ( 2003 ) 043505 ; phys . d * 70 * ( 2004 ) 063520 ; a. mazumdar and a. perez - lorenzana , phys . * 92 * ( 2004 ) 251301 . k. enqvist , s. kasuya and a. mazumdar , phys . * 90 * ( 2003 ) 091302 ; k. enqvist , a. jokinen , s. kasuya and a. mazumdar , phys . d * 68 * ( 2003 ) 103507 ; m. bastero - gil , v. di clemente and s. f. king , phys . d * 67 * ( 2003 ) 103516 ; m. postma , phys . d * 67 * ( 2003 ) 063518 ; j. mcdonald , phys . d * 69 * ( 2004 ) 103511 . s. kasuya , m. kawasaki and f. takahashi , phys . b * 578 * ( 2004 ) 259 ; k. enqvist , mod . a * 19 * ( 2004 ) 1421 . k. dimopoulos , d. h. lyth , a. notari and a. riotto , jhep * 0307 * ( 2003 ) 053 ; e. j. chun , k. dimopoulos and d. lyth , phys . rev . d * 70 * ( 2004 ) 103510 ; r. hofmann , nucl . phys . b * 740 * ( 2006 ) 195 . k. dimopoulos and g. lazarides , phys . d * 73 * ( 2006 ) 023525 . k. dimopoulos , g. lazarides , d. lyth and r. ruiz de austri , jhep * 0305 * ( 2003 ) 057 . k. dimopoulos and d. h. lyth , phys . d * 69 * ( 2004 ) 123509 . k. dimopoulos , d. h. lyth and y. rodriguez , jhep * 0502 * ( 2005 ) 055 ; k. dimopoulos , phys . b * 634 * ( 2006 ) 331 . j. p. derendinger , l. e. ibanez and h. p. nilles , phys . b * 155 * ( 1985 ) 65 ; e. witten , nucl . b * 474 * ( 1996 ) 343 . c. m. hull and p. k. townsend , nucl . b * 451 * ( 1995 ) 525 . s. watson , phys . d * 70 * ( 2004 ) 066005 . r. brustein , s. p. de alwis and e. g. novak , phys . d * 68 * ( 2003 ) 023517 . s. b. giddings , s. kachru and j. polchinski , phys . rev . d * 66 * ( 2002 ) 106006 . c. p. burgess , r. kallosh and f. quevedo , jhep * 0310 * ( 2003 ) 056 . s. hellerman , n. kaloper and l. susskind , jhep * 0106 * ( 2001 ) 003 ; w. fischler , a. kashani - poor , r. mcnees and s. paban , jhep * 0107 * ( 2001 ) 003 ; e. j. copeland , a. r. liddle and d. wands , phys . d * 57 * ( 1998 ) 4686 . j. m. cline , jhep * 0108 * ( 2001 ) 035 . u. j. lopes franca and r. rosenfeld , jhep * 0210 * ( 2002 ) 015 ; r. kallosh , a. linde , s. prokushkin and m. shmakova , phys . rev . d * 66 * ( 2002 ) 123503 ; r. kallosh and a. linde , jcap * 0302 * ( 2003 ) 002 ; a. kehagias and g. kofinas , class . * 21 * ( 2004 ) 3871 . d. blais and d. polarski , phys . d * 70 * ( 2004 ) 084008 .

quintessential inflation is studied using a string modulus as the inflaton quintessence field . it is assumed that the modulus crosses an enhanced symmetry point ( esp ) in field space . particle production at the esp temporarily traps the modulus resulting in a period of inflation . after reheating , the modulus freezes due to cosmological friction at a large value , such that its scalar potential is dominated by contributions due to fluxes in the extra dimensions . the modulus remains frozen until the present , when it can become quintessence .

transmission through multiple transmit and multiple receive antennas , or the so - called multiple - input multiple - output ( mimo ) technology , is known as an efficient means for providing enormous information rates in rich - scattering mobile environments @xcite-@xcite . characterization of the fading mimo channel capacity limits , under various assumptions on the transmitter - side and receiver - side channel - state information ( csi ) and channel - distribution information ( cdi ) , has motivated a great deal of valuable scholarly work ( e.g. , @xcite and references therein ) . in particular , the case where the csi is perfectly _ known _ at the receiver but _ unknown _ at the transmitter has drawn much interest due to its validity in many practical situations . this is because the presumption of perfect receiver - side csi is usually reasonable for wireless channels where the receiver can locally estimate the fading channel , while the complete csi feedback from the receiver to the transmitter is difficult or even impossible . consequently , many schemes that exploit various forms of _ partial _ channel feedback have been proposed in literature . among others , the _ transmit - covariance feedback _ scheme is known to be capable of achieving data rates close to the fading mimo channel ergodic capacity when the channel cdi exhibits some long - term consistent statistical properties , e.g. , constant channel mean and/or constant channel covariance matrix @xcite-@xcite . in the transmit - covariance feedback scheme , the receiver determines the transmit signal covariance matrix based on the cdi , and then sends it back to the transmitter through a feedback channel . in @xcite-@xcite , the feedback transmit covariance matrix for optimizing the channel ergodic capacity , and the conditions under which the beamforming the transmit covariance matrix has rank one is optimal , have been established for the single - user fading channel . in this scheme , the transmit covariance matrix is fixed as long as the cdi is not changed . therefore , this scheme requires much less feedback complexity and is also more robust to the delay of the feedback channel compared to other partial channel feedback schemes based on the instantaneous mimo channel realizations ( e.g. , @xcite , @xcite and references therein ) . this paper considers the fading mac with additive white gaussian noise ( awgn ) at the receiver , and assumes that the csi from each mobile terminal ( mt ) to the base station ( bs ) is unknown at each mt transmitter , but is perfectly known at the bs receiver . thus , the bs can acquire the channel cdi for each mt . this paper extends the transmit - covariance feedback scheme for the single - user fading mimo channel to the fading mimo multiple - access channel ( mimo - mac ) where multiple antennas are used by the bs and possibly by each mt . two multiple - access techniques are considered for scheduling transmissions from each mt to the bs at the same frequency : _ space - division multiple - access _ ( sdma ) and _ time - division multiple - access _ ( tdma ) . for sdma , all mts transmit simultaneously to the bs and their individual signals are decoded jointly at the bs while for tdma , each mt transmits independently to the bs during mutually orthogonal time slots and thus only single - user decoding is needed . the multiuser transmit - covariance feedback scheme is then described as follows . for sdma , the bs first jointly optimizes the transmit signal covariance matrices for all mts , based on the multiuser cdi as well as the rate requirement and the power budget of each mt , and then sends them back to each mt for transmission . this scheme has also been considered in @xcite , @xcite for characterizing the capacity region and establishing the conditions for the optimality of beamforming for the fading mimo - mac , respectively . in contrast , for tdma , the bs jointly optimizes the duration of transmission time slot for each mt along with their transmit covariance matrices . these optimized values are then sent back to each corresponding mt via the feedback channel . this paper studies the information - theoretic limits of the fading mimo - mac under the multiuser transmit - covariance feedback scheme when either sdma or tdma is employed . two commonly adopted means to measure the information - theoretic limits of multiuser channels are the _ capacity region _ and the _ power region_. the capacity region is defined as the constitution of all achievable rate - tuples for the users given their individual power constraints , while the power region consists of all possible power - tuples for the users under which a given rate - tuple is achievable . this paper is mainly concerned with the characterization of the multiuser power region . our motivations are justified as follows : first , characterization of the power region for the fading mimo - mac is a challenging problem . considering initially the case of sdma , the capacity region of a _ deterministic _ ( no fading ) gaussian mac with a single transmit and a single receive antenna ( siso - mac ) has the well - known _ polymatroid _ structure @xcite , which also holds for the fading mimo - mac . on the other hand , the power region of a _ deterministic _ siso - mac is known to have a _ contra - polymatroid _ structure @xcite . the polymatroid and the contra - polymatroid structures can be utilized to reduce significantly the computational complexity of finding the boundary points of the capacity region and the power region , respectively @xcite , @xcite . however , the contra - polymatroid structure is non - existent for the power region when the channel exhibits fading @xcite and/or the bs uses multiple antennas @xcite .. ] as a result , characterization of the power region for the fading mimo - mac under sdma is yet fully understood in literature . on the other hand , for tdma , given the duration of each mt transmission time slot ( e.g. , equal time - slot durations for all mts in the conventional tdma ) , the bs only needs to optimize the transmit covariance matrices for the mts independently such that their individual transmission powers are minimized for supporting their own target rates . however , with time - slot duration adjustment for each mt , the bs now needs to consider the more challenging problem of jointly optimizing the time - slot durations and the transmit covariance matrices for all mts . this joint optimization has been less studied in literature . secondly , characterization of the power region can potentially provide very useful insights on the resource allocation problem for wireless networks , e.g. , the wireless cellular network . in many practical situations , each bs in the cellular network controls the transmit power of each mt in its cell such that each individual mt rate demand transmission quality - of - service ( qos ) is satisfied ( e.g. , @xcite , @xcite ) . power control can be beneficial in many aspects , e.g. , to maintain fair power consumption among mts , to tailor for each mt s peak - power constraint , and to mitigate the co - channel interference between multiple cells operating at the same frequency . by exploiting the multiuser power region , the minimum power consumption in the network can be achieved under practical transmission constraints . the main contributions of this paper are summarized as follows : * for both sdma and tdma , the paper presents efficient algorithms for characterizing each boundary point of the power region for the fading mimo - mac . the developed algorithms are based on a lagrange primal - dual approach that exploits a novel dual relationship between the power region and the corresponding capacity region for the fading mimo - mac . for sdma , the proposed algorithm determines jointly the optimal transmit covariance matrices for all mts as well as their optimal decoding order at the receiver . for tdma , all mt transmit covariance matrices along with their assigned variable time - slot durations are jointly optimized . * in addition to the conventional way to characterize the boundary of the power region by solving a sequence of user _ weighted _ sum - power minimization problems subject to fixed user rate constraints , this paper proposes an alternative means for such characterization by considering the user sum - power minimization problem under different _ user power - profile _ constraints , where the user power - profile regulates the power consumption of users under some given proportional fairness . the remainder of this paper is organized as follows . section [ sec : system model ] introduces the system model for the fading mimo - mac and describes the proposed multiuser transmit - covariance feedback scheme under sdma and tdma . section [ sec : power region ] provides the definition of the power region for the fading mimo - mac , together with two problem formulations for characterization of the power region , based on the user weighted sum - power minimization and the user power - profile vector , respectively . section [ sec : sdma ] and section [ sec : tdma ] study the power region in the case of sdma and tdma , respectively , and present efficient algorithms for characterizing the power region in each case . section [ sec : simulations ] provides numerical results to verify the usefulness of the proposed algorithms under realistic channel parameters . finally , section [ sec : conclusions ] concludes the paper . notations _ : scalar signals are denoted by lower - case letters , e.g. , @xmath0 . bold - face lower - case letters are used for vector signals , e.g. , @xmath1 , and bold - face upper - case letters for matrices , e.g. , @xmath2 , @xmath3 . @xmath4 denotes the determinant , @xmath5 the inverse and @xmath6 the trace of a square matrix @xmath2 . for any general matrix @xmath3 , @xmath7 and @xmath8 denote its transpose and conjugate transpose , respectively . @xmath9 denotes the identity matrix . @xmath10 $ ] denotes statistical expectation . @xmath11 denotes the space of @xmath12 matrices with complex entries . @xmath13 denotes the @xmath14-dimensional real euclidean space and @xmath15 is the nonnegative orthant . the distribution of a circular symmetric complex gaussian ( cscg ) vector with the mean vector @xmath16 , and the covariance matrix @xmath17 is denoted by @xmath18 , and @xmath19 means `` distributed as . '' the sign @xmath20 denotes the generalized inequality @xcite and for a square matrix @xmath2 , @xmath21 means that @xmath2 is positive semi - definite . @xmath22 denotes the minimum between two real numbers @xmath23 and @xmath24 . this paper considers a narrow - band flat fading mimo - mac with @xmath25 receive antennas at the bs and @xmath26 mts equipped with @xmath27 transmit antennas , respectively . all mts transmit synchronously to the bs by sharing a common frequency band . it is assumed that the space of fading states is continuous and infinite , and the fading process is stationary and ergodic . under the assumption that the transmitted symbol period is equal to the inverse of the common transmission bandwidth for all mts , at each fading state @xmath28 , the equivalent discrete - time mac is given by @xmath29 where @xmath30 denotes the received signal vector , @xmath31 and @xmath32 denote , respectively , the transmitted signal vector and the channel matrix of mt @xmath33 , @xmath34 ; @xmath35 denotes the vector of additive noise at the receiver , and it is assumed that @xmath36 . this channel model also assumes that the csi is perfectly known at the bs but is unknown at each mt . with the csi available , the bs can acquire the long - term csi statistics ( or equivalently , the cdi ) of each mt . based on the multiuser cdi , the bs determines the transmit signal covariance matrices for all mts jointly according to their individual rate requirement and power budget , and then sends them back to each mt for transmission . this paper refers to this scheme as _ multiuser transmit - covariance feedback_. let the transmit covariance matrix of mt @xmath33 be @xmath37 $ ] , where the expectation is taken over the code - book and @xmath38 . @xmath39 is assumed to be _ fixed _ for all fading states @xmath28 . [ fig : mac model ] illustrates the system model considered in this paper . since this paper is concerned with the information - theoretic limits of a gaussian mac , the optimal gaussian code - book is assumed for each mt , i.e. , @xmath40 . the transmit covariance matrix of mt @xmath33 can be expressed by its eigenvalue decomposition as @xmath41 @xmath42 is known as the _ precoding matrix _ where @xmath43 , and @xmath44 . @xmath45 is usually referred to as the _ spatial multiplexing gain _ as it measures the number of degrees of transmission freedom ( or equivalently , the number of data streams ) in the spatial domain . if @xmath45 is equal to one , the associated transmission scheme is usually referred to as _ beamforming_. @xmath46 is a @xmath47 diagonal matrix with positive diagonal elements that provide the _ power - loading _ to the corresponding transmitted data streams . the transmitter of each mt can be implemented as the cascade of the following operations : encoding the information bits by the optimal gaussian code - book , interleaving coded symbols randomly into each data stream , power - loading and then jointly precoding all data streams . next , two multiple - access techniques considered in this paper are illustrated , namely , sdma and tdma . for sdma , all mts transmit simultaneously to the bs . in this paper , it is assumed that the bs receiver uses the optimal ( capacity - achieving ) multiuser detection . for a fixed set of @xmath48 , @xmath34 , all the rate - tuples in the set , @xmath49 , defined below , are achievable ( e.g. , @xcite , @xcite ) : @xmath50 \ \forall \mathcal{j}\subseteq \{1,\ldots , k\ } \right\}.\end{aligned}\ ] ] the code - book of mt @xmath33 should satisfy @xmath51 , where @xmath52 denotes the vector of average transmit powers for the mts . on the other hand , for tdma , the bs divides the total transmission time into multiple transmission blocks of equal duration @xmath53 . each transmission block is then further divided into @xmath26 non - overlapping time slots assigned to the mts . these time slots are assumed fixed over all blocks . during the time slot of mt @xmath33 , only this mt communicates with the bs and other mts are silent , i.e. , @xmath54 in ( [ eq : mimo mac ] ) . let @xmath55 denote the time - slot duration assigned to mt @xmath33 , where @xmath56 and @xmath57 . the bs determines jointly the slot duration for each mt and their transmit covariance matrices , and then sends them back to each mt . for a fixed set of @xmath58 and @xmath48 , @xmath34 , each mt transmits over a single - user fading mimo channel studied in e.g. , @xcite , @xcite , and thus the following rate - tuples in the set , @xmath59 , defined below are achievable : @xmath60 \ \forall k\in \{1,\ldots , k\ } \right\}.\end{aligned}\ ] ] and , again , @xmath52 denotes the average transmit powers for the mts . notice that for tdma , the actual transmit power for mt @xmath33 during its assigned time - slot duration @xmath55 is @xmath61 , but the average transmit power @xmath62 over each block duration @xmath53 is @xmath63 , the same as sdma . in this work , for both sdma and tdma , we consider each user s achievable rate in the `` ergodic '' sense , i.e. , averaged over all ergodic fading states . in the case of fast - fading channel , the resultant ergodic capacity can be achievable by assigning each mt a constant - rate code - book for which the codeword length is sufficiently long so as to exploit the ergodicity of the channel . in contrast , in the case of slow - fading channel , each mt s codeword might not be able to span over all possible fading states because of practical transmission delay constraint . however , if each mt uses multiple code - books with variable rates , the bs , based on the instantaneous channel , can determine the transmission rate of each mt and then sends back the corresponding code - book index to each mt for transmission . as in the fast - fading case , the same ergodic capacity ( sometimes known as the expected capacity ) can be achieved for each mt via time - averaging its transmission rates over different fading states . the multiuser _ power region _ for the fading mac is defined as the constitution of all user power - tuples under which a given set of rates is achievable for all the mts . let @xmath64 denote the vector of rate requirements for the mts . the power region is then defined as @xmath65 for sdma , and @xmath66 for tdma . it is not hard to show that the power regions for both sdma and tdma are convex sets . the power region is illustrated in fig . [ fig : power region defined ] for a two - user fading mac under either sdma or tdma . the solid line in fig . [ fig : power region defined ] represents the boundary points of the power region , which correspond to all _ pareto _ optimal power - tuples each minimizing a weighted - sum of the powers among the mts . each boundary point , e.g. , point a as indicated in fig . [ fig : power region defined ] , might be characterized by two alternative means described as follows . first , because of the convexity of the power region , each boundary point can be expressed as the solution to a _ weighted sum - power minimization _ ( w - spmin ) problem stated below , for some nonnegative user weights , @xmath67 , @xmath34 . for sdma , the w - spmin problem can be expressed as [ prob : w - spmin sdma ] @xmath68 notice that @xmath69 are auxiliary variables . for tdma , this w - spmin problem is given by [ prob : w - spmin tdma ] @xmath70 by definition of the power region for a given target rate - tuple @xmath71 , each boundary point of the power region , @xmath72 , under either sdma or tdma , can be expressed as a power - tuple supporting the target rates in @xmath71 that has the minimum weighted - sum , @xmath73 , for some weight vector @xmath74 , among all the power - tuples that can support @xmath71 . alternatively , the connection between the power - tuple @xmath72 on the boundary of the power region and the target rate @xmath71 can be established by employing the capacity region concept . [ fig : power capacity dual ] gives an illustration for this important observation . consider the power region of a two - user mac under the rate constraint @xmath75 , as shown in fig . [ fig : power capacity dual ] ( a ) . given @xmath76 and @xmath77 , the solution to the w - spmin problem is denoted by @xmath78 , which is represented by point a in fig . [ fig : power capacity dual ] ( a ) , and satisfies @xmath79 , where @xmath80 is the minimum value of the w - spmin problem . thus , the required minimum power - pair is @xmath78 for @xmath75 . on the other hand , it is not hard to verify that the rate - pair @xmath75 is on the boundary of the capacity region for the same mac under the _ weighted sum - power constraint _ given by @xmath81 . this is shown by point b in fig . [ fig : power capacity dual ] ( b ) . moreover , because of the convexity of the capacity region , @xmath82 must be the solution to a _ weighted sum - rate maximization _ ( w - srmax ) problem for a given nonnegative user weight vector @xmath83 , as shown in fig . [ fig : power capacity dual ] ( b ) . the above observation has an important consequence , i.e. , each power - region boundary point as the solution to the w - spmin problem can be equivalently characterized as a boundary point of the corresponding capacity region under the same weighted sum - power constraint . as will be shown later in this paper , this result also motivates the proposed algorithms for the w - spmin problem under both sdma and tdma . alternatively , each boundary point of the power region can also be considered geometrically as the intersection of a line passing through the origin ( the power - tuple with all zeros ) and the boundary of the power region ( see point a in fig . [ fig : power region defined ] ) . let each line be characterized as @xmath84 , for @xmath85 and @xmath86 . the vector @xmath87 is referred to as the _ user power - profile _ vector in this paper , and it is assumed that @xmath88 . the point where the line specified by @xmath89 intersects the power region boundary can be then obtained as the solution to the following optimization problem , referred to as the _ sum - power minimization under the power - profile constraint _ ( spmin - ppc ) . for sdma , this problem can be expressed as [ prob : spmin - pf sdma ] @xmath90 and similarly , the spmin - ppc problem can be defined for tdma . the characterization of the power region via some prescribed power - profile vector is useful for the bs to regulate the power consumption of mts in a desired _ proportionally - fair _ manner . it is not hard to show that both problems , w - spmin and spmin - ppc , are convex for either sdma or tdma , and hence , they can be solved by applying convex optimization techniques . the following two sections present the solutions to these problems for sdma and tdma , respectively . this subsection considers problem [ prob : w - spmin sdma ] , the w - spmin problem in the case of sdma . for the special case of a deterministic ( no fading ) siso - mac , the w - spmin can be simplified using the contra - polymatroid structure as proposed in @xcite , @xcite . however , as shown later in this subsection , the approach in @xcite , @xcite can not be applied here to handle the more general case of the fading mimo - mac . thus , an alterative approach is proposed . * deterministic siso - mac : * consider a deterministic siso - mac where the channel gain for each mt in ( [ eq : mimo mac ] ) is a positive constant , i.e. , @xmath91 . in this case , the solutions to the w - spmin problem are obtained as follows . [ theorem : siso - mac ] for a deterministic siso - mac consisting of @xmath26 users with channel gains @xmath92 , and rate requirements @xmath93 , the solutions to the w - spmin problem under sdma are given by ( * ? ? ? * lemma 3.2 ) : @xmath94 where the permutation @xmath95 indicates the optimal decoding order ( user @xmath96 is decoded last and user @xmath97 is decoded first ) at the receiver according to @xmath98 fig . [ fig : power capacity siso ] shows the connection between each boundary point on the power region and the corresponding capacity region earlier described in section [ sec : power region ] for a two - user deterministic siso - mac . [ fig : power capacity siso ] ( a ) shows the power region given rate constraint @xmath99 . consider a vertex @xmath100 of this power region , and arbitrary positive weights @xmath101 such that @xmath102 . according to theorem [ theorem : siso - mac ] , @xmath100 is the optimal solution to the w - spmin problem , and is achievable by decoding order @xmath103 ( user 2 s message is decoded before user 1 s ) . for this given choice of @xmath76 and @xmath77 , the boundary curve of the capacity region under the _ weighted - sum power constraint _ @xmath104 , shown in fig . [ fig : power capacity siso ] ( b ) , can be represented as the union of all rate regions , @xmath105 defined in ( [ eq : definition c sdma ] ) , over all values of @xmath106 and @xmath107 that satisfy @xmath108 . each @xmath105 is a pentagon with two vertices , each corresponding to one of the two possible decoding orders among the users @xcite . the fact that user 2 s message must be decoded first to achieve @xmath75 can be justified by the following observation . it is seen that the decoding order @xmath103 always achieves higher rates for both users than the other decoding order @xmath109 , hence , it must be the optimal decoding order to achieve the rate target @xmath99 on the boundary of this capacity region . in general , for a deterministic siso - mac , all rate - tuples on the boundary of the capacity region under a weighted sum - power constraint can be achieved by a _ unique _ decoding order for each user . this result is consistent with theorem [ theorem : siso - mac ] , i.e. , given @xmath67 s and @xmath110 s , the optimal decoding order of users can be first resolved by ( [ eq : decoding orders ] ) , and then their minimum powers can be found by ( [ eq : optimal powers ] ) . * fading mimo - mac : * unfortunately , the contra - polymatroid structure is non - existent for the mac when there is fading ( e.g. , the fading siso - mac ) or there are multiple antennas at the receiver ( e.g. , the deterministic simo - mac ) , and hence , it is non - existent for the general fading mimo - mac . equivalently , the rate - tuples on the boundary of the capacity region under the weighted sum - power constraint for a fading mimo - mac , unlike the case of a deterministic siso - mac , might not correspond to a unique decoding order for each user . [ fig : capacity region fading simo ] illustrates this fact by showing the capacity region of a two - user fading simo - mac ( @xmath111 ) under a sum - power constraint @xmath112 ( i.e. , @xmath113 ) . in this case , the channels @xmath114 and @xmath115 are assumed to be independent vectors each having independent entries distributed as @xmath116 . the capacity region for this case is shown to be symmetric . the dashed line and the dotted line show how two vertices of the constituting @xmath105 sweep on the boundary of the capacity region as @xmath106 and @xmath107 vary while their sum is kept equal to 10 . it is observed that the boundary rate - tuples of this capacity region indeed correspond to different decoding orders ; e.g. , point a corresponds to the decoding order @xmath109 , while point d does for @xmath117 . there is also a part of the capacity region ( e.g. , point b is in this region ) that does not consist of any vertices . this part of the region is referred to as the _ time - sharing _ region and consists of the -45 degree boundary lines of the constituting @xmath105 . hence , any point in the time - sharing region is not achievable by successive decoding given any fixed decoding orders , and time - sharing the transmission rates and the decoding orders among two users is required . as a result , unlike the deterministic siso - mac , given @xmath67 s and @xmath118 s , the optimal decoding order for the w - spmin problem can not be resolved directly for this channel . one heuristic approach ( e.g. , @xcite-@xcite ) for solving w - spmin problem for the fading mimo - mac under sdma might be searching through all possible @xmath119 decoding orders and then finding the optimal decoding order that gives the minimum weighted sum - power to support the target rates . this approach might work for some special cases , but as explained below , is problematic in general . for any fixed decoding order @xmath95 , problem [ prob : w - spmin sdma ] can be written as @xmath120\geq r_{\pi(k ) } \ \ \forall k \label{eq : rate given decoding order } \\ & & { \mbox{\boldmath{$ s $ } } } _ k \succeq 0 \ \ \ \forall k,\end{aligned}\ ] ] where the left - hand - side ( lhs ) of ( [ eq : rate given decoding order ] ) is the achievable rate for mt @xmath121 , denoted as @xmath122 , under the decoding order @xmath95 . except for mt @xmath96 , the rate constraints in ( [ eq : rate given decoding order ] ) are not convex , rendering the above optimization problem non - convex in general and , hence , it can not be solved efficiently . a suboptimal method that approximately solves this problem is described as follows . starting from the last decoded mt @xmath96 , the method minimizes the power required to maintain the target rate for that mt , while considering mts that have yet been decoded as interference . for example , for the two - user case and the decoding order of mt 2 followed by mt 1 , the method first determines @xmath123 with the minimum @xmath106 that satisfies latexmath:[$\mathbb{e}_{\nu}[\frac{1}{2 } \log and then fixes @xmath123 and determines @xmath125 with the minimum power @xmath107 that satisfies @xmath126 \geq r_1+r_2 $ ] . each of these two optimizations are convex and , hence , they both can be solved efficiently . the above algorithm is referred to as the _ greedy _ algorithm since each mt simply minimizes its own transmit power . for the special case of the fading siso - mac and simo - mac where each @xmath39 is simply a scalar and is equal to @xmath62 , the obtained @xmath106 and @xmath107 via the greedy algorithm are in fact optimal and minimize the weighted sum - power for any weights @xmath74 under the given decoding order . this is because from ( [ eq : rate given decoding order ] ) it can be shown that the minimum @xmath127 required to support @xmath128 for user @xmath121 is an increasing function of the powers for the not - yet - decoded users , @xmath129 . however , for the general fading mimo - mac , the above greedy algorithm might not be optimal because each mt now can adjust its covariance matrix to balance between minimizing its own transmit power and reducing the interference it causes to the users decoded earlier in the order . nevertheless , even if the w - spmin problem can be solved for each decoding order , the obtained powers that have the minimum weighted - sum among all decoding orders might still be suboptimal . this can occur when the target rate - tuple does not correspond to a unique optimal decoding order , e.g. , the rate - pair b in fig . [ fig : capacity region fading simo ] that is on the boundary of the time - sharing region . from the above discussions , it follows that for the fading mimo - mac in general , the decoding order and the transmit covariance matrices for the mts need to be jointly optimized for solving the w - spmin problem under sdma . this motivates the algorithm presented next . * proposed algorithm : * the proposed algorithm for problem [ prob : w - spmin sdma ] is based on its lagrangian @xcite , which is given below : @xmath130 where @xmath131 denotes the vector of dual variables associated with the rate inequality constraints in ( [ eq : vr constraint sdma ] ) . the variables , @xmath48 and @xmath69 belong to the set denoted by @xmath132 , which is specified by the remaining constrains in ( [ eq : domain def_1 sdma ] ) and ( [ eq : domain def_2 sdma ] ) . then the lagrange dual function @xcite is defined as @xmath133 the dual function serves as a lower bound on the optimal value of the original ( primal ) problem , denoted by @xmath134 , i.e. , @xmath135 @xcite . the dual problem is then defined as @xmath136 @xcite . let the optimal value of the dual problem be denoted by @xmath137 that is achievable by the optimal dual variables @xmath138 , i.e. , @xmath139 . for a convex optimization problem , the slater s condition states that the duality gap , @xmath140 , is indeed zero if the primal problem has a feasible solution set @xcite . by using sufficiently large user powers , the set @xmath49 in ( [ eq : domain def_1 sdma ] ) can be made arbitrarily large to contain any finite rate target @xmath71 as an interior point . in other words , we can always find a feasible set @xmath48 that satisfies any given rate constraint @xmath71 for problem [ prob : w - spmin sdma ] . thus , the slater s condition holds and the duality gap is zero for problem [ prob : w - spmin sdma ] . this result suggests that @xmath141 can be obtained by first minimizing the lagrangian @xmath142 to obtain the dual function @xmath143 for some given @xmath144 , and then maximizing @xmath143 over all possible values @xmath144 . first , consider the minimization of @xmath142 to obtain the dual function @xmath143 . in this case , @xmath144 is fixed and the variables are @xmath48 and @xmath69 . from ( [ eq : lagrange power min sdma ] ) , it follows that the minimization of @xmath142 can be rewritten as the following equivalent problem : @xmath145 from the definition of @xmath146 in ( [ eq : definition c sdma ] ) , there are @xmath147 rate constraints implied by ( [ eq : polymatroid constraint ] ) , which are difficult to be incorporated directly into the optimization . in order to simplify the problem , the following theorem is utilized to remove these constraints in ( [ eq : polymatroid constraint ] ) : [ theorem : polymatroid ] for any permutation @xmath95 over @xmath148 and a fixed set of covariance matrices @xmath48 , @xmath149 defined as @xmath150\end{aligned}\ ] ] is a _ vertex _ of the polymatroid @xmath49 in @xmath151 . furthermore , for any @xmath152 , the solution to the following w - srmax problem : @xmath153 is attained by a vertex @xmath154 , where @xmath155 is such that @xmath156 . please refer to ( * ? ? ? * lemma 3.10 ) . notice that theorem [ theorem : polymatroid ] holds for any given @xmath157 . furthermore , since minimization of @xmath158 is equivalent to maximization of @xmath159 , using theorem [ theorem : polymatroid ] , the rate constraints in ( [ eq : polymatroid constraint ] ) can be safely removed and the problem in ( [ eq : general minpower mac ] ) can be simplified as @xmath160 \label{eq : mac polymatroid new } \\ \mathtt { subject \ to } & & { \mbox{\boldmath{$ s $ } } } _ k \succeq 0 \ \ \forall k,\end{aligned}\ ] ] where @xmath95 is a permutation such that @xmath161 . by rearranging the terms regarding user rates in ( [ eq : mac polymatroid new ] ) , the above problem becomes the minimization of @xmath162,\end{aligned}\ ] ] with only optimization variables @xmath163 , @xmath164 , and @xmath165 . since the above problem is convex with a twice differentiable objective function and positive semi - definite constraints , it can be solved numerically , e.g. , by the interior - point method @xcite . next , the dual function @xmath143 is maximized over all possible values @xmath144 . the search of @xmath144 towards its optimal value @xmath138 can be done , e.g. , by the ellipsoid method @xcite , which utilizes the fact that the vector @xmath166 , defined as @xmath167 for @xmath34 , is a _ sub - gradient _ of @xmath143 at any @xmath144 , where @xmath168 } and @xmath169 are the minimizers of @xmath170 obtained via solving ( [ eq : mac polymatroid ] ) , i.e. , @xmath171 . it is noted that the algorithm proposed in ( * ? ? ? * algorithm 5.3 ) can also be modified to determine @xmath172 for the problem at hand . however , from programming implementations , it is observed that this method may exhibit oscillation when some of @xmath173 s happen to be equal . in contrast , the ellipsoid method is more suitable because of its robust and superior convergence behavior . the complete algorithm for problem [ prob : w - spmin sdma ] in the case of sdma is summarized below . [ algorithm : minpower mac sdma ] * * given * an ellipsoid @xmath174\subseteq \mathbb{r}^k$ ] , centered at @xmath175 $ ] and containing the optimal dual solution @xmath172 . * * set * @xmath176 . * * repeat * * * for given @xmath177 $ ] , solve the problem given in ( [ eq : mac polymatroid ] ) to obtain an optimal solution set @xmath178\}$ ] and @xmath179\}$ ] that minimizes @xmath180)$ ] over @xmath132 ; * * update the ellipsoid @xmath181 $ ] based on @xmath182 $ ] and the sub - gradients @xmath183 = r_k - r_k[i ] , k=1,\ldots , k$ ] . set @xmath184 $ ] as the center of the new ellipsoid @xmath181$];. ] * * set @xmath185 . * * until * the stopping criteria for the ellipsoid method is met . one possible method to obtain the initial ellipsoid @xmath174 $ ] that contains the optimal dual solution @xmath172 is given in appendix [ appendix : initial ellipsoid ] . the primal - dual approach used for solving problem [ prob : w - spmin sdma ] can be explained by the connection between the power region and the capacity region as described in section [ sec : power region ] . in fig . [ fig : power capacity dual ] , it is observed that each power - tuple on the boundary of the power region for a given target rate - tuple , as the solution to problem [ prob : w - spmin sdma ] for a given weight vector @xmath74 , defines a capacity region that contains the target rate - tuple on its boundary . consequently , the target rate - tuple can be expressed as the solution to a w - srmax problem for some unknown user weight vector @xmath83 over this capacity region . clearly , the primal - dual approach establishes the above connection by finding the optimal dual variable @xmath138 that is simply one candidate for the unknown weight vector @xmath83 in this w - srmax problem . from ( [ eq : mac polymatroid ] ) , it follows that the optimal decoding order of each mt is given by the magnitude of @xmath173 , i.e. , the optimal decoding order @xmath95 satisfies @xmath186 . hence , the proposed algorithm successfully jointly optimizes the decoding order and the transmit covariance matrices of the mts by exploiting the duality between the power region and the capacity region . * uniqueness of solutions : * so far , the proposed algorithm determines the optimal value of the primal problem @xmath141 ( equal to that of the dual problem @xmath137 ) , the corresponding primal variables @xmath187 and @xmath188 ( it is yet claimed that these primal variables are the primal optimal solutions ) , and the dual optimal solutions @xmath172 that satisfy , @xmath189 in the following , the issue on the uniqueness of these solutions is addressed . while uniqueness of the dual optimal variables @xmath172 is not an issue in the convergence of the proposed algorithm , can be viewed as the weight vector @xmath83 that attains the given rate requirements @xmath71 as the solution to the w - srmax problem over the corresponding capacity region . for the fading mimo - mac , as shown in fig . [ fig : capacity region fading simo ] , in general there is no `` sharp '' vertex with multiple tangent lines on the boundary of the capacity region under a weighted sum - power constraint . as a result , it can be inferred that @xmath172 is unique for any rate - tuple on the boundary and , hence , the uniqueness of @xmath138 is in general ensured . ] uniqueness of the primal variables , @xmath187 and @xmath188 , plays an important role in obtaining valid solutions for the w - spmin problem . since a primal - dual approach is used , the obtained primal variables that minimize the lagrangian at @xmath138 might not satisfy the rate constraints in ( [ eq : vr constraint sdma ] ) . notice that these variables are minimizers of the lagrangian and are not necessarily the primal optimal solutions . however , according to karush - kuhn - tucker ( kkt ) optimality conditions @xcite , the primal optimal solutions also minimize the lagrangian at @xmath138 . hence , if these lagrangian minimizers can be proven to be unique , it follows that they satisfy the rate constraints in ( [ eq : vr constraint sdma ] ) automatically . [ theorem : uniqueness of sk ] the primal optimal solutions @xmath187 for problem [ prob : w - spmin sdma ] under sdma is unique . please refer to appendix [ appendix : proof sk ] . if all @xmath173 s are positive and distinct , @xmath190 ( e.g. , shown by point a in fig . [ fig : capacity region fading simo ] ) that maximizes @xmath191 over @xmath192 will be one unique vertex of @xmath193 , which itself is also unique according to theorem [ theorem : uniqueness of sk ] . in this case , from the kkt conditions , @xmath194 s automatically satisfy the rate constraints in ( [ eq : vr constraint sdma ] ) . however , if @xmath173 s in some subset @xmath195 are positive and equal , @xmath194 s for the users in the set @xmath196 may not be unique and consequently they may not satisfy the rate constraints in ( [ eq : vr constraint sdma ] ) . this can be shown by point b , c and d in fig . [ fig : capacity region fading simo ] where @xmath197 . any point on the straight line between point c and d maximizes @xmath191 over @xmath198 because all these rate - pairs have the same sum - rate . hence , if point b is the rate demand for our problem , because the simplified optimization in ( [ eq : mac polymatroid ] ) always tends to use a vertex of the capacity region as the solution for @xmath190 , the proposed algorithm would converge to either point c or d as @xmath190 , which clearly does not satisfy the rate constraints . however , this is not an issue in the convergence of the proposed algorithm . as far as @xmath187 is unique , the target rate - pair is ensued to be some convex combination of at most @xmath26 vertices of the unique @xmath199 . this subsection presents the solutions to problem [ prob : spmin - pf sdma ] , the spmin - ppc problem under sdma , where the transmit power of each mt is regulated according to a given power - profile vector @xmath89 . the proposed algorithm is also based on a lagrange primal - dual approach . the lagrangian of the primal problem in ( [ eq : power profile min ] ) can be written as @xmath200 where @xmath131 and @xmath201 are dual variables associated with the inequality constraints in ( [ eq : vr constraint_new ] ) and ( [ eq : power constraint ] ) , respectively . let @xmath202 denote the set of primal variables specified by the remaining constraints in ( [ eq : domain def_1 new ] ) , ( [ eq : domain def_2 new ] ) and ( [ eq : power positive constraint ] ) , the lagrange dual function can be then expressed as @xmath203 from ( [ eq : lagrange dual power min new ] ) , it is necessary that @xmath204 for the dual function to be bounded from below . in the case of @xmath205 , the optimal @xmath206 that minimizes the lagrangian over @xmath202 can take any positive value ; while in the case of @xmath207 , the optimal @xmath206 must be equal to zero . thus , in both case , the term associated with @xmath206 in ( [ eq : lagrange dual power min new ] ) is indeed zero and , hence , can be removed from this point onward . the optimal value of @xmath206 can be then obtained as @xmath208 similar algorithm like algorithm [ algorithm : minpower mac sdma ] for problem [ prob : w - spmin sdma ] can be readily developed for solving this problem . it can be shown that for the problem at hand , @xmath209 has sub - gradients , @xmath166 and @xmath210 for @xmath144 and @xmath211 , respectively , which are defined as @xmath212 and @xmath213 , @xmath34 , where @xmath214 and @xmath169 are the lagrangian minimizers that satisfy @xmath215 . the convergence of the algorithm for this problem and the uniqueness of the solutions are similar as algorithm [ algorithm : minpower mac sdma ] . let @xmath216 and @xmath217 denote the corresponding primal and dual optimal solutions . it is worth mentioning here that the obtained primal solution for this problem @xmath218 might not be equal to the user sum - power @xmath219 since some of the power constraints in ( [ eq : power constraint ] ) , @xmath220 , may not be active in general . ( a ) , if the given power - profile vector @xmath89 for problem [ prob : spmin - pf sdma ] is such that the intersected boundary power - tuple @xmath221 of the power region is located on the vertical ( or horizontal ) boundary segment of the power region , the solutions to problem [ prob : spmin - pf sdma ] will converge to the upper ( or lower ) vertex @xmath78 of the power region for which , clearly , @xmath222 , but @xmath223 . thus , @xmath224 . ] actually , @xmath225 shown as follows . since @xmath226 , it can be verified from ( [ eq : lagrange dual power min new ] ) that @xmath227 . moreover , by the kkt conditions for the power constraints , @xmath228 . hence , @xmath229 . the primal - dual approach used for solving the spmin - ppc problem is also based on the connection between the power region and the corresponding capacity region , similar as that for the w - spmin problem . however , their difference lies in that for the spmin - ppc problem , with reference to fig . [ fig : power capacity dual ] , the weight vector for characterizing the solution on the boundary of the power region , @xmath74 , and the weight vector for the corresponding capacity region , @xmath83 , are both unknown and , hence , they need to be found under the given power - profile vector @xmath89 as the corresponding optimal dual solution @xmath217 and @xmath138 , respectively . in contrast , for the w - spmin problem , only the unknown @xmath83 needs to be found as the optimal dual solution @xmath138 because @xmath74 is already given . this section considers the characterization of the power region defined in ( [ eq : power def tdma ] ) for the fading mimo - mac under tdma , and for brevity only the w - spmin problem ( problem [ prob : w - spmin tdma ] ) is investigated . the alternative means for characterizing the power region based on the power - profile vector , i.e. , the spmin - ppc problem under tdma , is omitted since it can be readily obtained given the techniques developed in section [ subsec : power profile ] for the case of sdma . the proposed algorithm for problem [ prob : w - spmin tdma ] is also based on the lagrange primal - dual approach . first , it is noted that the constraints in ( [ eq : vr constraint tdma ] ) and ( [ eq : domain def_1 tdma ] ) in problem [ prob : w - spmin tdma ] can be combined and thus simplified , given the fact that in ( [ eq : definition c tdma ] ) , the inequality constraints are always satisfied with equalities for power minimization . hence , the lagrangian of the primal problem can be written as @xmath230-r_k\right),\end{aligned}\ ] ] where @xmath131 denotes the dual variables associated with the constraints in ( [ eq : vr constraint tdma ] ) . the variables , @xmath48 and @xmath58 , belong to the set denoted by @xmath231 , which is specified by the remaining constrains in ( [ eq : domain def_2 tdma ] ) , ( [ eq : domain def_3 tdma ] ) and ( [ eq : domain def_4 tdma ] ) . also note that the lagrangian is a convex function of both @xmath48 and @xmath58 . by changing the variables as @xmath232 , the lagrangian can be rewritten as @xmath233-r_k\right).\end{aligned}\ ] ] for @xmath34 , define @xmath234,\end{aligned}\ ] ] and @xmath235 from ( [ eq : lagrange power min tdma rewritten ] ) , ( [ eq : zk ] ) , and ( [ eq : zk hat ] ) , the lagrange dual function can be expressed as @xmath236 the optimal value of the primal problem , denoted by @xmath134 , can be then obtained as @xmath237 [ theorem : tdma ] if @xmath238 are the optimal primal solutions and @xmath239 are the optimal dual solutions for problem [ prob : w - spmin tdma ] under the strictly positive weight vector @xmath74 and rate target @xmath71 , they must satisfy @xmath240 where @xmath241 , @xmath242=r_k$ ] , and @xmath243 is a constant . since @xmath71 and @xmath74 are both strictly positive , then so are the obtained solutions @xmath244 . as a result , it is necessary to have @xmath245 , otherwise the minimization in ( [ eq : lagrange dual power min tdma ] ) must lead to only one user assigned with the total time slot , i.e. , @xmath246 and @xmath247 where @xmath248 . using this equality and also ( [ eq : zk ] ) , ( [ eq : zk hat ] ) , the proof is completed . using theorem [ theorem : tdma ] , the algorithm for problem [ prob : w - spmin tdma ] can be obtained as follows : in each iteration , the algorithm updates the dual variables @xmath144 such that @xmath249 . it then checks whether the obtained rates can support more than the target rates , and increases @xmath250 if they do or decreases it otherwise in the next iteration , until the rate targets are exactly met and @xmath250 converges to @xmath243 . the details for the proposed algorithm are presented below . [ algorithm : minpower mac tdma ] * * given * @xmath251 . * * repeat * * * @xmath252 . * * for each @xmath33 , obtain the optimal solutions @xmath253 and @xmath254 such that @xmath255 . , @xmath256 can be obtained by minimizing @xmath257 over @xmath258 as in ( [ eq : zk hat ] ) by means of a convex optimization method , e.g. , the interior - point method @xcite . the quantity @xmath254 for which @xmath259 can then be obtained by a bisection search over @xmath260 using the fact that @xmath256 is a decreasing function of @xmath260 . ] do the above for @xmath34 . * * compute @xmath261 such that @xmath262 $ ] for @xmath34 . * * if @xmath263 , @xmath264 ; otherwise @xmath265 . * * until * @xmath266 where @xmath267 is a small positive constant that controls the algorithm accuracy . since @xmath268 , it follows that @xmath269 . thus , we can take @xmath270 . similar as appendix [ appendix : initial ellipsoid ] , we can obtain the upper bounds @xmath271 s on the optimal dual solutions @xmath173 s . from ( [ eq : lagrange power min tdma ] ) and using the fact that @xmath272 , it is easy to show that @xmath273 . thus , we can take @xmath274 . at last , the convergence of the above algorithm as well as the uniqueness of the obtained solutions are ensured by the uniqueness of @xmath243 in theorem [ theorem : tdma ] . this section presents the power region for a fading mimo - mac with @xmath275 receive antennas and @xmath276 mts each equipped with @xmath277 transmit antennas . it is assumed that the receive antennas at the bs are sufficiently separated that they experience independent fading , while the fading levels are correlated across the transmit antennas because of their realistic size limitations . under this assumption , the employed channel model for mt @xmath33 is given by @xmath278 for @xmath279 , where @xmath280 denotes the transmit antenna correlation matrix for mt @xmath33 and is assumed to be constant over all fading states of @xmath28 . @xmath281 denotes the rayleigh - fading channel matrix that is independent across two mts and across all fading states , and has independent entries distributed as @xmath116 . similar as the proof given in @xcite and @xcite , it can be shown that the expressions in ( [ eq : mac polymatroid ] ) and ( [ eq : zk ] ) , for sdma and tdma , respectively , are both minimized when the transmit signal covariance matrix , @xmath39 , has the same set of eigenvectors as @xmath282 , @xmath279 , i.e. , if @xmath283 is the eigenvalue decomposition of @xmath282 , the optimal @xmath39 then takes the form of @xmath284 for some diagonal matrix @xmath46 . this observation reduces the number of ( real ) variables from @xmath285 in @xmath39 to @xmath286 in @xmath46 for mt @xmath33 and in turn reduces the total algorithm complexity . monte - carlo simulation with 5000 independent realizations of the random channels is used to approximate the actual expectation over the fading states . the simulation assumes that the target rate is @xmath287^t$ ] nats / sec / hz for two mts , and @xmath288 , \ \ \ \ \ \ \ \ \ \ \ { \mbox{\boldmath{$ q $ } } } _ { 2 } = \left[\begin{array}{cc } 1 & 0.5 \\ 0.5 & 1 \\ \end{array}\right].\end{aligned}\ ] ] fig . [ fig : power region sdma ] and fig . [ fig : power region tdma ] show the power region for this fading mac under sdma and tdma obtained by solving problem [ prob : w - spmin sdma ] and problem [ prob : w - spmin tdma ] , respectively . for the sdma case as is shown in fig . [ fig : power region sdma ] , there are two corner points on the boundary of the power region , denoted by point a and b , which can be obtained by the greedy algorithm described in section [ subsec : weighted sum power ] . recall that in this greedy algorithm , the bs first picks one possible decoding order for the mts , and then starting from the last decoded mt , it minimizes the power required to maintain the target rate for each mt while considering the mts that have yet been decoded as interference . in this figure , the power - pair a corresponds to the decoding order @xmath103 , while the power - pair b does for the reversed decoding order . the greedy algorithm achieves the optimal power - pairs for the w - spmin problem under some weight vectors , e.g. , point a for @xmath289 and point b for @xmath290 , but might be suboptimal for weight vectors other than these extreme choices . for example , the minimum value of @xmath291 to support the target rate @xmath292^t$ ] is 11.5 while the greedy algorithm leads to 12.8 and 13.3 units of power for power - pair a and b , respectively . moreover , the boundary curve of the power region is not attainable by simply time - sharing these two corner points , as shown by the dashed line in fig.[fig : power region sdma ] . the power region under tdma for this mac is shown in fig . [ fig : power region tdma ] . for comparison , the power region under sdma is also included in this figure . this figure uses the log scale for the powers and , hence , the power region under sdma looks as a non - convex set . the power savings achieved by sdma are observed to be substantial compared to tdma , even in this case where the number of transmit antennas at each mt is equal to that of the receive antennas at the bs , i.e. , both sdma and tdma have the same number of degrees of transmission freedom in the spatial domain , which is two in this case . the power - pairs on the boundary of the power region under tdma correspond to different time - slot durations , @xmath293 , @xmath279 , assigned to each mt . the power - pair a shown in fig . [ fig : power region tdma ] is achieved by assigning equal duration of time slot for both mts , i.e. , @xmath294 as in the conventional tdma . clearly , this power - pair is optimal for the w - spmin problem under a unique weight vector @xmath74 , and for the spmin - ppc problem under a unique power - profile vector @xmath89 , but it is suboptimal in all the other cases . for example , the minimum sum - power @xmath295 to achieve the target rate @xmath292^t$ ] is 42 units of power for @xmath296 and @xmath297 , as compared to 69 units of power for @xmath294 . this paper characterizes the power region for the fading mimo - mac . motivated by a general relationship between the power region and the corresponding capacity region , the lagrange primal - dual approach is employed to characterize all pareto optimal power - tuples on the boundary of the power region . these optimal power - tuples provide different power tradeoff among the mts and also ensure the fairness of power consumption among them . the algorithms developed in this paper can be used in the wireless cellular network for the bs to control the transmit powers from the mts in the uplink transmission . two multiple access techniques , namely , sdma and tdma , are considered in this paper . it is observed that substantial power savings can be obtained by using sdma compared to tdma . this observation provides an important information - theoretic guidance for practical system designs , i.e. , if the complexity for implementing the optimal sdma can be tailed for , an enormous capacity gain is still possible over the conventional tdma - based network . the multiuser transmit - covariance feedback scheme studied in this paper optimizes the transmit covariance matrices of all the mts based on their long - term cdi . hence , this scheme reduces significantly the feedback complexity compared to other feedback schemes based on the instantaneous channel realizations . as a result , this scheme is practically suitable for wireless channels that exhibit some consistent long - term channel statistics . the results obtained in this paper can provide insightful guidelines to many applications in wireless networks including resource allocation , partial channel feedback , and multiuser space - time code design . in the appendix , one possible method to obtain the initial ellipsoid @xmath174 $ ] for algorithm [ algorithm : minpower mac sdma ] is presented . first , we obtain an upper bound @xmath298 on @xmath299 for any given @xmath300 . let @xmath301 be any set of transmit covariance matricies that achieve @xmath302 given by @xmath303 from the definition of the dual function given by ( [ eq : lagrange dual power min sdma ] ) , we have @xmath304 since @xmath305 , it follows that @xmath306 thus , @xmath307 . similar upper bounds can be found for all other @xmath308 . next , @xmath174 $ ] can be chosen to cover the hyper - cube in @xmath309 specified by @xmath298 s . this appendix proves the uniqueness of the solutions for the optimal transmit covariance matrices @xmath187 in problem [ prob : w - spmin sdma ] . without loss of generality , it is assumed that @xmath310 . if @xmath311 and @xmath312 are two sets of optimal solutions for problem [ prob : w - spmin sdma ] , from ( [ eq : lagrange power min sdma ] ) and by using theorem [ theorem : polymatroid ] , it follows that @xmath313,\end{aligned}\ ] ] for @xmath314 . since the problem at hand is convex , for any @xmath315 $ ] , @xmath316 is also an optimal solution to satisfy ( [ eq : two optimal sk ] ) , where @xmath317 . this fact together with the concavity of the @xmath318 function implies that @xmath319=0,\end{aligned}\ ] ] where @xmath320 . let @xmath321 denote the function on the lhs of the above equation , then @xmath322 , for all @xmath323 . because @xmath321 is twice continuously differentiable , both of its first and second derivatives must vanish , i.e. , @xmath324=0 , \ \ \beta.\end{aligned}\ ] ] for every @xmath28 , the matrix in @xmath325 of the above equation is a positive semi - definite matrix and , hence , it has a nonnegative trace . since the expectation of a nonnegative random variable is zero , it must be zero _ a.s . _ , or @xmath326 _ , which implies that @xmath327 . s. a. jafar and a. goldsmith , `` transmitter optimization and optimality of beamforming for multiple antenna systems with imperfect feedback , '' _ ieee trans . wireless commun . 3 , no.4 , pp . 1165 - 1175 , jul . 2004 . e. jorswieck and h. boche , `` channel capacity and capacity - range of beamforming in mimo wireless systems under correlated fading with covariance feedback , '' _ ieee j. select . areas commun . 5 , pp . 1543 - 1553 , sep . 2004 . r. zhang , y. c. liang , r. narasimhan , and j. m. cioffi , `` approaching mimo - ofdm capacity with per - antenna power and rate feedback , '' _ ieee j. sel . areas commun . _ , vol . 25 , no . 7 , pp . 1284 - 1297 , sep . 2007 . a. soysal and s. ulukus , `` transmit directions and optimality of beamforming in mimo - mac with partial csi at the transmitters , '' in _ proc . conference on inform . , science , and systems ( ciss ) _ , mar . d. tse and s. hanly,``multi - access fading channels - part i : polymatroid structure , optimal resource allocation and throughput capacities , '' _ ieee trans . inf . theory _ , 44 , no . 7 , pp . 2796 - 2815 , nov . 1998 . h. viswanathan , s. venkatesan and h. c. huang , `` downlink capacity evaluation of cellular networks with known interference cancellation , '' _ ieee j. sel . areas commun . 802 - 811 , jun . 2003

this paper is concerned with the fading multiple - input multiple - output multiple - access channel ( mimo - mac ) with multiple receive antennas at the base station ( bs ) and multiple transmit antennas at each mobile terminal ( mt ) . two multiple - access techniques are considered for scheduling transmissions from each mt to the bs at the same frequency , which are _ space - division multiple - access _ ( sdma ) and _ time - division multiple - access _ ( tdma ) . for sdma , all mts transmit simultaneously to the bs and their individual signals are resolved at the bs via multiple receive antennas while for tdma , each mt transmits independently to the bs during mutually orthogonal time slots . it is assumed that the channel - state information ( csi ) of the fading channel from each mt to the bs is _ unknown _ at each mt transmitter , but is perfectly _ known _ at the bs receiver . thereby , the bs can acquire the long - term channel - distribution information ( cdi ) for each mt . this paper extends the well - known _ transmit - covariance feedback _ scheme for the point - to - point fading mimo channel to the fading mimo - mac , whereby the bs jointly optimizes the transmit signal covariance matrices for all mts based on their cdi , and then sends each transmit covariance matrix back to the corresponding mt via a feedback channel . the main goal of this paper is to characterize the so - called _ multiuser power region _ under the multiuser transmit - covariance feedback scheme for both sdma and tdma . the power region is defined as the constitution of all user transmit power - tuples that can achieve reliable transmissions for a given set of user target rates . simulation results show that sdma can achieve substantial power savings over tdma for the fading mimo - mac , even when the number of antennas at the bs is equal to that at each mt . moreover , this paper demonstrates the usefulness of the multiuser power region for maintaining _ proportionally - fair _ power consumption among the mts . multiple - input multiple - output ( mimo ) , multi - antenna systems , gaussian multiple - access channel ( mac ) , fading channel , capacity region , power region , partial channel feedback , space - division multiple - access ( sdma ) , time - division multiple - access ( tdma ) , proportional fairness , resource allocation , convex optimization . [ section ]

in vancouver a popular form of chinese luncheon is `` dim sum '' in which small quantities of a large variety of foods may be tasted . this review is a `` dim sum '' of parity violation experiments . as with a luncheon , my selection is biased by my personal taste and experience . i start with @xmath0 parity violation experiments , concentrating on the the triumf 221 mev @xmath0 experiment , then discuss @xmath3 parity violation experiments with details of the los alamos @xmath4 experiment now being installed at lansce . finally , i discuss @xmath6 parity violation experiments , particularly the gzero experiment at jefferson lab . i refer those interested in more background to specific reviews on nucleon - nucleon @xcite and @xmath6 @xcite experiments . figure [ pptypes ] shows typical @xmath0 parity violation experiments . they scatter a longitudinally polarized beam of protons from a hydrogen target and measure the difference in cross section for right - handed and left - handed proton helicities . the intermediate and high energy experiments use transmission geometry in which the change in scattering cross section is deduced from the change in transmission through the target . low energy experiments , where energy loss limits the target thickness , use scattering geometry , in which the detectors measure the scattered protons directly . both types of experiments measure the parity violating longitudinal analyzing power , @xmath9 , where @xmath10 and @xmath11 are the scattering cross sections for positive and negative helicity . .summary of @xmath0 parity violation experiments . the long times taken to achieve small uncertainties reflects the time taken to understand and correct for systematic errors . in cases where authors reported both statistical and systematic uncertainties , this table shows the quadrature sum of the two . [ cols="<,<,^ , > " , ] [ epexp ] the gzero experiment completed a successful commissioning run of the forward angle configuration in fall 2002 and january 2003 and all major systems are now fully operational . running will continue with an engineering run october to december , 2003 , and production running is scheduled to start in 2004 . parity violation experiments provide a way to study effects of the weak interaction in the presence of the much stronger electromagnetic and strong nuclear interactions . the polarized beam experiments i have described use similar experimental techniques and face similar problems controlling systematic errors . the physics addressed by these experiments can , however be quite diverse . @xmath3 experiments constrain the weak pion - nucleon coupling constant , @xmath5 . @xmath0 parity violation experiments are sensitive to the shorter range part of the nucleon - nucleon force and constrain the combinations @xmath12 and @xmath13 . finally , @xmath6 parity violation experiments , such as the jlab gzero experiment , offer the opportunity to measure the contribution of strange quark - antiquark pairs to the proton charge and magnetism . et al . _ , in _ proceedings of the 3rd international conference on high energy beams and polarized targets _ ( argonne , 1978 ) , edited by l.h . thomas , aip conference proceedings 51 , new york 1979 , p. 224 . p. von rossen _ et al . _ , in _ proceedings of the 5th international symposium on polarization phenomena in nuclear physics _ ( santa fe , 1980 ) , edited by g.g . et al . _ , aip conference proceedings 69 , new york , 1981 , p. 1442 . et al . _ , in _ proceedings of the international workshop on polarized beams and polarized gas targets _ ( cologne , 1995 ) , edited by h.p . schieck and l. sydow ( world scientific , singapore , 1996 ) , p. 120 ; a.n . zelenski , _ ibid . _ , p. 111 . g. batigne in _ proceedings of the 4th international conference on perspectives in hadronic physics _ ( trieste , 2003 ) , g0 report g0 - 03 - 075 , ( this report and others are available from http://www.npl.uiuc.edu/exp/g0/docs/ ) .

i start by reviewing existing @xmath0 measurements with particular emphasis on the recent 221 mev @xmath0 measurement at triumf which permitted the weak meson - nucleon coupling constants @xmath1 and @xmath2 to be determined separately for the first time . i then review @xmath3 experiments , with specific details of the @xmath4 experiment now under preparation at los alamos national laboratory . this experiment will provide a clean measurement of the weak pion nucleon coupling , @xmath5 . finally , i discuss @xmath6 parity violation experiments , particularly the gzero experiment under way at jefferson lab in virginia . this experiment will measure the weak form factors @xmath7 and @xmath8 , allowing the distribution of strange quarks in the quark sea to be determined .

the first two exact solution of einstein s field equations were obtained by schwarzschild [ 1 ] , soon after einstein introduced general relativity ( gr ) . the first solution describes the geometry of the space - time exterior to a prefect fluid sphere in hydrostatic equilibrium . while the other , known as interior schwarzschild solution , corresponds to the interior geometry of a fluid sphere of constant ( homogeneous ) energy - density , @xmath0 . the importance of these two solutions in gr is well known . the exterior solution at a given point depends only upon the total mass of the gravitating body and the radial distance as measured from the centre of the spherical symmetry , and not upon the ` type ' of the density distribution considered inside the mass . however , we will focus on this point of crucial importance later on in the present paper . on the other hand , the interior schwarzschild solution provides two very important features towards obtaining configurations in hydrostatic equilibrium , compatible with gr , namely - ( i ) it gives an absolute upper limit on compaction parameter , @xmath1 , mass to size ratio of the entire configuration in geometrized units ) @xmath2 for any static and spherical solution ( provided the density decreases monotonically outwards from the centre ) in hydrostatic equilibrium [ 2 ] , and ( ii ) for an assigned value of the compaction parameter , @xmath3 , the minimum central pressure , @xmath4 , corresponds to the homogeneous density solution ( see , e.g. , [ 3 ] ) . regarding these conditions , it should be noted that the condition ( i ) tells us that the values higher than the limiting ( maximum ) value of @xmath5 can not be attained by any static solution . but , what kinds of density variations are possible for a mass to be in the state of hydrostatic equilibrium ? , the answer to this important question could be provided by an appropriate analysis of the condition ( ii ) , and the necessary conditions put forward by exterior schwarzschild solution . despite the non linear differential equations , various exact solutions for static and spherically symmetric metric are available in the literature [ 4 ] . tolman [ 5 ] obtained five different types of exact solutions for static cases , namely - type iii ( which corresponds to the constant density solution obtained earlier by schwarzschild [ 1 ] ) , type iv , type v , type vi , and type vii . the solution independently obtained by adler [ 6 ] , adams and cohen [ 7 ] , and kuchowicz [ 8 ] . buchdahl s solution [ 9 ] for vanishing surface density ( the `` gaseous '' model ) . the solution obtained by vaidya and tikekar [ 10 ] , which is also obtained independently by durgapal and bannerji [ 11 ] . the class of exact solutions discussed by durgapal [ 12 ] , and also durgapal and fuloria [ 13 ] solution . knutsen [ 14 ] examined various physical properties of the solutions mentioned in references ( [ 6 - 8 ] , [ 10 - 11 ] , and [ 13 ] ) in great detail , and found that these solutions correspond to nice physical properties and also remain stable against small radial pulsations upto certain values of @xmath3 . tolman s v and vi solutions are not considered physically viable , as they correspond to singular solutions [ infinite values of central density ( that is , the metric coefficient , @xmath6 at @xmath7 ) and pressure for all permissible values of @xmath3 ] . except tolman s v and vi solutions , all other solutions mentioned above are known as regular solutions [ finite positive density at the origin ( that is , the metric coefficient , @xmath9 at @xmath10 ) which decreases monotonically outwards ] , which can be further divided into two categories : ( i ) regular solutions corresponding to a vanishing density at the surface together with pressure ( like , tolman s vii solution ( mehra [ 15 ] , durgapal and rawat [ 16 ] , and negi and durgapal [ 17 , 18 ] ) , and buchdahl s `` gaseous '' solution [ 9 ] ) , and ( ii ) regular solutions correspond to a non - vanishing density at the surface ( like , tolman s iii and iv solutions [ 5 ] , and the solutions discussed in the ref.[6 - 8 ] , and [ 10 - 13 ] respectively ) . the stability analysis of tolman s vii solution with vanishing surface density has been undertaken in detail by negi and durgapal [ 17 , 18 ] and they have shown that this solution also corresponds to stable ultra - compact objects ( ucos ) which are entities of physical interest . this solution also shows nice physical properties , such as , pressure and energy - density are positive and finite everywhere , their respective gradients are negative , the ratio of pressure to density and their respective gradients decrease outwards etc . the other solution which falls in this category and shows nice physical properties is the buchdahl s solution [ 9 ] , however , knutsen [ 19 ] has shown that this solution turned out to be unstable under small radial pulsations . all these solutions ( with finite , as well as vanishing surface density ) discussed above , in fact , fulfill the criterion ( i ) , that is , the equilibrium configurations pertaining to these solutions always correspond to a value of compaction parameter , @xmath3 , which is always less than the schwarzschild limit , i. e. , @xmath11 , but , this condition alone does not provide a necessary condition for hydrostatic equilibrium . nobody has discussed until now , whether these solutions also fulfill the condition ( ii ) ? which is necessary to satisfy by any static and spherical configuration in the state of hydrostatic equilibrium . recently , by using the condition ( ii ) , we have connected the compaction parameter , @xmath3 , of any static and spherical configuration with the corresponding ratio of central pressure to central energy - density @xmath12 and worked out an important criterion which concludes that for a given value of @xmath13 , the maximum value of compaction parameter , @xmath14 , should always correspond to the homogeneous density sphere [ 20 ] . an examination of this criterion on some well known exact solutions and equations of state ( eoss ) indicated that this criterion , in fact , is fulfilled only by those configurations which correspond to a vanishing density at the surface together with pressure [ 20 ] , or by the singular solutions with non - vanishing surface density [ section 5 of the present study ] . this result has motivated us to investigate , in detail , the various exact solutions available in the literature , and disclose the reason ( s ) behind non - fulfillment of the said criterion by various regular analytic solutions and eoss corresponding to a non - vanishing finite density at the surface of the configuration . in this connection , in the present paper , we have examined various exact solutions available in the literature in detail . it is seen that tolman s vii solution with vanishing surface density [ 15 , 17 , 18 ] , buchdahl s `` gaseous '' solution [ 9 ] , and tolman s v and vi singular solutions pertain to a value of @xmath3 which always turns out to be less than the value , @xmath15 , of the homogeneous density sphere for all assigned values of @xmath13 . on the other hand , the solutions having a finite non - zero surface density ( that is , the pressure vanishes at the finite surface density ) do not show consistency with the structure of the general relativity , as they correspond to a value of @xmath3 which turns out to be greater than @xmath15 for all assigned values of @xmath13 , and thus violate the criterion obtained in [ 20 ] . one may ask , for example , what could be , in fact , the reason(s ) behind non - fulfillment of the criterion obtained in [ 20 ] by various exact solutions ( corresponding to a finite , non - zero density at the surface ) ? we have been able to pin point ( which is discussed under section 3 of the present study ) the main reason , namely , the ` actual ' total mass @xmath16 which appears in the exterior schwarzschild solution , in fact , can not be attained by the configurations corresponding to a regular density variation with non - vanishing surface density . the metric inside a static and spherically symmetric mass distribution corresponds to @xmath17 where @xmath18 and @xmath19 are functions of @xmath20 alone . the resulting field equations for the metric governed by eq.(1 ) yield in the following form @xmath21 + 1/r^2 , \\ - 8\pi t^1_{1 } & = & 8\pi p = e^{-\lambda } [ ( \nu'/r ) + ( 1/r^2 ) ] - 1/r^2 , \\ - 8\pi t^2_{2 } & = & - 8\pi t^3_{3 } = 8\pi p = e^{-\lambda } [ ( \nu''/2 ) + ( \nu'^2/4)-(\nu'\lambda'/4)+(\nu'- \lambda')/2r ] . \end{aligned}\ ] ] where the primes represent differentiation with respect to @xmath20 , the speed of light @xmath22 and the universal gravitation constant @xmath23 are chosen as unity ( that is , we are using the geometrized units ) . @xmath24 and @xmath0 represent , respectively , the pressure and energy - density inside the perfect fluid sphere related with the non - vanishing components of the energy - momentum tensor , @xmath25 = 0 , 1 , 2 , and 3 respectively . eqs.(2 ) - ( 4 ) represent second - order , coupled differential equations which can be reduced to the first - order , coupled differential equations , by eliminating @xmath26 from eq . ( 3 ) with the help of eqs . ( 2 ) and ( 4 ) in the well known form ( namely , tov equations ( tolman [ 5 ] ; oppenheimer & volkoff [ 21 ] ) , governing hydrostatic equilibrium in general relativity ) @xmath27/r(r - 2m(r)),\ ] ] @xmath28 and @xmath29 where prime denotes differentiation with respect to @xmath20 , and @xmath30 is defined as the mass - energy contained within the radius @xmath31 , that is @xmath32 the equation connecting metric parameter @xmath19 with @xmath30 is given by @xmath33 = 1 - ( 8\pi / r)\int_{0}^{r } er^2 dr.\ ] ] the three field equations ( or tov equations ) mentioned above , involve four variables , namely , @xmath34 and @xmath19 . thus , in order to obtain a solution of these equations , one more equation is needed [ which may be assumed as a relation between @xmath24 , and @xmath0 ( eos ) , or can be regarded as an algebraic relation connecting one of the four variables with the radial coordinate @xmath20 ( or an algebraic relation between the parameters ) ] . for obtaining an exact solution , the later approach is employed . notice that eq.(9 ) yields the metric coefficient @xmath35 for the assumed energy - density , @xmath0 , as a function of radial distance @xmath31 . once the metric coefficient @xmath35 or mass @xmath30 is defined for assumed energy - density by using eqs.(9 ) or ( 8) , the pressure , @xmath24 , and the metric coefficient , @xmath36 , can be obtained by solving eqs.(5 ) and ( 6 ) respectively which yield two constants of integration . these constants should be obtained from the following boundary conditions , in order to have a proper solution of the field equations : ( b1 ) in order to maintain hydrostatic equilibrium throughout the configuration , the pressure must vanish at the surface of the configuration , that is @xmath37 where @xmath38 is the radius of the configuration . ( b2 ) the consequence of eq.(10 ) ensures the continuity of the metric parameter @xmath39 , belonging to the interior solution with the corresponding expression for well known exterior schwarzschild solution at the surface of the fluid configuration , that is : @xmath40 [ where @xmath41 is the total mass of the configuration ] . however , the exterior schwarzschild solution guarantees that : @xmath42 , which means that the matching of the metric parameter @xmath43 is also ensured at the surface of the configuration together with @xmath39 , that is @xmath44 irrespective of the condition that the surface density , @xmath45 , is vanishing with pressure or not , that is @xmath46 together with eq . ( 10 ) , or @xmath47 where @xmath48 is the ` compaction parameter ' of the configuration defined earlier , and @xmath16 is defined as [ eq . ( 8) ] @xmath49 thus , the analytic solution for the fluid sphere can be explored in terms of the only free parameter @xmath48 by normalizing the metric coefficient @xmath43 [ yielding from eq.(11 ) ] at the surface of the configuration [ that is , @xmath50 at @xmath51 after obtaining the integration constants by using eqs.(10 ) [ that is , @xmath52 at @xmath53 and ( 11 ) [ @xmath54 at @xmath51 respectively . however , at this place we recall the well known property of the exterior schwarzschild solution ( which follows directly from the definition of the mass , @xmath55 , appears in this solution ) , namely - at a given point outside the spherical distribution of mass , @xmath55 , it depends only upon @xmath55 , and not upon the ` type ' of the density variation considered inside the radius , @xmath56 , of this sphere . it follows , therefore , that the dependence of mass , @xmath55 , upon the ` type ' of the density distribution plays an important role in order to fulfill the requirement set up by exterior schwarzschild solution . the relation , @xmath57 , immediately tells us that for an assigned value of the compaction parameter , @xmath3 , the mass , @xmath55 , depends only upon the radius , @xmath56 , of the configuration which may either depend upon the surface density , or upon the central density , or upon both of them , depending upon the ` type ' of the density variation considered inside the mass generating sphere . we argue that this dependence should occur in such a manner that the definition of mass , @xmath55 , is not violated . we infer this definition as the ` type independence ' property of the mass , @xmath55 , which may be defined in this manner : `` the mass , @xmath55 , which appears in the exterior schwarzschild solution , should either depend upon the surface density , or upon the central density , and in any case , not upon both of them so that from an exterior observer s point of view , the ` type ' of the density variation assigned for the mass should remain unidentified '' . we may explain the ` type independence ' property of mass , @xmath55 , mentioned above in the following manner : the mass , @xmath55 , is called the coordinate mass , that is , the mass as measured by some external observer , and from this observer s point of view , if we are ` measuring ' a sphere of mass , @xmath55 , we can not know , by any means , the way in which the matter is distributed from the centre to the surface of this sphere , that is , if we are measuring , @xmath55 , with the help of non - vanishing surface density [ obviously , by calculating the ( coordinate ) radius , @xmath56 , from the expression connecting the surface density and the compaction parameter , and by using the relation , @xmath58 , we can not measure it , by any means , from the knowledge of the central density ( because , if we can not know , by any means , the way in which the matter is distributed from the centre to the surface of the configuration , then how can we know about the central density ? ) , and this is possible only when there exist no relation connecting the mass , @xmath55 , and the central density , that is , the mass , @xmath55 , should be independent of the central density , meaning thereby that _ the surface density should be independent of the central density for configurations corresponding to a non - vanishing surface density_. however , if we are measuring the mass , @xmath55 , by using the expression for central density ( in the similar manner as in the previous case , by calculating the radius , @xmath56 , from the expression of central density , and using the relation , @xmath59 ) , we can not calculate it , by any means , from the knowledge of the surface density ( in view of the ` type independence ' property of the mass , @xmath55 ) , and this is possible only when there exist no relation connecting the mass , @xmath55 , and the surface density , meaning thereby that the central density should be independent of the surface density . from the above explanation of ` type independence ' property of mass @xmath16 , it is evident that _ the ` actual ' total mass @xmath16 which appears in the exterior schwarzschild solution should either depend upon the surface density , or depend upon the central density of the configuration , and in any case , not upon both of them_. however , the dependence of mass @xmath16 upon both of the densities ( surface , as well as central ) is a common feature observed among all regular solutions having a non - vanishing density at the surface of the configuration [ see , for example , eqs.(21 ) , ( 25 ) , ( 29 ) , and ( 33 ) respectively , belonging to the solutions of this category which are discussed under sub - sections ( a ) - ( d ) of section 5 of the present study ] . thus , it is evident that the surface density of such solutions is _ dependent _ upon the central density and vice - versa , that is , the total mass , @xmath55 , depends upon _ both _ of the densities , meaning thereby that _ the ` type ' of the density distribution considered inside the sphere of mass , @xmath55 , is known to an external observer _ which is the violation of the definition of mass , @xmath55 ( defined as the ` type independence ' property of mass @xmath16 above ) , such structures , therefore , do not correspond to the ` actual ' total mass @xmath16 required by the exterior schwarzschild solution to ensure the condition of hydrostatic equilibrium . this also explains the reason behind non - fulfillment of the ` compatibility criterion ' by them which is discussed under section 5 of the present study . however , it is interesting to note here that there could exist only one solution in this regard for which the mass @xmath16 depends upon both , but the same , value of surface and centre density , and for regular density distribution the structure would be governed by the homogeneous ( constant ) density throughout the configuration ( that is , the homogeneous density solution ) . note that the requirement , ` type independence ' of the mass would be obviously fulfilled by the regular structures corresponding to a vanishing density at the surface together with pressure , because the mass @xmath16 will depend only upon the central density ( surface density is always zero for these structures ) [ see , for example , eqs.(37 ) and ( 41 ) , discussed under sub - section ( e ) and ( f ) for buchdahl s `` gaseous '' model and tolman s vii solution having a vanishing density at the surface , respectively ] . furthermore , the demand of ` type independence ' of mass , @xmath55 , is also satisfied by the singular solutions having a non - vanishing density at the surface , because such structures correspond to an infinite value of central density , and consequently , the mass @xmath16 will depend only upon the surface density [ see , for example , eqs.(46 ) and ( 50 ) , discussed under sub - section ( g ) and ( h ) for tolman s v and vi solutions , respectively ] . both types of these structures are also found to be consistent with the ` compatibility criterion ' as discussed under section 5 of the present study . the discussion regarding various types of density distributions considered above is true for any single analytic solution or equation of state comprises the whole configuration . at this place , we are not intended to claim that the construction of a regular structure with non - vanishing surface density is impossible . it is quite possible , provided we consider a two - density structure in such a manner that the mass @xmath16 of the configuration turns out to be independent of the central density so that the property ` type independence ' of the mass @xmath16 is satisfied . examples of such two - density models are also available in the literature ( see , e.g. , ref . [ 22 ] ) , but in the different context . however , it should be noted here that _ the fulfillment of ` type independence ' condition by the mass @xmath16 for any two - density model will represent only a necessary condition for hydrostatic equilibrium , unless the ` compatibility criterion ' [ 20 ] is satisfied by them , which also assure a sufficient and necessary condition for any structure in hydrostatic equilibrium _ ( this issue is addressed in the next section of the present study ) . the above discussion can be summarized in other words as : _ although the exterior schwarzschild solution itself does not depend upon the type of the density distribution or eos considered inside a fluid sphere in the state of hydrostatic equilibrium , however , it puts the important condition that only two types of the density variations are possible inside the configuration in order to fulfill the condition of hydrostatic equilibrium : ( 1 ) the surface density of the configuration should be independent of the central density , and ( 2 ) the central density of the configuration should be independent of the surface density_. obviously , the condition ( 1 ) will be satisfied by the configurations pertaining to an infinite value of the central density ( that is , the singular solutions ) , and/or by the two - density ( or multiple - density ) distributions corresponding to a surface density which turns out to be independent of the central density ( because , the regular configurations governed by a single exact solution or eos pertaining to this category are not possible ) . whereas , the condition ( 2 ) will be fulfilled by the configurations corresponding to a surface density which vanishes together with pressure [ the configurations in this category will include : the density variation governed by a single exact solution or eos , as well as the two - density ( or multiple - density ) distributions ] . however , the point to be emphasized here is that a two - density distribution in any of the two categories mentioned here will fulfill only a _ necessary _ condition for hydrostatic equilibrium unless the ` compatibility criterion ' [ 20 ] is satisfied by them which also assure a necessary and sufficient condition for any structure in the state of hydrostatic equilibrium as mentioned above . the criterion obtained in [ 20 ] can be summarized in the following manner : for an assigned value of the ratio of central pressure to central energy - density @xmath60 $ ] , the compaction parameter of homogeneous density distribution , @xmath14 should always be larger than or equal to the compaction parameter @xmath61 of any static and spherical solution , compatible with the structure of general relativity . that is @xmath62 in the light of eq . ( 15 ) , let us assign the same value , @xmath55 , for the total mass corresponding to various static configurations in hydrostatic equilibrium . if we denote the density of the homogeneous sphere by @xmath63 , we can write @xmath64 where @xmath65 denotes the radius of the homogeneous density sphere . if @xmath66 represents the radius of any other regular sphere for the same mass , @xmath55 , the average density , @xmath67 , of this configuration would correspond to @xmath68 eq . ( 15 ) indicates that @xmath69 . by the use of eqs . ( 16 ) and ( 17 ) we find that @xmath70 that is , for an assign value of @xmath13 , the average energy - density of any static spherical configuration , @xmath67 , should always be less than or equal to the density , @xmath63 , of the homogeneous density sphere for the same mass , @xmath55 . although , the regular configurations with finite non - vanishing surface densities , represented by a single density variation can not exist , because for such configurations the necessary condition set up by exterior schwarzschild solution can not be satisfied . however , we can construct regular configurations composed of core - envelope models corresponding to a finite central with vanishing and non - vanishing surface densities , such that the necessary conditions imposed by the schwarzschild s exterior solution at the surface of the configuration are appropriately satisfied . however , it should be noted that the necessary conditions satisfied by such core - envelope models at the surface may not always turn out to be sufficient for describing the state of hydrostatic equilibrium [ because for an assigned value of @xmath13 , the average density of such configurations may not always turn out to be less than or equal to the density of the homogeneous density sphere for the same mass , as indicated by eqs.(16 ) and ( 17 ) respectively ( it would depend upon the types of the density variations considered for the core and envelope regions and the the matching conditions at the core - envelope boundary ) ] . thus , it follows that the criterion obtained in [ 20 ] is able to provide a _ necessary _ and _ sufficient _ condition for any regular configuration to be consistent with the state of hydrostatic equilibrium . the future study of such core - envelope models [ see , for example , the models described in [ 22 ] and [ 23 ] ] , based upon the criterion obtained in [ 20 ] could be interesting regarding two density structures of neutron stars and other stellar objects compatible with the structure of gr . we have considered the following exact solutions expressed in units of compaction parameter @xmath71 , mass to size ratio in geometrized units ] , and radial coordinate measured in units of configuration size , @xmath72 $ ] , for convenience . the other parameters which will appear in these solutions , are defined at the relevant places . in these equations , @xmath24 and @xmath0 represent , respectively , the pressure and energy - density inside the configuration . the surface density is denoted by @xmath73 , and the central pressure and central energy - density are denoted by @xmath74 , and @xmath75 , respectively . the regular exact solutions which pertain to a non - vanishing value of the surface density are given under the sub - sections ( a ) - ( d ) , while those correspond to a vanishing value of the surface density are described under the sub - sections ( e ) and ( f ) respectively . sub - sections ( g ) and ( h ) represent the singular solutions having non - vanishing values of the surface densities . \(a ) tolman s iv solution @xmath76 @xmath77 by the use of eq.(20 ) , we can obtain the relation connecting central and surface densities in the following form @xmath78 eq.(21 ) shows that the surface density is _ dependent _ upon the central density , and vice - versa . by using eqs . ( 19 ) and ( 20 ) , we obtain @xmath79 this solution finds application [ it can be seen from eq . ( 19 ) ] for the values of @xmath80 . \(b ) adler [ 6 ] , adams and cohen [ 7 ] , and kuchowicz [ 8 ] solution : @xmath81,\ ] ] @xmath82 where @xmath83 , and @xmath84 . eq.(24 ) gives the relation , connecting central and surface densities of the configuration in the following form @xmath85 thus , the surface density _ depends _ upon the central density and vice - versa . equations ( 23 ) and ( 24 ) give @xmath86.\ ] ] it is seen from eq.(23 ) that this solution finds application for values of @xmath87 . \(c ) vaidya and tikekar [ 10 ] , and durgapal and bannerji [ 11 ] solution @xmath88},\ ] ] @xmath89 where the variable @xmath90 , and the constants @xmath91 and @xmath92 are given by @xmath93 ; and @xmath94 by the use of eq.(28 ) , we find that the surface and central densities are connected by the following relation @xmath95 it is evident from eq.(29 ) that the surface density is _ dependent _ upon the central density , and vice - versa . by the use of eqs.(27 ) and ( 28 ) , we obtain @xmath96 this solution finds application for the values of @xmath97 , as shown by eq.(27 ) . \(d ) durgapal and fuloria [ 13 ] solution @xmath98 \bigr ] , \ ] ] @xmath99 where @xmath92 is a constant and @xmath100 . the variables @xmath101 and @xmath102 are given by @xmath103,\ ] ] @xmath104,\ ] ] where the arbitrary constant @xmath105 is given by @xmath106}\ ] ] and @xmath107 . eq.(32 ) gives the relation , connecting central and surface densities as @xmath108 eq.(33 ) indicates that the surface density is _ dependent _ upon the central density , and vice - versa . by the use of equations ( 31 ) and ( 32 ) , we get @xmath109 \bigr ] , \ ] ] where @xmath110 and @xmath111 are given by @xmath112,\ ] ] @xmath113 and @xmath114 as seen from eq.(31 ) , this solution is applicable for the values of @xmath115 . \(e ) buchdahl s `` gaseous '' solution [ 9 ] @xmath116 @xmath117 where @xmath118 ; \,\ , \ , n = 2(1 - u)$ ] , and @xmath119};\,\,\ , 0 \leq z \leq \pi\ ] ] eq.(36 ) shows that the surface density vanishes together with pressure , thus , the central density will become _ independent _ of the surface density , given by the equation @xmath120 by using equations ( 35 ) and ( 36 ) , we obtain @xmath121 it is evident from eq.(35 ) that this solution is applicable for the values of @xmath87 . \(f ) tolman s vii solution with vanishing surface density @xmath122 @xmath123 where @xmath75 is the central energy - density given by @xmath124 and @xmath125 @xmath126 @xmath127^{1/2};\,\ , na^2 = u(5 - 3x),\ ] ] @xmath128/3u)^{1/2}],\ ] ] @xmath129^{1/2}],\ ] ] @xmath130 by using eqs . ( 39 ) and ( 40 ) , we get @xmath131,\ ] ] where @xmath132 is given by @xmath133.\ ] ] it follows from eq.(40 ) that the surface density is always zero , hence the central density is always _ independent _ of the surface density . eq.(39 ) indicates that his solution is applicable for the values of @xmath134 . \(g ) tolman s v solution @xmath135 and @xmath136y^q \bigr ] } { ( 2n + 1 - n^2)y^2},\ ] ] where @xmath137 is given by @xmath138 and @xmath139 is defined as @xmath140 eq.(44 ) shows that the central density is always infinite ( for @xmath141 ) together with central pressure ( eq.43 ) , however , their ratio ( @xmath142 ) is finite at all points inside the configuration , and at the centre , yields in the following form @xmath143 the consequence of the infinite central density is that the surface density will become _ independent _ of the central density , given by the equation @xmath144 it is evident from eq.(45 ) that this solution is applicable for a value of @xmath145 [ i. e. , for a value of @xmath87 ] . \(h ) tolman s vi solution @xmath147},\ ] ] @xmath148 eqs.(47 ) and ( 48 ) indicate that the central pressure and central density are always infinite , however , their ratio ( @xmath142 ) is finite at all points inside the structure , and at the centre , reduces into the following form @xmath149 and the surface density ( obviously , _ independent _ of the central density ) would be given by the equation @xmath150 where @xmath139 is defined as @xmath151 eq.(49 ) indicates that this solution is applicable for a value of @xmath152 . let us denote the compaction parameter of the homogeneous density configuration by @xmath15 , and for the exact solutions corresponding to the sub - sections ( a ) - ( d ) by @xmath153 and @xmath154 respectively . the compaction parameters of the exact solutions described under sub - section ( e ) and ( f ) are denoted by @xmath155 and @xmath156 respectively , and those discussed under sub - sections ( g ) and ( h ) are denoted by @xmath157 and @xmath158 respectively . solving these analytic solutions for various assigned values of the ratio of central pressure to central energy - density @xmath159 $ ] , we obtain the corresponding values of the compaction parameters as shown in table 1 and table 2 respectively . it is seen that for each and every assigned value of @xmath13 , the values represented by @xmath153 and @xmath154 respectively ( table 1 ) , turn out to be higher than @xmath15 ( that is , @xmath153 and @xmath160 ) , while those represented by @xmath161 , and @xmath158 respectively ( table 2 ) correspond to a value which always remains less than @xmath15 ( that is , @xmath162 , and @xmath163 ) . thus , we conclude that the configurations defined by @xmath153 and @xmath154 respectively , do not show compatibility with the structure of general relativity , while those defined by @xmath161 , and @xmath158 respectively , show compatibility with the structure of general relativity . however , this type of characteristics [ that is , the value of compaction parameter larger than the value of @xmath15 for some or all assigned values of @xmath13 ] can be seen for any regular exact solution having a finite non - vanishing surface density [ because such exact solutions ( having finite central densities ) with non - vanishing surface densities can not possess the actual mass , @xmath55 , required to fulfill the boundary conditions at the surface ] . on the other hand , the value of compaction parameter for a regular solution with vanishing surface density , and a singular solution with non - vanishing surface density will always remain less than the value of @xmath15 for all assigned values of @xmath13 [ because such solutions naturally fulfill the definition of the actual mass , @xmath55 , required for the hydrostatic equilibrium ] . therefore , it is evident that the findings based upon the ` compatibility criterion ' carried out in this section are fully consistent with the definition of the mass , @xmath55 ( defined as the ` type independence ' property under section 3 of the present study ) . we have investigated the criterion obtained in the reference [ 20 ] which states : for an assigned value of the ratio of central pressure to central energy - density , @xmath165 , the compaction parameter , @xmath166 , of any static and spherically symmetric solution should always be less than or equal to the compaction parameter , @xmath15 , of the homogeneous density distribution . we conclude that this criterion is fully consistent with the reasoning discussed under section 3 which states that in order to fulfill the requirement set up by exterior schwarzschild solution ( that is , to ensure the condition of hydrostatic equilibrium ) , the total mass , @xmath55 , of the configuration should depend either upon the surface density ( that is , independent of the central density ) , or upon the central density ( that is , independent of the surface density ) , and in any case , not upon both of them . an examination , based upon this criterion , show that among various exact solutions of the field equations available in the literature , the regular solutions corresponding to a vanishing surface density together with pressure , namely - ( i ) tolman s vii solution with vanishing surface density , and ( ii ) buchdahl s `` gaseous '' solution , and the singular solutions with non - vanishing surface density , namely - tolman s v and vi solutions are compatible with the structure of general relativity . the only regular solution with finite non - vanishing surface density which could exist in this regard is described by constant ( homogeneous ) density distribution . this criterion provides a necessary and sufficient condition for any static spherical configuration to be compatible with the structure of general relativity , and may be used to construct core - envelope models of stellar objects like neutron stars with vanishing and non - vanishing surface densities , such that for an assigned value of central pressure to central density , the average density of the configuration should always remain less than or equal to the density of the homogeneous sphere for the same mass . this criterion could provide a convenient and reliable tool for testing equations of state ( eoss ) for dense nuclear matter and models of relativistic star clusters , and may find application to investigate new analytic solutions and eoss . r. c. tolman , _ phys . _ * 55 * , 364 ( 1939 ) . r. j. adlar , _ j. math . phys . _ * 15 * , 727 ( 1974 ) . r. c. adams and j. m. cohen , _ astrophys . _ * 198 * , 507 ( 1975 ) . b. kuchowicz , _ astrophys . & space science _ * l13 * , 33 ( 1975 ) . h. a. buchdahl , _ astrophys . j. _ * 147 * , 310 ( 1967 ) . p. c. vaidya and r. tikekar , _ j. astrophys . _ , * 3 * , 325 ( 1982 ) . m. c. durgapal and r. bannerji , _ phys . _ * d27 * , 328 ( 1983 ) . erratum * d28 * , 2695 . m. c. durgapal , _ _ * 15 * , 2637 ( 1982 ) . m. c. durgapal and r. s. fuloria , _ gen . _ * 17 * , 671 ( 1985 ) . h. knutsen , _ astrophys . & space science _ * 140 * , 385 ( 1988 ) ; h. knutsen , _ mon . not . soc . _ * 232 * , 163 ( 1988 ) ; h. knutsen , _ astrophys . & space science _ * 162 * , 315 ( 1989 ) . a. l. mehra , _ j. aust * 6 * , 153 ( 1966 ) . m. c. durgapal and p. s. rawat , _ mon . not . * 192 * , 659 ( 1980 ) . p. s. negi and m. c. durgapal , _ astrophys . & space science _ * 245 * , 97 ( 1996 ) . p. s. negi and m. c. durgapal , _ gen . _ , * 31 * , 13 ( 1999 ) . h. knutsen , _ gen . grav . _ * 20 * , 317 ( 1988 ) . p. s. negi and m. c. durgapal , _ gravitation & cosmology _ * 7 * , 37 ( 2001)/astro - ph/0312516 . j. r. oppenheimer and g. m. volkoff , _ phys rev . _ * 55 * , 374 ( 1939 ) . p. s. negi , a. k. pande and m. c. durgapal , _ class . quantum grav . _ * 6 * , 1141 ( 1989 ) ; p. s. negi , a. k. pande and m. c. durgapal , _ gen . rel . _ * 22 * , 735 ( 1990 ) . p. s. negi and m. c. durgapal , _ gravitation & cosmology _ * 5 * , 191 ( 1999 ) . p. s. negi and m. c. durgapal , _ astron . astrophys . _ * 353 * , 641 ( 2000 ) .

we examine various well known exact solutions available in the literature to investigate the recent criterion obtained in ref . [ 20 ] which should be fulfilled by any static and spherically symmetric solution in the state of hydrostatic equilibrium . it is seen that this criterion is fulfilled only by ( i ) the regular solutions having a vanishing surface density together with the pressure , and ( ii ) the singular solutions corresponding to a non - vanishing density at the surface of the configuration . on the other hand , the regular solutions corresponding to a non - vanishing surface density do not fulfill this criterion . based upon this investigation , we point out that the exterior schwarzschild solution itself provides necessary conditions for the types of the density distributions to be considered inside the mass , in order to obtain exact solutions or equations of state compatible with the structure of general relativity . the regular solutions with finite centre and non - zero surface densities which do not fulfill the criterion [ 20 ] , in fact , can not meet the requirement of the ` actual mass ' set up by exterior schwarzschild solution . the only regular solution which could be possible in this regard is represented by uniform ( homogeneous ) density distribution . the criterion [ 20 ] provides a necessary and sufficient condition for any static and spherical configuration ( including core - envelope models ) to be compatible with the structure of general relativity . thus , it may find application to construct the appropriate core - envelope models of stellar objects like neutron stars and may be used to test various equations of state for dense nuclear matter and the models of relativistic stellar structures like star clusters . _ pacs nos . : 04.20.jd ; 04.40.dg ; 97.60.jd._ * exact solutions of einstein s field equations * + .25 in p. s. negi _ department of physics , kumaun university , nainital 263 002 , india _

for reliable realization of quantum feedback control , it is indispensable to take into consideration some real - world limitations , such as incomplete knowledge of the physical systems and poor performance of the control devices . various efforts on these issues have been undertaken in these few years , see e.g. , @xcite for the system parameter uncertainty . among such limitations , time delays in the feedback loop , which happen due to the finite computational speed of _ classical _ controller devices , are extremely serious , since their effect may completely lose the benefit of feedback control @xcite . to avoid the time delays , one can think to use the markovian feedback control , in which the measurement results are directly fed back @xcite . however , while these experimental simplification has been extensively studied , theoretical ways to evaluate the effect of the time delays have not been proposed so far . in this paper , we investigate the effect of the time delays on the control performance , which is defined in terms of the cost function optimized by feedback control . this investigation provides theoretical guidelines for the feedback control experiment . as the controlled object , the linear quantum systems are considered . in order to prepare the tool for the analysis , we first consider the optimal lqg control problem subject to the constant time delay . the optimal controller is obtained via the existing results in the classical control theory @xcite . further , these results allow us to obtain the formula for the optimal value of the cost . the obtained formula enables us to examine the relation between the optimal control performance and the time delay both in an analytical and a numerical ways . then , the intrinsic stability of the systems is dominant for the performance degradation effect . if the system is stable , the degradation effect converges to some value in the large time delay limit . otherwise , the performance monotonically deteriorates as the delay length becomes larger . based on this fact , we perform the analysis stated above for several physical systems that possess different stability properties . in addition to the controller design , we examine the relationship between the measurement apparatus and the best achievable performance . based on this , we propose a detector parameter tuning policy for feedback control of the time - delayed systems . this paper is organized as follows . linear quantum control systems are introduced in the next section . in section iii , we state the control problem for dealing with the time delay issue , and provide its optimal solution . in section iv , we investigate the effect of the time delay in quantum feedback control based on two typical examples possessing different stability properties . section v concludes the paper . we use the following notation . for a matrix @xmath0 , @xmath1 , @xmath2 and @xmath3 are defined by @xmath4 , @xmath5 and @xmath6 , respectively , where the matrix element @xmath7 may be an operator and @xmath8 denotes its adjoint . the symbols @xmath9 and @xmath10 denote the real and imaginary parts of @xmath11 , respectively , i.e. , @xmath12 and @xmath13 . all the rules above are applied to any rectangular matrix . consider a quantum system which interacts with a vacuum electromagnetic field through the system operator @xmath14 where @xmath15{^{\sf t}}$ ] and @xmath16\in{{\mathbb c}}^{1\times 2}$ ] . when the system hamiltonian is denoted by @xmath17 , this interaction is described by a unitary operator @xmath18 obeying the following quantum stochastic differential equation called the _ hudson - parthasarathy equation _ @xcite : @xmath19u_t,\ ] ] where @xmath20 is the identity operator . the field operators @xmath21 and @xmath22 are the creation and annihilation operator processes , which satisfy the following quantum it rule : @xmath23 further , suppose that the system is trapped in a harmonic potential , and that a linear potential is an input to the system . the system hamiltonian @xmath24 at time @xmath25 is given by @xmath26 where @xmath27 is the control input at time @xmath25 , the system parameters @xmath28 and @xmath29 are a symmetric matrix and a column vector , and @xmath30 is given by @xmath31}.\ ] ] then , by defining @xmath32{^{\sf t}}=[u_t q u_t^\dagger , u_t p u_t^\dagger]{^{\sf t}}$ ] and by using the commutation relation @xmath33=i$ ] and the quantum it formula , we obtain the following linear equation : @xmath34 where @xmath35 $ ] . measurement processes are described as follows . suppose that the field observable @xmath36 is measured by the perfect homodyne detector , where @xmath37 denotes the detector parameter that the experimenter can change @xcite . then , the output signal @xmath38 is obtained by @xmath39 the simple calculation yields the infinitesimal increment of the observable @xmath38 as follows : @xmath40 in the following section , we refer to ( [ system ] ) and ( [ output ] ) as the system dynamics and the output equation , respectively . as stated in the introduction , the effect of time delays is significant in feedback control of quantum systems . those delays are mainly originated from the computational time for a controller and the transition delay of signals . thus , they should be modelled practically as input - output delays in the feedback loop , i.e. , at time @xmath25 , the signal @xmath41 works as a control input for the system and the information @xmath42 is available in the controller , where we assume that @xmath43 and @xmath44 are constants . without loss of generality , when we consider the optimal control problem for such a system , the total delay time can be simply put together into one input ( or output ) delay . then , the system dynamics are modified as follows : @xmath45 here , the real constant @xmath46 denotes the total time delay in the feedback loop , i.e. , @xmath47 . note here that @xmath48 should be determined by @xmath49 . we consider the optimal control problem for the system described by ( [ output ] ) and ( [ delaysystem ] ) . the following system expression is convenient for exploiting results in the classical control theory . let us define a quantum noise vector @xmath50}.\ ] ] it is shown that the quantum noise vector satisfies the following properties : @xmath51 where @xmath52 denotes the expectation and @xmath53 is the non - negative hermitian matrix given by @xmath54}.\ ] ] also , we define the matrix @xmath55 . by substituting the terms of the field observables @xmath22 , @xmath21 with the noise vector @xmath56 , we obtain the following equations : @xmath57 here , @xmath58 is an additional output signal defined to evaluate the system performance , and @xmath59 and @xmath60 are matrices freely tunable in controller design . the other system matrices are defined as follows : @xmath61}\right),\\ b_2&\hspace{-1mm}:=\hspace{-1mm}&b,\\ c_2&\hspace{-1mm}:=\hspace{-1mm}&e^{-i\phi}c+e^{i\phi}c^\ast,\\ d_{21}&\hspace{-1mm}:=\hspace{-1mm}&{\left [ \begin{array}{cc } 1&0 \end{array}\right]}.\end{aligned}\ ] ] as depicted in fig . [ fig : loop ] , we investigate the feedback loop consisting of the system and a controller implemented by classical devices , such as analogue or digital circuits . then , the optimal control problem is stated as follows . [ problem]for the linear quantum system ( [ gplant ] ) , find the causal , linear and time - invariant control law @xmath62 that minimizes the cost functional @xmath63 and determine the minimum value of @xmath64 . we make the following assumption , which is standard in the classical control theory ; see @xcite for the details . [ assumption ] 1 . @xmath65 is stabilizable and @xmath66 is detectable . 2 . for any @xmath67 , @xmath68 } , { \left [ \begin{array}{cc } a - j\zeta i&b_1\\ c_2&d_{21 } \end{array}\right]}\ ] ] are row- and column - full rank , respectively . 3 . @xmath69 and @xmath70 are nonsingular . ( 70,46 ) ( 0,8 ) ( 10,43)@xmath58 ( 7,30)@xmath38 ( 60,13)@xmath48 ( 35.5,39)(0,0)plant ( [ gplant ] ) ( 35.5,35)(0,0)(quantum ) ( 35.2,16)(0,0)controller ( 35.2,11)(0,0)(classical ) ( 11.7,22.3)(0,0)delay @xmath43 ( 58,22.3)(0,0)delay @xmath44 as in the other results for linear quantum systems , the solution of problem [ problem ] can be obtained by slightly modifying the derivation of the classical result in @xcite . here , we only provide the minimum value of the cost @xmath64 , which is of importance for the later discussion . for the specific form of the optimal controller , see the appendix . [ maintheorem ] consider problem [ problem ] with assumption [ assumption ] . let @xmath71 , @xmath72 be the solutions of the matrix riccati equations @xmath73 and @xmath74 with @xmath75 and @xmath76 such that @xmath77 , @xmath78 are stable . then , the optimal value of the cost functional @xmath64 is given by @xmath79 where @xmath80 is the optimal value of @xmath64 when @xmath81 . note that the existence of the riccati solutions @xmath71 and @xmath72 follows from assumption [ assumption ] . in the experiment of the feedback control , it is of importance to reduce the time delay by carefully setting up the experimental devices and achieve the best performance possible @xcite . however , some quantity of the time delay remains in practice . in this section , we investigate how the time delay deteriorates the optimal control performance by using the formula ( [ performance ] ) . in addition to the algorithm in the controller , we have the tunable parameter in the measurement apparatus . thus , we do the analysis taking the detector parameter tuning into consideration . it should be noted that the optimal measurement technique was first introduced by wiseman and doherty @xcite . their technique is only for the delay - free systems , i.e. , the optimization of the value of @xmath82 . first of all , notice that the performance degradation effect is mainly determined by the exponential term in ( [ performance ] ) . this means that the degradation is largely related to the system s intrinsic stability , i.e. , the eigenvalues of the matrix @xmath11 . thus , it is obvious from the exponential growth of @xmath83 that the unstable system easily deteriorates as the time delay length increases , and that their control is significantly difficult . on the other hand , the remaining two classes , i.e. , stable and marginally stable systems , are relatively insensitive to the time delay and it is worth to analyze them in detail . in order to provide some guidelines for the experiments , we analyze the two physical systems that frequently appear in the context of quantum feedback control . in the following , we choose the matrices @xmath84 and @xmath85 as @xmath86},\ d_{12}={\left [ \begin{array}{c } 1\\#2 \end{array}\right]}.\ ] ] * stable system - * consider a damped cavity with an on - threshold parametric down converter . the system hamiltonian and the coupling operator are given by @xmath87 where @xmath88 and @xmath89 are constant parameters . if they satisfy @xmath90 , the system is stable . in this case , clearly , @xmath83 converges as @xmath91 since the real part of every eigenvalue of @xmath11 is negative . when we choose the parameter as @xmath92 and @xmath93 , the optimal performance curves with the different detector parameters @xmath94 are given by fig . [ fig : damped ] . it is certainly confirmed that the performance degradation converges in the limit of @xmath91 for any detector parameters . the dashed line depicts the value of @xmath64 when there is no control input field , i.e. , @xmath95 for any @xmath96 . we can see from the figure that even in the large delay limit , the appropriate measurement strategy significantly enhances the control performance compared to the uncontrolled case . in order to examine the effective detector parameter tuning , let us look at figs . [ fig : damped ] and [ fig : opt_detector ] . these show that the optimal detector parameter hardly fluctuates over every delay length . in fact , we can see from fig . [ fig : damped ] that the difference between two performance curves around the optimal one , i.e. , the curve corresponding to @xmath97 , is very small . further , it is shown that the optimal measurement strategy in the delay - free case is perfectly the same as that in the large delay case . consider problem [ problem ] with @xmath24 and @xmath98 defined by ( [ parametric ] ) . let @xmath99 denote the detector parameter that minimizes @xmath83 for the fixed delay length @xmath46 . then , the following relation holds : @xmath100 * proof . * the goal is to show that the minimal values of @xmath82 and @xmath101 are achieved with the same detector parameter value . firstly , we compute the value of @xmath102 . note that @xmath103 where @xmath104,\ y=\left[\begin{array}{cc}y_{11}&y_{12}\\y_{12}&y_{22}\end{array}\right].\ ] ] then , since @xmath71 is independent of @xmath94 , we obtain @xmath105 besides , the relation @xmath106 provides us with @xmath107 where we used @xmath108 thus , with the attention to ( [ jphi ] ) , the following equation is obtained : @xmath109 hence , we obtain @xmath110 where the first equality follows from the pointwise convergence of @xmath83 and the uniform convergence of @xmath111 on the domain of @xmath94 . this completes the proof.@xmath112 + from the discussion above , we can conclude that wiseman s measurement strategy ( the optimal tuning for delay - free case ) is valid for the stable delay systems in that @xmath113 for any @xmath114 . ( 70,46 ) ( 7.5,3 ) when @xmath95 . three solid lines correspond to the optimal control performances with the detector parameters @xmath115 , @xmath116 , @xmath117 , from the top downwards , respectively , where @xmath97 is the optimal detector parameter.,title="fig:",width=219,height=143 ] ( 0,3)@xmath118 ( 0,12)@xmath119 ( 0,21.3)@xmath120 ( 0,30.5)@xmath121 ( 0,39.7)@xmath122 ( 7.3,0)@xmath123 ( 18.4,0)@xmath124 ( 29.7,0)@xmath125 ( 41.2,0)@xmath126 ( 52.8,0)@xmath127 ( 63.4,0)@xmath128 ( 35.5,-3)@xmath46 ( -7,32)@xmath83 ( 70,46 ) ( 7.5,5 ) ( 3.5,4.8)@xmath123 ( 1,10.7)@xmath129 ( 1,16.5)@xmath130 ( 1,22.2)@xmath131 ( 1,28.3)@xmath132 ( 1,34)@xmath133 ( 1,40)@xmath134 ( 7.3,2)@xmath123 ( 18.4,2)@xmath124 ( 29.7,2)@xmath125 ( 41.4,2)@xmath126 ( 52.8,2)@xmath127 ( 63.4,2)@xmath128 ( 35.5,-1)@xmath46 ( -6,33)@xmath99 * marginally stable system - * the next system is a single particle trapped in the harmonic potential and coupled to the probe field via the position operator . the system hamiltonian and the coupling operator are given by @xmath135 where @xmath136 and @xmath137 are the mass of the particle and the angular frequency of the harmonic potential , respectively . for this system , the shape of the optimal performance curves is analytically calculated . [ main_evaluation ] consider problem [ problem ] with @xmath24 and @xmath98 defined by ( [ harmonic ] ) . then there exist constants @xmath138 , @xmath139 and @xmath140 such that the best achievable performance is given by @xmath141 moreover , @xmath138 and @xmath139 are independent of the choice of @xmath94 . * proof . * notice that the riccati solution @xmath72 is dependent on the parameter @xmath94 . to make this dependence explicit , we write @xmath72 as @xmath142 and , similarly , @xmath143 as @xmath144 throughout the proof . then , by theorem [ maintheorem ] , the best achievable performance is given by @xmath145 where @xmath82 is the positive constant given by @xmath146 on the other hand , direct computation yields @xmath147 where @xmath148 $ ] , @xmath149{^{\sf t}}$ ] and @xmath150 satisfies @xmath151 by combining this with ( [ bap ] ) , we obtain the first claim . it should be emphasized that @xmath94 contributes to @xmath152 and @xmath153 only through @xmath154 hence , it is sufficient to show the second claim that ( [ coeff ] ) does not depend on @xmath94 . when defining @xmath155,\ ] ] simple calculation yields @xmath156 then , we obtain the following : @xmath157 this completes the proof.@xmath112 ( 70,46 ) ( 7.5,3 ) , @xmath158 , @xmath159 , respectively.,title="fig:",width=219,height=143 ] ( 3,2.5)@xmath123 ( 3,7.2)@xmath160 ( 3,12)@xmath124 ( 3,16.6)@xmath161 ( 3,21.3)@xmath125 ( 3,25.9)@xmath162 ( 3,30.2)@xmath126 ( 3,34.8)@xmath163 ( 3,39.6)@xmath127 ( 7.3,0)@xmath123 ( 18.4,0)@xmath124 ( 29.7,0)@xmath125 ( 41.2,0)@xmath126 ( 52.8,0)@xmath127 ( 63.4,0)@xmath128 ( 35.5,-3)@xmath46 ( -5,30)@xmath83 roughly speaking , the first statement says that @xmath83 increases linearly with the oscillation as the delay length @xmath46 becomes large . this is a natural result of the fact that the matrix @xmath11 has only pure imaginary eigenvalues . on the other hand , the second statement gives us a nontrivial insight : the growth rate @xmath138 and the oscillation amplitude @xmath139 are independent of the detector parameter @xmath94 . hence , depending on the delay length , the importance of the choice of the measurement apparatus differs . if the delay length is small , the improvable performance level is sensitive to the value of @xmath46 . on the other hand , if the system suffers from the large delay , the apparatus adjustment is not significant , since the improvable performance level is relatively small compared to the value of @xmath83 . thus , if the time delay can be made sufficiently small , experimenters have to adjust the detector parameter depending on the resulting delay length . for the illustration , see fig.[fig : harmonic ] for @xmath164 , which illustrates the optimal performance curve @xmath83 . in this paper , we investigated performance degradation effects due to time delays in optimal control of linear quantum systems . the analysis was performed by the optimal control performance formula from the classical control theory . the obtained remarks are strongly related to the intrinsic stability of the physical systems . in particular , we performed intimate evaluations for two typical systems with different types of stability . these results are expected to give useful guidelines for the future experiments . consider problem [ problem ] . with the same notation as that in theorem [ maintheorem ] , the optimal feedback controller is given by @xmath165 @xmath166 and the finite - time integration system @xmath167 it should be noted that the implementation of this controller involves infinite - dimensional elements . thus , computers with the finite memory can not implement it in a precise sense . fortunately , however , it is known that the approximation method proposed in @xcite permits the control with high accuracy . m. r. james , phys . a * 69 * , 032108 ( 2004 ) . m. r. james , j. opt . b : quantum semiclass * 7 * , 198 ( 2005 ) . m. r. james , h. i. nurdin , and i. r. petersen , arxiv : quant - ph/0703150v2 ( 2007 ) . n. yamamoto , phys . a * 74 * , 032107 ( 2006 ) . r. van handel , j. k. stockton , and h. mabuchi , ieee trans . automatic control * 50 * , 768 ( 2005 ) . d. a. steck , k. jacobs , h. mabuchi , s. habib , and t. bhattacharya , phys . a * 74 * , 012322 ( 2006 ) . j. stockton , m. armen , and h. mabuchi , j. opt . b * 19 * , 3019 ( 2002 ) . h. m. wiseman and g. j. milburn , phys . lett . * 70 * , 548 ( 1993 ) . h. m. wiseman , s. mancini , and j. wang , phys . . a * 66 * , 013807 ( 2002 ) . l. mirkin , ieee trans . automatic control * 48 * , 543 ( 2003 ) . r. l. hudson and k. r. parthasarathy , commun . * 93 * , 301 ( 1984 ) . l. bouten , r. van handel , and m. r. james , siam j. control and optimization * 46 * , 2199 ( 2007 ) . k. zhou , j. c. doyle , and k. glover , _ robust optimal control _ ( prentice - hall , 1995 ) . h. m. wiseman and a. c. doherty , phys . lett . * 94 * , 070405 ( 2005 ) . l. mirkin , system and control letters * 51 * , 331 ( 2004 ) .

we investigate feedback control of linear quantum systems subject to feedback - loop time delays . in particular , we examine the relation between the potentially achievable control performance and the time delays , and provide theoretical guidelines for the future experimental setup in two physical systems , which are typical in this research field . the evaluation criterion for the analysis is given by the optimal control performance formula , the derivation of which is from the classical control theoretic results about the input - output delay systems .

the central regions of the dwarf spiral galaxy ngc 4395 show most of the common signatures of nuclear activity . the optical and ultraviolet ( uv ) spectra reveal prominent high - ionization forbidden lines on top of a nearly featureless continuum , and broad wings corresponding to gas velocities in excess of @xmath010@xmath7 km s@xmath8 are detected in the permitted lines ( filippenko & sargent 1989 ) . contrary to the objects of the same kind , the emission - line properties , the optical to x - ray variability pattern and the inferred accretion rate of ngc 4395 are those typical of the seyfert class , of which this source is usually considered to represent the least luminous member . different methods have been employed in the last years to derive the mass of its central black hole : the estimate obtained through reverberation mapping is @xmath9 ( peterson et al . 2005 ) , but the lack in this galaxy of a significant bulge and the stringent upper limit of 30 km s@xmath8 on its velocity dispersion suggest an even lower value , of the order of @xmath010@xmath1010@xmath1 ( filippenko & ho 2003 ) . anyhow , the engine of ngc 4395 falls somewhere between the stellar - mass black holes found in galactic x - ray binaries and the supermassive black holes residing inside active galactic nuclei ( agn ) . as such , it can provide critical information about the relationship between these two populations and the physics of accretion systems in general . + in the light of all these pieces of observational evidence , ngc 4395 is a true scaled - down version of an ordinary seyfert galaxy , the only difference with respect to its high - luminosity counterparts being the much smaller mass of the central black hole . on the other hand , the x - ray observations of this source indicate an unusual spectral hardness at @xmath0210 kev , and suggest a wide range of variations for the intrinsic photon index . the most extreme states ( @xmath11 ; moran et al . 2005 ) are even difficult to interpret within the standard two - phase model ( haardt & maraschi 1991 ) , posing serious questions on the production mechanism of the x - ray emission itself . the presence of undetected absorption effects has been frequently invoked as a possible explanation . indeed , the stronger flux variability characterizing systematically the soft x - ray bands can be attributed to a complex , multi - zone warm absorber , whose properties have been discussed in detail in several works ( iwasawa et al . 2000 ; shih , iwasawa & fabian 2003 ; dewangan et al . here we review the two highest - quality observations of ngc 4395 , performing in both cases an accurate time - resolved analysis mainly focused on neutral absorption , in order to test whether changes of its column density and/or covering factor play a role in the apparent x - ray spectral hardness of this source . the longest _ xmm - newton _ monitoring of ngc 4395 started on 2003 november 30 , for a total duration of @xmath0113 ks . after the subtraction of high - background periods , the useful exposure declines to 91.4 ks . we also take into account the deep _ suzaku _ observation , which was carried out on 2007 june 25 over a span of @xmath0230 ks , corresponding to a net integration time of 101.3 ks . we have followed the standard procedures for the reduction of the event files , and extracted the source and background spectra from circular regions with radii of 30(_xmm - newton _ ) and 2(_suzaku _ ) . in the first case , for the sake of clarity only epic - pn data are presented here and plotted in the figures , even though the mos spectra have been checked throughout and give fully consistent results ; in the second one , instead , only the data from the front - illuminated detectors of the x - ray imaging spectrometer ( xis ) have been examined , after merging the xis0 and xis3 spectra . the spectral analysis has been performed using the xspec v12.6 fitting package . all the uncertainties are given at the 90 per cent confidence level ( @xmath12 ) for the single parameter of interest . we first restricted our analysis to the energies above 1 kev , following the customary approach of fitting the spectrum averaged over the whole observation to obtain a benchmark model . we achieved a fully acceptable fit ( @xmath13 for 1390 degrees of freedom , with no obvious structure in the residuals ) through a simple model consisting of a power law with photon index @xmath14 , a narrow iron emission line with equivalent width @xmath15 ev , and a partial neutral absorber whose column density and covering fraction are @xmath16 @xmath5 and @xmath17 , respectively . @xmath5 ( kalberla et al . 2005 ) . ] no reflection component is strictly required for the continuum : however , by adding a pexrav model ( magdziarz & zdziarski 1995 ) for physical consistency and forcing its strength to match the width of the narrow iron line , the fit is slightly improved ( @xmath18 with the loss of one d.o.f . ) and the power - law photon index steepens to @xmath19 . both cases above yield a formally satisfactory fit , yet the flat slope of the intrinsic continuum deserves a thorough investigation . this result would actually confirm one of the most remarkable features of ngc 4395 , which had been previously caught by _ chandra _ even in harder states ( see moran et al . 2005 ) . in any case , the typical value of @xmath201.5 is still unusually low , compared with both the distribution of photon indices found in local agn ( e.g. bianchi et al . 2009 ) and the average 15150 kev spectrum of the source measured by the burst alert telescope ( bat ) onboard _ swift _ ( @xmath21 ; fig . [ bm ] ) . + a viable explanation for such a discrepancy is the presence of a complex ( variable ) absorber , strongly modifying the observed spectral shape of ngc 4395 below @xmath010 kev . the extrapolation of the basic model down to 0.5 kev supports this working assumption : the apparent extra emission at 0.50.7 kev , in fact , hints at absorption effects in the @xmath00.71.5 kev range rather than at a genuine soft excess like the one detected in a large fraction of seyfert galaxies ( e.g. boller , brandt & fink 1996 ; porquet et al . 2004 ) . moreover , a clearly time - dependent behaviour of the x - ray source is revealed not only by the overall flux light curve but also by the hardness ratio ( hr ) evolution . it is well established that ngc 4395 is characterized by strong variations of the x - ray flux on time - scales as short as a few hundreds of seconds ( fig . [ lc ] , upper panel ) : the fractional rms variability amplitude during the _ xmm - newton _ observation under review is exceptionally high ( @xmath090 per cent at 0.52 kev ; see vaughan et al . 2005 ) . .,width=321 ] on top of these fluctuations of the intrinsic x - ray brightness , mainly related to the workings of the primary source , significant _ spectral _ variations are also evident from the visual inspection of the hr light curve ( fig . [ lc ] , lower panel ) . the observed hr pattern can be accounted for through either changes of the column density or , alternatively , oscillations of the continuum slope : this proves that a time - averaged spectral analysis is not sufficient to fully understand the nature of the x - ray emission of ngc 4395 . the hardness ratio is defined here as the ratio between the 510 kev and the 25 kev flux , thus it is especially sensitive to values of @xmath22 of the order of 10@xmath310@xmath4 @xmath5 . as already mentioned , the spectral complexity and the larger fractional rms variability below @xmath02 kev are likely due instead to a multi - zone warm absorber , whose presence has been revealed since the early _ asca _ observations of the source ( iwasawa et al . 2000 ; shih et al . we anyway note that this is not a critical point , since the hr evolution is only used to select the appropriate intervals for the subsequent time - resolved spectral analysis , which is crucial to validate or dismiss the suggested interpretation . + the shape of the hr light curve reveals two periods of sudden spectral hardening around @xmath025 and 90 ks ; this hints at increased opacity , which would mainly affect the 25 kev band . according to fig . [ lc ] , six different regimes of the hardness ratio can be roughly defined . in this framework , the minor hr fluctuations on very short time - scales can be due to a flickering of the photon index in response to variations of the physical conditions in the disc / corona system , but the statistics is not enough for a complete investigation of these aspects . regardless of this , our aim is to check whether a reasonable configuration of the neutral absorber allows us to recover a steeper intrinsic photon index for the x - ray emission of ngc 4395 . + in order to obtain an adequate description of the entire 0.510 kev spectral range , we included in the reference model above also an apec component for the soft thermal emission ( smith et al . 2001 ) and a two - zone warm absorber , whose complex effects have been already pointed out in several previous works ( see also dewangan et al . 2008 ) . in our analysis we have assumed that the soft emission , warm absorption and reflection features do not vary in the course of a @xmath0100 ks long observation : since the light crossing time over a distance of a gravitational radius ( @xmath23 ) is of the order of one second , the reflection component is expected to change significantly in response to the primary continuum only if the scattering material is located well within @xmath010@xmath24 from the centre ( as in the disc reflection scenarios ; e.g. nardini et al . 2011 ) . similarly , the soft emission likely arises from a very extended region , while the warm absorber , in principle , could lie much closer to the x - ray source ( but see blustin et al . 2005 ) : if this is the case , we are just sampling its average properties . on the other hand , @xmath25 , @xmath22 and @xmath26 were initially left free to evolve among the different intervals . ( 1 ) or @xmath22(2 ) , and plotted respectively in red , green and blue.,width=321 ] in spite of the possible degeneracies , it turns out that the photon index and the column density are subject to very limited variations : in particular , @xmath27 . this suggests that @xmath25 and @xmath22 can be roughly treated as constant parameters as well , hence their values have been tied in all the six spectra : only the covering fraction of the cold absorber is allowed to vary with time . + on sheer statistical grounds , this model already gives a very good fit ( @xmath28 , with @xmath29 and @xmath30 @xmath5 ) , proving that , in first approximation , @xmath26 alone would be able to account for all the observed spectral variability : its qualitative trend across the six intervals , in fact , fairly correlates with the shape of the hr light curve . there are some problems with the reflection efficiency , though . it is not obvious how to assess the latter quantity in the time - resolved analysis , anyway by comparing the pexrav amplitude to the highest flux level of the power - law continuum we obtain a reflection strength of @xmath31 , which is completely inconsistent with the moderate equivalent width of the companion iron line ( george & fabian 1991 ) . moreover , when extrapolated at higher energies , such component would give rise to a very flat spectrum with a prominent compton hump at @xmath030 kev , which is not detected by _ swift_/bat . given that the time - scales are extremely different , it is still possible that the source is caught in a state of large reflection during the _ xmm - newton _ observation , but in this case the fluorescent iron line should be much stronger as well . forcing @xmath32 to have a standard value delivers a poor spectral description , and no improvement is achieved through the self - consistent reflionx table models ( ross & fabian 2005 ) . + as a consequence , we attempted to introduce a second partial covering cold absorber , assuming again that @xmath25 and @xmath22(1,2 ) are constant , and @xmath26(1,2 ) variable . the best fit quality is significantly improved ( @xmath33 ) , while the photon index @xmath34 is still in excellent agreement with the average high - energy slope of the source . the presence of an additional absorber with column density of @xmath010@xmath4 @xmath5 removes all the previous limitations , reducing substantially the required strength of the reflection continuum ( now @xmath35 ) . this is also clear from fig . [ sc ] , where the different model components are disentangled to display their relative contribution in a single time segment . the basic parameters of this final model are listed in table [ t1 ] , while the six individual spectra are shown in fig . [ xmm ] . . the data are rebinned for plotting purposes only.,width=321 ] .main parameters of the _ xmm - newton _ and _ suzaku _ best - fitting models , common to all the intervals on which our time - resolved spectral analysis has been performed . @xmath25 : photon index ; @xmath36 : temperature of the soft emission in ev ; @xmath37 , @xmath38 : column density in @xmath5 and ionization parameter in erg cm s@xmath8 of the warm absorption components ; ew@xmath39 : time - averaged equivalent width of the 6.4-kev iron line in ev ; @xmath32 : strength of the reflected continuum ; @xmath22 : column density of the neutral absorbers in @xmath5 ; @xmath40 , @xmath41 : average observed and intrinsic 0.510 kev flux in erg @xmath5 s@xmath8 . [ cols="<,^,^ " , ] the cold absorption model provides a good interpretation of the spectral variability of the x - ray source during both the _ xmm - newton _ and _ suzaku _ observations . moreover , considering the entries of table [ t1 ] , the basic physical quantities appear to be in fair agreement . this is a strong confirmation of the validity of the scenario explored in this work , where the changes are driven by the evolution of the covering fractions . the @xmath26(1 ) and @xmath26(2 ) progression is listed in table [ t2 ] . due to the smaller uncertainties , we focus our discussion on the _ xmm - newton _ case . we first point out that our model is defined in such a way that mutually exclusive regions of the x - ray source are affected by the two cold absorbers : in other words , we only consider a single - layer configuration of the neutral gas along the line of sight . all the @xmath26 sequences should be taken with some caution , though . the low ionization stage in one of the warm absorption component impairs to some extent the determination of @xmath26(1 ) , while possible changes in the reflection strength on that of @xmath26(2 ) . however , even from the face values obtained under our assumptions , some interesting considerations can be drawn . it is now @xmath42 that shows a tighter correlation with the hr light curve : the peaks during periods 2 and 5 might be explained in terms of _ eclipses _ from individual _ clouds _ ( e.g. lamer , uttley & mchardy 2003 ; risaliti et al . given that the duration of such events is @xmath010 ks , a typical dimension of the x - ray source of @xmath010@xmath43 corresponds to a transverse velocity of the putative clouds of @xmath010@xmath7 km s@xmath8 , placing the obscuring gas at the broad - line region scale . the size and the shape of these blobs are not precisely known . yet , our time - resolved analysis suggests that different column densities are present along the line of sight at the same time . the range of variations in @xmath26 implies that the number of intervening clouds is very limited ( a few at most , see fig [ blr ] ) , and that their size is comparable to the dimensions of the x - ray source . the extreme case entails a single irregular and inhomogeneous absorber . + the physical situation is then expected to be rather complex . in this view , the x - ray absorbers that we have included should be only regarded as a linear combination of the _ real _ ones , whose exact geometrical structure can not be probed with the present data quality . the large disparity ( roughly an order of magnitude ) between the column densities involved , and specifically the fact that @xmath44 , hints at a multi - layer configuration of the cold absorber , where the two phases have a different location and are superimposed in part on one another when seen in projection . such a scheme can be envisaged as in fig . [ abs ] : only the broad - line region component , with @xmath45 @xmath5 , has a time - dependent covering fraction ( @xmath460.8 ) . conversely , the absorption system with lower column density is constant ( or at least variable over much longer scales ) , and can be associated with a more distant gaseous component , within a narrow - line or torus - like region . indeed , this would be consistent with both the optical classification of ngc 4395 as a type 1.51.8 seyfert galaxy ( ho et al . 1997 ; panessa et al . 2006 ) and the large covering factor of the emission - line regions with respect to the central source ( kraemer et al . the broad- and narrow - line regions are expected to be responsible for some of the uv to x - ray absorption detected in ngc 4395 ( crenshaw et al . 2004 ) , and evidence for the identification of the rapidly variable x - ray absorber with the broad - line emitting clouds have been recently found in other seyfert galaxies ( e.g. maiolino et al . 2010 ; risaliti et al . 2011 ) . in this perspective , the frequency and amplitude of the variations in @xmath22 and/or @xmath26 would be linked to the degree of clumpiness of the circumnuclear environment at the different scales . + once the _ suzaku _ data are taken into account , all the considerations made above still hold , even though the uncertainties on the key parameters are quite large , and the smaller difference between @xmath22(1 ) and @xmath22(2 ) makes the covering fraction patterns less meaningful . apart from the speculations on the physical and geometrical structure of the cold absorption system , the scenario outlined in this work has the other great advantage of strengthening the correlation between ngc 4395 and the high - luminosity agn population . first , it gives reasons for the x - ray spectral hardness of this source , reconciling the estimate of its photon index with the typical values found among agn ; secondly , neutral absorption variability systematically occurs in a significant fraction of active galaxies , the prototypical case being another type 1.8 seyfert , ngc 1365 ( e.g. risaliti et al . our findings then represent a further point of contact between ngc 4395 and its more massive and luminous counterparts . incidentally , it is also worth noting that the intrinsic 0.510 kev emission of ngc 4395 implied in this cold absorption scheme is larger than the observed one by just a factor of @xmath02 ( table [ t1 ] ) . taking advantage of the _ swift_/bat spectral constraints , the resulting 0.5100 kev luminosity is @xmath06@xmath47 erg s@xmath8 , which exceeds the usual estimates of the bolometric luminosity ( see also moran et al . 2005 ; iwasawa et al . depending of the exact value of the black hole mass , the eddington ratio of ngc 4395 could be much closer to @xmath00.01 than previously thought . . the number of clouds simultaneously crossing the line of sight and their size are limited by the rapid changes of @xmath26 . even a single blob with irregular shape and non - uniform physical properties is consistent with the observed variability pattern . ( the observer is located towards the bottom of the page).,width=321 ] , but an alternative configuration is assumed : the absorption component with lower column density has a constant covering fraction of @xmath48 and is external to the system of clouds responsible of the x - ray spectral variability ( see the discussion in the text).,width=321 ] we have discussed a possible interpretation of the x - ray spectral hardness usually observed in the low - luminosity active galaxy ngc 4395 . this source harbours in its centre a black hole with estimated mass of @xmath010@xmath1 , and it is one of the few known objects whose study can shed light on the links between the physics of accretion processes in galactic black hole binaries and agn . in spite of being a genuine seyfert galaxy in many respects , ngc 4395 remains a somewhat puzzling source because of the inferred flatness of its primary x - ray continuum . the existence of complex absorption effects has often been proposed as a likely explanation . here we have provided for the first time an example of these effects based on observational evidence , by reviewing the two highest - quality looks of ngc 4395 taken by _ xmm - newton _ and _ suzaku _ : in both cases , a time - resolved analysis shows that the spectral evolution of the source can be interpreted by means of a two - phase neutral absorber with variable covering factor . as a first approximation , this cold absorber is identified with the system of broad - line clouds , allowing for a double - peaked @xmath22 distribution . the low column density component can otherwise be attributed to an external narrow - line or torus - like region , with nearly constant @xmath26 . this is presumably an oversimplification of the physical and geometrical structure of the circumnuclear environment ( which is known to comprise also a complex , multi - zone warm absorber ) , but even the existence of a single partial - covering cold screen can not be completely ruled out on statistical grounds . + we stress that the absorption variability scenario presented here is not unique , and different models based on intrinsically flat x - ray continua ( with significant changes of either the photon index or the reflection strength ) can describe the observed behaviour of this source equally well . this interpretation , however , fits into the analogy between the properties of ngc 4395 and those of _ standard _ high - luminosity seyfert galaxies , many of which are systematically affected by x - ray absorption variability due to the clumpiness of their circumnuclear regions . moreover , it allows us to retrieve in a natural way a @xmath49 photon index below 10 kev , in perfect agreement with the _ swift_/bat spectral slope and the usual values measured among agn . no alternative explanation would then be required for the intrinsic 210 kev flatness of ngc 4395 . this work has been partly supported by nasa grant nnx08an48 g . we thank the anonymous referee for constructive and useful comments which significantly improved the content of this paper . bianchi s. , guainazzi m. , matt g. , fonseca bonilla n. , ponti g. , 2009 , a&a , 495 , 421 blustin a. j. , page m. j. , fuerst s. v. , branduardi - raymont g. , ashton c. e. , 2005 , a&a , 431 , 111 boller t. , brandt w. n. , fink h. , 1996 , a&a , 305 , 53 crenshaw d. m. , kraemer s. b. , gabel j. r. , schmitt h. r. , filippenko a. v. , ho l. c. , shields j. c. , turner t. j. , 2004 , apj , 612 , 152 dewangan g. c. , mathur s. , griffiths r. e. , rao a. r. , 2008 , apj , 689 , 762 filippenko a. v. , sargent w. l. w. , 1989 , apj , 342 , l11 filippenko a. v. , ho l. c. , 2003 , apj , 588 , l13 george i. m. , fabian a. c. , 1991 , mnras , 249 , 352 haardt f. , maraschi l. , 1991 , apj , 380 , l51 ho l. c. , filippenko a. v. , sargent w. l. w. , peng c. y. , 1997 , apjs , 112 , 391 iwasawa k. , tanaka y. , gallo l. c. , 2010 , a&a , 514 , a58 kalberla p. m. w. , burton w. b. , hartmann d. , arnal e. m. , bajaja e. , morras r. , pppel w. g. l. , 2005 , a&a , 440 , 775 kraemer s. b. , ho l. c. , crenshaw d. m. , shields j. c. , filippenko a. v. , 1999 , apj , 520 , 564 lamer g. , uttley p. , mchardy i. m. , 2003 , mnras , 342 , l41 magdziarz p. , zdziarski a. a. , 1995 , mnras , 273 , 837 maiolino r. , et al . , 2010 , a&a , 517 , a47 moran e. c. , eracleous m. , leighly k. m. , chartas g. , filippenko a. v. , ho l. c. , blanco p. r. , 2005 , aj , 129 , 2108 nardini e. , fabian a. c. , reis r. c. , walton d. j. , 2011 , mnras , 410 , 1251 panessa f. , bassani l. , cappi m. , dadina m. , barcons x. , carrera f. j. , ho l. c. , iwasawa k. , 2006 , a&a , 455 , 173 peterson b. m. , et al . , 2005 , apj , 632 , 799 porquet d. , reeves j. n. , obrien p. , brinkmann w. , 2004 , a&a , 422 , 85 risaliti g. , elvis m. , fabbiano g. , baldi a. , zezas a. , salvati m. , 2007 , apj , 659 , l111 risaliti g. , et al . , 2009 , mnras , 393 , l1 risaliti g. , nardini e. , salvati m. , elvis m. , fabbiano g. , maiolino r. , pietrini p. , torricelli - ciamponi g. , 2011 , mnras , 410 , 1027 ross r. r. , fabian a. c. , 2005 , mnras , 358 , 211 shih d. c. , iwasawa k. , fabian a. c. , 2003 , mnras , 341 , 973 smith r. k. , brickhouse n. s. , liedahl d. a. , raymond j. c. , 2001 , apj , 556 , l91 vaughan s. , iwasawa k. , fabian a. c. , hayashida k. , 2005 , mnras , 356 , 524

we present a new x - ray analysis of the dwarf seyfert galaxy ngc 4395 , based on two archival _ xmm - newton _ and _ suzaku _ observations . this source is well known for a series of remarkable properties : one of the smallest estimated black hole masses among active galactic nuclei ( of the order of @xmath010@xmath1 ) , intense flux variability on very short time - scales ( a few tens of seconds ) , an unusually flat x - ray continuum ( @xmath2 over the 210 kev energy range ) . ngc 4395 is also characterized by significant variations of the x - ray spectral shape , and here we show that such behaviour can be explained through the partial occultation by circumnuclear cold absorbers with column densities of @xmath010@xmath310@xmath4 @xmath5 . in this scenario , the primary x - ray emission is best reproduced by means of a power law with a standard @xmath6 photon index , consistent with both the spectral slope observed at higher energies and the values typical of local agn . [ firstpage ] galaxies : active galaxies : individual : ngc 4395 x - rays : galaxies .

crosslinked polymer hydrogels are completely interconnected polymer networks that combine high water content with high porosity , forming macroscopic molecules that have applications in many fields . their wide range of pore sizes makes such gels ideal for separation of biological macromolecules by electrophoresis or chromatography @xcite . the crosslinking process induces a reorganization of the polymer structure , resulting in inhomogeneities in the spatial density @xcite . these inhomogeneities affect the surface configurations of the gels , but only in recent years have advances in atomic force microscopy ( afm ) made possible the imaging of soft material surfaces in an aqueous environment @xcite . the effects of the crosslinking density , temperature , pressure , and sample thickness on the surface topography have been studied by suzuki _ et al_. @xcite . their results indicate that the structural features of the surface on both the micrometer and nanometer scales depend on these factors . it has been suggested @xcite that control of the characteristic length scale of the gel surface using external stimuli may have applications in a variety of fields , such as regulation of adsorption and release of specific molecules by the intermolecular forces between the surface and the molecule . thus , gel surfaces provide a unique opportunity to explore the interplay between phenomena on the macroscopic and nanoscopic scales . templated polyacrylamide gels are formed by polymerizing acrylamide with a crosslinker in the presence of a surfactant . the surfactant molecules form monodisperse micelles of a size roughly comparable with the crosslinker separation , and the presence of these micelles alters the gel pore structure , enhancing the gel s separation properties @xcite . in a recent work the surface morphology of templated polyacrylamide gels was extensively studied by afm and scaling analysis of the resulting images @xcite . this study indicates that the gel surfaces are self - affine on short length scales , with roughness ( hurst ) exponents on the order of 0.81 . in the absence of surfactant a cross - over length , above which the surface is no longer self - affine , was estimated to be on the order of 300 nm , and the saturation value of the interface width was on the order of 1 nm . both values increased dramatically with the introduction of surfactant . detailed kinetic lattice models of the polymerization of crosslinked polymer gels have previously been constructed @xcite . however , they emphasize the kinetics of polymerization and are extremely computationally intensive . inspired by the experiments reported in ref . @xcite , the aim of the present work is rather to construct and study a simple continuum model that can reproduce some of the observed scaling characteristics of the templated gel surfaces at a more modest ( but still substantial ) computational cost by concentrating on the elastic structure of the gel . some preliminary results were presented in ref . @xcite . the remainder of this paper is organized as follows . in sec . [ sec : model ] we introduce the model and detail some aspects of the numerical calculations . in sec . [ sec : scal ] we recall some scaling concepts associated with the analysis of surfaces . in sec . [ sec : results ] we show the results of our calculations , in particular the scaling properties of the surface width and increment correlation function along with their dependence on the size of the sample . we also comment on the relevance of our results for the interpretation of recent afm experiments . finally , in sec . [ sec : conc ] we present our conclusions . the model consists of a two - dimensional network of nodes interconnected by massless springs . the nodes represent crosslinker molecules , and the connecting springs represent polymer chains . the network topology consists of a triangular lattice of nodes , each of which is connected by harmonic springs to its six nearest neighbors ( except at the top and bottom surfaces , where each node has only four connections ) . the model has no excluded - volume interactions . a triangular lattice was chosen to ensure geometrical stability in two dimensions without introducing bond - angle constraints . the corresponding unrealistically high crosslinker functionality should not significantly influence our results . the network has periodic boundary conditions in the horizontal ( @xmath0 ) direction , free boundary conditions at the top ( @xmath3 ) layer , and the nodes in the bottom layer are fixed at @xmath4 , corresponding to bonding to a rigid substrate . there are @xmath5 nodes in the horizontal direction and @xmath6 nodes in the vertical direction . the total energy of the network is @xmath7 , where @xmath8 , @xmath9 , and @xmath10 , are the spring constant , the actual length , and the equilibrium length under zero external force of the @xmath11th spring , respectively . ( all quantities in this paper are given in dimensionless units . ) the equilibrium length of each spring is independent of the other springs and is randomly chosen with probability density function ( pdf ) @xmath12 where @xmath13 is proportional to the inverse of the average number of monomers between crosslinkers . this pdf corresponds to the case that the equilibrium distance between crosslinkers is proportional to the square root of an exponentially distributed number of monomers . it is consistent with the picture that crosslinkers are distributed randomly along the polymer chains , and that a spring of equilibrium length @xmath10 corresponds to a polymer of the same average end - to - end distance in the random - coil collapsed phase @xcite . the average equilibrium length of a spring is @xmath14 . in agreement with a mean - field spin - chain approximation for the elastic properties of a collapsed polymer chain @xcite , we require that the elastic constant of the @xmath11th spring should be inversely proportional to its equilibrium length , @xmath15 in our dimensionless units . with the above distribution of equilibrium lengths , the pdf for the force @xmath16 exerted by a spring of length @xmath17 is @xmath18 \label{eq : fprob}\ ] ] for @xmath19 , while it vanishes for @xmath20 . the average force exerted by a spring of length @xmath17 is then @xmath21 initially , springs with equilibrium lengths drawn independently according to @xmath22 are placed on the bonds of a regular triangular lattice of unit lattice constant , such that all of them are stretched or compressed to a length of one . in order that this initial configuration should not be globally stressed , we impose the condition that the average force exerted by a spring of unit length should be zero . by eq . ( [ eq : avgf ] ) , this is satisfied for @xmath23 . however , the initial configuration is _ locally _ stressed , since @xmath24 for all @xmath11 while @xmath25 . the system is frustrated in the sense that there is no configuration that can simultaneously minimize the energies of all the springs . as a result there is a large number of configurations corresponding to local energy minima with similar energies . the network was relaxed using the limited - memory broyden - fletcher - goldfarb - shanno ( l - bfgs ) quasi - newton minimization algorithm @xcite until a locally stable configuration was reached , corresponding to a local minimum in the global energy landscape . by minimizing several statistically equivalent realizations of the random spring configuration , we checked that the local minima reached by this procedure are representative of typical relaxed configurations . although the total energies of different relaxed configurations were not completely identical , all the structural results were the same to within our numerical accuracy . the l - bfgs subroutine requires the calculation of the total energy and its gradients , i.e. , the local forces in the vertical and horizontal directions for each node . by monitoring the total energy and the absolute value of the force we found that the subroutine must be called about @xmath26 times to ensure that the system reaches a local minimum since each relaxation step requires a new evaluation of the total energy and the local forces , the l - bfgs algorithm is quite computationally intensive , taking about 1 - 2 days on a current 600 mhz dual - processor pentium iii pc with 512 mb of shared memory to relax a typical system with about @xmath27 degrees of freedom . however , it was far more efficient for our application than a nave steepest - descent method , requiring about an order of magnitude less computer time while leading to essentially equivalent final energies and structures . after relaxation , we examined the bulk and surface properties of the network . images of a typical relaxed network are shown in figs . [ fig : zooms ] and [ fig : bulk ] . inhomogeneities in the spatial crosslinker density are evident in these figures . the scaling properties associated with the height fluctuations of rough surfaces have been the object of a great deal of interest . vapor deposition , epitaxial growth , polymer growth , and many other phenomena exhibit scale invariance @xcite . in a statistical sense , the height of a single - valued self - affine random surface , @xmath1 , satisfies the scaling relation @xmath28 where @xmath29 is a dimensionless scaling parameter and @xmath30 is the so - called hurst exponent , which in this case is identical to the roughness exponent @xcite . consequently , the increment correlation function @xmath31 of a translationally invariant , self - affine surface scales as @xmath32 ^ 2 \rangle \sim x^{2h}\ ; , \label{eq : cx}\ ] ] where the average is taken over @xmath33 . many surfaces present a mixed behavior characterized by @xmath34 where @xmath35 is a cross - over length scale separating the two behaviors . however , there exist surfaces for which the scaling can not be described by a single exponent ; instead an infinite hierarchy of characteristic exponents has to be introduced . these surfaces are called multi - affine , and they are occasionally observed in nature @xcite . molecular - beam epitaxy ( mbe ) models @xcite , polished metal surfaces @xcite , geomorphological features @xcite , and time series of financial data @xcite provide some examples of multi - affine scaling . the characteristic exponents of a multi - affine function can be calculated from the @xmath36th - order increment correlation function , defined as @xmath37 with the generalized hurst exponent @xmath38 depending continuously on @xmath36 , at least in some region of @xmath36 values . thus @xmath39 is identical to the usual increment correlation function @xmath31 defined in eq . ( [ eq : cx ] ) . like self - affine surfaces , multi - affine surfaces can show a cross - over to scale - independent behavior , analogous to eq . ( [ eq : cross ] ) . in the next section we present our results on the structure of the bulk and the scaling behavior of the surface , calculated after the spring network has been relaxed to a representative local energy minimum . we analyzed spring networks of sizes @xmath40 . in order to simulate experimental measurements of the scaling behavior of surfaces of gels in which pores have been introduced by polymerizing acrylamide plus a crosslinker templated with surfactants that are later removed @xcite , we created voids in the networks . the voids were introduced by selecting a proportion of the nodes at random and removing these nodes and the springs associated with them . next , the hoshen - kopelman algorithm @xcite was applied to identify and discard clusters that were not connected to the fixed substrate . this last step is particularly important near the site - percolation limit , which for the triangular lattice is at a vacancy concentration of 50% @xcite . note that by vacancy concentration " we always mean the initial concentration of vacant nodes before removal of the disconnected clusters . the total volume fraction of the void structure resulting from the cluster removal is of course larger . we define the one - dimensional surface as the set of surface heights , @xmath41 , at equally spaced discrete points , @xmath42 , considering the polymer chains as straight line segments between nearest - neighbor connected nodes . to calculate the surface height at @xmath43 , we first identify the set of all springs which intersect a vertical line at @xmath43 . from this set of springs , we define @xmath44 } , \label{eq : surfdef}\ ] ] where @xmath45 is the vertical location of the intersection of the @xmath11th spring at @xmath43 . we use the sign convention that @xmath4 for the fixed bottom nodes and the free surface is located at positive values of @xmath46 . in fig . [ fig : zooms ] we show details of the bulk and free surface of systems with vacancy concentrations of @xmath47% , @xmath48% , and @xmath49% , while in fig . [ fig : bulk ] we show a magnified view of a section of the bulk of a vacancy - free system . to obtain an estimate of the finite - size effects and select an appropriate system size for the production runs , we first considered vacancy - free spring networks of sizes @xmath50 , with @xmath51 , 1024 , and @xmath52 , and @xmath53 , 384 , 512 , 640 , and @xmath54 . we calculated the root - mean - square width of the surface ( also known as the rms roughness ) on a length scale @xmath55 , @xmath56 , by recursively subdividing the surface into @xmath57 nonoverlapping segments of length @xmath58 , with @xmath57 an integer between 2 and @xmath59 . the mean - square width of the @xmath60th segment is @xmath61 where @xmath62 denotes averaging within the segment . the rms width is defined as the square root of the average of @xmath63 over the @xmath57 segments , @xmath64 . the statistical errors in @xmath56 were estimated in the usual way from the standard deviation of @xmath63 . at this point we emphasize that the order in which the average and the square root are taken in the definition of @xmath56 is important . for a simple self - affine surface it is irrelevant whether the square root is taken before or after the average ; in either case one obtains the same exponent . however , for a multi - affine surface the results are quite different . in this case the exponent obtained from @xmath56 is equal to that calculated from the increment correlation function @xmath65 only when the average is calculated _ before _ taking the square root , as it is done here . the results for @xmath56 , averaged over four independent configurations for each system size , are shown in fig . [ fig : size ] . the data shown in fig . [ fig : size ] indicate that , for the values of @xmath5 and @xmath6 considered , @xmath56 is independent of the system size . however , when comparing the average height for the @xmath66th layer in the relaxed system , @xmath67 , with its original value in the regular triangular lattice , we found a slight contraction limited to approximately the top 250 layers . to minimize finite - size effects due to interference of the finite thickness @xmath6 with this skin effect , while maintaining @xmath5 as large as possible so as to reach large horizontal length scales and reduce statistical uncertainties in the estimates of @xmath56 , we chose a system size of @xmath68 for the production runs . both the bulk and surface structures were analyzed as discussed in the following two subsections . in this subsection , we discuss the bulk properties of the spring network far from both the fixed and free surfaces . we only present results for a gel without vacancies . for gels with high vacancy concentrations , it is difficult to define a bulk region , since the width of the surface region increases with increasing vacancy concentration . to study the bulk structure in this case , simulations with periodic boundary conditions would be preferable . the presence of inhomogeneities in the spatial crosslinker distribution , where the crosslinkers and the polymer chains ( shown as line segments ) have greater densities in some regions ( clumps ) , can be seen in fig . [ fig : zooms ] and fig . [ fig : bulk ] . the effect is more pronounced for higher vacancy concentrations . this is a desirable feature of the model since static inhomogeneities are characteristic of crosslinked polyacrylamide gels and have been related to a freezing - in of the topological structure during polymerization @xcite . for gels without vacancies , fig . [ fig : zooms](a ) , the clumping is not as obvious to the eye . thus , we show in fig . [ fig : bulk ] a magnified image of a small region of the bulk of a vacancy - free system , in which the clumping can be clearly seen . to quantify the clumping , we characterize the bulk scaling properties of the gel by a method analogous to the width - scaling method of eq . ( [ eq : rmsdef ] ) . we define @xmath69 to be the density of nodes at a particular point @xmath70 of the bulk . here @xmath71 is the two - dimensional dirac delta function , and the sum runs over all nodes @xmath72 in the system . we calculate the mean and the variance of the number of nodes inside a box of side @xmath55 , @xmath73 and @xmath74 , respectively , by averaging @xmath75 over non - overlapping boxes . when there are no correlations within the gel , corresponding to a purely random distribution of node positions , the number of nodes in a box obeys a poisson distribution . this yields @xmath76 . however , when the nodes have a more regular distribution , the denominator is reduced and the ratio increases . in fig . [ fig : avgbulk ] @xmath77 , averaged over eight independent realizations of the system , is shown vs @xmath55 from @xmath78 to @xmath79 . for small @xmath55 , the function tends to zero , indicating that nodes which are very close together are essentially randomly placed . the function goes through a maximum near the average spring length ( @xmath80 ) , corresponding to a more regular distribution of the nodes on this scale . we calculated the @xmath36th root of the @xmath36th - order increment correlation function , @xmath81 , for @xmath82 , 1 , 1.5 , 2 , and @xmath83 . the results for systems with @xmath47% , @xmath48% , and @xmath49% vacancies are shown in fig . [ fig : q0](a ) , ( b ) , and ( c ) , respectively . for each of the two lower vacancy concentrations the data were averaged over ten independent realizations of the system , while thirteen realizations were used for the highest concentration . for all three vacancy concentrations , the surface is seen to be multi - affine with @xmath36-dependent slopes for @xmath55 below a crossover length , @xmath35 . [ for the highest vacancy concentration ( fig . [ fig : q0](c ) ) , two scaling regimes with different , nonzero @xmath38 are seen for @xmath84 . ] for @xmath85 , @xmath86 reaches a saturation value @xmath87 . both of these length scales , @xmath35 and @xmath87 , depend on the vacancy concentration and @xmath36 in ways that are discussed below . by numerically calculating the logarithmic derivatives of the correlation functions with respect to @xmath55 we determine the scaling exponents , @xmath88 effective values for @xmath38 for different @xmath55 and @xmath36 were estimated as two - point derivatives over intervals corresponding to a doubling of @xmath55 . the results are shown in fig . [ fig : dq0](a ) , ( b ) , and ( c ) for systems with @xmath47% , @xmath48% , and @xmath49% vacancies , respectively . the figures clearly indicate that on length scales below @xmath35 , the surfaces are multi - affine with @xmath36-dependent scaling exponents @xmath38 . in this scaling regime the effective exponents are approximately independent of @xmath55 . scaling exponents @xmath38 for different values of the vacancy concentration are presented in table [ tab : exp ] . the exponents shown in this table were obtained from eq . ( [ eq : hq ] ) at fixed @xmath89 , in the small-@xmath55 scaling regime . in fig . [ fig : hq ] we show how these values of @xmath38 change with @xmath36 and the vacancy concentration . for @xmath90 , @xmath38 decreases with increasing vacancy concentration , while it increases for @xmath91 . for @xmath92 , @xmath38 is slightly below unity and independent of the vacancy concentration to within our numerical accuracy . multi - affinity is relatively rare in natural surfaces , and , in fact , it is not observed in the experimental system that inspired the present study @xcite . it is therefore reasonable to inquire about the source of the multi - affinity observed in our model system . we find that it is caused by the finite density of discontinuities , which result from overhangs in the simulated surfaces ( see fig . [ fig : zooms ] ) . this effect is illustrated in two ways in fig . [ fig : discont ] . in fig . [ fig : discont](a ) we show on a log - log scale @xmath93 for different values of @xmath36 for surfaces generated by fitting linear line segments of unit length to self - affine surfaces with a hurst exponent of 0.75 , which were generated by the method of successive additions @xcite . these surfaces are seen to be self - affine with a hurst exponent which crosses over from near unity to approximately 0.75 at about @xmath94 . in the same figure we also show ( by thin curves ) the corresponding generalized increment correlation functions for surfaces generated from the above ones by the insertion of a finite density of vertical discontinuities . in contrast to the result for the original surfaces , the slopes of the latter curves depend on @xmath36 , indicating that the discontinuous surfaces are multi - affine . conversely , in fig . [ fig : discont](b ) we illustrate the effect of removing the discontinuities from a surface generated by our model with 0% vacancies . while the original surface is multi - affine ( signified by the @xmath36-dependent slopes ) , the surface from which the discontinuities are removed is self - affine with a hurst exponent of approximately 0.98 , corresponding to the straight line segments used to represent the polymer chains . these numerical results clearly show that the multi - affinity in our model surfaces is due to the vertical discontinuities caused by the overhangs . possible methods to improve the agreement of the scaling behavior of the model surfaces with that of the experimental system by removing the discontinuities are discussed in sec . [ sec : conc ] . the large-@xmath55 saturation values , @xmath95 , are summarized in fig . [ fig : sat ] . for a fixed concentration of vacancies , the saturation value increases with increasing @xmath36 . the saturation value also increases with the vacancy concentration , quite dramatically so when the concentration approaches the site - percolation limit of 50% . figure [ fig : sat](a ) shows @xmath96 vs the vacancy concentration on a linear - log scale , while fig . [ fig : sat](b ) shows the same quantity vs the relative distance from the percolation limit , @xmath97 , on a log - log scale . the straight line in fig . [ fig : sat](b ) is proportional to @xmath98 with @xmath99 being the exact value for the connectivity - length exponent @xcite . the figure indicates that @xmath96 appears to be coupled to the connectivity length for vacancy percolation in the bulk , at least for @xmath100 . for a given vacancy concentration , all the curves in fig . [ fig : dq0 ] coincide for @xmath101 , indicating that the surface structure is not scale - dependent on these large length scales . due to the difficulty in obtaining a precise estimate for @xmath35 , we chose to define it as the value of @xmath55 at which @xmath102 . this value is shown vs the vacancy concentration in fig . [ fig : lcross ] which , like fig . [ fig : sat ] , is divided into two parts . in fig . [ fig : lcross](a ) @xmath103 is shown vs the vacancy concentration on a linear - log scale , while in fig . [ fig : lcross](b ) it is shown vs @xmath104 on a log - log scale . the increase in the vicinity of the percolation threshold is similar to that of @xmath95 in fig . [ fig : sat ] , except that @xmath103 has a significant background value which dominates at and below a vacancy concentration of 40% ( @xmath105 ) . thus our data indicate that both these characteristic surface lengths are coupled to the bulk connectivity length as percolation is approached . similar behavior to that observed for the saturation value of @xmath106 is also seen for the saturation value of @xmath56 . in the templated polyacrylamide gels studied in ref . @xcite , a strong increase in the crossover length and surface width at surfactant concentrations above 20% by weight coincides with a change from an optically clear material to a white , opaque one . such a change in the light scattering intensity is consistent with an increase in the bulk correlation length from much smaller than , to on the order of the wavelength of visible light . considering the numerical and experimental evidence together , we conclude that the dramatic increases in the surface - related length scales that occur with increasing volume fraction of voids are related to a percolation transition in the system of vacancies in the bulk , both in our model and in the experimental system by which it was inspired . in this paper we have proposed and studied a simple frustrated spring - network model which , with a relatively modest computational effort , reproduces aspects of the scaling behavior of the surfaces of surfactant - templated polyacrylamide gels , recently observed in afm experiments @xcite . these results suggest that simple models that incorporate elastic properties of the gel can provide a useful framework for understanding the structural characteristics of these materials , which are otherwise extremely difficult to simulate due to their complexity . the main similarities and differences between the surface structures of the model and the experimental system are as follows . _ the main similarity _ is a nontrivial scaling behavior corresponding to a power - law form of the increment correlation function at small and intermediate length scales . this power - law behavior terminates at a crossover length scale , above which the rms surface width and the increment correlation functions both reach saturation values that appear to be independent of the system size . the origin of this crossover behavior is explained in terms of the structural inhomogeneities of the gel . on large length scales the scaling behavior reflects the average network structure , while on small length scales the microscopic density fluctuations of the gel network play the most important role . the quantitative dependence of the crossover on the bulk vacancy concentration is discussed further below . _ the main difference _ between the model and the experimental surfaces lies in the details of the scaling behavior at length scales below the crossover length : the model surfaces are multi - affine , while the experimental surfaces are self - affine . the multi - affinity in the model is caused by the finite density of vertical discontinuities in the simulated surfaces ( see fig . [ fig : zooms ] ) , which are not seen in the experimental system ( see fig . 7 of ref . the discontinuities in the simulated surfaces as they are defined here result from overhangs in the spring configuration at the surface , and modification of the model to improve the detailed agreement with the scaling behavior of the experimental system should aim at eliminating these overhangs . ( it is also possible that overhangs do in fact exist in the undisturbed experimental surfaces , but are so mechanically weak that they are smoothed out by the force of the afm tip . ) in the model , the number and size of overhangs could be reduced or eliminated in several ways . one possibility is to simulate the smoothing action of the afm tip , _ e.g. _ , by a running - average method . another possibility is to include a weak gravitational field ( the specific gravity of the polymers is greater than that of water ) , which would bend overhanging sections down into the main part of the interface . it should also be noted that the larger coordination number in three dimensions most likely would reduce the importance of overhangs in a three - dimensional version of the model , even without further modifications . ( extension to three dimensions would also reduce the number of disconnected pieces , which are partly responsible for the large overhangs seen in the case of high vacancy densities . however , the computational cost would be large . ) a more exhaustive numerical and analytical study of the influence of discontinuities on surface scaling behavior is currently underway @xcite . the experimental surfactant templating was simulated by introducing a nonzero volume fraction of voids in the network through deleting lattice nodes and their associated springs . both the horizontal crossover length and the vertical surface thickness ( as measured by the saturation values of both the rms surface width and the generalized increment correlation functions ) increase dramatically as the vacancy concentration approaches the site - percolation limit of @xmath107% . an analogous increase of the characteristic horizontal and vertical length scales with surfactant concentration is observed in the experimental system , where it coincides with a dramatic increase in the optical opacity . the numerical results for our model thus support the hypothesis that this increase in the experimental system is caused by a percolation transition of the voids created by removal of the surfactants . in conclusion , despite the differences in the details of the scaling behavior at small and intermediate length scales , our simplified random - spring model successfully mimics several of the surface characteristics of an important new class of porous media , including the dramatic increase in surface roughness that occurs with increasing surfactant concentration . efforts to further improve the agreement between the small - scale scaling behavior of our theoretical model and that of the experimental system by which it was inspired are left for future study . these efforts should in particular involve investigations of the dependence of the scaling properties of the model on external forces due to gravity or the afm tip , as well as on the statistics of the local spring distribution and the dimensionality of the model . we acknowledge useful conversations with m. chakrapani , s. p. lewis , p. meakin , m. a. novotny , r. h. swendsen , and d. h. van winkle . we thank i. m. navon for bringing to our attention the l - bfgs algorithm and ref . @xcite , and j. cardy for telling us about ref . this work was supported in part by national science foundation grant no . dmr-9981815 , and by florida state university through the center for materials research and technology and the school of computational science and information technology . the computer simulations were performed on the florida state university physics department s beowulf cluster . g. m. buenda , s. j. mitchell , and p. a. rikvold , in _ computer simulation studies in condensed matter physics xiv _ , edited by d. p. landau , s. p. lewis , and h .- b . schttler ( springer , berlin , 2002 ) . e. g. richards and c. j. temple , nature physical science * 230 * , 92 ( 1971 ) .

we propose and study a simplified model for the surface and bulk structures of crosslinked polymer gels , into which voids are introduced through templating by surfactant micelles . such systems were recently studied by atomic force microscopy [ m. chakrapani _ et al . _ , e - print cond - mat/0112255 ] . the gel is represented by a frustrated , triangular network of nodes connected by springs of random equilibrium lengths . the nodes represent crosslinkers , and the springs correspond to polymer chains . the boundaries are fixed at the bottom , free at the top , and periodic in the lateral direction . voids are introduced by deleting a proportion of the nodes and their associated springs . the model is numerically relaxed to a representative local energy minimum , resulting in an inhomogeneous , `` clumpy '' bulk structure . the free top surface is defined at evenly spaced points in the lateral ( @xmath0 ) direction by the height of the topmost spring , measured from the bottom layer , @xmath1 . its scaling properties are studied by calculating the root - mean - square surface width and the generalized increment correlation functions @xmath2 . the surface is found to have a nontrivial scaling behavior on small length scales , with a crossover to scale - independent behavior on large scales . as the vacancy concentration approaches the site - percolation limit , both the crossover length and the saturation value of the surface width diverge in a manner that appears to be proportional to the bulk connectivity length . this suggests that a percolation transition in the bulk also drives a similar divergence observed in surfactant templated polyacrylamide gels at high surfactant concentrations .

the off - shell three - gluon vertex has been under investigation for more than three decades . by an analysis of the nonabelian gauge ward identities , ball and chiu@xcite in 1980 found a form factor decomposition of this vertex which is valid at any order in perturbation theory , with the only restriction that a covariant gauge be used . at the one - loop level , they also calculated the vertex explicitly for the case of a gluon loop in feynman gauge . later cornwall and papavassiliou@xcite applied the pinch technique to the non - perturbative study of this vertex . davydychev , osland and sax @xcite calculated the massive quark contribution of the one loop three - gluon vertex . binger and brodsky@xcite calculated the one - loop vertex in the pinch technique and found the following susy - related identity between its scalar , spinor and gluon loop contributions , @xmath0 in this talk , i present a recalculation of the scalar , spinor and gluon loop contributions to the three - gluon vertex using the worldline formalism @xcite . the vertex is shown in fig . 1 ( for the fermion loop case ) . following the notation of @xcite , we write @xmath1 the gluon momenta are ingoing , such that @xmath2 . there are actually two diagrams differing by the two inequivalent orderings of the three gluons along the loop . those diagrams add to produce a factor of two . the ball - chiu decomposition of the vertex can be written as @xmath3 here the @xmath4 , @xmath5 and @xmath6 functions are symmetric in the first two arguments , @xmath7 antisymmetric , and h(s ) are totally ( anti)symmetric with respect to interchange of any pair of arguments . note that the @xmath6 and @xmath8 functions are totally transverse , i.e. , they vanish when contracted with any of @xmath9 , @xmath10 or @xmath11 . the path integral ( [ bk ] ) is gaussian so that its evaluation requires only the standard combinatorics of wick contractions and the appropriate green s function , @xmath20 in this formalism structural simplification can be expected from the removal of all second derivatives @xmath21 s , appearing after the wick contractions , by suitable integrations by part ( ibp ) . after doing this we have ( see@xcite for the combinatorial details of the wick contraction and ibp procedure ) @xmath22)\int_{0}^{\infty } \frac{dt}{t^{\frac{d}{2}}}{\rm e}^{-m^2 t}\int_{0}^{t}d\tau_{1}\int_0^{\tau_{1}}d\tau_{2}\ , q_3 ^ 3\vert_{\tau_3=0}~ { \rm e}^{(g_{b12}p_{1}\cdot p_{2}+g_{b13}p_{1}\cdot p_{3}+g_{b23}p_{2}\cdot p_{3})}\nonumber\\ \gamma_{\rm scalar}^2 & = & \gamma_{\rm scalar}^3(q_3 ^ 3\to q_3 ^ 2 ) \nonumber\\ \gamma_{\rm scalar}^{\rm bt } & = & - { \mbox tr}(t^{a_{1}}[t^{a_{2}},t^{a_{3 } } ] ) \int_{0}^{\infty } \frac{dt}{t^{\frac{d}{2}}}{\rm e}^{-m^2 t}\int_{0}^{t}d\tau_{1 } \dot{g}_{b12}\dot{g}_{b21 } \bigl\lbrack\varepsilon_3\cdot f_1\cdot\varepsilon_2~ { \rm e}^{g_{b12}p_{1}\cdot ( p_{2}+p_{3 } ) } + { \rm 2\,perm } \bigr\rbrack\nonumber\\ q_{3}^3&=&\dot{g}_{b12}\dot{g}_{b23}\dot{g}_{b31}{\mbox tr}(f_1f_2f_3 ) \nonumber\\ q_{3}^2&=&\frac{1}{2}\dot{g}_{b12}\dot{g}_{b21}{\mbox tr } ( f_1f_2 ) \sum_{k=1,2}\dot{g}_{b3k}\varepsilon_{3}\cdot p_{k}+{\rm 2\,perm}\nonumber\\ \label{q3}\end{aligned}\ ] ] the abelian field strength tensors @xmath23 appear automatically in the ibp procedure . the @xmath24 s are boundary terms of the ibp . we rescale to the unit circle , @xmath25 and rewrite these integrals in term of the standard _ feynman / schwinger _ parameters , related to the @xmath26 by @xmath27 for the scalar case , we find @xmath28)(\gamma_{\rm scalar}^3 + \gamma_{\rm scalar}^2 + \gamma_{\rm scalar}^{\rm bt})\nonumber\\ \gamma_{\rm scalar}^3 & = & \gamma\bigl(3-\frac{d}{2}\bigr)\mbox{tr } ( f_1f_2f_3 ) i^d_{3,b}(p_1 ^ 2,p_2 ^ 2,p_3 ^ 2 ) \nonumber\\ \gamma_{\rm scalar}^2 & = & \frac{1}{2}\gamma\bigl(3-\frac{d}{2}\bigr)\bigl\lbrack \mbox{tr } ( f_1f_2)\bigl(\varepsilon_3\cdot p_1 i^d_{2,b}(p_1 ^ 2,p_2 ^ 2,p_3 ^ 2)-\varepsilon_3\cdot p_2 i^d_{2,b}(p_2 ^ 2,p_1 ^ 2,p_3 ^ 2)\bigr)+{\rm 2\,perm}\bigr\rbrack\nonumber\\ \gamma_{\rm scalar}^{\rm bt } & = & -\gamma\bigl(2-\frac{d}{2}\bigr ) \bigl\lbrack\varepsilon_3 \cdot f_1\cdot\varepsilon_2 i^d_{{\rm bt},b}(p_1 ^ 2)+{\rm 2\,perm } \bigr\rbrack\nonumber\\ \label{gammas0fin}\end{aligned}\ ] ] where @xmath29 by an off - shell generalization of the bern - koswer replacement rules @xcite , whose correctness for the case at hand we have verified , one can get the results for the spinor and gluon loop from the scalar loop one simply by replacing @xmath30 where the @xmath31 s are three integrals similar to the @xmath32 s above ( for the spinor loop one must also multiply by a global factor of @xmath33 ) . from ( [ gamma ] ) we immediately recover the binger - brodsky identity eq.([gammabb ] ) . @xmath39)(f}+ig[a , a])\nonumber\\ \label{comparison}\end{aligned}\ ] ] in our recalculation of the scalar , spinor and gluon contributions to the one - loop three gluon vertex we have achieved a significant improvement over previous calculations both in efficiency and compactness of the result . this improvement is in large part due to the replacement rules ( [ gamma ] ) whose validity off - shell we have verified . details and a comparison with the ball - chiu decomposition will be presented elsewhere . we believe that along the lines presented here even a first calculation of the four - gluon vertex would be feasible . 99 j. s. ball and t. w. chiu , phys . d * 22 * , 2550 ( 1980 ) . j. m cornwall and j. papavassiliou , phys . d * 40 * , 3474 ( 1989 ) . a. i. davydychev , p. osland and l. saks , jhep 0108:050 ( 2001 ) . m. binger and s. j. brodsky , phys . d * 74 * , 054016 ( 2006 ) . z. bern and d. a. kosower , phys . lett . * 66 * , 1669 ( 1991 ) ; nucl . b * 379 * , 451 ( 1992 ) . m. j. strassler , nucl . b * 385 * , 145 ( 1992 ) . m. reuter , m. g. schmidt and c. schubert , ann . ( n.y . ) * 259 * , 313 ( 1997 ) . c. schubert , phys . rept . * 355 * , 73 ( 2001 ) .

the three - gluon vertex is a basic object of interest in nonabelian gauge theory . it contains important structural information , in particular on infrared divergences , and also figures prominently in the schwinger - dyson equations . at the one - loop level , it has been calculated and analyzed by a number of authors . here we use the worldline formalism to unify the calculations of the scalar , spinor and gluon loop contributions to the one - loop vertex , leading to an extremely compact representation . the susy - related sum rule found by binger and brodsky follows from an off - shell extension of the bern - kosower replacement rules . we explain the relation of the structure of our representation to the low - energy effective action . = 11.6pt

understanding physical properties of the high-@xmath11 galaxies beyond @xmath13 is very important , because such galaxies are not only building blocks of lower-@xmath11 galaxies and local galaxies but also the leading candidate of ionizing sources for the cosmic reionization . such galaxies are much fainter objects because they are very distant objects and less massive systems . thus , in order to observe these galaxies , ultra deep surveys are necessary . the wide field camera ( wfc3 ) installed on the hubble space telescope ( hst ) in 2009 have detected many @xmath0 galaxy candidates using the dropout ( lyman break ) technique @xcite . the combination of gravitational lensing by foreground galaxy clusters and the wfc3 camera have also discovered even intrinsically fainter high-@xmath11 galaxy candidates @xcite . many spectroscopic surveys for the ly@xmath6 line of @xmath0 galaxies have been attempted to confirm spectroscopic redshift of these galaxies ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? now , the highest confirmed redshift using ly@xmath6 line is @xmath14 @xcite which is the first and only example of a successful ly@xmath6 line observation at @xmath15 . other observations so far have not succeeded in the ly@xmath6 line detection at @xmath12 @xcite . moreover , the spectroscopic surveys for ly@xmath6 line revealed that the fraction of high-@xmath11 galaxies with detectable ly@xmath6 emission decreases from @xmath16 ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? these observational results imply that ly@xmath6 attenuation by the neutral hydrogen in the intergalactic medium ( igm ) is significant because the neutral fraction of the hydrogen is high at such higher-@xmath11 before the completion of the cosmic reionization . the ly@xmath6 line detection at @xmath15 may be very difficult even by future telescopes such as the james webb space telescope ( jwst ) , the european extremely large telescope ( e - elt ) , the giant magellan telescope ( gmt ) and the thirty meter telescope ( tmt ) . thus , it appears to be important to consider the detectability of alternative lines instead of the ly@xmath6 line . furthermore , the detection of high-@xmath11 metal lines helps to understand when and how the metal enrichment proceeds at very early universe . the surveys for the far - infrared ( fir ) lines using the atacama large millimetre / submillimetre array ( alma ) can be very powerful tool to confirm redshifts of very high-@xmath11 galaxies . the [ c ii ] @xmath17 line is a well known brightest line in the fir region . the line has been detected in many qsos and submm galaxies ( smgs ) at various redshifts @xcite . [ c ii ] @xmath17 line in some high-@xmath11 lbgs has also been discovered @xcite . interestingly , almost all survey have failed to detect [ c ii ] @xmath17 line in high-@xmath11 lyman @xmath6 emitters ( laes ) @xcite . this fact disadvantages to explore high-@xmath11 [ c ii ] @xmath17 line because the fraction of young galaxies like laes increases with redshift . the [ c ii ] @xmath17 is thought to be originated mainly from photo - dissociation regions ( pdrs ) , the modelling of which is very complicated . this implies that the reliable theoretical prediction for the [ c ii ] @xmath17 line is very difficult . in our previous work @xcite , we concluded that the [ o iii ] @xmath18 line from h ii regions is potentially the best line to measure redshifts of galaxies in the early universe . the modelling of this line is simpler than that of the [ c ii ] @xmath17 line and the alma band 7 covers the line at @xmath19 . there are also various strong lines in the rest - frame uv to optical region . in the high - redshift emission line survey ( hizels ) with narrow - band filters in the @xmath20 , @xmath21 and @xmath22 bands , many h@xmath6 , [ o ii ] @xmath4 and [ o iii ] @xmath23 emitters have been discovered at @xmath24 @xcite . the spectroscopic surveys have also been performed and discovered many lines such as c iii ] @xmath2 , [ o ii ] @xmath4 , [ o iii ] @xmath23 and h@xmath6 in the uv to the optical region at @xmath25 @xcite . @xcite detected c iii ] @xmath2 emission in two high-@xmath11 galaxies at @xmath26 and @xmath27 . the c iii ] @xmath2 emission in unusually luminous galaxy at @xmath28 have been detected by @xcite . in addition , @xcite discovered the c iv ] @xmath1 and o iii ] @xmath29 emission lines in @xmath30 galaxy confirmed by the ly@xmath6 emission line . more recently , a strong he ii @xmath31 was found in @xmath32 laes ( cr7 , * ? ? ? interestingly , the cr7 has no metal lines . @xcite suggested the existence of metal - free stars so - called pop iii stars and/or the formation of a direct collapse black hole . now , the source of strong he ii @xmath31 in the cr7 is hotly debated . ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the uv to optical lines are very useful for studying the physical properties of galaxies such as the gas metallicity , the radiation filed strength and the gas density ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? nevertheless , there have been only a few line surveys at higher redshift ( @xmath33 ) using the ground - based telescopes due to the observational difficulty . this is because the wavelength of these lines in the high-@xmath11 universe is redshifted into the ir region . the next generation telescopes such as the jwst , the e - elt , the gmt and the tmt can perform a deeper survey than the current ones in the ir regime . thus , the metal lines can be good targets for these future telescopes and useful to diagnose the physical properties of very high-@xmath11 galaxies ( @xmath0 ) . unfortunately , only small number of bright galaxies at @xmath0 appropriate for the follow - up survey can be discovered by these telescopes due to their narrow field - of - view ( fov ) ( @xmath34 ) . the wide - field infrared survey telescope ( wfirst ) and the first light and reionization explorer ( flare ) , which are also future telescopes , are the ir deep and wide - field imaging surveyors . the fov of these telescopes are much larger than the tmt , the e - elt , or the jwst . many bright high-@xmath11 galaxies or very rare objects , which are good targets for spectroscopic follow - up with the tmt , the e - elt , or the jwst , will be discovered . many theoretical studies have been made on very high-@xmath11 galaxies ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . in these papers , they investigated the physical properties of star - forming galaxies at high-@xmath11 such as the halo mass , the stellar mass and the uv magnitude . they also revealed the evolution of the star - formation density , the stellar mass functions ( smfs ) , the uv luminosity function ( lfs ) and the metal enrichment in the igm . there are some studies for high-@xmath11 galaxies with nebular emission consideration . @xcite studied nebular emission effects in the spectral energy distribution ( sed ) . they showed that the estimated stellar mass , the age and the extinction strength change in the cases of with or without the nebular emission . @xcite performed galaxy sed calculations in the uv - to - optical wavelength range with the nebular emission using the photo - ionisation code cloudy @xcite . in the paper , they discussed criteria of equivalent widths of some lines and broad - band colours to select extremely metal - poor and metal - free galaxies . @xcite took into account the effect of the nebular emission on observational colour selections using a large cosmological hydrodynamical simulation . they claimed that the photometric redshift estimation is strongly affected by the strength of the nebular emission . although there are many studies for very high-@xmath11 galaxies with the nebular emission , studies about the emission line detectability in these high-@xmath11 galaxies are few . in this study , we examine whether some emission lines in the rest - frame uv to optical wavelength of galaxies at @xmath13@xmath35 are detectable with the current and future telescopes . next , we discuss the detectability of the lines at even @xmath8 in the future surveys . in section 2 , we describe our numerical simulations and calibrate the parameters therein . we virtually observe the galaxies located in our simulation through a light - cone output and select simulated galaxies by applying a relevant magnitude limit . in section 3 , we present some line fluxes in the rest - frame uv to optical wavelength of selected galaxies in our simulation . in section 4 , we make predictions using the flare , the wfirst , jwst and tmt telescopes . the final section is devoted to our conclusion . throughout this paper , we adopt a @xmath36cdm cosmology with the matter density @xmath37 , the cosmological constant @xmath38 , the hubble constant @xmath39 in the unit of @xmath40 and the baryon density @xmath41 . the matter density fluctuations are normalised by setting @xmath42 @xcite . all magnitudes are quoted in the ab system @xcite . the assumed initial mass function ( imf ) in the observational data and in our simulation is always the chabrier imf with the mass range of 0.1100 @xmath43 @xcite . in this section , we describe our cosmological hydrodynamic simulation , calculation of the spectral energy distribution ( sed ) of the simulated galaxies and the treatment of the dust attenuation for these galaxies . we performed high - resolution numerical simulations with an updated version of the tree - pm smoothed particle hydrodynamics ( sph ) code gadget-3 which is a successor of tree - pm sph code gadget-2 @xcite . we implemented relevant physical processes to galaxy formation such as the star formation , the supernova ( sn ) feedback and the chemical enrichment following papers @xcite . in our previous work @xcite , the radiation pressure and the agn feedback were newly implemented . we consequently reproduced various observational quantities such as the stellar mass functions ( smfs ) , the cosmic star formation history , the galaxy downsizing , and the relation between stellar mass and metallicity from @xmath44 to @xmath45 . the details of these processes are found in the above references . here , we briefly introduce settings of our simulations . we employ a total of @xmath46 particles for dark matter and gas in a comoving volume of @xmath47 cube . the mass of a dark matter particle is @xmath48 and that of a gas particle is initially @xmath49 , respectively . the softening length for the gravitational force is set to be @xmath50 in comoving unit . the sph gas particles can spawn star particles when they satisfies a set of standard criteria for star formation , and then , their mass is reduced . sph particles around evolved star particles get the mass including metal elements due to sne and the stellar mass loss in the agb phase , resulting in the mass reduction of the star particles . in order to identify simulated galaxies , we run a friends - of - friends ( fof ) group finder with a comoving linking length of 0.2 in units of the mean particle separation to identify groups of dark matter particles as haloes . then , we find gravitationally bound groups of at least 32 total ( dark matter + sph + star ) particles as substructures ( subhaloes ) in each fof group using subfind algorithm developed by @xcite . we regard substructures that contain at least 10 star particles as our simulated galaxies . we note that the lowest dark halo and stellar masses are around @xmath51 and @xmath52 , respectively . in order to calibrate parameters in the code such as the sn feedback efficiency , we compare our model with the observed stellar mass function ( smf ) of galaxies . [ smf ] represents our smf at @xmath13 to @xmath53 . we also plot the observational results ( the points with error bars ) at @xmath13 and @xmath54 taken from @xcite , @xcite and @xcite . our model reproduces the observed smf within error - bars . at the massive end of smfs ( @xmath55 ) , our model may underestimate the smf because our simulation box is not sufficiently large . on the other hand , our simulation tends to overproduce less massive galaxies ( @xmath56 ) when comparing the observational data at @xmath57 . this may indicate that the sfr of less massive galaxies in our simulation is @xmath58 times higher than real ones . in our previous work @xcite , we argue that the shape of the smf depends on the definition of the stellar mass , i.e. , with or without the remnant mass as the stellar mass . the effect is important for low-@xmath11 galaxies , but , it is negligible at the high-@xmath11 discussed in this paper . after the identification of the simulated galaxies , we calculate their seds . the sed of each star particle , which has its own age , metallicity , and mass , is calculated by using the population synthesis code pgase2 @xcite . then , we sum up the seds of the star particles composing a simulated galaxy to obtain the total intrinsic sed of the galaxy . we also adopt the nebular continuum and the hydrogen recombination lines ( ly@xmath6 , h@xmath6 , h@xmath7 , and h@xmath59 ) calculated by pgase2 with the case b approximation . moreover , the metal lines emitting from h ii regions are calculated based on @xcite and @xcite , in which expected line fluxes are calculated by the photo - ionization code cloudy @xcite . these line luminosity ( @xmath60 ) are proportional to h@xmath7 luminosity ( @xmath61 ) , @xmath62 where @xmath63 is the lyman continuum escape fraction and @xmath64 is the metallicity dependent emission efficiency for each line which is calculated in @xcite and @xcite , respectively . measurements of @xmath63 of an order of 0.010.1 have been obtained at @xmath65 ( e.g. , * ? ? ? * ; * ? ? ? * ) and @xmath66 at @xmath67 has been favoured for cosmic reionization @xcite . however , we simply assume @xmath68 in this paper , yielding the maximum fluxes of nebular emission lines . it is worth describing effect of a different choice of stellar population synthesis ( sps ) model . one of important point for line emissivity calculation is the lyman continuum photon production efficiency per nonionizing uv ( @xmath69 ) because line emissivity strongly correlates with the efficiency . the difference between pgase2 , starburst99 @xcite , maraston @xcite and bc03 @xcite is very small ( e.g. , * ? ? ? * ) . however , the efficiency of bpass / binary @xcite is about a factor of 2 higher than others ( see also * ? ? ? * ) ) . hence , uncertainty in the line flux estimation due to the choice of sps models is also a factor of 2 . after calculating the intrinsic seds of simulated galaxies , then , we consider the dust attenuation for the continuum and the lines of them . the procedure of dust attenuation for the continuum is same as our previous work @xcite . we adopt the calzetti law @xcite for the shape of the attenuation law . for the attenuation amplitude , we calculate the escape probability of uv photons at @xmath70 ( @xmath71 ) . we apply the sandwich model for @xmath71 @xcite , @xmath72 where @xmath73 is a parameter whose value is from @xmath74 to @xmath75 and @xmath76 is the uv ( 1500@xmath77 ) optical depth of the simulated galaxies . the parameter @xmath73 is the fraction of the thickness of the central star+dust slab in the total thickness . for the case of @xmath78 , eq . [ eq : ep ] corresponds to the well - mixed dusty slab geometry . on the other hand , it is the case with a central infinitely thin dusty sheet when @xmath73 is zero . we calculate the optical depth @xmath76 using the following equation : @xmath79 where @xmath80 and @xmath81 are a typical size and the material density of dust grains , respectively . we set @xmath82 and @xmath83 motivated by sne dust production models @xcite . the dust surface density is defined by the following equation : @xmath84 where @xmath85 and @xmath86 are the total dust mass and the effective radius of the dust distribution in a galaxy , respectively . these values are directly obtained from our simulations as follows . the dust mass @xmath85 and @xmath86 are assumed to be proportional to the metal mass , @xmath87 , and the half stellar mass radius , @xmath88 , of each simulated galaxy , respectively . therefore , @xmath85 and @xmath86 are expressed as @xmath89 and @xmath90 , where @xmath91 and @xmath92 are the proportional constants of the dust mass and the effective radius , respectively . these two parameters can be reduced to one parameter . thus , equation ( [ eq : dsmd ] ) is reduced to @xmath93 where @xmath94 is a global constant for all the galaxies in our simulation . we calibrate @xmath95 and @xmath73 so as to reproduce the observed uv luminosity function ( lf ) at @xmath13 , and we keep these values even at higher redshifts . in this study , we adopt @xmath96 and @xmath97 , respectively . when we calculate band magnitudes , we also apply the igm absorption for the blue side of @xmath98 following @xcite , @xcite with an extrapolation for @xmath0 . the top panel of fig . [ uvlf ] represents the uv lfs from @xmath13 to @xmath99 . observational results are also shown in the same panel . our model reproduces the observations well up to @xmath100 . moreover , we predict the uv lfs beyond @xmath99 in the bottom panel of fig . [ uvlf ] . next , we consider the dust attenuation for various emission lines . in this study , we apply the same attenuation law as the continuum to the emission lines but with a different normalization @xcite . according to @xcite , the attenuation for emission lines is a factor of about 2 larger than that for the continuum at the same wavelength , although different factors were suggested by recent studies ( e.g. , * ? ? ? the line escape probability ( @xmath101 ) at @xmath70 which is the normalization of the attenuation law for emission lines is @xmath102 where @xmath71 and 2.27 are the normalization of the attenuation law for the continuum ( eq . [ eq : ep ] ) and the attenuation enhancement for emission lines , respectively . after calculation of the dust attenuation for the continuum and lines , we finally calculate the dust thermal emission in the same manner as @xcite and @xcite . one difference from our previous work @xcite is that not only the continuum photon energy but also the line photon energy absorbed by dust is converted into the dust luminosity . we assume that all the energy absorbed by the dust is re - emitted in the ir emission . the monochromatic luminosity ( @xmath103 ) at frequency @xmath104 is written as @xmath105 where @xmath106 , @xmath107 and @xmath108 are the absorption coefficient , the planck function and the dust grain temperature , respectively . we assume that 50 % of the total metal mass in a simulated galaxy is converted into the dust mass ( @xmath85 ) . the absorption coefficient @xmath106 in the fir is well described by a power - law @xmath109 with @xmath110 . in this study , we apply @xmath111 based on observational results @xcite . the typical dust temperature of our simulated galaxies is about @xmath112 . after these procedures , we obtain realistic sed of simulated galaxies from uv to ir / fir wavelength . fig . [ sed ] represents an example sed from uv to ir of a simulated galaxy . as described our previous work @xcite , the [ o iii ] @xmath18 line is the most prominent line . in the uv to optical range , many bright lines are also seen . we note that , in this study , the pdr / molecular cloud lines are not considered and the pah features are not included in our seds , either . these are put into our future work . here , we explore whether our simulation is reasonable to predict line detectability . we check essential physical properties of our simulated galaxies such as the stellar mass , the sfr , the dust attenuation and the metallicity which are necessary to calculate the line fluxes . in order to directly compare our model with the observations , we make galaxies distribution on a light - cone from @xmath16 to @xmath113 the same as in our previous work @xcite . the field - of - view ( fov ) in this study is ( @xmath114 ) which is about 150 times wider than the hst fov . we regard objects brighter than @xmath115 as our selected galaxies . then , we divide them into four redshift subsets at @xmath116 and @xmath117 . the star formation rate ( sfr ) of simulated galaxies is very important because the line luminosities of various elements are essentially proportional to the sfr . fig . [ sm - sfr_hst ] shows the star formation rates as a function of the stellar mass . the large triangle , square , circle and star points are the median values of sfr in the stellar mass bins for @xmath118 and @xmath100 , receptively . the boxes and error - bars ranges show @xmath119 percentile and @xmath120 percentile in the distributions . we also plot the observational data at @xmath121 ( cross points with error - bars ) for comparison @xcite . the sfr strongly correlates with their stellar mass . interestingly , the trend is similar to that of the star forming galaxies at lower-@xmath11 universe known as the main sequence distribution ( e.g. , * ? ? ? * ; * ? ? ? we find that the distribution of @xmath57 galaxies is almost same as the observation at @xmath121 . moreover , the evolution from @xmath57 to @xmath100 is very weak . this implies that line luminosities are also proportional to their stellar mass . [ band - auv_hst ] represents the dust attenuation @xmath122 at @xmath70 as a function of @xmath123 band magnitude . the large triangle , square , circle and star points are the median values of @xmath122 in @xmath123 band magnitude bins for @xmath118 and @xmath100 , receptively . the boxes and error - bars ranges show @xmath119 percentile and @xmath120 percentile in the distributions . clearly , there are a few galaxies strongly affected by the dust attenuation . the typical value of the dust attenuation correlates with the uv magnitude even though there is a large scatter . we note that the typical value is smaller than that we find in our previous work @xcite . in the new code which we use in this study , the radiation pressure and the agn feedback are newly included in addition to the sn feedback @xcite . especially , the radiation pressure and the sn feedback effectively suppress the star formation in small mass galaxies than the case of the sn feedback only simulation . as a result , less dust attenuation is required to reproduce the uv lfs . next , we calculate the uv slope @xmath7 with the exactly same formulae as that in the observations @xcite : @xmath124 for @xmath57 galaxies , @xmath125 for @xmath126 galaxies , @xmath127 for @xmath128 galaxies and @xmath129)- 2.0,\ ] ] for @xmath100 galaxies , respectively . @xmath130 and @xmath131 are wfc3 band filters equipped on the hubble space telescope and @xmath132 $ ] is the spitzer irac band1 [ uv_slope_hst ] shows the slope @xmath7 as a function of the absolute uv magnitude ( without dust correction ) . the large triangle , square , circle and star points are the median values of @xmath7 in the absolute uv magnitude bins for @xmath118 and @xmath100 , receptively . the boxes and error - bars ranges show @xmath119 percentile and @xmath120 percentile in the distributions . the large cross points are also shown as the observational data @xcite . as described our previous paper , the large scatter of the observational data may be due to the observational error and/or uncertainty . the uv brighter galaxies have larger @xmath7 values . this is because the uv brighter galaxies tend to be more evolved and have a smaller fraction of young stellar populations than the uv fainter ones . moreover , such uv brighter galaxies rapidly proceed metal enrichment . this trend can be seen in high-@xmath11 galaxy observations @xcite . at @xmath128 , the @xmath7 distribution of our model shows the large scatter and different trend from the other redshift ones . this is because ly@xmath6 break or line of some @xmath133 simulated galaxies enters in @xmath131 band . if we adopted the same @xmath7 formula as @xmath100 for @xmath128 , we would get a similar @xmath7 distribution for @xmath128 to other redshifts . this suggests that some modifications are necessary for @xmath128 @xmath7 estimator . in the fiducial line emission model of @xcite , the line emissivities are proportional to the metallicity in the case of @xmath134 where @xmath135 and @xmath136 are the gas metallicity and the solar metallicity ( @xmath137 ) . then , when @xmath138 , the dependence becomes opposite because of lower lyman continuum ( lyc ) emissivity for higher metallicity . thus , in order to detect the line emissions of very high-@xmath11 galaxies , the follow - up survey for @xmath139 systems is efficient . [ sm - metal_hst ] represents the nebular metallicity of the simulated galaxies as a function of the stellar mass . the definition of the metallicity is the same as our previous work @xcite in which the metallicity is defined by the lyc luminosity weighted average metallicity called the nebular metallicity rather than the mass weighted average metallicity . the nebular metallicity of the simulated galaxies is proportional to their stellar mass . the metallicity of some galaxies reaches @xmath139 even at @xmath100 . this means that the metal enrichment or the chemical evolution proceeds rapidly in the early universe . interestingly , the redshift dependence of the metallicity is very weak or almost no evolution . thus , the each line emissivity at the redshift range varies little if their stellar mass is similar values . in this section , using our mock galaxies , we explore expected fluxes of some emission lines . we investigate the observed line fluxes of our simulated galaxies and discuss the detectability of these lines at @xmath1310 . then , we predict the detectability in @xmath8 galaxies . here , we show some bright and important metal lines for the diagnostics of the gas metallicity . we also present hydrogen recombination lines such as h@xmath6 and h@xmath7 . the @xmath140-index defined by ( [ o iii ] @xmath141+[o iii ] @xmath142+[o ii ] @xmath143)/h@xmath7 is well known as a good indicator for gas metallicity . moreover , the n2 ( @xmath144([n ii ] @xmath145/h@xmath6 ) ) and o3n2 ( @xmath144(([o iii ] @xmath23/h@xmath7)/([n ii ] @xmath145/h@xmath6 ) ) ) indices are also used to estimate the oxygen abundances . @xcite suggests that the line ratio [ ne iii ] @xmath146/[o ii ] @xmath4 is also a useful metallicity indicator for high-@xmath11 galaxies , especially when the r23-index or other diagnostics involving [ o iii ] @xmath23 or [ n ii ] @xmath145 are not available . fig . [ band - line_hst ] represents the observed line fluxes of those important emission lines in the rest - frame uv to optical wavelength as a function of observed @xmath123 band magnitude . we calculate all elements for the diagnostics ( r23-index , n2 and o3n2 ) of the gas metallicity . the small triangle , square , circle and star points are the expected line fluxes of the simulated galaxies at @xmath57 , 8 , 9 and 10 , respectively , while the large points with error - bars represent the average values and their standard deviations in @xmath123 band magnitude bins . we also plot the observational results by @xcite , @xcite and @xcite as the two cross and plus - sign points ( the fainter point was demagnified by the quoted lensing magnification ) in the civ @xmath1 and the c iii ] @xmath147 panels , which are very consistent with our prediction . we explore whether the existing facilities such as the vlt / x - shooter and the keck / mosfire can detect the emission lines considered in this paper . the wavelength range of these instruments is about 12.5 @xmath148 . although , the detection limit is not deep enough ( a few@xmath149 $ ] ) under a practical observational condition , the civ @xmath1 and the c iii ] @xmath147 lines of the brighter galaxies ( @xmath150 ) are detectable at @xmath0 without a significant magnification by gravitational lensing . recently , @xcite reported no c iii ] @xmath147 line with @xmath151 was detected under the condition of the detection limit @xmath152 $ ] . this is also consistent with our result because typical line fluxes with @xmath151 are @xmath153 $ ] which is below the observational detection limit . in order to detect these nebular lines , we need the future facilities with a wider wavelength coverage ( the rest - frame uv to optical ) and a stronger light - gathering power . the jwst and the tmt have potential to detect the nebular line of very high-@xmath11 galaxies . we consider the detectability of the nebular lines using the jwst and the tmt . the jwst / nirspec , the jwst / miri and the tmt / irms provide spectroscopy over the wavelength range of 0.65 @xmath148 , 528 @xmath148 and 0.82.5 @xmath148 , respectively . the point - source detection limits of the jwst / nirspec , jwst / miri and tmt / irms , which depend on the observed wavelength are a few@xmath154 $ ] , a few@xmath149 $ ] and a few@xmath155 $ ] , respectively , if the integration time is @xmath156 $ ] and the signal to noise ratio ( s / n ) is 5 . we note that only c iv @xmath157 and the c iii ] @xmath147 lines at @xmath1310 can be available for the tmt / irms due to the wavelength coverage shorter than 2.5 @xmath148 . interestingly , the redshift evolution of the each line flux against the apparent @xmath123 band magnitude is not seen . this is because the sfr and metallicity evolutions of the simulated galaxies are weak as shown in fig . [ sm - sfr_hst ] and fig . [ sm - metal_hst ] . moreover , the evolution of the dust attenuation is also weak as shown in fig . [ band - auv_hst ] . almost all these lines are detectable by these instruments if the band magnitude are brighter than about 28 mag . this means that we may detect some lines even at @xmath8 galaxies if the galaxies are bright enough ( @xmath158 ) and we can confirm the spectroscopic redshift using these lines . moreover , we can study the gas metallicity of bright high-@xmath11 galaxies ( @xmath159 ) through the line ratio diagnostics . finally , we comment on heii @xmath160 in the cr7 @xcite . we also calculate the line and search the cr7 candidates around @xmath32 . however , we can not find any cr7 like heii @xmath160 bright galaxies in our simulation . this suggests that in order to reproduce such an object , we should consider the first star formation and/or intermediate black - hole formation ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? this is beyond the scope of this paper . it will be our future work . the equivalent width ( ew ) of lines is also very interesting to discuss detectability of lines in high-@xmath11 galaxies because the value strongly depends on the star formation history and the imf . [ flux - ew_hst ] represents the rest - frame ews of various lines in the rest - frame uv to optical wavelength as a function of line fluxes . the large triangle , square , circle and star points are the median values of ews in the observed line flux bin for @xmath118 and @xmath100 , receptively . the boxes and error - bars ranges show @xmath119 percentile and @xmath120 percentile in the distributions . we also plot the observational results by @xcite , @xcite and @xcite as the cross and plus - sign points in the civ @xmath1 and the c iii ] @xmath147 panels . the strong lines exceeding @xmath161 can be seen in some lines such as [ oiii ] @xmath5 , h@xmath6 and h@xmath7 . in the fainter line fluxes ( less massive galaxies ) regime ( @xmath162 $ ] ) , the ews are proportional to their line fluxes . the ew represents the current sfr divided by an average of sfr over a certain time duration , and the line flux is almost proportional to the sfr . the metallicity effect on the line emissivity is secondary because the simulated galaxies have more or less a metallicity about @xmath163 solar ( see fig . [ sm - metal_hst ] ) . if we choose galaxies emitting a weak line flux , they are in a low sfr phase , so that the ew is small . in the brighter line fluxes ( massive galaxies ) regime ( @xmath164 $ ] ) , ews are almost constant ( or very slowly decrease with increasing line fluxes ) . these galaxies are in an on - going star - formation episode ( or starburst ) and their ew ( or @xmath165 ) reaches an asymptotic value . interestingly , only the h@xmath6 and h@xmath7 lines show a decreasing ew as increasing the line flux rather than a constant . this is because stronger line flux galaxies tend to have a higher metallicity and the lyc photons production efficiency per nonionizing photon decreases with the metallicty increasing . on the other hand , for metal lines , the effect is cancelled out by an increase of the line emissivity due to increasing metallicity . finally , we mention higher h@xmath6 ew galaxies than theoretical expectation reported by @xcite and @xcite . in our simulation , there is no object which can reproduce such a very high ew . this implies that in order to reproduce such higher ew galaxies , we may need to include the metal - free star formation and/or a top - heavy imf which we are not considered in this study . we have presented the detectability of multiple emission lines in the rest - frame uv to optical wavelength at @xmath5710 . here , we discuss the detectability of these lines at even higher-@xmath11 , @xmath16715 . the number of bright galaxies decreases with increasing redshift , thus , a very wide survey is necessary to discover sufficiently bright very high-@xmath11 galaxies . in this study , we feature high-@xmath11 galaxies which will be detected with the wide - field infrared survey telescope ( wfirst ) and the first light and reionization explorer ( flare ) . the surveys with these facilities are potentially the most efficient one to detect very high-@xmath11 galaxies thanks to the wide survey area and the depth . we notice that the instrument design of these facilities has not been determined yet . in the survey strategy , they expect to detect many high-@xmath11 galaxy candidates at @xmath168 . in order to compare our model with the future observations , we make a mock galaxy catalogue at @xmath168 . we can identify a few galaxies at @xmath10 even though the fov of our simulation is very narrow ( @xmath169 ) . [ sm - sfr_high_z ] and [ sm - metal_high_z ] represent the evolution of sfr and nebular metallicity , respectively . these results suggest that a weak redshift evolution of the line flux magnitude relation even at @xmath170 as lower-@xmath11 discussed in the previous section . thus , some line emissions with k band magnitude @xmath171 mag can be detected even at @xmath10 with the jwst / nirspec and the jwst / miri . next we estimate the line fluxes of these selected galaxies . [ band - line_high_z ] represents the expected line fluxes in the rest - frame uv to optical wavelength as a function of k band magnitude . the point styles and their corresponding redshifts are noted in the panel . the points with error - bars represent the average values and the standard deviations in k band magnitude bins . the line flux magnitude relations of different redshifts are almost the same as found in the cases of @xmath172 galaxies as expected . we find that the c iv @xmath1 and the c iii ] @xmath2 lines of simulated galaxies with k band magnitude @xmath173 mag are detectable even at @xmath174 with the tmt / irms . on the other hand , in our simulation , there are no galaxies at @xmath10 that are bright enough . this implies that the survey area of our simulation ( fov@xmath169 ) is not wide enough to have galaxies emitting line fluxes bright enough to be detectable with the jwst or the tmt . therefore , a much wider survey area is imperative to discover line detectable galaxies beyond @xmath175 . we can detect some emission lines using the jwst or the tmt if the bright galaxy candidates are discovered at any redshifts because there is no evolution in the relation between apparent uv magnitude and line fluxes along the redshift . we also present the rest - frame ews as a function of observed line fluxes in fig . [ flux - ew_high_z ] . same trend as the @xmath176 cases can be seen . there are very ew strong lines exceeding 100 @xmath177 even @xmath8 . we have not discussed yet how many line emitting galaxies can be detected by the future telescopes . this topic is very important for future plans . we show cumulative luminosity functions ( lfs ) of various emission lines at @xmath178 in fig . [ line_lf ] . here , we discuss only the c iv @xmath1 and the c iii ] @xmath2 lines because the line have strong potential to be detected even at @xmath8 with the tmt and the jwst . typical detection limits of tmt / irms ( dashed line ) , jwst / nirspec ( solid line ) and jwst / miri are also shown in fig . [ line_lf ] if the integration time is @xmath179 $ ] and the signal - to - noise ratio ( s / the detection limit for tmt / irms is around @xmath180 $ ] . thus , one or more c iv @xmath1 and c iii ] @xmath2 emitters at @xmath181 can be detected by tmt / irms if the survey volume is about @xmath182 $ ] which is equivalent to @xmath183 fovs of the tmt and the jwst . on the other hand , the wavelength of the line at @xmath184 is beyond the coverage of the instrument . the jwst / nirspec can cover that wavelength range and the detection limit is around @xmath185 $ ] . in this case , a larger survey volume is necessary to detect the emitters because there is no detectable galaxy in our small fov ( @xmath169 ) simulation . we emphasise again that a wide imaging survey with the wfirst and the flare is very useful to discover very higher-@xmath11 line emitters . here , we mention detectability of galaxies magnified by gravitational lensing . recently , many surveys with lensing magnification have been actively performed . many faint galaxies have been discovered by the combination the hst / wfc3 and gravitational lensing . thanks to the magnification by gravitational - lensing , some lensed galaxies which are candidates of @xmath12 objects are detectable by even the current facilities such as the vlt / x - shooter and the keck / mosfire . let us discuss the detectability of the c iii ] @xmath2 emission line from @xmath12 galaxy candidates found by surveys for galaxy cluster lensing fields like the cluster lensing and supernova survey with hubble ( clash ) and the hubble frontier field ( hff ) with instruments currently available . the c iii ] @xmath2 line is the second strongest emission line after ly@xmath6 in the uv range and can be observed with ground - based 8 - 10 m class telescopes at @xmath181 . table [ table1 ] is a summary of the candidates to be discussed here . first , we estimate the range of their real apparent @xmath123 magnitudes by demagnifying with the quoted magnification factor @xmath186 and its uncertainty . next , we select the galaxies within the magnitude range and satisfying appropriate colour selection criteria from our simulation . then , we obtain the magnification factor for each simulated galaxy so as to be observed as the observed @xmath123 magnitude . finally , we magnify the c iii ] @xmath2 emission line flux by this factor , and obtain the distribution of the expected c iii ] @xmath2 line fluxes . fig . [ lensed_galaxy ] shows the result for the five @xmath12 candidates . the currently reachable limit is @xmath187 $ ] ( e.g. , * ? ? ? the c iii ] @xmath2 line from macs1149jd is detectable with a high probability , 73 or 85% for the two quoted magnification factors . the detectability of the line from macs0647jd1 is also relatively high ( 57% ) . the other three candidates have a detectability of @xmath188% . therefore , macs1149jd and macs0647jd1 are the two best targets for the follow - up spectroscopy aiming to detect the c iii ] @xmath2 line even with the current facilities . in order to study higher-@xmath11 ( @xmath0 ) line emitting galaxies , the combination of the gmt , the tmt or the jwst and gravitational lensing effect could be one of the most powerful ways . [ cols="<,^,^,^,^,^,<",options="header " , ] @xmath2 line for gravitationally - lensed @xmath12 galaxy candidates found so far . for macs1149jd , there are two estimates for the magnification factor and we show two cases ( solid and dotted histograms).,width=302 ] we have performed a large and high resolution cosmological hydrodynamic simulation to investigate the detectability of nebular lines in the rest - frame uv to optical wavelength range at @xmath0 . our new simulation code based on @xcite and @xcite can reproduce not only the stellar mass function , the downsizing , the mass - metallicity ratio but also the statistical properties of star forming galaxies such as lyman break galaxies ( lbgs ) , lyman @xmath6 emitters and sub - mm galaxies @xcite . first , we calibrate our model parameters so as to reproduce not only the stellar mass function at @xmath57 but also the uv luminosity function from @xmath57 to @xmath100 . then , we generate a light - cone output which extends from @xmath16 to @xmath113 using a number of simulation outputs . after making this light - cone output , we adopt the same colour selection criteria the same as in the observations with the hst / wfc3 camera @xcite . using the colour selected galaxies , we explore expected emission line fluxes in the rest - frame uv to optical wavelength such as c iv @xmath1 , c iii ] @xmath2 , [ o iii ] @xmath23 and h@xmath6 , etc . these lines are very important not only to confirm spectroscopic redshift of the galaxies but also to perform the diagnostics of the gas metallicity . we find that the redshift evolution of the line flux as a function of the @xmath123 magnitude is fairly weak because the metallicity evolution is very weak as shown in fig . [ sm - metal_hst ] . the c iv @xmath1 and the c iii ] @xmath2 lines from galaxies with the band magnitude @xmath171 ab at @xmath0 are detectable even by the current telescopes such as the vlt / x - shooter and the keck / mosfire . indeed , @xcite successfully detected the lines from two galaxies at @xmath57 and their detections are fully consistent with our prediction . many important lines are detectable with the jwst / nirspec , jwst / miri and the tmt / irms if the @xmath123 band magnitude is brighter than 28 mag . especially , the c iv @xmath1 , c iii ] @xmath2 , [ o iii ] @xmath5 and h@xmath7 lines are good targets for them . finally , we predict the detectability of nebular lines for @xmath189 galaxies satisfying k band magnitude @xmath190 mag . the redshift evolution of the line flux k band magnitude relation is very weak even at @xmath189 . we find that the c iv @xmath1 and the c iii ] @xmath2 lines are detectable by the jwst / nirspec , and tmt / irms even at @xmath174 . however , there are few galaxies at @xmath10 to exceed the detection limit of the future telescopes in our small fov simulation . this does not mean that we can not detect emission lines from @xmath10 galaxies . in fact , the jwst and the tmt can detect emission lines from galaxies with @xmath191 ab mag at any redshift which would be discovered , for example , by a @xmath192 survey planned with the future telescope such as the wfirst and the flare . we can detect @xmath10 galaxies with a help of magnification by gravitational lensing even narrow fovs . we predict that , according to our model , the c iii ] @xmath2 line in @xmath12 galaxy candidates ( macs1149jd and macs0647jd1 ) is detectable using even the current facilities such as the vlt / x - shooter and the keck / mosfire with high probability . the survey with the combination of gravitational lensing and the gmt , the tmt , or the jwst is a good way to study higher-@xmath11 ( @xmath0 ) line emitting galaxies . we thank our referee for his / her constructive comments . we would like to thank tohru nagao for helpful discussion . numerical simulations have been performed with cray xc30 in cfca at naoj and with coma cluster at centre for computational sciences in university of tsukuba . is acknowledges the financial support of grant - in - aid for young scientists ( a : 23684007 ) by mext , japan . to acknowledges the financial support of grant - in - aid for young scientists ( b : 24740112 ) . aki acknowledges the financial support of grant - in - aid for scientific research ( b : 26287034 ) . 200 agarwal , b. , johnson , j. l. , zackrisson , e. , labbe , i. , van den bosch , f. c. , natarajan , p. , khochfar , s. , 2015 , arxivl151001733 atek , h et al . 2015 , apj , 814 , 69 bouwens , r. j. , et al . , 2011 , nature , 469 , 504 bouwens , r. j. , et al . , 2011 , apj , 737 , 90 bouwens , r. j. , et al . , 2014a , apj , 793 , 115 bouwens , r. j. , et al . , 2014b , apj , 795 , 126 bouwens , r. j. , et al . , 2015 , apj , 803 , 34 bradley , l. d. , et al . , 2012 , apj , 760 , 108 brammer , g. b. , van dokkum , p. g. , illingworth , g. d. , bouwens , r. j. , labb i. , franx m. , momcheva i. , oesch p. a. , 2013 , apjl , 765 , 2 bruzual , g. , charlot , s. 2003 , mnras , 344 , 1000 bunker , a. j. , et al . , 2013 , mnras , 430 , 3314 calzetti , d. , armus l. , bohlin , r. c. , kinney , a. l. , koornneef , j. , storchi - bergmann , t. , 2000 , apj , 533 , 682 capak , p. l. , et al . , 2011 , nature , 470 , 233 capak , p. l. , faisst , a. , vieira , j. d. , tacchella , s. , carollo , m. , scoville , n. z. , 2013 , apj , 773 , 14 capak , p. l. et al . , 2015 , nature , 522 , 455 capak , p. l. , 2015 , arxiv:150307596 caruana , j. , bunker , a. j. , wilkins , s. m. , stanway , e. r. , lacy , m. , jarvis , m. j. , lorenzoni , s. , hickey , s. , 2012 , mnras , 427 , 3055 caruana , j. , bunker , a. j. , wilkins , s. m. , stanway , e. r. , lorenzoni , s. , jarvis , m. j. , ebert , h. , 2014 , mnras , 443 , 2831 chabrier g. , 2003 , pasp , 115 , 763 clay , s. , thomas , p. , wilkins , s. , henriques , b. , 2015 , mnras , 451 , 2692 coe , d. , et al . , 2013 , apj , 762 , 32 daddi , e. , et al . , 2007 , apj , 670 , 156 dayal , p. , dunlop , j. s. , maio , u. , ciardi , b. , 2013 , mnras , 434 , 1486 dunlop , j. s. , et al . , 2013 , mnras , 432 , 3520 dunne , l. , eales , s. a. , 2001 , mnras , 327 , 697 dunne , l. , eales , s. a. , edmunds , m. g. , ivison , r. j. , alexander , p. , clements , d. l. , 2000 , mnras , 315 , 115 ellis , r. s. , et al . , 2013 , apj , 763 , l7 feng , y. , di matteo , t. , croft , r. a. , bird , s. , battaglia , n. , wilkins , s. , 2016 , mnras , 455 , 2778 ferland , g. j. , et al . 2013 , rmxaa , 49 , 137 finkelstein , s. l. , et al . 2013 , nature , 502 , 524 finkelstein , s. l. , et al . 2014 , arxiv:1410.5439 finlator , k. , oppenheimer , b. d. , dav , r. , 2011 , mnras , 410 , 1703 fioc , m. , rocca - volmerange , b. , 1997 , a&a , 326 , 950 geach j. e. , smail , i. , best , p. n. , kurk , j. , casali , m. , ivison , r. j. , coppin , k. , 2008 , mnras , 388 , 1473 gonzlez , v. , labb , i. , bouwens , r. j. , illingworth , g. , franx , m. , kriek m. , 2011 , apj , 735 , 34 gonzaez - lopez , j. , et al . , 2014 , apj , 784 , 99 grazian , a. , et al . 2015 , a&a , 575 , a96 hartwig , t. , et al . , 2015 , arxiv:151201111 huang , k .- h . , et al . , 2015 , arxiv:1504.02099 inoue , a. k. , iwata , i. , deharveng , j. m. , buat , v. , burgarella , d. 2005 , a&a , 435 , 471 inoue , a. k. , iwata , i. , deharveng , j. m. , 2006 , mnras , 371 , 1 inoue , a. k. , shimizu , i. , tamura , y. , matsuo , h. , okamoto , t. , yoshida , n. , 2014 , apj , 780 , 18 inoue , a. k. , shimizu , i. , iwata , i. , tanaka , m. , 2014 , mnras , 442 , 1805 inoue , a. k. , 2011 , mnras , 415 , 2920 iono , d. , et al . , 2006 , apj , 645 , 97 ishigaki , m. , kawamata , r. , ouchi , m. , oguri , m. , shimasaku , k. , ono , y. , 2015 , apj , 799 , 12 iwata , i. , et al . , 2009 , apj , 692 , 1287 kaneker , n. , wagg , j. , ram chary , r. , carilli , c. 2013 , apj , 771 , l20 kashino , d. , et al . , 2014 , apj , 785 , 37 koekemoer , a. m. , 2013 , apjs , 209 , 3 leitherer , c. , et al . 1999 , apjs , 123 , 3 mcleod , d. j. , mclure , r. j. , dunlop , j. s. , 2016 , arxiv:1602.05199 madau , p. , 1995 , apj , 441 , 18 maiolino , r. , et al . , 2005 , a&a , 440 , 51 mannucci f. , cresci g. , maiolino r. , marconi a. , gnerucci a. , 2010 , mnras , 408 , 2115 maraston , c. 2005 , mnras , 362 , 799 matthee , j. j. a. , et al . , 2014 , mnras , 440 , 2375 mclure , r. j. , et al . , 2013 , mnras , 432 , 2696 nagao t. , maiolino r. , marconi a. , 2006 , a&a , 459 , 85 nakajima , k. , ouchi , m. , 2014 , mnras , 442 , 900 noeske , k. g. , et al . , 2007 , apj , 660 , 43 nozawa , t. , kozasa , t. , umeda , h. , maeda , k. , nomoto , k. , 2003 , apj , 598 , 785 oesch , p. a. , et al . , 2012 , apj , 759 , 135 oesch , p. a. , et al . , 2013 , apj , 773 , 75 oesch , p. a. , et al . , 2014 , arxiv1409.1228 oesch , p. a. , et al . , 2015 , arxiv1502.05399 okamoto , t. , frenk , c. s. , jenkins , a. , theuns , t. 2010 , mnras , 406 , 208 okamoto , t. , frenk , c. s. , 2009 , mnras , 399 , 174 okamoto , t. , nemmen , r. s. , bower , r. g. , 2008 , mnras , 385 , 161 okamoto , t. , shimizu , i. , yoshida , n. , 2014 , pasj , 66 , 70 oke , j. b. , 1990 , aj , 99 , 1621 ono , y. , et al . , 2012 , apj , 744 , 83 ono , y. , et al . , 2013 , apj , 777 , 155 ota , k. , et al . , 2014 , apj , 792 , 34 ouchi , m. , et al . , 2013 , apj , 778 , 102 pacucci , f. , ferrara , a. , volonteri , m. , dubus , g. , 2015 , mnras , 454 , 3771 pallottini , a et al . , 2015 , mnras , 453 , 2465 pentericci l. et al . , 2011 , apj , 743 , 132 pentericci l. et al . , 2014 , apj , 793 , 113 pirzkal , n , . et al . , 2015 , apj , 804 , 11 planck collaboration . , 2014 , a&a , 571 , 16 robertson , b. e. , ellis , r. s. , dunlop , j. s. , mclure , r. , 2010 , nature , 468 , 49 robertson , b. e. , et al . , 2013 , apj , 768 , 71 salmon , b. , et al . , 2015 , apj , 799 , 183 sanders , r. , et al . , 2015 , apj , 799 , 138 schaerer , d. , de barros , s. , 2009 , a&a , 502 , 423 schenker , m. a. , stark , d. p. , ellis , r. s. , robertson , b. e. , dunlop , j. s. , mclure , r. j. , kneib , j .- p . , richard , j. , 2012 , apj , 744 , 179 schenker , m. a. , et al . , 2013 , apj , 768 , 196 schenker , m. a. , ellis , r. s. , konidaris , n. p. , stark , d. p. , 2014 , apj , 795 , 20 shapley , a. e. , steidel , c. c. , pettini , m. , adelberger , k. l. , erb d. k. , 2006 , apj , 651 , 688 shibuya , t. , kashikawa , n. , ota , k. , iye , m. , ouchi , m. , furusawa , h. , shimasaku , k. , hattori , t. , 2012 , apj , 752 , 114 shim , h. , et al . 2011 , apj , 738 , 69 shimizu , i. , inoue , akio . k. , 2013 , pasj , 65 , 96 shimizu , i. , inoue , a. k. , okamoto , t. , yoshida , n. , 2014 , mnras , 440 , 731 shimizu , i. , yoshida , n. , okamoto , t. , 2011 , mnras , 418 , 2273 shimizu , i. , yoshida , n. , okamoto , t. , 2012 , mnras , 427 , 2866 siana , b. , et al . 2015 , apj , 804 , 17 silverman , j. , et al . , 2014 , arxiv1409.0447 smit , r. , bouwens , r. j. , labb , i. , et al . 2016 , arxiv:1511.08808 sobral , d. , et al . , 2009 , mnras , 398 , 75 sobral , d. , best , p. n. , smail , i. , mobasher , b. , stott , j. , nisbet , d. , 2014 , mnras , 437 , 3516 sobral , d. , et al . , 2015 , mnras , 450 , 630 sobral , d. , matthee , j. , darvish , b. , schaerer , d. , mobasher , b. , rottgering , h. j. a. , santos , s. , hemmati , s. , 2015b , apj , 808 , 139 song , m. et al . , 2015 , arxiv:1507.05636 springel , v. , white , s. d. m. , tormen , g. , kauffmann , g. , 2001 , mnras , 328 , 726 springel , v. , 2005 , mnras , 364 , 1105 stanway , e. r. , eldridge , j. j. , becker , g. d. 2015 , arxiv:1511.03268 stark , d. p. , ellis , r. s. , ouchi , m. 2011 , apjl , 728 , l2 stark , d. p. , et al . 2014 , mnras , 445 , 3200 stark , d. p. , et al . 2015 , mnras , 450 , 1846 stark , d. p. , et al . 2015 , mnras , 454 , 1393 stark , d. p. , et al . 2016 , arxiv:1606.01304 tilvi , v. , et al . , 2014 , apj , 794 , 5 todini , p. , ferarra , a. , 2001 , mnras , 325 , 726 treu , t. , schmidt , k. b. , trenti , m. , bradley , l. d. , stiavelli , m. , 2013 , apj , 775 , 29 vanzella , e. , et al.2014 , a&a , 569 , 78 vieira , j. d. , et al . , 2013 , nature , 495 , 344 walter , f. , et al . , 2012 , apj , 752 , 93 wang , r. , et al . , 2013 , apj , 773 , 44 watson , d. , christensen , l. , knudsen , k. k. , richard , j. , gallazzi , a. , michalowski , m. j. , 2015 , nature , 519 , 327 wilkins , s. m. et al . , 2013 , mnras , 435 , 2885 wilkins , s. m. , bouwens , r. j. , oesch , p. a. , labbe , i. , sargent , m. , caruana , j. , wardlow , j. , clay , s.,2016 , mnras , 455 , 659 xu , c. , buat , v. , 1995 , a&a , 293 , l65 yan , h. , et al . 2011 , apjl , 728 , l22 zheng , w. , et al . , 2012 , nature , 489 , 406 zitrin , a. , et al . , 2014 , apj , 793 , 12 zitrin , a. , ellis , r. s. , belli , s. , stark , d. p. , 2015 , apj , 805 , 7 zitrin , a. , et al . , 2015 ,

we have performed very large and high resolution cosmological hydrodynamic simulations in order to investigate detectability of nebular lines in the rest - frame ultraviolet ( uv ) to optical wavelength range from galaxies at @xmath0 . we find that the expected line fluxes are very well correlated with apparent uv magnitudes . the c iv @xmath1 and the c iii ] @xmath2 lines of galaxies brighter than @xmath3 are detectable with current facilities such as the vlt / xshooter and the keck / mosfire . metal lines such as c iv @xmath1 , c iii ] @xmath2 , [ o ii ] @xmath4 and [ o iii ] @xmath5 can be good targets for the spectroscopic observation with thirty meter telescope ( tmt ) , european extremely large telescope ( e - elt ) , giant magellan telescope ( gmt ) and james webb space telescope ( jwst ) . h@xmath6 and h@xmath7 lines are also expected to be detectable with these telescopes . we also predict detectability of the nebular lines for @xmath8 galaxies that will be found with jwst , wide - field infrared survey telescope ( wfirst ) and first light and reionization explorer ( flare ) @xmath9 ) . we conclude that the c iv @xmath1 , c iii ] @xmath2 , [ o iii ] @xmath5 and h@xmath7 lines even from @xmath10 galaxies can be strong targets for tmt , elt and jwst . we also find that the magnification by gravitational lensing is of great help to detect such high-@xmath11 galaxies . according to our model , the c iii ] @xmath2 line in @xmath12 galaxy candidates is detectable even using the current facilities . [ firstpage ] cosmology observations ; galaxies evolution ; galaxies formation ; galaxies high - redshift ; galaxies luminosity function , mass function ;

in future experiments , the determination of the leptonic cp phase @xmath4 is one of the most important aim in elementary particle physics . a lot of effort have been dedicated both from theoretical and experimental point of view in order to attain this aim , see @xcite and the references therein . the cp asymmetry , @xmath9 , is widely used as the index of the cp phase dependence . here , @xmath10 and @xmath11 are the oscillation probabilities for the transition @xmath12 and @xmath13 respectively . however , this index has to be improved on the following three points . the first one is that the fake cp violation due to matter effect @xcite can not be separated clearly in @xmath14 . the second one is that only the effect originated from @xmath15 is included in @xmath14 . the third one is that we need to observe the channels both in neutrino and anti - neutrino for calculating @xmath14 . in this letter , we introduce a new index of the cp phase dependence improved on the above three points . in arbitrary matter profile , we derive the maximal condition of this index exactly for @xmath12 transition . this index can be extended to the case for other channels and other parameters @xcite . we can simply find the situation that the cp phase effect becomes large by using this index . as an example , we demonstrate the following interesting phenomena . it is commonly expected that a large @xmath3 appearance signal is observed in the jparc experiment @xcite if the 1 - 3 mixing angle @xmath16 is relatively large @xmath2 and is determined by the next generation reactor experiments like the double chooz experiment @xcite and the kaska experiment @xcite . however , there is the possibility that @xmath3 appearance signal can not be observed in certain mass squared differences and mixing angles even the case for large @xmath16 . we call this `` @xmath16 screening '' . this occurs due to the almost complete cancellation of the large @xmath16 effect by the cp phase effect . if the background can be estimated precisely , we can obtain the information on the cp phase through the @xmath16 screening . this means that we can not neglect the cp phase effect , which is actually neglected in many investigations as the first approximation . at first , we write the hamiltonian in matter @xcite as @xmath17 by factoring out the 2 - 3 mixing angle and the cp phase , where @xmath18 is the rotation matrix in the 2 - 3 generations and @xmath19 is the phase matrix defined by @xmath20 . the reduced hamiltonian @xmath21 is given by @xmath22 where @xmath23 , @xmath24 , @xmath25 is the fermi constant , @xmath26 is the electron number density , @xmath27 is neutrino energy and @xmath28 is the mass of @xmath29 . the oscillation probability for @xmath12 is proportional to the @xmath30 and @xmath15 in arbitrary matter profile @xcite and can be expressed as @xmath31 here @xmath32 , @xmath33 and @xmath34 are determined by parameters other than @xmath4 and are calculated by @xmath35c_{23}s_{23 } \label{eq a } , \\ b&=&2{\rm im}[s_{\mu e}^{\prime * } s_{\tau e}^{\prime}]c_{23}s_{23 } , \\ c&=&|s_{\mu e}^{\prime}|^2c_{23}^2+|s_{\tau e}^{\prime}|^2s_{23}^2 \label{eq c},\end{aligned}\ ] ] where @xmath36_{\alpha\beta}$ ] , @xmath37 and @xmath38 is the cp dependent term and @xmath34 is the cp independent term . next , let us introduce a new index of the cp phase dependence @xmath0 . suppose that @xmath39 and @xmath40 as the maximal and minimal values when @xmath4 changes from @xmath41 to @xmath42 . then , we define @xmath0 as @xmath43 namely , the new index is expressed by the ratio of the coefficient of the cp dependent term to the cp independent term . @xmath0 is a useful tool to explore where is the most effective region in parameter spaces to extract the cp effect from long baseline experiments although @xmath0 is not an observable . @xmath14 is also similar one and this is an observable . however @xmath14 have to be expressed by @xmath4 though @xmath4 is still unknown parameter so that @xmath14 seems not to be so good index to make the exploration . on the other hand , @xmath0 is calculated without using @xmath4 . this is the main difference between these two indices and it is more effective to use @xmath0 . [ cols="^,^ " , ] fig.3 . cp dependence of @xmath3 appearance signal in the jparc - sk experiment . we use the parameters as in fig.1 ( right ) except for @xmath44 and @xmath45 . the condition ( [ screening - condition ] ) is satisfied in the top - left figure , and is not satisfied in other figures . the statistical error is also shown within the 1-@xmath46 level . we also show the value of @xmath0 calculated at @xmath47gev in figures . as expected from the oscillation probability in fig.2 , @xmath3 appearance signal will become almost zero around @xmath48 even during the five years data acquisition in the sk experiment in the top - left of fig . 3 . note that this occurs only when the maximal condition ( [ screening - condition ] ) is satisfied , namely @xmath1 . other panels in fig.3 show that the minimal value of @xmath3 appearance signal rise and is a little different from zero because ( [ screening - condition ] ) is not satisfied so precisely . we obtain the similar results in the case that @xmath49 or @xmath50 changes within the allowed region obtained from solar and the kamland experiments . let us here illustrate how the value of @xmath4 is constrained by the experiment . below , suppose that the atmospheric parameters have some uncertainties as @xmath51- @xmath52 and @xmath53-@xmath54 , while the solar parameters @xmath49 and @xmath50 and @xmath55 are fixed for simplicity . for example , if 15 appearance signals are observed in the experiment , we obtain the allowed region of @xmath4 as @xmath41-@xmath56 or @xmath57-@xmath58 or @xmath59-@xmath42 from four figures in fig.3 . namely , combined range of all allowed region is totally @xmath60 . next , we consider the case that no appearance signal is obtained . this gives the allowed region of @xmath4 as @xmath61-@xmath62 . namely , combined range of all allowed region is totally @xmath56 . thus , we found from above rough estimation that the stronger constraint is obtained in the case of @xmath16 screening even if the uncertainties of parameters except for @xmath4 are considered . in other words , we can also obtain the information on the atmospheric and solar parameters . although the precise estimation of the background is a difficult problem , it is interesting to have strong constraint for not only the value of the cp phase but also other parameters like @xmath44 and @xmath45 when the @xmath3 appearance signal is not observed and the 1 - 3 mixing angle has a comparatively large value @xmath2 . in summary , we introduced a new index of the cp phase dependence @xmath0 and derived their maximal condition in a simple and general form . in particular , we showed that @xmath1 is realized in a rather wide region in the @xmath27-@xmath63 plane at certain values of parameters . in the case that @xmath16 has a comparatively large value @xmath64 , ( namely @xmath16 will be observed in next generation reactor experiments ) nevertheless we can not observe @xmath3 appearance signal in the jparc experiment , we obtain the information on the cp phase as @xmath65 for @xmath66 without depending on the uncertainties of other parameters . also for @xmath67 , there is a possibility that the @xmath16 screening will occur . in this case , we need to consider the reason for the absence of @xmath3 appearance signal in the jparc experiment more carefully , in order to understand whether @xmath16 is small or the @xmath16 effect is canceled out by the cp phase effect . we also note that the @xmath16 screening may be realized for not only @xmath12 oscillation in super - beam experiments but also @xmath68 oscillation in neutrino factory experiments . we can also use the zero probability in the @xmath16 screening to explore new physics like non - standard interaction . we would like to thank prof . wilfried wunderlich ( tokai university ) for helpful comments and advice on english expressions .

we introduce a new index of the leptonic cp phase dependence @xmath0 and derive the maximal condition for this index in a simple and general form . @xmath1 may be realized even in the jparc experiment . in the case that the 1 - 3 mixing angle can be observed in the next generation reactor experiments , namely @xmath2 , and nevertheless @xmath3 appearance signal can not be observed in the jparc experiment , we conclude that the cp phase @xmath4 becomes a value around @xmath5 @xmath6 for @xmath7 @xmath8 without depending the uncertainties of solar and atmospheric parameters .

there have been a number of calculations @xcite using light - front coordinates @xcite as a convenient means of attacking field - theoretic problems nonperturbatively , particularly in 1 + 1 dimensions . efforts in more dimensions have generally been less successful , however , due to the need for regularization and renormalization . here two approaches that include consistent regularization are briefly reviewed in the context of specific model calculations @xcite . one approach is pauli villars ( pv ) regularization @xcite , where massive negative - metric particles are added to a theory to provide the necessary cancellations @xcite . the other is supersymmetry @xcite . these are not the only approaches available on the light front ; in particular , one can find the transverse lattice technique @xcite and similarity transformations @xcite discussed elsewhere in this volume . the primary numerical method is discrete light - cone quantization ( dlcq ) @xcite , in which one imposes a discrete momentum grid , with length scales @xmath0 and @xmath1 , as @xmath2 and @xmath3 , and approximates integrals in the mass - squared eigenvalue problem with trapezoidal sums . the continuum limit @xmath4 can be exchanged for a limit in terms of the integer resolution @xmath5 , because light - cone momentum fractions @xmath6 are measured in units of @xmath7 . for supersymmetric theories there is a supersymmetric version of dlcq ( sdlcq ) that preserves supersymmetry within the discrete approximation @xcite . this is accomplished by discretizing the supercharge @xmath8 and constructing the hamiltonian @xmath9 from the superalgebra via the anticommutator : @xmath10 . this @xmath9 and the dlcq @xmath9 are equivalent in the @xmath11 limit . the matrix eigenvalue problems that result from these discretizations are large but sparse . an efficient means for extracting a few lowest eigenvalues and their eigenvectors is the lanczos algorithm @xcite . in the case of pv - regulated theories , with their indefinite metrics , a special form @xcite based on the biorthogonal algorithm @xcite is required . in either case , the process is an iterative one that generates a tridiagonal matrix of much smaller size , which is easily diagonalized to yield approximate eigenvalues and eigenvectors . the number of iterations and the size of the tridiagonal matrix are determined by the rate of convergence . the primary difficulty that arises is that round - off error causes spurious copies to appear in the derived spectrum ; however , there are techniques for removing them @xcite . in the remaining sections we discuss an application of pv regularization to yukawa theory and a study of supersymmetric qcd ( sqcd ) with a chern simons ( cs ) term in the large-@xmath12 approximation . as the action of the pv - regulated yukawa theory , we take + & & . -m_0_0_0 -(_1^_-(__1)^ ) _ 1+m_1_1_1 -g(_0 + _ 1)(_0 + _ 1)(_0 + _ 1 ) ] , with the subscript 0 indicating physical fields and 1 indicating pv fields . when fermion pairs are excluded , this action provides a light - cone hamiltonian of the form [ eq : yukawap- ] + & & + _ i , j , k , sd d\ { b_j , s^ ( ) a_k^ ( ) b_i ,- s(+ ) . + & & . ' '' '' + b_j , s^ ( ) a_k^()b_i , s(+ ) + h.c . } , where @xmath13 creates a boson and @xmath14 a fermion , u_j ( , ) , v_2s ( , ) , _ 2s-(2s , i ) , and = ( -1)^i_ij ( - ) , \{b_i , s(),b_j , s^( ) } = ( -1)^i_ij _ s , s ( - ) . the eigenfunction for the dressed - fermion state is expanded in a fock basis as & & + & & + _ f_ijks(_1,_2 ) b_is^(-_1-_2 ) a_j^(_1)a_k^(_2)|0 + it is normalized according to @xmath15 . the wave functions satisfy a coupled system of equations , derived from the fundamental mass - squared eigenvalue problem @xmath16 to be m_i^2z_i&+ & _ i,j(-1)^i+j p^+ ^p^+ d \ { f_ij-()[v_+(-,)+v_-^ * ( , ) ] . + f_ij+()[u_i(-,)+u_i ( , ) ] } = m^2z_i,[eq : oneboson ] + & & ' '' '' . + z_i_s,+[u_i(-,)+u_i ( , ) ] } + & & + 2_i,kp^+ ^p^+-q^+ d \{f_ijk ,- s(, ) [ v_2s(--,)+v_-2s^*(-, ) ] . + & & .+f_ijks(, ) [ u_i(--,)+u_i(-, ) ] } = m^2f_ijs ( ) , + & & + _ i(-1)^ip^+ \{f_ij ,- s ( ) [ v_-2s^*(--,)+v_2s(- , ) ] . + & & + f_ijs ( ) [ u_i(--,)+u_i(- , ) ] + & & + f_ik ,- s ( ) [ v_-2s^*(--,)+v_2s(- , ) ] + & & . + f_iks ( ) [ u_i(--,)+u_i(- , ) ] } + = m^2f_ijks ( , ) . these equations can be approximated directly by dlcq . we now have a well - defined numerical problem . the pv particles are kept in the dlcq basis and provide the necessary counterterms . the range of the now - finite transverse integrations is cut off by imposing @xmath17 for each particle in a fock state , to reduce the matrix problem to a finite size . the transverse momentum indices @xmath18 and @xmath19 are limited by the transverse resolution @xmath20 . the bare parameters @xmath21 and @xmath22 are fixed by fitting `` physical '' constraints , such as specifying the dressed - fermion mass @xmath23 and its radius . the limits of infinite resolution , infinite ( momentum ) volume , and infinite pv masses can then be explored . this process can be studied analytically in the case of a one - boson truncation @xcite . the one - boson wave functions are f_ij+()&= & , + f_ij-()&= & ( _ k ( -1)^k+1z_k)v_+^*(- , ) . the bare - fermion amplitudes and the coupling satisfy a pair of algebraic equations [ eq : onefermion ] ( m^2-m_i^2)z_i & = & g^2_0 ^ 2 ( z_0-z_1)j+g^2 m_i(z_0m_0-z_1m_1 ) i_0 + & & + g^2_0[(z_0-z_1)m_i+z_0m_0-z_1m_1 ] i_1 , with i_n&= & _ jk , + j&= & _ jk = i_0 . the solution is [ eq : gofm ] g^2=- , = . @xcite contains many subsequent results . for a two - boson truncation , the solution is no longer analytic , but the coupled equations can be reduced to eight equations for the two - particle amplitudes only , which are of the form + & & + _ abs_0 ^ 1dydq_^2 j_ijs , abs^(0)(y , q_;y,q_)f_abs(y,q _ ) + & & + _ abs_0 ^ 1-ydydq_^2 j_ijs , abs^(2)(y , q_;y,q_)f_abs(y,q _ ) , with the angular dependence removed via @xmath24 and @xmath25 . here @xmath26 is a computable self - energy , @xmath27 is the kernel due to bare - fermion intermediate states , and @xmath28 is the kernel due to two - boson intermediate states . although one could impose a transverse cutoff and discretize these equations as per dlcq , alternative quadratures are more efficient . in particular , the transverse momentum @xmath29 can be mapped to a finite range that compresses the wave function s tail to a relatively small region , so that a gauss legendre quadrature can yield a good approximation . a comparison of results is given in fig . [ fig : g2 ] . , and the pv masses are @xmath30 . the longitudinal resolution is @xmath31 ; the transverse resolutions @xmath20 are specified in the legend.,width=491 ] as the action for ( 2 + 1)-dimensional @xmath32 sqcd - cs theory , we consider [ action ] s&=&d^3x\{-f_f^ + d_^d^+i| d_^. + & & .-g+^d_+^ + |}. the adjoint fields are the gauge boson @xmath33 ( gluons ) and a majorana fermion @xmath34 ( gluinos ) ; the fundamental fields are the dirac fermion @xmath35 ( quarks ) and a complex scalar @xmath36 ( squarks ) . the cs coupling @xmath37 induces a mass for the adjoint fields without breaking the supersymmetry ; this reduces formation of the long strings characteristic of super yang mills theory . the covariant derivatives are d_=_+ig[a _ , ] , d_=_+iga _ , d_=_+iga_. the supersymmetry transformations are a_= _ , = f_^ , = , = -^d_. we reduce this theory to 1 + 1 dimensions by taking the fields to be independent of the transverse coordinate @xmath38 . as usual , there are constraints and not all of the fields are dynamical . to separate the dynamical fields , we first introduce components for the fermi fields and the supercharge as = ( , ) ^t,=(,)^t , q=(q^+,q^-)^t . then , in light - cone gauge ( @xmath39 ) , the constraints are _ -&=&- ( [ a^2,]+i^-i^ ) , _ -=-a^2 + -/ , [ fcurrent ] + ^2_-a^-&=&g\{i[a^2,_-a^2]+ \ { , } + _ - a^2 -ih_-^+ i_-^+^}. when these constraints are used to eliminate the nondynamical fields , the supercharge becomes q^-&=&gdx^-\ { 2 ^ 3/4(i[a^2,_-a^2]-_-a^2 + \ { , } ) . + & & -(i _ -^-i_-^+2^ ) .-2(^a^2+^a^2)}. the mode expansions of the dynamical fields are [ expanda2 ] a^2_ij(0,x^-)&=&_k=1^ ( a_ij(k)e^-ikx^-/l+a^_ji(k)e^ikx^-/l ) , + [ expandlambda ] _ ij(0,x^-)&=&_k=1^ ( b_ij(k)e^-ikx^-/l+b^_ji(k)e^ikx^-/l ) , + [ expandxi ] _ i(0,x^-)&=&_k=1^ ( c_i(k)e^-ikx^-/l+c^_i(k)e^ikx^-/l ) , + [ expandpsi ] _ i(0,x^-)&=&_k=1^ ( d_i(k)e^-ikx^-/l+d^_i(k)e^ikx^-/l ) . the creation and annihilation operators obey the commutation relations ( for finite @xmath12 ) = ( _ il_kj-_ij_kl ) , \{b_ij , b^_kl}= ( _ il_kj-_ij_kl ) , = _ ij,=_ij,\{d_i , d^_j}=_ij \{d_i , d^_j}=_ij . the solutions that we obtain are meson - like states @xmath40 , where @xmath41 or @xmath42 , and glueball states @xmath43|0\rangle$ ] . because of the supersymmetry , either could be a boson or a fermion . because we work in the large-@xmath12 limit , there is no mixing between these states , and they are composed of single traces . this simplifies the calculation , particularly with respect to the size of the matrices that are diagonalized ; however , study of baryons will require finite @xmath12 or additional approximations . to help reduce the size of the calculation further , there is an additional @xmath44 symmetry @xcite @xmath45 , @xmath46 . this divides the states between those with even and odd numbers of gluons , and we diagonalize in each sector separately . a collection of results can be found in ref . the spectrum for mesons shows the existence of light states at strong coupling , a feature found previously in other supersymmetric theories @xcite . methods for the nonperturbative solution of multidimensional field theories are now available ; they do , however , require more development . a number of calculations will soon be undertaken to further demonstrate and improve this capability . in yukawa theory , the two - fermion sector is of interest , particularly for a pseudo - scalar coupling with which the deuteron might be modelled . quantum electrodynamics is immediately treatable with the same techniques ; the anomalous magnetic moment of the electron can be computed . the full ( 2 + 1)-dimensional sqcd - cs theory can be solved , including finite-@xmath12 corrections to allow baryons and meson - glueball mixing . this work should bring us closer to the goal of being able to solve for hadron properties in quantum chromodynamics . the work summarized here was done in collaboration with s.j . brodsky and g. mccartor , and with s.s . pinsky and u. trittmann , and was supported in part by the us department of energy and the minnesota supercomputing institute . brodsky , h .- c . pauli , and s.s . pinsky , phys . * 301 * ( 1997 ) 299 ; j.r . hiller , nucl . b ( proc . suppl . ) * 90 * ( 2000 ) 170 . dirac , rev . * 21 * ( 1949 ) 392 . brodsky , j.r . hiller , and g. mccartor , ann . * 305 * ( 2003 ) 266 [ hep - th/0209028 ] . hiller , s.s . pinsky , and u. trittmann , phys . d * 67 * ( 2003 ) 115005 [ hep - ph/0304147 ] ; nucl . b * 661 * ( 2003 ) 99 [ hep - ph/0302119 ] . for earlier work on fundamental matter in sdlcq , see o. lunin and s. pinsky , phys . d * 63 * ( 2001 ) 045019 . w. pauli and f. villars , rev . mod . * 21 * ( 1949 ) 4334 . brodsky , j.r . hiller , and g. mccartor , phys . d * 58 * ( 1998 ) 025005 [ hep - th/9802120 ] ; phys . d * 60 * ( 1999 ) 054506 [ hep - ph/9903388 ] ; ann . * 296 * ( 2002 ) 406 [ hep - th/0107246 ] . brodsky , j.r . hiller , and g. mccartor , phys . d * 64 * ( 2001 ) 114023 [ hep - ph/0107038 ] . o. lunin and s. pinsky , in _ new directions in quantum chromodynamics _ , edited by c .- r . ji and d .- min , aip conf . 494 ( aip , melville , ny , 1999 ) , p. 140 , [ hep - th/9910222 ] . m. burkardt and s. dalley , prog . * 48 * ( 2002 ) 317 [ hep - ph/0112007 ] ; s. dalley and b. van de sande , phys . d * 67 * ( 2003 ) 114507 [ hep - ph/0212086 ] . glazek and j. mlynik , phys . d * 67 * ( 2003 ) 045001 [ hep - th/0210110 ] ; st.d . glazek and m. wieckowski , phys . d * 66 * ( 2002 ) 016001 [ hep - th/0204171 ] . pauli and s.j . brodsky , phys . d * 32 * ( 1985 ) 1993 ; * 32 * ( 1985 ) 2001 . for earlier , more formal work , see t. maskawa and k. yamawaki , prog . * 56 * ( 1976 ) 270 . y. matsumura , n. sakai , and t. sakai , phys . d * 52 * ( 1995 ) 2446 [ hep - th/9504150 ] . c. lanczos , j. res . bur . stand . * 45 * ( 1950 ) 255 ; j. cullum and r.a . willoughby , _ lanczos algorithms for large symmetric eigenvalue computations _ ( birkhauser , boston , 1985 ) , vol . i and ii . j.h . wilkinson , _ the algebraic eigenvalue problem _ ( clarendon , oxford , 1965 ) ; y. saad , comput . * 53 * ( 1989 ) 71 ; s.k . kin and a.t . chronopoulos , j. comp . and appl . math * 42 * ( 1992 ) 357 . j. cullum and r.a . willoughby , in _ large - scale eigenvalue problems _ , edited by j. cullum and r.a . willoughby , math . * 127 * ( elsevier , amsterdam , 1986 ) , p. 193 . d. kutasov , nucl . b * 414 * ( 1994 ) 33 . hiller , s.s . pinsky , and u. trittmann , phys . * 89 * ( 2002 ) 181602 [ hep - th/0203162 ] ; j.r . hiller , s.s . pinsky , and u. trittmann , phys . b * 541 * ( 2002 ) 396 [ hep - th/0206197 ] .

two methods for the nonperturbative solution of field - theoretic bound - state problems , based on light - front coordinates , are briefly reviewed . one uses pauli villars regularization and the other supersymmetry . applications to yukawa theory and super yang mills theory with fundamental matter are emphasized .

the hungarian - made automated telescope network ( hatnet ; bakos et al . 2004 ) survey , has been one of the main contributors to the discovery of transiting exoplanets ( teps ) , being responsible for approximately a quarter of the @xmath0 confirmed teps discovered to date ( fig . 1 ) . it is a wide - field transit survey , similar to other projects such as super - wasp ( pollaco et al . 2006 ) , xo ( mccullough et al . 2005 ) , and tres ( alonso et al . the teps discovered by these surveys orbit relatively _ bright _ stars ( @xmath1 ) which allows for precise parameter determination ( e.g. mass , radius and eccentricity ) and enables follow - up studies to characterize the planets in detail ( e.g. studies of planetary atmospheres , or measurements of the sky - projected angle between the orbital axis of the planet and the spin axis of its host star ) . since 2006 , hatnet has announced twenty - six teps . below we highlight some of the exceptional properties of these planets ( section 2 ) , we then describe the procedures which we followed to discover them ( section 3 ) , and we conclude by summarizing what hatnet provides to the tep community with each discovery ( section 4 ) . hatnet - detected teps span a wide range of physical properties , including : two neptune - mass planets ( hat - p-11b , bakos et al . 2010a ; and -26b , hartman et al . 2010b ) ; planets with masses greater than @xmath2 ( -2b , bakos et al . 2007 ; and -20b , bakos et al . 2010b ) ; compact planets ( -2b , and -20b ) ; inflated planets ( -7b , pl et al . 2008 ; -8b , latham et al . 2009 ; -12b , hartman et al . 2009 ; -18b , and -19b , hartman et al . 2010a ) ; a planet with a period of just over one day ( -23b , bakos et al . 2010b ) ; planets with periods greater than 10 days ( -15b , kovcs et al . 2010 ; and -17b , howard et al . 2010 ) ; multi - planet systems ( -13b , c , bakos et al . 2009 ; and -17b , c ) ; and a number of eccentric planets ( -2b ; -11b ; -14b , torres et al . 2010 ; -15b ; -17b ; and -21b , bakos et al . we have also provided evidence for outer planets for 4 systems : hat - p-11c , -13c , -17c ( the latter two with almost closed orbits ) , and hat - p-19c . some of these discoveries were the first of their kind , and thus were important landmarks in exoplanet science . this includes : the first transiting heavy - mass planet ( -2b ) ; the first retrograde planet ( -7b ; narita et al . 2009 , winn et al . 2009 ) ; two of the first four transiting neptunes ; the first inflated saturn ( -12b ) ; the first and second multi - planet systems with transiting inner planets ; and two of the first six planets with periods longer than 10 days . the 26 hatnet teps were identified from a shortlist of 1300 hand - selected transit _ candidates _ culled from millions of light curves , which were , in turn , the result of diverse activities ranging from remote hardware operations to data analysis . here we briefly describe this process . hatnet utilizes 6 identical instruments , each with an 11 cm aperture f/1.8 lens and a @xmath3 front - illuminated ccd with 9pixels ( yielding a wide , @xmath4 field ) , attached to a horseshoe mount , protected by a clam - shell dome , and with all devices controlled by a single pc . each instrument , called a hat ( bakos et al . 2002 ) , can obtain per - image photometric precision reaching 4mmag at 3.5-min cadence on the bright end at @xmath5 , and 10mmag at @xmath6 . by collecting a light curve with @xmath0 or more points in transit , a transit with a depth of only a few mmag may be detected . we note that the original hatnet hardware employed @xmath7 front illuminated detectors with cousins @xmath8-band filters . this was replaced to @xmath3 front - illuminated ccds and cousins @xmath9 filters in 2007 september , and the filter was changed to sloan @xmath10 in 2008 july . four hat instruments are located at the smithsonian astrophysical observatory s ( sao ) fred lawrence whipple observatory ( flwo ) , and an additional two instruments are on the roof of the hangar servicing the antennae of sao s submillimeter array , at mauna kea observatory ( mko ) in hawaii . the network with its current longitude coverage has significant advantages in detecting teps with periods longer than a few days . the instruments are truly autonomous in the sense that they are aware of the observing schedule and the weather conditions , they prepare all the devices ( ccds , dome , telescope ) for the observations , acquire ample calibration frames ( biases , darks , skyflats ) , and then proceed to the science program of the night . for the purpose of monitoring bright stars for transits , the sky has been split up to 838 @xmath11 non - overlapping fields . fields are chosen for observation based on several factors such as optimal visibility at the given time of the year , proximity of the field to solar system objects , and various other factors . to date hatnet has observed @xmath12 fields ( 29% of the northern sky ) . typically a field is monitored for 3 months ; a given instrument will begin observations of the field after evening twilight and observe it continuously at a cadence of 3.5 minutes until the field sets . the instrument will then target a second field and continue observing it until morning twilight . all time between dusk and dawn is spent by exposing on the selected fields . a single field is typically assigned to a flwo instrument as well as a mko instrument to increase the duty cycle of the observations . based on operations since 2003 , we find that the effective duty cycle of hatnet is @xmath13 . the images are calibrated using standard techniques that take into account the problems raised by the wide fov , such as strong vignetting , distortions , sky - background changes , etc . the entire data flows to the cfa via fast internet . the astrometric solution is determined following pl & bakos ( 2006 ) , based on the two micron all sky survey ( 2mass ; skrutskie et al . 2006 ) catalog . we then perform aperture photometry at the fixed positions of the 2mass stars . an initial ensemble magnitude calibration is performed , and the remaining systematic variations are removed from the light curves by decorrelating against external parameters ( i.e. parameters describing the point spread function , the position of the star on the image , and others ) and by making use of the trend filtering algorithm ( tfa ; kovcs et al . 2005 ) . the resulting light curves typically reach a per - point photometric precision ( at 3.5min cadence ) of @xmath14mmag for the brightest non - saturated stars . we search the trend filtered light curves for periodic transit events using the box - fitting least squares algorithm ( bls ; kovcs et al . we then subject potential transit candidates to a number of automatic filters to select reliable detections which are consistent with a transiting planet - size object , and are not obviously eclipsing binary star systems or other types of variables . to go from the lengthy list of potential transit candidates to a much smaller number of confirmed and well - characterized teps , we first manually select a list of candidates to consider for follow - up observations , we gather and analyze spectroscopic and photometric observations to reject false positives and confirm bona fide teps , and we finally analyze and publish the confirmed tep systems . the automated transit candidate selection procedure provides a manageable list of potential candidates which must then be inspected by eye to select the most promising targets for follow - up . typically a few hundred to one thousand potential candidates per field are identified by the automated procedures , which are then narrowed down to a few dozen candidates deemed worthy of follow - up . at the end of this procedure , relative priorities are assigned to these candidates and appropriate facilities for follow - up are identified . candidates selected for follow - up then undergo a procedure consisting of three steps : reconnaissance spectroscopy , photometric follow - up observations , and high - precision spectroscopy . we have found that a very efficient method to reject the majority of false positive transit detections is to first obtain one or more high - resolution , low signal - to - noise ratio ( s / n@xmath15 per resolution element ) spectra using a @xmath16@xmath17 m telescope ( e.g. latham et al . double - lined eclipsing binaries , giant stars ( where the detected transit is most likely a blend between an eclipsing binary and the much brighter giant ) , and rapidly rotating stars or excessively hot stars ( where confirming a planetary orbit would be very difficult ) may be immediately rejected based on one spectrum . stars that are confirmed to be slowly rotating dwarfs are observed a second time at the opposite quadrature phase to look for a significant velocity variation . stars where the rv amplitude implies a stellar companion ( typically @xmath18kms@xmath19 ) are rejected . in some cases changes in the shape of the spectral line profile may also be detected , enabling us to reject the target as a stellar triple . we find that a significant fraction of the initial candidates selected for follow - up are rejected by this reconnaissance spectroscopy ( rs ) procedure , saving time on precious resources that we use for the final follow - up . to carry out the rs work we primarily use the flwo 1.5 m telescope , previously with the digital speedometer ( ds ) , and now with the tillinghast reflector echelle spectrograph ( tres ) , and to some extent the fiber - fed echelle spectrograph ( fies ) on the nordic optical telescope ( not ) at la palma . we have also made use of the echelle spectrograph on the du pont 2.5 m telescope at lco , the echelle spectrograph on the anu 2.3 m telescope at sso , and the coralie echelle spectrograph on the 1.2 m euler telescope at la silla observatory ( lso ) in chile . if a candidate passes the rs step , we then schedule photometric observations of the candidate over the course of a transit to confirm that the transit is real , and to confirm that the shape of the transit light curve is consistent with a tep . we also note that these follow - up light curves are essential for obtaining precise measurements of system parameters , such as the planet - star radius ratio or the transit impact parameter . in some cases the candidate is subjected to photometry follow - up without rs or with incomplete rs , when , e.g. , there is evidence that the star is a dwarf ( colors , proper motion , parallax ) , or when the transit events are rare ( long period ) . we primarily use the keplercam instrument on the flwo 1.2 m telescope , but have also made use of faulkes telescope north ( ftn ) of the las cumbres observatory global telescope ( lcogt ) on mauna haleakala , hawaii , and occasionally telescopes in hungary and israel . the final step in the confirmation follow - up procedure is to obtain high - resolution , high - s / n spectra with sufficient velocity precision to detect the orbital variation of the star due to the planet , confirm that the system is not a subtle blend configuration , and measure the effective temperature , surface gravity and metallicity of the star . by this stage we have already excluded the majority of false positives , so that roughly half of the candidates that reach this step are confirmed as teps . false positives that reach this stage generally are rejected after only a few spectra are obtained , so that @xmath20 of the time is spent observing teps . we primarily use the high resolution echelle spectrometer ( hires ) with the iodine cell on the 10 m keck i telescope at mko . we have also used the fies / not facility , the high dispersion spectrograph ( hds ) with the iodine cell on the subaru 8.2 m telescope at mko , and the sophie instrument on the 1.93 m telescope at the observatoire de haute - provence ( ohp ) , in france . once a planet is confirmed , we conduct a joint analysis of the available high - precision rv observations and photometric observations to determine the system parameters , including in particular the masses and radii of the star and planet(s ) ( e.g. bakos et al . 2010a ) . in cases where the spectral line bisector spans are inconclusive , we must also carry out a detailed blend - model of the system concurrently with the tep modeling to definitively prove that the object is a tep . the hatnet project strives to provide the following to the community for a given tep discovery : 1 . hatnet discovery data . 2 . high - precision , often multi - band , photometry follow - up . 3 . high - precision radial velocity follow - up . access to all of the data via the online tables , including raw and detrended values . characterization of the host star : stellar atmospheric and fundamental parameters ( from isochrone fitting ) . 6 . a blend analysis . 7 . accurate parameters of the planetary system . 8 . a publication available on arxiv with all the above when the planet is announced

we summarize the contribution of the hatnet project to extrasolar planet science , highlighting published planets ( hat - p-1b through hat - p-26b ) . we also briefly discuss the operations , data analysis , candidate selection and confirmation procedures , and we summarize what hatnet provides to the exoplanet community with each discovery .

a _ latin bitrade _ @xmath5 is a pair of partial latin squares which are disjoint , occupy the same set of non - empty cells , and whose corresponding rows and columns contain the same set of entries . one of the earliest studies of latin bitrades appeared in @xcite , where they are referred to as _ exchangeable partial groupoids_. latin bitrades are prominent in the study of _ critical sets _ , which are minimal defining sets of latin squares ( @xcite,@xcite,@xcite ) and the intersections between latin squares ( @xcite ) . we write @xmath6 when symbol @xmath7 appears in the cell at the intersection of row @xmath8 and column @xmath9 of the ( partial ) latin square @xmath10 . a @xmath3-homogeneous bitrade has @xmath3 elements in each row , @xmath3 elements in each column , and each symbol appears @xmath3 times . cavenagh @xcite obtained the following theorem , using combinatorial methods , as a corollary to a general classification result on @xmath3-homogeneous bitrades . [ theorem3homtransversals ] let @xmath5 be a @xmath3-homogeneous bitrade . then @xmath10 can be partitioned into three transversals . in this paper we provide an independent and geometric proof of cavenagh s result . in doing so we provide a framework for studying bitrades as tessellations in spherical , euclidean or hyperbolic space . in particular , bitrades can be thought of as finite representations of certain triangle groups . we let permutations act on the right , in accordance with computer algebra systems such as sage @xcite . graphs in this paper may contain loops or multiple edges ; otherwise our notation is standard and we refer the reader to diestel @xcite . some basic topological terms will be used ; for these we refer the reader to stillwell @xcite . finally , a good reference for hypermaps and graphs on surfaces is @xcite . a _ partial latin square _ @xmath11 of order @xmath12 is an @xmath13 array where each @xmath14 appears at most once in each row , and at most once in each column . a _ latin square _ @xmath15 of order @xmath12 is an @xmath13 array where each @xmath14 appears exactly once in each row , and exactly once in each column . it is convenient to use setwise notation to refer to entries of a ( partial ) latin square , and we write @xmath16 if and only if symbol @xmath7 appears in the intersection of row @xmath8 and column @xmath9 of @xmath11 . in this manner , @xmath17 for finite sets @xmath18 , each of size @xmath19 . it is also convenient to interpret a ( partial ) latin square as a multiplication table for a binary operator @xmath20 , writing @xmath6 if and only if @xmath21 . [ defnbitradea123 ] let @xmath10 , @xmath22 be two partial latin squares . then @xmath5 is a _ bitrade _ if the following three conditions are satisfied : * @xmath23 ; * for all @xmath24 and all @xmath25 , @xmath26 , @xmath27 , there exists a unique @xmath28 such that @xmath29 and @xmath30 ; * for all @xmath31 and all @xmath25 , @xmath26 , @xmath27 , there exists a unique @xmath32 such that @xmath29 and @xmath30 . conditions ( r2)and ( r3)imply that each row ( column ) of @xmath10 contains the same subset of @xmath33 as the corresponding row ( column ) of @xmath34 . a _ @xmath7-homogeneous bitrade _ @xmath5 has @xmath7 entries in each row of @xmath10 , @xmath7 entries in each column of @xmath10 , and each symbol appears @xmath7 times in @xmath10 . by symmetry the same holds for @xmath34 . a set @xmath35 is a _ transversal _ if @xmath36 intersects each row of @xmath10 in precisely one entry , each column in precisely one entry , and if the number of symbols appearing in @xmath36 is equal to @xmath37 . the latter condition can be written as @xmath38 . a bitrade @xmath5 is _ primary _ if whenever @xmath39 is a bitrade such that @xmath40 and @xmath41 , then @xmath42 . bijections @xmath43 , for @xmath44 , @xmath45 , @xmath3 , give an _ isotopic _ bitrade , and permuting each @xmath18 gives an _ autotopism_. in @xcite , drapal gave a representation of bitrades in terms of three permutations @xmath46 acting on a finite set . for @xmath47 , define the map @xmath48 where @xmath49 if and only if @xmath50 and @xmath51 for @xmath52 . by definition [ defnbitradea123 ] each @xmath53 is a bijection . then @xmath54 are defined by @xmath55 we refer to @xmath56 $ ] as the @xmath46 _ representation_. we write @xmath57 for the set of points that the ( finite ) permutation @xmath58 acts on . [ defnt1234 ] let @xmath59 , @xmath60 , @xmath61 be ( finite ) permutations and let @xmath62 . define four properties : 1 . @xmath63 ; 2 . if @xmath64 is a cycle of @xmath46 and @xmath65 is a cycle of @xmath66 then @xmath67 , for any @xmath68 ; 3 . each @xmath46 is fixed - point - free ; 4 . the group @xmath69 is transitive on @xmath70 . by letting @xmath18 be the set of cycles of @xmath46 , drapal obtained the following theorem , which relates definition [ defnbitradea123 ] and [ defnt1234 ] . [ theoremdrapaltaustructure ] a bitrade @xmath5 is equivalent ( up to isotopism ) to three permutations @xmath59 , @xmath60 , @xmath61 acting on a set @xmath70 satisfying ( t1 ) , ( t2 ) , and ( t3 ) . if ( t4)is also satisfied then the bitrade is primary . to construct the @xmath46 representation for a bitrade we simply evaluate equation . in the reverse direction we have the following construction : [ constructiontautobitrade ] let @xmath59 , @xmath60 , @xmath61 be permutations satisfying condition ( t1 ) , ( t2 ) , and ( t3 ) let @xmath62 . define @xmath71 for @xmath44 , @xmath45 , @xmath3 . now define two arrays @xmath10 , @xmath34 : @xmath72 by theorem [ theoremdrapaltaustructure ] @xmath5 is a bitrade . [ exampleintercalaterep ] the smallest bitrade @xmath5 is the _ intercalate _ , which has four entries . the bitrade is shown below : @xmath73 the @xmath46 representation is @xmath74 , @xmath75 , @xmath76 , where we have written @xmath77 for @xmath78 to make the presentation of the @xmath46 permutations clearer . by construction [ constructiontautobitrade ] with @xmath79 we can convert the @xmath46 representation to a bitrade @xmath80 : @xmath81 in this way we see that row @xmath82 of @xmath10 corresponds to row @xmath83 of @xmath84 , which is the cycle @xmath85 of @xmath59 , and so on for the columns and symbols . [ ex3hom ] the following @xmath3-homogeneous bitrade is pertinent to the proof of the main result of this paper : @xmath86 writing @xmath87 for @xmath88 , the @xmath46 representation is @xmath89 the bitrade has four rows so @xmath59 has four cycles ; similarly @xmath60 and @xmath61 each have four cycles . ( in general , a bitrade can have a different number of row , column , and symbol cycles . ) using construction [ constructiontautobitrade ] , the cell at row @xmath90 , column @xmath91 , will contain the symbol @xmath92 since these cycles intersect in @xmath93 . before showing how a bitrade can be represented as a graph embedded in a surface , we briefly review the theory of hypermaps . a _ combinatorial hypermap _ @xmath94 $ ] is made up of three permutations @xmath95 , @xmath96 , @xmath97 and a finite set @xmath70 such that @xmath98 and @xmath99 acts transitively on @xmath70 . the following construction takes a combinatorial hypermap to a _ hypermap _ , which is a bipartite graph embedded in a surface . for a proof of correctness see chapter 1 of @xcite and references therein , and for further examples see chapter 1 and 2 of @xcite . the representation of hypermaps as bipartite graphs was given by walsh @xcite . [ constrpairtobipartitegraph ] let @xmath94 $ ] be a combinatorial hypermap on the finite set @xmath70 . create vertex sets @xmath100 , @xmath101 and undirected edges @xmath102 : @xmath103 colour the vertices of @xmath100 black ( denoted @xmath104 ) and those of @xmath101 white ( denoted @xmath105 ) . when drawing the graph we usually label an edge @xmath106 with @xmath107 to save space . suppose that @xmath108 is a cycle of @xmath95 , and let @xmath109 be the associated black vertex with adjacent edges @xmath110 for @xmath111 . then order the edges adjacent to @xmath109 as @xmath112 , @xmath113 , , @xmath114 in the anticlockwise direction . apply the same process to each @xmath115 . this defines a _ rotation scheme _ for the vertices of the bipartite graph , and hence an embedding in a surface . [ exbipartitegraph ] let @xmath116 and define @xmath117 and @xmath118 . then there are two black vertices , two white vertices , and four edges : @xmath119 @xmath120 and @xmath121 . the graph embedding , with anticlockwise orientation , is shown in figure [ figbipartiteembedding ] . given a bipartite graph embedding , we often move to the _ canonical triangulation _ , as described in the following construction : [ constrfundamentaltriangulation ] let @xmath122 be a hypermap . place a new vertex @xmath123 in each face of the hypermap . connect this new vertex to each vertex that lies on the border of the face using dotted edges to @xmath104 vertices and dashed edges to @xmath105 vertices . the surface is now subdivided into triangles . each triangle has three types of vertices : @xmath104 , @xmath105 , and @xmath123 ; each triangle has three types of sides : a solid , dashed , or dotted line . from the inside of a triangle , we view its vertices according to the order @xmath104 , @xmath105 , @xmath123 , @xmath104 , and if we turn in the anticlockwise direction then the triangle is _ positive _ , otherwise it is _ negative_. we shade the positive triangles . . vertices are labelled from the set @xmath124 to aid in identification with the intercalate bitrade , and the action of @xmath95 is shown on a particular shaded triangle . ] since each ( shaded ) triangle is adjacent to precisely one solid edge in the canonical triangulation , we can identify the action of @xmath95 as the rotation of shaded triangles around their black vertex in an anticlockwise direction , as shown in figure [ figcanonical ] ( also , see @xcite ) . in general , the action of @xmath96 and @xmath97 correspond to rotations around white and star vertices , as indicated in figure [ figfundtriangulation ] . , @xmath96 , and @xmath97 on certain shaded triangles is shown.,title="fig : " ] the canonical triangulation of the bipartite graph embedding of example [ exbipartitegraph ] is shown in figure [ figcanonical ] . writing @xmath77 for the shaded triangle with vertex labels @xmath8 , @xmath9 , @xmath7 on black , white and star vertices , respectively , we see that @xmath125 , @xmath126 , and @xmath127 . as expected , the action of @xmath95 , @xmath96 , and @xmath97 in figure [ figcanonical ] is exactly the same as @xmath59 , @xmath60 , and @xmath61 of example [ exampleintercalaterep ] . applying euler s formula leads to the _ genus formula _ for hypermaps : @xmath128 where @xmath129 denotes the number of cycle of the permutation @xmath58 . [ lemmatorus ] a @xmath3-homogeneous bitrade @xmath5 defines a tessellation of shaded and unshaded triangles in the euclidean plane . each shaded triangle is edge - wise adjacent only to unshaded triangles ( and vice - versa ) . shaded and unshaded triangles correspond to the entries of @xmath10 and @xmath34 , respectively . black , white , and star vertices correspond to row , column , and symbol labels of @xmath10 . let @xmath5 be a @xmath3-homogeneous bitrade and let @xmath56 $ ] be the @xmath46 representation . by condition ( t1)and ( t4)we see that @xmath59 , @xmath60 , and @xmath61 satisfy the properties to be a combinatorial hypermap . let @xmath94 = [ \tau_1,\ , \tau_2,\ , \tau_3]$ ] and construct the associated hypermap using construction [ constrpairtobipartitegraph ] . apply construction [ constrfundamentaltriangulation ] so that the hypermap consists of shaded and unshaded triangles . since @xmath130 and @xmath131 it follows that @xmath132 so the underlying surface is the torus . the fundamental group of the torus is @xmath133 so the covering surface is the euclidean plane . by construction [ constrfundamentaltriangulation ] , each shaded triangle is adjacent , edge - wise , to precisely one unshaded triangle , and vice - versa . the permutation @xmath95 acts on shaded triangles while @xmath59 acts on elements of @xmath10 by equation . we set @xmath134 so shaded triangles correspond to elements of @xmath10 and unshaded triangles correspond to elements of @xmath34 . black vertices correspond to cycles of @xmath59 which , in turn , correspond to row labels of @xmath10 ( and similar for white and star vertices ) . figure [ fig3homtessellation ] shows the tessellation for the @xmath3-homogeneous bitrade of example [ ex3hom ] . identifying opposite sides of the parallelogram marked by thick grey lines gives the torus . with regards to theorem [ theorem3homtransversals ] , we can partition @xmath10 into three transversals @xmath135 where @xmath136 . these transversals may be located geometrically in figure [ fig3homtessellation ] : @xmath137 is made up of shaded triangles located directly above a @xmath104 vertex , @xmath138 is made up of shaded triangles located directly to the lower - left of a @xmath104 vertex , and @xmath139 is made up of shaded triangles located directly to the lower - right of a @xmath104 vertex . . identifying opposite sides of the solid grey parallelogram gives a torus . [ fig3homtessellation ] ] in this section we provide the geometric proof of theorem [ theorem3homtransversals ] . let @xmath5 be a @xmath3-homogeneous bitrade . apply lemma [ lemmatorus ] to obtain the labelled tessellation of the euclidean plane for @xmath5 . without loss of generality , let the triangles have unit length sides . let @xmath140 be an unshaded triangle in the tessellation . define three actions @xmath64 on @xmath140 : @xmath141 rotates @xmath140 by angle @xmath142 anticlockwise around its @xmath104 vertex ; @xmath143 rotates @xmath140 by angle @xmath142 anticlockwise around its @xmath105 vertex ; @xmath144 rotates @xmath140 by angle @xmath142 anticlockwise around its @xmath123 vertex . the plane is tessellated by hexagons similar to those in figure [ fig3homtessellation ] . since @xmath5 is @xmath3-homogeneous , it follows that there are three shaded triangles at each vertex , so @xmath145 for @xmath146 and @xmath147 . these @xmath64 induce a _ triangle group _ @xmath148 which acts on the set of equilateral triangles @xmath149 of the tessellation : @xmath150 if @xmath151 $ ] is the @xmath46 representation for the bitrade in question , then we define the _ cartographic group _ @xmath152 by : @xmath153 note that @xmath148 is an infinite group , acting on the tessellation of the euclidean plane , while @xmath152 is a finite permutation group , acting on the corresponding triangles on the identified surface ( the torus ) . the group @xmath152 has all of the defining relations for @xmath148 so it is natural to define a group homomorphism @xmath154 that sends @xmath64 to @xmath46 and the empty word @xmath155 to the identity @xmath156 . we then extend @xmath154 to an arbitrary word by @xmath157 where @xmath158 for @xmath159 . to relate the group actions @xmath160 and @xmath161 we form a map @xmath162 . fix a shaded triangle @xmath163 and an entry @xmath164 and set @xmath165 . then use @xmath154 to extend @xmath166 to any @xmath167 by defining @xmath168 where @xmath169 for some @xmath170 . let @xmath166 be defined as above for some fixed @xmath171 , @xmath172 . first we check that @xmath166 is a well - defined map , namely that the choice of @xmath173 for @xmath169 does not matter . suppose that @xmath174 . then @xmath175 so @xmath176 where @xmath177 ( we ca nt assume that @xmath178 is the identity in @xmath152 , only that it fixes @xmath172 ) . then @xmath179 so @xmath180 takes the same value whether @xmath181 or @xmath182 was chosen . hence @xmath166 is well defined . the tessellation lies on the euclidean plane and we are free to place the @xmath107 , @xmath191 axes as we wish . we will choose one of three placements : that shown in figure [ fig3homxyaxes ] , or the rotation of those axes by angle @xmath142 or @xmath192 . suppose , for contradiction , that there exists a triple @xmath197 such that @xmath198 . since @xmath59 has no fixed - point(condition ( t3)in the definition of a bitrade ) there must be a @xmath3-cycle @xmath199 in @xmath59 for some @xmath200 , @xmath201 . if @xmath202 for some shaded triangle @xmath203 then it must be that @xmath204 and @xmath205 . recalling that @xmath141 is rotation about a @xmath104 vertex in the anticlockwise direction , the labelled tessellation must have hexagons like those shown in figure [ fig3homconsistency ] . the hexagon @xmath206 has @xmath207 while the hexagon @xmath7 has @xmath208 . the order of @xmath25 , @xmath200 , @xmath140 is forced by @xmath166 . let each side of a triangle in the tessellation have unit length . without loss of generality we can place the @xmath107 , @xmath191 axes on the tessellation ( as in figure [ fig3homxyaxes ] ) so that the @xmath104 vertex of @xmath206 is at @xmath209 and the @xmath104 vertex of @xmath7 is at @xmath210 . then we have a euclidean distance @xmath211 which we assume to be minimal . we then show that there exists another pair of inconsistently labelled hexagons @xmath212 and @xmath213 such that @xmath214 except for a few cases in which contradictions arise with respect to the bitrade itself . in the limiting case we get @xmath215 which implies that @xmath59 has a fixed point @xmath25 . there are four main cases to check , each with three subcases * a * , * b * , and * c*. each of the * a * and * b * cases cover an infinite part of the plane so we use various constructions to find @xmath212 , @xmath213 such that @xmath214 . the * c * cases are finite and provide the required local contradictions . suppose that @xmath218 and consider the action of @xmath219 on the triangles labelled @xmath140 as shown in figure [ figcase1.1 ] . recall that @xmath141 is rotation to the next shaded triangle in an anticlockwise direction around a @xmath104 vertex , and @xmath143 is rotation around a @xmath105 vertex . we find @xmath220 and factor out the @xmath221 term : @xmath222 now @xmath223 since @xmath217 so @xmath214 . the last step is to observe that there must be a cycle @xmath224 in @xmath59 as shown above . then @xmath225 , so @xmath212 and @xmath213 are a closer pair of inconsistent hexagons . this completes case 1a . suppose that @xmath228 and consider the action of @xmath229 on the triangles labelled @xmath25 as in figure [ figcase1.2 ] . we calculate the distance between @xmath212 and @xmath213 : @xmath230 we need @xmath231 which simplifies to @xmath232 . by assumption , @xmath233 and this completes case 1b . here we deal with local cases which give rise to contradictions . note that some @xmath235 do not correspond to valid hexagon positions , e.g. there is no hexagon centred at @xmath236 . the valid cases are as follows : ( i ) @xmath237 , @xmath238 : : : in this case @xmath206 and @xmath7 are the same hexagon , implying that @xmath239 so @xmath59 has a fixed - pointwhich contradicts ( t3)in the definition of a bitrade ( ii ) @xmath237 , @xmath240 : : : in this case @xmath61 has a fixed - point@xmath25 , contradicting ( t3)(see figure [ figcase1.3 ] ) . ( iii ) @xmath241 , @xmath242 : : : here we find a fixed - pointof @xmath60 , contradicting ( t3)(see figure [ figcase1.4 ] ) . ( iv ) @xmath241 , @xmath243 : : : if @xmath228 then the action of @xmath229 on the triangles labelled @xmath25 shows that @xmath59 has a fixed - pointas shown in figure [ figcase1.5 ] . cases 2 , 3 , and 4 are very similar . for each sub - case of type * a * and * b * we state the @xmath244 word along with the corresponding action as @xmath245 . each sub - case of type * c * gives an immediate contradiction in the form of a fixed point for some @xmath59 , @xmath60 , or @xmath61 . suppose that there exists @xmath197 such that @xmath265 . since @xmath59 is fixed - point - freeit must contain a @xmath3-cycle @xmath199 for some @xmath200 , @xmath201 . then the inconsistent hexagons are as shown in figure [ fig3homhexagon2 ] . we see that @xmath266 , so by lemma [ 3hompartition12 ] we have a contradiction . the case where @xmath267 is similar . by corollary [ 3hompartition123 ] the @xmath195 sets are mutually disjoint . by assumption , the group @xmath269 acts transitively on @xmath10 . the group homomorphism @xmath154 given by @xmath270 is actually a group epimorphism . with lemma [ 3homcommutes ] it follows that each @xmath197 will be an element of some @xmath195 set . hence @xmath271 is a partition of @xmath10 . in lemma [ 3hompartition12 ] and corollary [ corpartition ] we constructed the partition by dividing up elements of @xmath10 around a @xmath104 vertex , so it is impossible for any @xmath195 to have more than one element from a row of @xmath10 ( in particular this would imply a fixed point of @xmath59 ) . conversely , suppose that a cycle @xmath199 exists in @xmath60 and is labelled as shown in figure [ fig3homrowcol ] . now @xmath272 , @xmath273 , @xmath274 according to the labelling induced around @xmath104 vertices . if another hexagon centred at a @xmath105 vertex was inconsistently labelled then we would have an inconsistent labelling around @xmath104 vertices , contradicting corollary [ corpartition ] . similarly , labellings around @xmath123 vertices are consistent . corollary [ corpartition ] and lemma [ 3homtransversal ] give the main result , theorem [ theorem3homtransversals ] . we note that , in general , the covering surface will be spherical , euclidean , or hyperbolic . most ( large ) bitrades will be hyperbolic , and we expect that future work will derive combinatorial properties of hyperbolic bitrades from their geometrical representation .

a latin bitrade @xmath0 is a pair of partial latin squares that define the difference between two arbitrary latin squares @xmath1 and @xmath2 of the same order . a @xmath3-homogeneous bitrade @xmath0 has three entries in each row , three entries in each column , and each symbol appears three times in @xmath4 . cavenagh @xcite showed that any @xmath3-homogeneous bitrade may be partitioned into three transversals . in this paper we provide an independent proof of cavenagh s result using geometric methods . in doing so we provide a framework for studying bitrades as tessellations in spherical , euclidean or hyperbolic space . additionally , we show how latin bitrades are related to finite representations of certain triangle groups .

instrumentation is a technique whereby existing code is modified in order to observe or modify its behaviour . it has a lot of different applications , such as profiling , coverage analysis and cache simulations . one of its most interesting features is however the ability to perform automatic debugging , or at least assist in debugging complex programs . after all , instrumentation code can intervene in the execution at any point and examine the current state , record it , compare it to previously recorded information and even modify it . debugging challenges that are extremely suitable for analysis through instrumentation include data race detection @xcite and memory management checking @xcite . these are typically problems that are very hard to solve manually . however , since they can be described perfectly using a set of rules ( e.g. the memory must be allocated before it is accessed , or no two threads must write to the same memory location without synchronising ) , they are perfect candidates for automatic verification . instrumentation provides the necessary means to insert this verification code with little effort on the side of the developer . the instrumentation can occcur at different stages of the compilation or execution process . when performed prior to the execution , the instrumentation results in changes in the object code on disk , which makes them a property of a program or library . this is called static instrumentation . if the addition of instrumentation code is postponed until the program is loaded in memory , it becomes a property of an execution . in this case , we call it dynamic instrumentation . examples of stages where static instrumentation can be performed are directly in the source code @xcite , in the assembler output of the compiler @xcite , in the compiled objects or programs ( e.g. eel @xcite , atom @xcite , alto @xcite ) . the big advantage of static instrumentation is that it must be done only once , after which one can perform several executions without having to reinstrument the code every time . this means that the cost of instrumenting the code can be relatively high without making such a tool practically unusable . the larges disadvantage of static instrumentation is that it requires a complex analysis of the target application to detect all possible execution paths , which is not always possible . additionally , the user of a static instrumentation tool must know which libraries are loaded at run time by programs he wants to observe , so that he can provide instrumented versions of those . finally , every time a new type of instrumentation is desired , the application and its libraries must be reinstrumented . most of the negative points of static instrumentation are solved in its dynamic counterpart . in this case , the instrumentation is not performed in advance , but gradually at run time as more code is executed . since the instrumentation can continue while the program is running , no prior analysis of all possible execution paths is required . it obviously does mean that the instrumentation must be redone every time the program is executed . this is somewhat offset by having to instrument only the part of the application and its libraries that is covered by a particular execution though . one can even apply dynamic optimization techniques @xcite to further reduce this overhead . when using dynamic instrumentation , the code on disk is never modified . this means that a single uninstrumented copy of an application and its libraries suffices when using this technique , no matter how many different types of instrumentation one wants to perform . another consequence is that the code even does not have to exist on disk . indeed , since the original code is read from memory and can be instrumented just before it is executed , even dynamically loaded and generated code pose no problems . however , when the program starts modifying this code , the detection and handling of these modifications is not possible using current instrumentation techniques . yet , being able to instrument self - modifying code becomes increasingly interesting as run time systems that exhibit such behaviour gain more and more popularity . examples include java virtual machines , the .net environment and emulators with embedded just - in - time compilers in general . these environments often employ dynamic optimizing compilers which continuously change the code in memory , mainly for performance reasons . instrumenting the programs running in such an environment is often very easy . after all , the dynamic compiler or interpreter that processes said programs can do the necessary instrumentation most of the time . on the other hand , observing the interaction of the environments themselves with the applications on top and with the underlying operating system is much more difficult . nevertheless , this ability is of paramount importance when analysing the total workload of a system and debugging and enhancing these virtual machines . even when starting from a system that can already instrument code on the fly , supporting self - modifying code is a quite complex undertaking . first of all , the original program code must not be changed by the instrumentor , since otherwise the program s own modifications may conflict with these changes later on . secondly , the instrumentor must be able to detect changes performed by the program before the modified code is executed , so that it can reinstrument this code in a timely manner . finally , the reinstrumentation itself must take into account that an instruction may be changed using multiple write operations , so it could be invalid at certain points in time . in this paper we propose a novel technique that can be used to dynamically instrument self - modifying code with an acceptable overhead . we do this by using the hardware page protection facilities of the processor to mark pages that contain code which has been instrumented as read - only . when the program later on attempts to modify instrumented code , we catch the resulting protection faults which enables us to detect those changes and act accordingly . the described method has been experimentally evaluated using the _ diota _ ( dynamic instrumentation , optimization and transformation of applications @xcite ) framework on the linux / x86 platform by instrumenting a number of javagrande @xcite benchmarks running in the sun 1.4.0 java virtual machine . the paper now proceeds with an overview of dynamic instrumentation in general and _ diota _ in particular . next , we show how the detection of modified code is performed and how to reinstrument this code . we then present some experimental results of our implementation of the described techniques and wrap up with the conclusions and our future plans . dynamic instrumentation can be be done in two ways . one way is modifying the existing code , e.g. by replacing instructions with jumps to routines which contain both instrumentation code and the replaced instruction @xcite . this technique is not very usable on systems with variable - length instructions however , as the jump may require more space than the single instruction one wants to replace . if the program later on transfers control to the second instruction that has been replaced , it will end up in the middle of this jump instruction . the technique also wreaks havoc in cases of data - in - code or code - in - data , as modifying the code will cause modifications to the data as well . the other approach is copying the original code into a separate memory block ( this is often called _ cloning _ ) and adding instrumentation code to this copy @xcite . this requires special handling of control - flow instructions with absolute target addresses , since these addresses must be relocated to the instrumented version of the code . on the positive side , data accesses still occur correctly without any special handling , even in data - in - code situations . the reason is that when the code is executed in the clone , only the program counter ( pc ) has a different value in an instrumented execution compared to a normal one . this means that when a program uses non - pc - relative addressing modes for data access , these addresses still refer to the original , unmodified copy of the program or data . pc - relative data accesses can be handled at instrumentation time , as the instrumentor always knows the address of the instruction it is currently instrumenting . this way , it can replace pc - relative memory accesses with absolute memory accesses based on the value the pc would have at that time in a uninstrumented execution . _ diota _ uses the cloning technique together with a cache that keeps track of already translated instruction blocks . it is implemented as a shared library and thus resides in the same address space as the program it instruments . by making use of the ld_preload environment variable under linux , the dynamic linker ( ld.so ) can be forced to load this library , even though an application is not explicitly linked to it . the init routines of all shared libraries are executed before the program itself is started , providing _ diota _ an opportunity to get in control . as shown in figure [ fig : diota_operation ] , the instrumentation of a program is performed gradually . first , the instructions at the start of the program are analysed and then copied , along with the desired instrumentation code , to the _ clone _ ( a block of memory reserved at startup time , also residing in the program s address space ) . during this process , direct jumps and calls are followed to their destination . the instrumentation stops when an instruction is encountered of which the destination address can not be determined unequivocally , such as an indirect jump . at this point , a _ trampoline _ is inserted in the clone . this is a small piece of code which will pass the actual target address to _ diota _ every time the corresponding original instruction would be executed . for example , in case of a jump with the target address stored in a register , the trampoline will pass the value of that specific register to _ diota _ every time it is executed . diota _ is entered via such a trampoline , it will check whether the code at the passed address has already been instrumented . if that is not the case , it is instrumented at that point . next , the instrumented version is executed . figure [ fig : translationtable ] shows how _ diota _ keeps track of which instructions it has already instrumented and where the instrumented version can be found . a marker consisting of illegal opcodes is placed after every block of instrumented code ( aligned to a 4-byte boundary ) , followed by the translation table . such a translation table starts with two 32 bit addresses : the start of the block in the original code and its counterpart in the clone . next , pairs of 8 bit offsets between two successive instructions in the respective blocks are stored , with an escape code to handle cases where the offset is larger than 255 bytes ( this can occur because we follow direct calls and jumps to their destination ) . in addition to those tables , an avl tree is constructed . the keys of its elements are the start and stop addresses of the blocks of original code that have been instrumented . the values are the start addresses of the translation tables of the corresponding instrumented versions . every instruction is instrumented at most once , so the keys never overlap . this means that finding the instrumented version of an instruction boils down to first searching for its address in the avl tree and if found , walking the appropriate translation table . to speed up this process , a small hash table is used which keeps the results of the latest queries . a very useful property of this system is that it also works in reverse : given the address of an instrumented instruction , it is trivial to find the address of corresponding original instruction . first , the illegal opcodes marker is sought starting from the queried address and next the table is walked just like before until the appropriate pair is found . this ability of doing two - way translations is indispensable for the self - modifying code support and proper exception handling . since the execution is followed as it progresses , code - in - data and code loaded or generated at run time can be handled without any problems . when a trampoline passes an address to _ diota _ of code it has not yet instrumented , it will simply instrument it at that time . it is irrelevant where this code is located , when it appeared in memory and whether or not it doubles as data _ diota _ has several modes of operation , each of which can be used separately , but most can be combined as well . through the use of so - called backends , the different instrumentation modes can be activated and the instrumentation parameters can be modified . these backends are shared libraries that link against _ diota _ and which can ask to intercept calls to arbitrary dynamically linked routines based on name or address , to have a handler called whenever a memory access occurs , when a basic block completes or when a system call is performed ( both before and after the system call , with the ability to modify its parameters or return value ) . several backends can be used at the same time . other features of the _ diota _ framework include the ability to handle most extensions to the 80x86 isa ( such as mmx , 3dnow ! and sse ) and an extensible and modular design that allows easy implementation of additional backends and support for newly introduced instructions . this paper describes the support for instrumenting self - modifying code in _ diota_. for other technical details about _ diota _ we refer to @xcite . an aspect that is of paramount importance to the way we handle self - modifying code , is the handling of exceptions ( also called signals under linux ) . the next section will describe in more detail how we handle the self - modifying code , but since it is based on marking the pages containing code that has been instrumented as read - only , it is clear that every attempt to modify such code will cause a protection fault ( or _ segmentation fault _ ) exception . these exceptions and those caused by other operations must be properly distinguished , in order to make sure that the program still receives signals which are part of the normal program execution while not noticing the other ones . this is especially important since the java virtual machine that we used to evaluate our implementation uses signals for inter - thread communication . when a program starts up , each signal gets a default handler from the operating system . if a program wants to do something different when it receives a certain signal , it can install a signal handler by performing a system call . this system call gets the signal number and the address of the new handler as arguments . since we want to instrument these user - installed handlers , we have to intercept these system calls . this can be achieved by registering a system call analyser routine with _ diota_. this instructs _ diota _ to insert a call to this routine after every system call in the instrumented version of the program . if such a system call successfully installed a new signal handler , the analyser records this handler and then installs a _ diota _ handler instead . next , when a signal is raised , _ diota _ s handler is activated . one of the arguments passed to a signal handler contains the contents of all processor registers at the time the signal occurred , including those of the instruction pointer register . since the program must not be able to notice it is being instrumented by looking at at that value , it is translated from a clone address to an original program address using the translation tables described previously . finally , the handler is executed under control of _ diota _ like any other code fragment . once the execution arrives at the sig_return or sig_rt_return system call that ends this signal s execution , _ diota _ replaces the instruction pointer in the signal context again . if the code at that address is not yet instrumented , the instruction pointer value in the context is replaced with the address of a trampoline which will transfer control back to _ diota _ when returning from the signal s execution . otherwise , the clone address corresponding to the already instrumented version is used . dynamically generated and loaded code can already be handled by a number of existing instrumentors @xcite . the extra difficulty of handling self - modifying code is that the instrumentation engine must be able to detect modifications to the code , so that it can reinstrument the new code . even the reinstrumenting itself is not trivial , since a program may modify an instruction by performing two write operations , which means the intermediate result could be invalid . there are two possible approaches for dealing with code changes . one is to detect the changes as they are made , the other is to check whether code has been modified every time it is executed . given the fact that in general code is modified far less than it is executed , the first approach was chosen . the hardware page protection facilities of the processor are used to detect the changes made page . once a page contains code that has been instrumented , it will be write - protected . the consequence is that any attempt to modify such code will result in a segmentation fault . an exception handler installed by _ diota _ will intercept these signals and take the appropriate action . since segmentation faults must always be caught when using our technique to support self - modifying code , _ diota _ installs a dummy handler at startup time and whenever a program installs the default system handler for this signal ( which simply terminates the process if such a signal is raised ) , or when it tries to ignore it . apart from that , no changes to the exception handling support of _ diota _ have been made , as shown in figure [ fig : exceptions ] . whenever a protection fault occurs due to the program trying to modify some previously instrumented code , a naive implementation could unprotect the relevant page , perform the required changes to the instrumented code inside the signal handler , reprotect the page and continue the program at the next instruction . there are several problems with this approach however : * on a cisc architecture , most instructions can access memory , so decoding the instruction that caused the protection fault ( to perform the change that caused the segmentation fault in the handler ) can be quite complex . * it is possible that an instruction is modified by means of more than one memory write operation . trying to reinstrument after the first write operation may result in encountering an invalid instruction . * in the context of a jit - compiler , generally more than one write operation occurs to a particular page . an example is when a page was already partially filled with code which was then executed and thus instrumented , after which new code is generated and placed on that page as well . a better way is to make a copy of the accessed page , then mark it writable again and let the program resume its execution . this way , it can perform the changes it wanted to do itself . after a while , the instrumentor can compare the contents of the unprotected page and the the buffered copy to find the changes . so the question then becomes : when is this page checked for changes , how long will it be kept unprotected and how many pages will be kept unprotected at the same time . all parameters are important for performance , since keeping pages unprotected and checking them for changes requires both processing and memory resources . the when - factor is also important for correctness , as the modifications must be incorporated in the clone code before it is executed again . on architectures with a weakly consistent memory model ( such as the sparc and powerpc ) , the program must make its code changes permanent by using an instruction that synchronizes the instruction caches of all processors with the current memory contents . these instructions can be intercepted by the instrumentation engine and trigger a comparison of the current contents of a page with the previously buffered contents . on other architectures , heuristics have be used depending on the target application that one wants to instrument to get acceptable performance . for example , when using the sun jvm 1.4.0 running on a 80x86 machine under linux , we compare the previously buffered contents of a page to the current contents whenever the thread that caused the protection fault does one of the following : * it performs a kill system call . this means the modifier thread is sending a signal to another thread , which may indicate that it has finished modifying the code and that it tells the other thread that it can continue . * it executes a ret or other instruction that requires a lookup to find the appropriate instrumentation code . this is due to the fact that sometimes the modifying and executing threads synchronise using a spinlock . the assumption here is that before the modifying thread clears the spinlock , it will return from the modification routine , thus triggering a flush . although this method is by no means a guarantee for correct behaviour in the general case , in our experience it always performs correctly in the context of instrumenting code generated by the sun jvm 1.4.0 . the unprotected page is protected again when it has been checked n successive times without any changes having been made to it , or when another page has to be unprotected due to a protection fault . note that this optimisation only really pays off in combination with only checking the page contents in the thread that caused the initial protection fault . the reason is that this ensures that the checking limit is not reached prematurely . otherwise , the page is protected again too soon and a lot of extra page faults occur , nullifying any potential gains . finally , it is possible to vary the number of pages that are being kept unprotected at the same time . possible strategies are keeping just one page unprotected for the whole program in order to minimize resources spent on buffering and comparing contents , keeping one page unprotected per thread , or keeping several pages unprotected per thread to reduce the amount of protection faults . which strategy performs best depends on the cost of a page fault and the time necessary to do a page compare . different code fragments in the clone are often interconnected by direct jumps . for example , when while instrumenting we arrive at an instruction which was already instrumented before , we generate a direct jump to this previously instrumented version instead of instrumenting that code again . this not only improves efficiency , but it also makes the instrumentation of modified code much easier , since there is only one location in the clone we have to adapt in case of a code modification . because of these direct jump interconnections , merely generating an instrumented version of the modified code at a different location in the clone is not enough . even if every lookup for the instrumented version of the code in that fragment returns one of the new addresses in the clone , the old code is still reachable via de direct jumps from other fragments . removing the direct jumps and replacing them with lookups results in a severe slowdown . another solution would be keeping track of to which other fragments each fragment refers and adapting the direct jumps in case of changes . this requires a lot of bookkeeping however , and changing one fragment may result in a cascade effect , requiring a lot of additional changes elsewhere in the clone . for these reasons , we opted for the following three - part strategy . the optimal way to handle the modifications , is to reinstrument the code in - place . this means that the previously instrumented version of the instructions in the clone are simply replaced by the new ones . this only works if the new code has the same length as ( or is shorter than ) the old code however , which is not always the case . a second way to handle modifications can be applied when the instrumented version of the previous instruction at that location was larger than the size of an immediate jump . in this case , it is possible to overwrite the previous instrumented version with a jump to the new version . at the end of this new code , another jump can transfer control back to rest of the original instrumentation code . finally , if there is not enough room for an immediate jump , the last resort is filling the room originally occupied by the instrumented code with breakpoints . the instrumented version of the new code will simply be placed somewhere else in the code . whenever the program then arrives at such a breakpoint , _ diota _ s exception handler is entered . this exception handler has access to the address where the breakpoint exception occurred , so it can use the translation table at the end of the block to look up the corresponding original program address . next , it can lookup where the latest instrumented version of the code at that address is located and transfer control there . [ cols= " < , > , > , > , > , > , > " , ] we evaluated the described techniques by implementing them in the _ diota _ framework . the performance and correctness were verified using a number of tests from the javagrande @xcite benchmark , running under the sun jvm 1.4.0 on a machine with two intel celeron processors clocked at 500mhz . the operating system was redhat linux 7.3 with version 2.4.19 of the linux kernel . several practical implementation issues were encountered . the stock kernel that comes with redhat linux 7.3 , which is based on version 2.4.9 of the linux kernel , contains a number of flaws in the exception handling that cause it to lock up or reboot at random times when a lot of page protection exceptions occur . another problem is that threads in general only have limited stack space and although _ diota _ does not require very much , the exception frames together with _ s overhead were sometimes large enough to overflow the default stacks reserved by the instrumented programs . therefore , at the start of the main program and at the start of every thread , we now instruct the kernel to execute signal handlers on an alternate stack . _ diota _ s instrumentation engine is not re - entrant and as such is protected by locks . since a thread can send a signal to another thread at any time , another problem we experienced was that sometimes a thread got a signal while it held the instrumentation lock . if the triggered signal handler was not yet instrumented at that point , _ diota _ deadlocked when it tried to instrument this handler . disabling all signals before acquiring a lock and re - enabling them afterwards solved this problem . the problem with only parts of instructions being modified , which happens considerably more often than replacing whole instructions , was solved by adding a routine that finds the start of the instruction in which a certain address lies and starting the reinstrumentation from that address . most modifications we observed were changes to the target addresses of direct calls . the heuristics regarding only checking for changes in the thread that caused the initial unprotection of the page , reduced the slowdown caused by the instrumentation by 43% relative to a strategy where pages are checked for changes every time a system call occurs and every time a lookup is performed , regardless of the involved threads . the limit on the number of checks done before a page is protected again ( with n set between 3 and 5 ) provided an additional speed increase of 22% . a large part of the overhead stems from the fact that the elf binary format ( which is used by all modern linux applications ) permits code and data to be on the same page . the result is that once a code fragment on such a page has been instrumented , a lot of page faults occur and unnecessary comparisons have to be performed whenever such data is modified afterwards . a possible solution is not marking pages belonging to the elf binary and the standard libraries loaded at startup time as read - only , but this could compromise the correctness of the described technique . however , it could be done if execution speed is of great concern and if one is certain that no such code will be modified during the execution . table [ tab : jvg ] shows the measured timings when running a number of tests from sections 2 and 3 of the sequential part of the javagrande benchmark v2.0 @xcite , all using the sizea input set . the first column shows the name of the test program . the second and third columns show the used cpu time ( as measured by the time command line program , expressed in seconds ) of an uninstrumented resp . instrumented execution , while the fourth column shows the resulting slowdown factor . the fifth column contains the the amount of protection faults divided by the uninstrumented execution time , so it is an indication of the degree in which the program writes to pages that contain already executed code . the last column shows the number of lookups per second of uninstrumented execution time , where a lookup equals a trip to _ diota _ via a trampoline to get the address of the instrumented code at the queried target address . the results have been sorted on the slowdown factor . regression analysis shows us that the overhead due to the lookups is nine times higher than that caused by the protection faults ( and page compares , which are directly correlated with the number of protection faults , since every page is compared n times after is unprotected due to a fault ) . this means that the page protection technique has a quite low overhead and that most of the overhead can be attributed to the overhead of keeping the program under control . the cause for the high cost of the lookups comes from the fact that the lookup table must be locked before it can be consulted , since it is shared among all threads . as mentioned before , we have to disable all signals before acquiring a lock since otherwise a deadlock might occur . disabling and restoring signals is an extremely expensive operation under linux , as both operations require a system call . we have verified this by instrumenting a program that does not use use any signals nor self - modifying code using two different versions of _ diota _ : one which does disable signals before acquiring a lock and one which does not . the first version is four times slower than the second one . we have described a method which can be used to successfully instrument an important class of programs that use self - modifying code , specifically java programs run in an environment that uses a jit - compiler . the technique uses the hardware page protection mechanism present in the processor to detect modifications made to already instrumented code . additionally , a number of optimisations have already been implemented to reduce the overhead , both by limiting the number of protection faults that occurs and the number of comparisons that must to be performed . in the near future , a number of extra optimisations will be implemented , such as keeping more than one page unprotected at a time and the possibility to specify code regions that will not be modified , thus avoiding page protection faults caused by data and code being located on the same page . additionally , we are also adapting the _ diota _ framework in such a way that every thread gets its own clone and lookup table . this will greatly reduce the need for locking and disabling / restoring signals , which should also result in a significant speedup for the programs that perform a large number of lookups . derek bruening , evelyn duesterwald , and saman amarasinghe . design and implementation of a dynamic optimization framework for windows . in _ proceedings of the 4th acm workshop on feedback - directed and dynamic optimization ( fddo-4 ) _ , austin , texas , december 2001 . s. eggers , d. keppel , e. koldinger , and h. levy . techniques for efficient inline tracing on a shared - memory multiprocessor . in _ sigmetrics conference on measurement and modeling of computer systems _ , volume 8 , may 1990 . barton p. miller , mark d. callaghan , jonathan m. cargille , jeffrey k. hollingsworth , r. bruce irvin , karen l. karavanic , krishna kunchithapadam , and tia newhall . . 28(11):3744 , november 1995 . special issue on performance evaluation tools for parallel and distributed computer systems . jonas maebe , michiel ronsse , and koen de bosschere . : dynamic instrumentation , optimization and transformation of applications . in _ compendium of workshops and tutorials , held in conjunction with pact02 : international conference on parallel architectures and compilation techniques _ , charlottesville , virginia , usa , september 2002 . michiel ronsse and koen de bosschere . jiti : a robust just in time instrumentation technique . volume 29 of _ series computer architecture news _ , chapter proceedings of workshop on binary translation - 2000 , pages 4354 . acm press , march 2001 . k. scott , n. kumar , s. velusamy , b. childers , j. w. davidson , and m. l. soffa . retargetable and reconfigurable software dynamic translation . in _ proceedings of the international symposium on code generation and optimization 2003 _ , san francisco , california , march 2003 .

adding small code snippets at key points to existing code fragments is called instrumentation . it is an established technique to debug certain otherwise hard to solve faults , such as memory management issues and data races . dynamic instrumentation can already be used to analyse code which is loaded or even generated at run time . with the advent of environments such as the java virtual machine with optimizing just - in - time compilers , a new obstacle arises : self - modifying code . in order to instrument this kind of code correctly , one must be able to detect modifications and adapt the instrumentation code accordingly , preferably without incurring a high penalty speedwise . in this paper we propose an innovative technique that uses the hardware page protection mechanism of modern processors to detect such modifications . we also show how an instrumentor can adapt the instrumented version depending on the kind of modificiations as well as an experimental evaluation of said techniques . elis , ghent university , sint - pietersnieuwstraat 41 , 9000 gent , belgium @ifundefined

this is the fifth in a series of papers@xcite investigating the properties of strongly correlating liquids,@xcite i.e. , liquids that have strong correlations between their constant - volume equilibrium fluctuations of potential energy , @xmath4 , and virial@xcite @xmath5 where @xmath6 is the position of particle @xmath7 at time @xmath8 . as is well known , the average virial @xmath0 gives the configurational contribution to the pressure : @xmath9 letting @xmath10 denote instantaneous deviations from equilibrium mean values , the @xmath3 correlation is quantified by the correlation coefficient @xmath11 ( with @xmath12 denoting equilibrium average ) : @xmath13 perfect correlation gives @xmath14 . as a pragmatic definition we have chosen `` strongly correlating liquids '' to designate liquids that have @xmath15 in the @xmath2 ensemble ( constant volume , @xmath16 , and temperature , @xmath17 ) . strongly correlating liquids have simpler physics than liquids in general , an observation that has particular significance for the highly viscous phase.@xcite thus it has been shown that strongly correlating viscous liquids to a good approximation have all eight frequency - dependent thermoviscoelastic response functions@xcite given in terms of just one@xcite ( i.e. , are single - parameter liquids in the sense of having dynamic prigogine - defay ratio@xcite close to unity@xcite ) . strongly correlating viscous liquids moreover obey density scaling@xcite to a good approximation , i.e. , their dimensionless relaxation time @xmath18 ( where @xmath19 is the average particle mass ) depends on density @xmath20 and temperature as @xmath21.@xcite paper i@xcite presented computer simulations of 13 different systems , showing that van der waals type liquids are strongly correlating , whereas hydrogen - bonding liquids like methanol or water are not . strongly correlating liquids include@xcite , for instance , the standard lennard - jones ( lj ) liquid , the kob - andersen binary lj ( kablj ) mixture , an asymmetric rigid - bond dumbbell model , a seven - site united - atom toluene model , and the lewis - wahnstrm otp model . paper ii@xcite analyzed the cause of @xmath3 correlations with a focus on the lj potential . the strong correlations were related to the well - known fact that an inverse power - law ( ipl ) pair potential , @xmath22 where @xmath23 is the distance between two particles , @xcite implies perfect @xmath3 correlation@xcite , @xmath24 with @xmath25 . around the potential energy minimum , the lj potential is well described by an `` extended '' inverse power - law potential ( eipl),@xcite @xmath26 . at constant volume the linear term contributes little to the virial and potential - energy fluctuations : when one nearest - neighbor interatomic distance increases , others decrease in such a way that the sum is almost constant . thus systems interacting via the lj potential inherit strong @xmath3 correlations from an underlying inverse power - law - they have a `` hidden scale invariance''.@xcite paper iii@xcite gave further numerical evidence for the explanation for strong @xmath3 correlations presented in paper ii , and theoretical results were given on the statistical mechanics and thermodynamics of the hidden scale invariance that characterizes strongly correlating liquids . it was also shown that strong virial - potential energy correlations are present even in out - of - equilibrium situations - the hidden scale invariance is a property of the potential energy surface , not just of the equilibrium states . paper iv@xcite introduced the concept of `` isomorphs '' in the phase diagram of a strongly correlating liquid . starting from a single assumption a number of isomorph invariants were derived . in particular , structure and dynamics were shown to be invariant on isomorphs when reduced units are used . in the present paper further simulation results supporting the isomorph predictions are presented using systems interacting with the multicomponent generalized lj potential : @xmath27 where @xmath28 is an ipl potential acting between the two particles @xmath7 and @xmath29 : @xmath30 for systems interacting via a generalized lj potential , a prediction for the shape of the isomorphs in the @xmath3 phase diagram is derived in section iii and demonstrated to fit well to simulation results.@xcite interestingly , the isomorph shape depends only on the exponents @xmath19 and @xmath31 . thus , e.g. , all 12 - 6 lj systems have the same isomorphs in the @xmath3 phase diagram . finally we briefly present two applications of the theory . one tests the paper iv prediction that solid - liquid coexistence lines are isomorphs . the second application gives an approximate equation of state for systems interacting via generalized lj potentials ; this is arrived at by combining the present theory with rosenfeld and tarazona s expression for the isochoric temperature dependence of the potential energy . we term a microscopic configuration `` physically relevant '' if its influence on the thermodynamics and dynamics of the sustem is not _ a priori _ negligible ( paper iv ) . for instance , any configuration with very strong particle overlap is physically irrelevant ; note , however , that even unlikely configurations like transition states are relevant . two state points ( 1 ) and ( 2 ) with temperatures @xmath32 and @xmath33 and densities @xmath34 and @xmath35 , respectively , are defined to be _ isomorphic _ ( paper iv ) if they obey the following : any two physically relevant configurations of state points ( 1 ) and ( 2 ) , @xmath36 and @xmath37 , which trivially scale into one another , @xmath38 have proportional configurational boltzmann statistical weights : @xmath39 here @xmath40 is the potential energy function and it is understood that the constant @xmath41 depends only on the state points ( 1 ) and ( 2 ) . the property of being isomorphic defines a mathematical equivalence relation on the set of state points . the corresponding equivalence classes are smooth curves in the phase diagram termed isomorphs or isomorphic curves . equation ( [ eq : isomorphdef ] ) implies identity of the normalized boltzmann probabilities for scaled relevant configurations of isomorphic state points . as detailed in paper iv this identity implies that several quantities are invariant along an isomorph . examples include the configurational entropy , the isochoric specific heat , n - particle distribution functions in reduced units ( in particular , the radial distribution function(s ) ) , reduced - unit dynamics ( both for newtonian and stochastic dynamics ) , normalized reduced - time autocorrelation functions , reduced - unit transport coefficients , etc . it was further shown in paper iv that a jump from an equilibrium state to an isomorphic state point brings the system instantaneously to equilibrium ( see also ref . ) . ipl potentials are euler homogeneous functions , i.e. , obey @xmath42 . such purely repulsive potentials which do not describe real intermolecular interactions are the only potentials that have 100% correlation between virial and potential energy fluctuations ; this reflects the well - known ipl virial - theorem identity @xmath43 . likewise , only ipl liquids have exact isomorphs ; they satisfy eq . ( [ eq : isomorphdef ] ) with @xmath44 for all configurations of two state points obeying @xmath45 where @xmath25 . appendix a of paper iv showed that strongly correlating liquids generally have isomorphs to a good approximation . consequently , to a good approximation these liquids inherit a number of the exact invariants that ipl liquids have along their curves given by @xmath46 , namely all the ipl invariants that do not rely on the ipl identity @xmath44 . examples include the above mentioned thermodynamic , static , and dynamic isomorph invariants . ipl invariants that are not inherited by strongly correlating liquids in general include , e.g. , the helmholtz free energy over temperature , the potential energy over temperature , and the reduced - unit compressibility . the fact that strongly correlating liquids have isomorphs to a good approximation should be understood as follows . any liquid has curves in its phase diagram of constant excess entropy , curves of constant isochoric specific heat , curves of constant reduced relaxation time , curves of constant reduced - unit transport coefficients , etc . for a strongly correlating liquid these curves are almost identical , and they define its ( approximate ) isomorphs . points on these isomorphs have approximately identical structure and approximately identical dynamics as probed , e.g. , by normalized reduced - time - autocorrelation functions . for clarity of presentation below we shall not repeatedly mention that the existence of isomorphs is not exact for generalized lj systems . accordingly , we shall speak of isomorphs as a unique , well - defined concept . in other words : all properties of isomorphs derived below in generalized lj systems would be exact if these systems did have exact isomorphs , e.g. , fulfilled eq . ( [ eq : isomorphdef ] ) ; since this is not the case , the properties are in principle as well as practice approximate . the structure in reduced units ( @xmath47 ) is predicted to be invariant along an isomorph ( paper iv ) . the excess entropy , @xmath48 , depends only on structure and was shown in paper iv also to be invariant along an isomorph ( here and henceforth `` excess '' refers to the quantity in question in excess to that of an ideal gas with same density and temperature ) . in the below simulations we generate an isomorph as a set of state points in the phase diagram with constant @xmath49 . to change density and temperature keeping @xmath49 constant , the following identity is used ( paper iv ) : @xmath50 where @xmath51 and a maxwell relation was applied . utilizing the standard fluctuation formulae this leads ( paper iv ) to : @xmath52 the procedure applied to generate isomorphs numerically can be summarized as : 1 ) an equilibrium @xmath2 simulation was performed at a given state point ; 2 ) @xmath53 was calculated from the fluctuations using eq . ( [ eq : gammasex ] ) ; 3 ) the density @xmath54 was changed slightly ( of order 1% ) , and @xmath53 was used to calculate the corresponding change in temperature in order to keep @xmath55 constant ( eq . ( [ eq : gammasex ] ) ) ; 4 ) a simulation at the new state point was performed , and the procedure was repeated . figure [ fig : structure](a ) demonstrates the isomorph invariance of the radial distribution function for the large ( a ) particles of the kob - andersen binary lj mixture @xcite ( kablj ) . for comparison , [ fig : structure](b ) shows the same on an isotherm with a similar ( but smaller ) density change . clearly , structure is to a good approximation invariant on the isomorph , whereas this is not the case on the isotherm . ) of the kablj mixture ; time is given in reduced units : @xmath56 and the q - vector is kept constant in reduced units . b ) same as in a ) , but along an isotherm ( density variation @xmath57).,title="fig:",scaledwidth=45.0% ] ) of the kablj mixture ; time is given in reduced units : @xmath56 and the q - vector is kept constant in reduced units . b ) same as in a ) , but along an isotherm ( density variation @xmath57).,title="fig:",scaledwidth=45.0% ] taking the logarithm of eq . ( [ eq : isomorphdef ] ) shows that the potential energy surface in reduced units ( @xmath58 ) is the same for two isomorphic state points , except for an additive constant . this constant does not influence the dynamics , and consequently the dynamics in reduced units is the same for the two isomorphic state points . figure [ fig : dynamics](a ) demonstrates the isomorph invariance of the incoherent intermediate scattering function of the a particles of the kablj mixture . for comparison , [ fig : dynamics](b ) shows the same property on the @xmath59 isotherm . like the structure , the dynamics is also invariant to a good approximation on the isomorph - both regarding average relaxation time and the shape of the relaxation function ; this is far from the case on the corresponding isotherm . calculated from equilibrium fluctuations via eq . ( [ eq : gammasex ] ) for the kablj system . red squares : state points belonging to the isomorph presented in figs . [ fig : structure ] and [ fig : dynamics ] . black circles : isochoric state points . straight lines are linear regression fits . standard errors on slopes are 0.06 ( isomorph ) and 0.04 ( isochore ) . , scaledwidth=45.0% ] the exponent @xmath53 calculated from eq . ( [ eq : gammasex ] ) generally depends on the state point . figure [ fig : gammasex ] shows ( as red squares ) @xmath53 along the isomorph of figs . 1(a ) and 2(a).temperature changes by almost a factor of two , but @xmath53 changes less than 5% . paper iv argued that for a system with good isomorphs @xmath53 depends only on density . this is supported by fig . [ fig : gammasex ] in which @xmath53 on an isochore is shown to vary only @xmath60% when temperature is changed by a factor of two ( black circles ) . , @xmath61 , eq . ( [ eq : genlj ] ) ) and parameters chosen to place the minimum of the potential at the same distance and depth as in the 12 - 6 kablj system . likewise , 18 - 9 kablj is using the exponents @xmath62 and @xmath63 . finally 12 - 6 wblj is the wahnstrm 50:50 mixture@xcite using the standard 12 - 6 lj potential . straight lines are linear regression fits . the inset shows a zoom on the data for the two 12 - 6 lj systems . for each system the simulated temperatures ranges from where non - exponential relaxation sets in ( @xmath64 ) to a viscous state with two - step relaxation ( @xmath65 ) . , scaledwidth=45.0% ] it follows from eq . ( [ eq : gammasex1 ] ) that the slope of isochores in the @xmath3 phase diagram is given by @xmath53 : @xmath66 the insignificant change of @xmath53 along isochores ( black circles in fig . [ fig : gammasex ] ) means that these to a good approximation are straight lines in the @xmath3 phase diagram . this is illustrated in fig . [ fig : uwischores ] for four different binary generalized lj mixtures . what is the shape of an isomorph in the @xmath3 phase diagram ? this section answers this question in generality for the multi - component generalized lj potential . recall that this potential ( eq . ( [ eq : genlj ] ) ) is the sum of two ipl s . correspondingly , the potential energy and virial can be expressed as sums of two ipl terms : @xmath67 for pair interactions the virial is given by@xcite @xmath68 where the prime denotes the derivative with respect to @xmath69 . from this we get @xmath70 for any point in the @xmath3 phase diagram we can solve eqs . ( [ eq : u ] ) and ( [ eq : w ] ) for @xmath71 : @xmath72 phase diagram for systems interacting via the standard 12 - 6 lj potential . included in this figure are also two theoretical isomorphs , plotted using eq . ( [ eq : parabola ] ) . the accessible points are those above both dashed lines ( see text ) . the part of the isomorphs between the dashed lines in the @xmath73 quadrant are thus not accessible to the standard 12 - 6 lj ( they are accessible to the purely repulsive 12 - 6 lj potential with @xmath74.),scaledwidth=45.0% ] these equations allow one to determine which regions of the @xmath3 phase diagram are accessible to the system . adopting the convention @xmath75 , we restrict ourselves to the case @xmath76 in order to have a repulsive core for all interactions . this implies @xmath77 for all configurations , and therefore also for all thermodynamic averages . combining this with eq . ( [ eq : um ] ) implies @xmath78 . if all interactions have an attractive part ( @xmath79 ) we have @xmath80 , which in combinations with eq . ( [ eq : un ] ) implies @xmath81 . as an example , the region of the @xmath3 phase diagram accessible to the standard 12 - 6 lj potential is given by @xmath82 and @xmath83 ( the later inequality being relevant when @xmath84 ) . see fig . [ fig : genericwu ] for an illustration . along an isomorph the structure is invariant in reduced units ( fig . [ fig : structure ] ) . from this it follows that @xmath85 is invariant for any @xmath86 along an isomorph ( @xmath47 ) : @xmath87 if we let `` * '' denote a reference point , @xmath88 , we can use eqs . ( [ eq : um ] ) and ( [ eq : un ] ) to get @xmath89 and find for other state points on the same isomorph ( where @xmath90 ) : @xmath91 combining eq . ( [ eq : scaled_u ] ) with eqs . ( [ eq : u ] ) and ( [ eq : w ] ) we obtain a parametric description of an isomorph in the @xmath3 phase diagram : @xmath92 once the numbers @xmath93 and @xmath94 have been determined from a reference state point using eqs . ( [ eq : um ] ) and ( [ eq : un ] ) , the entire isomorph to which this state point belongs is thus traced out in the wu phase diagram using eqs . ( [ eq : isomorphu ] ) and ( [ eq : isomorphw ] ) . alternatively , given a collection of isomorphic state point generated , e.g. , as described in section iic , the predicted relation between @xmath1 , @xmath0 , and @xmath95 can be tested by linear regression : @xmath96 this is done in fig . [ fig : isomorphuwtest ] for the 12 - 6 kablj isomorph presented in figs . [ fig : structure ] and [ fig : dynamics ] . the potential energies were fitted by linear regression by means of eq . ( [ eq : isomorphu_lin ] ) . subsequently , the virials were compared to eq . ( [ eq : isomorphw_lin ] ) using the parameters estimated from the potential energies ( eq . ( [ eq : isomorphu_lin ] ) ) , i.e. , without performing a new fit . and [ fig : dynamics ] ) . top panel : u fitted to eq . ( [ eq : isomorphu_lin ] ) using @xmath97 ( i.e. , @xmath98 ) . bottom panel : w compared to eq . ( [ eq : isomorphw_lin ] ) , using the parameters @xmath93 and @xmath94 found from @xmath1 ( top panel).,scaledwidth=45.0% ] phase diagram . the points give simulation results , the full curves the predictions of eqs . ( [ eq : isomorphu ] ) and ( [ eq : isomorphw ] ) . two isomorphs belong to the kablj system , the third result from simulations of wahnstrm binary lj liquid@xcite ( wblj ) , and the fourth ( sclj ) is the single component lj liquid . , scaledwidth=43.0% ] by scaling with @xmath99 defined as the virial on each isomorph when @xmath100 . the points give simulation results , the dashed curve the prediction of eq . ( [ eq : parabola ] ) . , scaledwidth=45.0% ] ) and ( [ eq : isomorphw ] ) ) . for reference , also the isomorph prediction for 12 - 6 lj systems is included . , scaledwidth=45.0% ] we note here two important consequences of the above . 1 ) if the reference state point @xmath101 generates the isomorph @xmath102 , then the reference state point @xmath103 generates the isomorph @xmath104 . this means that all isomorphs for a given system have the same shape in the @xmath105 diagram and can be scaled onto one another . 2 ) the shape of the isomorphs depends only on the exponents , @xmath19 and @xmath31 . this means that all 12 - 6 lj systems have the same isomorphs in the @xmath3 phase diagram they may be scaled onto a single `` master isomorph '' . figure [ fig : isomorph ] shows four isomorphs in the @xmath3 phase diagram for three different 12 - 6 lj systems.@xcite figure [ fig : figure10 ] shows the same isomorphs scaled onto the 12 - 6 master isomorph , figure [ fig : isomorph_18_6 ] shows an isomorph for the 18 - 6 kablj system and compares it to a 12 - 6 lj isomorph . overall , the agreement between simulation results and predicted shapes is quite good . combining eq . ( [ eq : scaled_u ] ) with eqs . ( [ eq : um ] ) and ( [ eq : un ] ) we get @xmath106 eliminating density and rearranging , we find an invariant for the isomorph : @xmath107 choosing the reference point with zero potential energy @xmath108 ( this reference point is guaranteed to exist if all interactions have attraction ) , we get : @xmath109 @xmath110 is a unique number identifying the isomorph . equation ( [ eq : invariant ] ) implies that the isomorph is the solution to @xmath111 here it can be seen directly that - as argued above - for a given set of exponents @xmath112 all isomorphs have the same shape , suggesting the name `` master isomorph '' ; @xmath110 is the parameter that determines the scale of each isomorph . for @xmath113 ( e.g. , the standard 12 - 6 lj potential ) the solution to this ( quadratic ) equation is : @xmath114 this has real solution(s ) whenever @xmath115 , where equality corresponds to @xmath116 . figure [ fig : genericwu ] plots eq . ( [ eq : parabola ] ) with @xmath61 for two values of @xmath110 . ) . left panel : eq . ( [ eq : isomorphu_lin ] ) , right panel : eq . ( [ eq : isomorphw_lin ] ) ; data points are simulation results from ahmed and sadus , @xcite straight lines are linear regression fits . the lower panel plots the fit - parameters from eqs . ( [ eq : isomorphu_lin ] ) and ( [ eq : isomorphw_lin ] ) against each other , demonstrating that the fits to energies and virials , respectively , are consistent . , title="fig:",scaledwidth=23.5% ] ) . left panel : eq . ( [ eq : isomorphu_lin ] ) , right panel : eq . ( [ eq : isomorphw_lin ] ) ; data points are simulation results from ahmed and sadus , @xcite straight lines are linear regression fits . the lower panel plots the fit - parameters from eqs . ( [ eq : isomorphu_lin ] ) and ( [ eq : isomorphw_lin ] ) against each other , demonstrating that the fits to energies and virials , respectively , are consistent . , title="fig:",scaledwidth=23.5% ] ) . left panel : eq . ( [ eq : isomorphu_lin ] ) , right panel : eq . ( [ eq : isomorphw_lin ] ) ; data points are simulation results from ahmed and sadus , @xcite straight lines are linear regression fits . the lower panel plots the fit - parameters from eqs . ( [ eq : isomorphu_lin ] ) and ( [ eq : isomorphw_lin ] ) against each other , demonstrating that the fits to energies and virials , respectively , are consistent . , title="fig:",scaledwidth=30.0% ] , plotting potential energy versus density ( left panel ) and virial versus density ( right panel ) . dashed lines : isomorphic predictions ( eqs . ( [ eq : isomorphu ] ) and ( [ eq : isomorphw ] ) ) , using the parameters determined in fig . [ fig : uwcoexist_lin ] . , title="fig:",scaledwidth=23.5% ] , plotting potential energy versus density ( left panel ) and virial versus density ( right panel ) . dashed lines : isomorphic predictions ( eqs . ( [ eq : isomorphu ] ) and ( [ eq : isomorphw ] ) ) , using the parameters determined in fig . [ fig : uwcoexist_lin ] . , title="fig:",scaledwidth=23.5% ] ) ) with @xmath117 and @xmath118 respectively . open circles indicate results from simulations in the solid and liquid phase respectively . both phases are ( very ) strongly correlating with @xmath119 . inset shows the solid - liquid phase behavior in a @xmath120-diagram with logarithmic axes . dashed red lines are the isomorph prediction @xmath121 using @xmath53 found using eq . ( [ eq : gammasex ] ) from the simulations of the two phases ( open circles ) giving @xmath122 and 5.18 respectively . full black lines are free fits giving @xmath123 and 6.49 respectively . , scaledwidth=45.0% ] in paper iv it was argued briefly that eq . ( [ eq : isomorphdef ] ) predicts that solid - liquid coexistence lines are isomorphs . in more detail the argument goes as follows . first one notes that an isomorph can not cross a solid- liquid coexistence line : recall that the isomorph concept relates to equilibrium ensemble probabilities . for a pair of isomorphic state points all pairs of microscopic configurations related by eq . ( [ scaledef ] ) have identical boltzmann probabilities according to eq . ( [ eq : isomorphdef ] ) . consequently , the two isomorphic state points must also have the same macroscopic phase behavior . in other words , if an isomorph were to cross the coexistence line , part of the isomorph would have liquid microstates as the most likely and part of it would have mixed crystalline - liquid microstates as the most likely . this would violate the proportionality of boltzmann statistical weights for scaled states , the property that defines an isomorph . given that an isomorph can not cross the solid - liquid coexistence line , the next point is to note that because all state points belong to some isomorph a curve infinitesimally close to the coexistence line must be an isomorph . this argument applies in the @xmath124 phase diagram , as well as in the density - temperature or the pressure - temperature phase diagrams . in the first two cases there is a multitude of isomorphs in the coexistence region ; for instance one particular isomorph will be characterized by 53% liquid and 47% crystal , etc . in the pressure - temperature phase diagram these `` coexistence '' isomorphs collapse into one . the prediction that the melting line , both slightly to the liquid side and slightly to the solid side , are isomorphs has a number of consequences and it sheds new light on well - known phenomenological melting rules ( as well as exceptions to these rules for non - strongly correlating liquids , paper iv ) . thus , for instance the observation that the lindemann melting criterion is pressure independent for a given liquid follows from the isomorph property of the melting line ( whereas the existence of a universal lindemann criterion does not follow ) . indeed , as is easy to show from paper iv the isomorph theory implies ross `` generalized lindemann melting law '' from 1969 @xcite , according to which the reduced configurational partition function of the crystalline phase @xmath125 $ ] is invariant along the melting line ( here @xmath1 is the potential energy , @xmath126 is the potential energy of the crystal with all atoms located exactly at their lattice sites , and the tilde denotes reduced coordinates . further well - known melting rules , which the isomorph theory explains , include ( paper iv ) : the reduced viscosity , the reduced surface tension , the reduced diffusion constant , and the reduced heat conductivity are all invariant in the liquid phase along the melting line . likewise , the reduced - unit static structure factor is invariant along the melting curve . these rules have been confirmed for several liquids in both experiment and simulation ( see references of paper v ) . like other isomorph predictions following from eq . ( [ eq : isomorphdef ] ) , the prediction that solid - liquid coexistence curves are isomorphs is only expected to be exact for systems that obey eq . ( [ eq : isomorphdef ] ) exactly , i.e. , perfectly correlating systems . indeed , it straightforward to show that in ipl systems the phase behavior depends only on @xmath127 , and thus the co - existence line is an ( exact ) isomorph . the interesting question now is : how well do strongly correlating systems , e.g. , generalized lj systems , follow the prediction ? note that we do not expect the liquid - gas co - existence lines to be isomorphs ( and indeed they are not ) ; the gas - phase is not strongly correlating and thus eq . ( [ eq : isomorphdef ] ) can not be fulfilled in the relevant part of configuration space . in contrast , at the solid - liquid coexistence both phases involved are strongly correlating for lj systems ( paper i ) , and we thus in the following test to what degree the isomorph prediction is fulfilled . ahmed and sadus used in a recent paper @xcite a novel numerical method for determining solid - liquid coexistence . they reported results for the generalized lj potential with the repulsive exponent @xmath19 varying from 12 to 7 and fixed attractive exponent @xmath61 ( ahmed and sadus termed the repulsive exponent @xmath31 , not @xmath19 ; we keep , however , the above notation ) . figure [ fig : uwcoexist_lin ] shows the state points reported by ahmed and sadus for the liquid side of the solid - liquid coexistence curve compared to the isomorphic predictions as expressed in eqs . ( [ eq : isomorphu_lin ] ) and ( [ eq : isomorphw_lin ] ) . the lower panel shows that the parameters found by fitting to eqs . ( [ eq : isomorphu_lin ] ) and ( [ eq : isomorphw_lin ] ) respectively are consistent . figure [ fig : uwcoexist ] shows the same data plotting , respectively , potential energy ( left panel ) and virial ( right panel ) versus density . figure [ fig : uwcoexistphase ] shows the 12 - 6 lj solid - liqud phase behavior in the wu - diagram and in the @xmath128-diagram ( inset ) , including here also data from mastny and de pablo.@xcite the prediction that solid - liquid coexistence lines are isomorphs agrees well on the liquid side with the simulation results of ahmed and sadus@xcite as well as with the results for 12 - 6 lj from mastny and de pablo@xcite . on the solid side the agreement is less convincing ( see fig . [ fig : uwcoexistphase ] ) ; in particular the logarithmic derivative in the @xmath128 diagram ( slope in the inset ) differs by 25% . at present we do not have an explanation for this . it might be related to inter - particle distances longer than the first peak of the radial distribution function playing a bigger role for the phase behavior on the solid side@xcite than on the liquid side . ) ) plotting potential energy versus @xmath129 on isochores for the systems investigated in fig . [ fig : uwischores ] . , scaledwidth=45.0% ] phase diagram using @xmath130 as reference density . curves and lines : theoretical predictions ( eqs . ( [ eq : ustar ] ) to ( [ eq : isomorph_scaling_w ] ) ) for three isochores , an isomorph , and an isotherm . data points : simulations of the 12 - 6 kablj mixture . the parameters of the equation of state were found from the two state points indicated by circles , giving @xmath131 . by comparison , the corresponding parameters found by fitting to the @xmath130 isochore in figs . [ fig : uwischores ] and [ fig : urosenfeld ] are @xmath132 , illustrating the robustness of the fitted equation of state . applying the fitting procedure to the simulated state points on the @xmath133 isotherm gives the parameters ( 38.5 , 5.19 , -8.63 , 2.61 ) . lower panel : same data as above in a @xmath134 diagram . , title="fig:",scaledwidth=45.0% ] phase diagram using @xmath130 as reference density . curves and lines : theoretical predictions ( eqs . ( [ eq : ustar ] ) to ( [ eq : isomorph_scaling_w ] ) ) for three isochores , an isomorph , and an isotherm . data points : simulations of the 12 - 6 kablj mixture . the parameters of the equation of state were found from the two state points indicated by circles , giving @xmath131 . by comparison , the corresponding parameters found by fitting to the @xmath130 isochore in figs . [ fig : uwischores ] and [ fig : urosenfeld ] are @xmath132 , illustrating the robustness of the fitted equation of state . applying the fitting procedure to the simulated state points on the @xmath133 isotherm gives the parameters ( 38.5 , 5.19 , -8.63 , 2.61 ) . lower panel : same data as above in a @xmath134 diagram . , title="fig:",scaledwidth=45.0% ] equations ( [ eq : isomorphu ] ) and ( [ eq : isomorphw ] ) give the potential energy and the virial as functions of density along an isomorph . the temperature variation along the isomorph can be found by integrating eq . ( [ eq : gammasex ] ) . since @xmath53 generally changes only little ( fig . [ fig : gammasex ] ) , in many situations it is a good approximation to assume it constant , leading to the relation @xmath135 along an isomorph , where @xmath136 is the temperature at the reference point . taken together eqs . ( [ eq : isomorphu ] ) , ( [ eq : isomorphw ] ) , and ( [ eq : gammasex ] ) thus imply that from a single reference point @xmath137 we have a prediction for @xmath138 on the isomorph to which the reference point belongs . figure [ fig : uwischores ] demonstrated that in the @xmath3 phase diagram , isochores to a good approximation are straight lines with slope @xmath53 , i.e. , that one can write @xmath139 . the only ingredient missing to generate a full equation of state expressed as @xmath140 and @xmath141 is a relation between temperature and potential energy or virial along an isochore . rosenfeld and tarazona @xcite derived from density functional theory an expression for the potential energy on an isochore : @xmath142 rosenfeld and tarazona noted that the expression `` ... provides a good estimate of the thermal energy ( ... ) near freezing densities only for predominant repulsive interactions '' , and confirmed their expression by simulations of ipl potentials with different exponents . equation ( [ eq : urosenfeld ] ) has since been shown , however , to be an excellent approximation also for several models with attraction , including the kablj mixture.@xcite this was regarded as a bit of a mystery , but can now be understood as a consequence of the fact that dynamics and heat capacity of strongly correlating liquids are well reproduced by a corresponding ipl system@xcite a consequence of the hidden scale invariance discussed briefly in the introduction ( see papers i - iii ) . figure [ fig : urosenfeld ] tests the rosenfeld - tarazona prediction for the systems investigated in fig . [ fig : uwischores ] . the prediction works very well , but less so for the supercooled wahnstrm blj system , for which the formation of extended `` frank - kasper '' clusters at low temperatures affects the temperature dependence of the potential energy . @xcite combining eq . ( [ eq : urosenfeld ] ) with our results on isomorphs and isochores we can construct an equation of state for generalized lj systems . the idea is to map any state point @xmath138 to the corresponding isomorphic state point @xmath137 on a reference isochore ( where it is implecitely understood that the parameters ( @xmath143 ) are evaluated at the reference isochore @xmath144 ) : @xmath145 from ( @xmath146 ) we can calculate the contribution to the potential energy from the two ipl terms of the potential ( compare eqs . ( [ eq : um ] ) and([eq : un ] ) ) : @xmath147 finally , we use isomorphic scaling ( eqs . ( [ eq : isomorphu ] ) and ( [ eq : isomorphw ] ) ) to go back to the @xmath138 state point : @xmath148 the new equation of state given by eqs . ( [ eq : umseos ] ) to ( [ eq : isomorph_scaling_w ] ) contains four parameters ( @xmath143 ) evaluated at a reference density @xmath144 . the parameters can be calculated from @xmath149 , and @xmath150 at a single state point . a more convenient way of estimating the parameters might be : for a collection of @xmath138 state points , at least two of which are non - isomorphic , use isomorphic scaling ( eqs . ( [ eq : isomorphu])and ( [ eq : isomorphw ] ) ) to get the corresponding @xmath101 at the chosen reference density , and fit them to eq . ( [ eq : wisochor ] ) to determine @xmath151 and @xmath53 . next , using the fitted value for @xmath53 , fit to eq . ( [ eq : ustar ] ) in order to find @xmath152 and @xmath153 . to put the new equation of state to a test , the procedure described above was used to determine the four parameters ( @xmath143 ) for the 12 - 6 kablj system using merely two state points . the resulting equation of state is shown as full lines in fig . [ fig : uwstate ] , where it is compared to simulation results shown as data points . the agreement is good . the largest deviations are seen when changing density away from the reference density , most evident for the isotherm in the @xmath3 diagram and the isomorph in the @xmath134 diagram . better agreement is expected if the density dependence of @xmath53 is taken into account , but such a correction will not be attempted here . to summarize , combining the isomorph theory with two further approximations , a state - point independent exponent @xmath53 and the rosenfeld - tarazona expression , a realistic equation of state with four parameters is arrived at . paper iv demonstrated that strongly correlating liquids have isomorphs , i.e. , curves in the phase diagram along which structure and dynamics to a good approximation are invariant in reduced units . the main new results in this paper are predictions specific to generalized lennard - jones systems and the supporting numerical evidence . starting from the invariance of structure along isomorphs , a prediction for the shape of these in the @xmath3 phase diagram was presented and shown to agree well with simulation results . it was shown that for a given system , the isomorphs all have the same shape in the @xmath3 phase diagram ; they are simply scaled versions of each other . furthermore , the isomorph shape depends only on the exponents @xmath19 and @xmath31 , i.e. , all systems with , say , 12 - 6 lj interactions have same isomorphs in the @xmath3 phase diagram . what differs from system to system is the values of the density and temperature along an isomorph . throughout papers i - v the @xmath3 phase diagram was referred to extensively . this particular phase diagram has not been used much before , but it is evident that for strongly correlating liquids several features stand out when presented in the @xmath3 phase diagram . prominent examples are that isochores are straight lines ( with slope @xmath53 ) and the existence of the `` master '' isomorph discussed above for generalized lj systems . the class of strongly correlating liquids includes , we believe , all van der waals liquids and metallic liquids ( but excludes hydrogen - bonding , covalent , as well as strongly ionic liquids ) . more simulations , as well as experiments , are needed to substantiate that the class of strongly correlating liquids is this large , but preliminary simulations of simple molecular models like@xcite the asymmetric dumbbell and the lewis - wahnstrom otp three - site model are encouraging . in particular , the isomorph properties of more complex molecule models need to be simulated , as well as theoretically contemplated ; the scaling properties of molecules with fixed bond lengths are not trivial . it should be noted that even a quite complex system like a phospholipid membrane can be strongly correlating with respect to its slow degrees of freedom,@xcite the interactions of which are dominated by van der waals forces . another interesting question is to study in detail the properties of liquids where the exponent @xmath53 varies significantly throughout the phase diagram , for instance the weeks - chandler - andersen cutoff variant of the generalized binary lj liquid that was recently investigated om detail by coslovich and roland , @xcite as well as berthier and tarjus.@xcite another system that deserves further investigation is the 200 - 100 lj system@xcite which we find to be an exception to the general observation that generalized lj systems are strongly correlating presumably as a consequence of its extremely narrow range of interactions . this series of papers investigated in depth the properties of strongly correlating liquids . we believe to have demonstrated that strongly correlating liquids are simpler than liquids in general . hopefully this insight will turn out to be useful in furthering the understanding of the properties of liquids in general . acknowledgments + urp is supported by the danish council for independent research in natural sciences . the center for viscous liquid dynamics `` glass and time '' is sponsored by the danish national research foundation ( dnrf ) . unless stated otherwise , simulations are performed using a newly developed molecular dynamics code optimized for nvidia graphics cards . the code is available as open source at http://rumd.org . potentials were cut and shifted at @xmath154 . reported potentials and virials do not include contributions beyond the cut - off . simulations reported in figs . 7 and 8 were performed using gromacs ; h. j. c. berendsen , d , van der spoel , and r. van drunen , comp . comm . * 91 * , 43 ( 1995 ) ; e. lindahl , b. hess , and d. van der spoel , j. mol . mod . * 7 * , 306 ( 2001 ) .

this series of papers is devoted to identifying and explaining the properties of strongly correlating liquids , i.e. , liquids with more than 90% correlation between their virial @xmath0 and potential energy @xmath1 fluctuations in the @xmath2 ensemble . paper iv [ n. gnan _ et al . _ , j. chem . phys . * 131 * , 234504 ( 2009 ) ] showed that strongly correlating liquids have `` isomorphs '' , which are curves in the phase diagram along which structure , dynamics , and some thermodynamic properties are invariant in reduced units . in the present paper , using the fact that reduced - unit radial distribution functions are isomorph invariant , we derive an expression for the shapes of isomorphs in the @xmath3 phase diagram of generalized lennard - jones systems of one or more types of particles . the isomorph shape depends only on the lennard - jones exponents ; thus all isomorphs of standard lennard - jones systems ( with exponents 12 and 6 ) can be scaled onto to a single curve . two applications are given . one is testing the prediction that the solid - liquid coexistence curve follows an isomorph by comparing to recent simulations by ahmed and sadus [ j. chem . phys . * 131 * , 174504 ( 2009 ) ] . excellent agreement is found on the liquid side of the coexistence , whereas the agreement is worse on the solid side . a second application is the derivation of an approximate equation of state for generalized lennard - jones systems by combining the isomorph theory with the rosenfeld - tarazona expression for the temperature dependence of potential energy on isochores . it is shown that the new equation of state agrees well with simulations .

plasma tails of comets and their time variations potentially provide crucial information on solar winds and magnetic fields in the solar system ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? short - time variations in plasma tails , however , are not yet fully understood . indeed , most previous studies observed tails and structures at far distances ( @xmath3 km from the nucleus ) with a time resolution of an order of an hour . regarding the speed of movement along the tail , @xcite studied 72 disconnection events ( des ) of various comet tails and found @xmath4 100 km s@xmath2 at @xmath5 km from the nuclei . their initial speeds before des are around 44 km s@xmath2 and the typical acceleration is 21 cm s@xmath6 . @xcite analyzed a knot in the plasma tail of comet 1p / halley and derived its average velocity of 58 km s@xmath2 at 49 @xmath1 km from the nucleus . @xcite observed c/1996 b2(hyakutake ) and measured the speed of a knot of 99.2 km s@xmath2 at 5.0 @xmath7 km from the nucleus . @xcite investigated de of c/1995 o1 ( hale - bopp ) and obtained the speed of @xmath0 500 km s@xmath2 at @xmath0 7@xmath8 km from the nucleus . @xcite analyzed several knots in comets c/2001 q4(neat ) and c/2002 t7(linear ) using the solar mass ejection imager . they found the speed to be 50100 km s@xmath2 around @xmath9 km from the nucleus . these previous studies did not catch the moment immediately after the formation of knots or the detachment of knots from the tail . the initial speed at these critical times was only an extrapolation from later observations relatively far away . in this letter , we report detections of knots in the plasma tail 3@xmath1 km away from the nucleus of c/2013 r1(lovejoy ) and a direct measurement of their initial motions . we adopt the ab magnitude system throughout the paper . the comet was observed on 2013 december 4 ( ut ) using the subaru prime focus camera ( suprime - cam ; * ? ? ? * ) mounted on the subaru telescope at mauna kea ( observatory code 568 ) . the camera consists of a 5 @xmath10 2 array of 2k @xmath10 4k ccds . the pixel scale is 0.202 arcsec pixel@xmath2 . the field of view is about 35@xmath1028 arcmin . we used two broadband filters : w - c - ic ( i - band : center=7970 , full width at half maximum ( fwhm)=1400 ) and w - j - v ( v - band : center=5470 , fwhm=970 ) filters . both bands trace the plasma tail . i - band includes predominantly h@xmath11o@xmath12 line emissions , while v - band includes co@xmath12 and h@xmath11o@xmath12 line emissions . the observation log is given in table [ tab : obslog ] . the total observing time of 24 minutes was spent after main science targets of the observing run were set . the start time of the exposures have an uncertainty of about 1 second . the position angle and the pointing offset were adopted to catch the comet nucleus at the bottom - left corner and to have the tail run diagonally across the field - of - view so that the maximum extent of the tail is framed in each exposure . the subaru telescope s non - sidereal tracking mode@xcite was used so that the comet was always observed at the same position on the ccd array . for the observing run , the comet s coordinates were calculated using the nasa / jpl horizons system with the orbit element of jpl#22 . we used an ephemeris of one - minute step . unfortunately , the values of jpl#22 were not recorded . in the following , we instead used newer orbital elements ( jpl#55 ) . the values are given in table [ tab : orbit ] . the positions calculated from jpl#22 and jpl#55 have an offset by 0.34 arcsec in right ascension and -1.88 arcsec in declination but show no drift during the observation . the offset of the absolute celestial coordinate does not affect in this study , since our analysis is on relative position of structures inside of the comet . at the time of the observations , the observercentric and heliocentric distance to the comet were 0.55230.5526 and 0.88120.8811 au , respectively . in the sky projection , the conversion from the angular to the physical scales was 400.6400.8 km arcsec@xmath2 . considering the phase angle ( sun - target - observer angle ) of 83.5 degrees at the time of the observations , we adopt the physical scale along the tail of 403.3 km arcsec@xmath2 in the following discussion . the heliocentric ecliptic coordinate of the comet nucleus was ( @xmath13,@xmath14)=(87.5,30.7 ) . the comet was located before the perihelion passage , and its heliocentric velocity was -12.6 km s@xmath2 . the seeing size is estimated from short ( two - second ) exposures and was 1.0 and 1.1 arcsec in i and v - bands , respectively . the movement of the comet in the celestial coordinate was about ( dra@xmath10cos(d)/dt , d(d)/dt)=(283 284,-115 ) arcsec hr@xmath2 during the observations . due to the non - sidereal tracking , stars move in an exposure , but this motion does not affect the measurement of the seeing size significantly since the shift is only about 0.17 arcsec in a two - second exposure . the data was reduced in a standard manner ; the steps include overscan subtraction , crosstalk correction @xcite , flat fielding using twilight flat , and distortion correction . the relative flux and relative position among the ccds are calibrated using other dithered datasets taken in the same night . the mosaicked image of v1 ( table [ tab : obslog ] ) is shown as figure [ fig : v1 ] as an example . the flux was calibrated against stars in the field using the eighth data release of the sloan digital sky survey ( sdss dr8 ; * ? ? ? * ) catalog in the same way as in @xcite . we first used sextractor @xcite for object detection in the suprime - cam images and astrometric calibration was performed to the center of the elongated stars against the guide star catalog 2.3.3 @xcite . then , we measured aperture fluxes at each position of sdss star whose r - band magnitude is 16@xmath15r@xmath1520 . the aperture radius was 40 , 50 , and 60 pixels , and we used @xmath16\ ] ] as the background - corrected aperture flux of an elongated star , where @xmath17 is the aperture flux of @xmath18 radius . photometric zero point was estimated from the instrument flux @xmath19 and the catalog magnitude converted to the suprime - cam ab system . the color conversion coefficients from sdss to suprime - cam magnitudes are given in table [ tab : colorcoeff ] . the k - correct @xcite v4 offset for sdss is applied ; @xmath20=0.012 , 0.010 and 0.028 for @xmath21 , @xmath18 and @xmath22 , respectively . the number of stars used for calibration was @xmath23 in v and @xmath24 in i. the measured photometric zero - point of chip 2 ( reference chip ) was 27.36(i ) and 27.15(v ) ab mag per 1 adu s@xmath2 , and the peak - to - peak variation among exposures were smaller than 0.02 mag in each band . these zero points are comparable to the ones derived typically under photometric conditions , and thus we regard the data as having been obtained under photometric conditions . to investigate the fine structures in the plasma tail , additional image processing was applied . first , we rotate the image counterclockwise by 50 degrees so that the plasma tail aligns to the y - axis , and the region around the plasma tail within full width of 1000 pixels ( 3.4 arcmin ) was extracted . the position angle of the image to the north is the same among the nine images ( 160.0 degree ) . in the coordinate , the sun lay in the direction of ( x , y)=(-0.014,+1.000 ) . the nuclei position was measured in short exposures ( i2 and v2 ) . the relative positional consistency among the images depends on the accuracy of the non - sidereal tracking mode of the subaru telescope . as we used the ephemeris of a one - minute step , enough accuracy is guaranteed . we then applied an unsharp masking technique to each image ; we convolved the extracted images with various sizes of gaussian filter ( @xmath25=10 , 20 , 30 , 50 , 75 pixels ) and subtracted the smoothed image from the original ones . the details are presented in appendix a. this processing removes larger structures and enhances fine structures . to differentiate the processing with different kernel sizes , we refer to each as highpass@xmath26 , e.g. , highpass10 when processed with the gaussian with @xmath25=10 pixels . bright stars in the background contaminate small features associated physically with the cometary tail ( e.g. , knots ) , and hence , we iteratively masked them . finally , we binned the image by 5@xmath105 pixels ( 1.01 arcsec square ) . figure [ fig : knots ] shows six longer - exposure images after the processing . these images show the gaps between ccd chips as gray bands . brightening / darkening around chip edges are artifacts . these do not affect the following analyses . we can visually see two knots moved downstream around the distance of @xmath0 3@xmath1 km from the nucleus ( indicated by arrows in figure [ fig : knots ] ) . the closeup images are shown as figure [ fig : closeup ] . one of the knots first appears to be connected to the global filamentary structures running along the tail . it later becomes detached from the structures . the positions of the knots are measured by running sextractor @xcite on the 5@xmath105 binned unsharp masked images . we measured in highpass30 , highpass50 , highpass70 , and highpass90 images . as examined in appendix b , an unsharp masking with a smaller kernel may introduce larger positional errors . meanwhile , the knot may not be detected after an unsharp masking with a large kernel because it is buried in global features . we therefore adopted the position from the largest kernel in which the peak of the knot is detected . the root mean squares ( rms ) of the position was @xmath27 km . distances from the nucleus ( along the tail ; y - axis ) and offsets from the tail axis ( perpendicular to the tail ; x - axis ) are plotted as a function of time in figure [ fig : speed ] . 20 and 25 km s@xmath2 along the tail and 3.8 and 2.2 km s@xmath2 across the tail , respectively . the motion should make the shape of the knots elongated in a two - minute exposure by 67 arcsec along the tail and 1.3 or 0.7 arcsec across the tail . since the size of the knots is at least 1015 arcsec in diameter , elongation by the motion had little effect on the size estimation . these knot motions show tilts from the central axis of the tail by about 0.10.2 radians ( 611 degrees ) . we note that the direction to the sun is 0.014 radians ( 0.8 degrees ) from the axis , and therefore , the knot motions are not in a perfect alignment to the direction to the sun . the distances to the knots from the nucleus change almost linearly with time in figure [ fig : speed ] . the effect of acceleration is therefore small in the short duration of our observations ( 23 minutes ) . the clear detection of the spatial motions , however , indicates the potential for direct measurements of the acceleration in future . if we can take the accelerations of previous measurements of other comets by @xcite ( 21 cm s@xmath6 ) , the expected change in the speed is only 0.3 km s@xmath2 in 23 minutes . if we adopt 17 cm s@xmath6 , which is calculated from figure 4 of @xcite as 2.4 km s@xmath2 increment is seen in four hours , the expected change in the speed during our observation is even smaller ( 0.2 km s@xmath2 ) . therefore , a sequence of few hours observations of a knot would permit us to measure the acceleration . we found that the width of the plasma tail also rapidly changed . in the analysis , the original images before the unsharp masking were used . in figure [ fig : itail ] and figure [ fig : width ] show spatial surface brightness profiles across the tail and their time variation . since we have measured the motions of the knots ( presumably the motion of the tail ) , we can track the variation in the tail width at a comoving position as the tail would flow at the speed of @xmath0 22 km s@xmath2 . in a sequence of images ( i1 , i3 , and i4 ) , we measured a profile at the distance of 6@xmath1010@xmath28 km in i1 . we then tracked the location of the material initially at this distance , but at later times using the flow speed , and plotted the profiles of ( presumably ) the same material in i3 and i4 . the tail width is determined at the surface brightness of 21.3 ab magnitude arcsec@xmath6 in i - band . it narrowed from 2.4@xmath29 to 2.0@xmath29 , and then to 1.8@xmath29 km for i1 , i3 , and i4 , respectively . from this measurement , the speed of the narrowing motion across the tail is @xmath0 8 km s@xmath2 at each edge . the center of the light distribution , meanwhile , did not show any notable change ( figure [ fig : width ] ) . this narrowing speed is larger than those of the knots in the same direction perpendicular to the tail ( 3.8 and 2.2 km s@xmath2 ; section [ sec : knots ] ) . this apparent difference could be attributed to a projection of the motion of the knots in the sky . if we assume every fine structure ( filamentary structure ) in the tail is moving at @xmath0 8 km s@xmath2 perpendicular to the tail , the amount of spatial shift would be about 2.5 arcsec within the 120-second exposures . on the other hand , the fine structures appear to be a typical 5 10 arc sec width in short exposures ( i2 , v2 ) . therefore , the blur due to the motions during a single 120-second exposure should not affect the positional measurements of the structures much . the initial speeds of two moving knots ( @xmath0 22 km s@xmath2 ) are significantly slower than the ones measured by @xcite ( 44 km s@xmath2 ; rms of 10.9 km s@xmath2 ) . this speed is also smaller than that measured in comet halley by @xcite ( 58 km s@xmath2 ) , who suggested that the velocity was constant at 49 @xmath1 km from the nucleus . though it is not clear what the dominant factor of the initial speed of knots is , we can compare several parameters of the comets . as the data used by @xcite include information from various comets , we compare parameters with the single case of the comet halley by @xcite . at the observation by @xcite , the heliocentric distance to comet halley was 1.016 au , the heliocentric ecliptic coordinate was ( @xmath13,@xmath14)=(30.1,8.8 ) , and the heliocentric velocity was -26.5 km s@xmath2 . compared with c/2013 r1(lovejoy ) in this study , a part of the difference in the initial speed may be explained by the difference in the heliocentric velocity of the nucleus , -12.6 km s@xmath2 versus -26.5 km s@xmath2 , if we assume that the initial speed of the knots might be comparable in heliocentric frame . it , however , does not fully explain @xmath0 40 km s@xmath2 difference . another difference is heliocentric ecliptic latitude , 30.7 vs 8.8 , which may result in the difference in the speed of the solar wind at the comet position . yet another point is that we have compared the speeds of our relatively faint and small knots with more prominent knots / kinks and des in the previous studies . the relevance of this comparison might be debated in light of future studies . in addition , we analyzed only one comet tail observed in a relatively short duration . a more systematic investigation is obviously needed as to the distribution of the initial speed of the tail as a function of the heliocentric velocity , the heliocentric distance , and the ecliptic position of the comet . in summary , we found two knots that were just formed at 3@xmath1 km from the nucleus of c/2013 r1(lovejoy ) . their initial speed was smaller than the ones measured in previous studies , and a physical interpretation requires a more statistically significant sample at various heliocentric positions . we also found a rapid variation in the tail width in seven minutes , which implies a rapid change in ambient solar winds and magnetic field . these results strongly suggest that the variations in comet plasma tails , especially in their fine structures , require high time resolution observations with a large aperture telescope such as subaru . this work is based on observations obtained with the subaru telescope . we acknowledge david thilker , alexandre y. k. bouquin , fumiaki nakata , and yutaka komiyama for their help . we thank miriam forman for suggestive comments . we also thank anonymous referee for useful comments . this work has made use of sdss dr8 database , horizons ephemerides service in nasa / jpl , and the computer systems at astronomical data analysis center of naoj . jk is supported by the nsf through grant ast-1211680 and by nasa through grants nnx09af40 g , nnx14af74 g , a herschel space observatory grant , and a hubble space telescope grant .

we report short - time variations in the plasma tail of c/2013 r1(lovejoy ) . a series of short ( two to three minutes ) exposure images with the 8.2-m subaru telescope shows faint details of filaments and their motions over 24 minutes observing duration . we identified rapid movements of two knots in the plasma tail near the nucleus ( @xmath0 3@xmath1 km ) . their speeds are 20 and 25 km s@xmath2 along the tail and 3.8 and 2.2 km s@xmath2 across it , respectively . these measurements set a constraint on an acceleration model of plasma tail and knots as they set the initial speed just after their formation . we also found a rapid narrowing of the tail . after correcting the motion along the tail , the narrowing speed is estimated to be @xmath0 8 km s@xmath2 . these rapid motions suggest the need for high time - resolution studies of comet plasma tails with a large telescope .

this paper concerns the problem of the high probability state transfer ( hpst ) @xcite between different nodes of the spin 1/2 chain , which becomes a popular problem due to the development of the quantum communication systems and quantum computing . by `` state transfer '' we mean the following phenomenon @xcite . consider the chain of spins 1/2 with dipole - dipole interactions in the strong external magnetic field . let all spins be directed along the external magnetic field except the @xmath6th one whose initial state is arbitrary , @xmath7 , where @xmath8 and @xmath9 mean the spin directed along and opposite the external magnetic field respectively . let the energy of the ground state ( all spins are aligned along the magnetic field ) be zero . if the state of @xmath10th node becomes @xmath11 with @xmath12 at time moment @xmath13 then we say that the state has been transfered from the @xmath6th to the @xmath10th node with the phase shift @xmath14 . since @xmath15 , all other spins are directed along the field at @xmath16 , i.e. their states are @xmath17 . function @xmath18 is the transition amplitude of an excited state @xmath19 from the @xmath6th to the @xmath10th node . note that if all nodes of the chain have equal larmor frequencies and we are interested in the state propagation between two nodes , say between @xmath20th to @xmath21th nodes , then the shift @xmath22 may be simply removed by the proper choice of the constant magnetic field value @xcite , so that @xmath23 , i.e. the state is perfectly transfered . there exists a wide literature studying the state transfer along the spin chains in the strong external magnetic field . for instance , propagation of the spin waves in homogeneous chains was considered in @xcite . it was demonstrated that the state may be perfectly transfered between two end nodes @xcite as well as between two symmetrical inner nodes @xcite in the inhomogeneous chain . state propagation along the alternating chains was studied in @xcite . it was shown in @xcite that the chains with week end bonds provide the state transfer from one to another side of the chain . end - to - end entanglement in both alternating chains and chains with weak end bonds has been studied in@xcite . some aspects of the entanglement between remote nodes of the chain have been studied in @xcite . however , all these references consider the state transfer between two end nodes ( or between two symmetrical nodes ) which is required for the construction of the communication channels where the state must be transfered from one object to another . meanwhile , the quantum computation requires such systems which have hpsts among many different nodes and , as a consequence , may distribute information among these nodes . such systems may be candidates for the quantum register . emphasise that , as it was indicated above , the excited state may be transfered from the initial @xmath6th node to some @xmath10th node with proper phase shift @xmath24 . however , we will show that all these shifts may be removed in a simple way introducing the time dependent magnetic field , see sec.[section : xyz ] ( remember that the single phase shift can be removed by a constant magnetic field , like it was done in the case of the state propagation between two nodes @xcite ) . thus , in general , the phase shifts @xmath24 do not create serious obstacles for the quantum communications . the only problem is the organization of the state transfers with big values of @xmath25 . a simple variant of such systems is suggested in our paper . namely , we construct the chain of @xmath1 nodes which has set @xmath26 of @xmath27 nodes @xmath28 , @xmath29 , @xmath30 with the hpsts between any two of them , i.e. if the unknown state @xmath31 is generated in any particular node @xmath32 from the set @xmath26 ( while the initial states of all other nodes are @xmath8 ) then this state may be detected with high probability in any other node @xmath33 from the set @xmath26 after appropriate time intervals . hereafter we will use the notation @xmath34 for the chain shown in fig.[fig : gen ] . here @xmath35 and @xmath36 are the numbers of nodes in @xmath37 and @xmath38 respectively . let us clarify the structure of this scheme . each of the chains @xmath37 , @xmath39 , allows the hpst between its end nodes @xmath40 and @xmath41 . chains @xmath42 collect all inner nodes of @xmath37 and consequently have @xmath43 nodes . two chains @xmath37 and @xmath44 are connected by the `` week bond '' through the chain @xmath38 , @xmath45 . by `` week bond '' between @xmath37 and @xmath44 we mean the following necessary inequality among the coupling constants : @xmath46 emphasize that , also the hpst is organized between end nodes of each particular chain @xmath37 ( taken out of the general chain ) , the whole chain @xmath47 does not provide the hpsts between all @xmath28 ( @xmath48 ) in general case . however , we are interested in the particular form of the chain @xmath47 which does provide the hpst among all nodes @xmath28 . first of all , such chain must be symmetrical and may be written as ( @xmath49 ) @xmath50 where @xmath51 , @xmath52 , see fig.[fig : d1 ] . let the chain have @xmath1 nodes . we use notations @xmath53 , @xmath54 and @xmath55 for the probability of the exited state transfer between @xmath56th and @xmath57th nodes , for the time interval required for this transfer and for the phase shift of the transfered exited state , @xmath58 : @xmath59 due to the symmetry of the chain , we have the following identities : @xmath60 because of the wide spread of the coupling constants , the time interval @xmath61 needed for the state transfer between @xmath56 and @xmath57 significantly depends on the values @xmath32 and @xmath33 . in general , @xmath62 i.e. the state may be transfered between two nodes much faster if both nodes are placed in the same half of the chain . thus , an important characteristic of such chain is the interval @xmath63 hereafter the state transfer between the nodes @xmath56 and @xmath57 will be referred to as hpst if @xmath64 the value @xmath65 is conventional . we take @xmath66 in examples of sec.[section:4n ] and in example 1 of sec.[section : def ] and @xmath67 in examples of sec.[section:8n ] and in example 2 of sec.[section : def ] . the set of all possible hpsts between any two nodes from the list @xmath26 will be referred to as hpst@xmath68 , where @xmath1 is the total number of nodes in the chain . we call the parameters of hpst@xmath68 in such chain the set of parameters @xmath69 since the organization of the hpsts among different nodes of the set of @xmath70 nodes is an essential property of the quantum register , the chains constructed in this paper may be candidates for this role . while the nodes @xmath28 may serve as the q - bits of the quantum register , the chains @xmath42 ( and @xmath38 ) serve to decrease the time intervals @xmath71 ( and @xmath72 ) required for the state transfer between the nodes @xmath40 and @xmath41 ( and between the nodes @xmath41 and @xmath73 ) separated by the long distance as it happens in the communication channels . namely , if @xmath37 consists of two nodes @xmath40 and @xmath41 then the time interval @xmath74 may be reduced putting additional chain @xmath42 with properly adjusted coupling constants between these two nodes . similarly , the time interval @xmath72 required to transfer the excited state between the last node of @xmath37 ( i.e. node @xmath41 ) and the first node of @xmath44 ( i.e. node @xmath73 ) may be reduced putting chain @xmath38 with properly adjusted coupling constants between @xmath37 and @xmath44 , see , for instance , chain @xmath75 in fig.[fig : d6_4n_mod_ex ] . thus the above spin chain @xmath76 combines properties of both quantum register and communication channel . finally we note that , constructing the spin chain , we want @xmath77 this paper is organized as follows . in sec.[section : xyz ] we show that @xmath78 matrix representation of the xxz hamiltonian may be reduced to the @xmath79 matrix representation for a certain type of problems . sec.[section:4n ] describes the general structures of the spin chains having the set @xmath26 of four nodes ; simple examples are represented . similar study of the eight node chain is given in sec.[section:8n ] . deformations of the above chains decreasing parameter @xmath80 are discussed in sec.[section : def ] . conclusions are given in sec.[section : conclusions ] . we study the hpsts @xcite among nodes of the spin 1/2 chain in strong external magnetic field @xmath82 with the @xmath81 hamiltonian @xmath83 where @xmath84 is the gyromagnetic ratio , @xmath85 is the distance between @xmath6th and @xmath10th spins , @xmath86 is the projection operator of the @xmath6th spin on the @xmath87 axis , @xmath88 , @xmath89 is the @xmath90-projection operator of the total spin , @xmath91 are the spin - spin coupling constants . this hamiltonian describes the secular part of the dipole - dipole interaction in the strong external magnetic field @xcite . for our convenience , we call @xmath92 , @xmath93 . it is obvious that all @xmath94 may be arbitrary by definition . although the approximation of the hamiltonian ( [ hamiltonian ] ) by the nearest neighbour interaction is very popular in study of the quantum state transfer , one can show that it is not applicable to the inhomogeneous chains which have a wide spread of coupling constants . in fact , this approximation is applicable if @xmath95 , @xmath96 @xmath97 . only in this case we may disregard terms with coupling constants @xmath91 ( @xmath97 ) in the hamiltonian . however , this is not possible in general . consider , for instance , the chain in fig.[fig : chain ] . in this chain , the approximation by the nearest neighbour interaction takes into account the term with coupling constant @xmath98 and disregards the term with coupling constant @xmath99 in the hamiltonian , i.e. we neglect the term which is bigger then the term which is taken into account . @xmath100 the approximation by the nearest neighbour interaction requires , in particular , @xmath101 however , we see that the second condition is not satisfied for the case shown in fig.[fig : chain ] . comparison of the coupling constants in the chains considered in secs.[section:4n]-[section : def ] confirms that the approximation of the hamiltonian by the nearest neighbour interaction is not applicable to these chains . thus , hereafter we consider the total hamiltonian ( [ hamiltonian ] ) . it is convenient to take the eigenvectors of @xmath89 as the basis of the matrix representation of the hamiltonian ( [ hamiltonian ] ) . this is possible since the hamiltonian ( [ hamiltonian ] ) commutes with @xmath89 : @xmath102=0,\end{aligned}\ ] ] so that both @xmath103 and @xmath89 have the common set of eigenvectors . in general the dimensionality of the matrix representation of the hamiltonian is @xmath78 . as usual , we write the eigenvectors of the operator @xmath89 in terms of the dirac notations . let latexmath:[\ ] ] where @xmath114 is @xmath79 identity matrix . thus , instead of @xmath78 matrix representation of the hamiltonian we have @xmath79 block @xmath112 and scalar block @xmath111 , which significantly reduces all calculations . it was shown @xcite that effectiveness of the quantum communication channel may be measured by the fidelities of the state transfers between @xmath56th and @xmath57th nodes : @xmath115 where the amplitudes @xmath116 and the phases @xmath117 are defined as follows ( @xmath118 ) : @xmath119 here @xmath120 , @xmath121 , are components of the eigenvector @xmath122 corresponding to the eigenvalue @xmath123 of the matrix @xmath124 : @xmath125 . the effectiveness of the state transfer is characterized by the set of parameters @xmath126 , @xmath118 . it is evident , that these parameters take maximum values if @xmath127 which may be considered as the system of equation defining ( not uniquely ) the time dependence @xmath128 ( which must be positive over the interval @xmath129 ) : @xmath130 then the values of the fidelities @xmath131 are defined by @xmath132 or , equivalently , by the probabilities @xmath53 . for this reason , namely probabilities @xmath53 will be studied in the rest of this paper . phases @xmath133 ( which are independent on @xmath128 ) will be taken as parameters of the hpst@xmath68 instead of the parameters @xmath134 in the set ( [ parametershpst ] ) , see tables 1 - 6 . remark that , deriving eqs.([f_nm]-[varphi ] ) we took into account that the energy of the ground state @xmath135 is not zero : @xmath136 . an explicite example of the function @xmath128 constructed in accordance with eqs.([gamma_b ] ) with @xmath137 will be given in sec.[section:4n ] , see example 1 . [ section:4n ] the hpsts from the end node to ( some of ) the inner nodes of the spin chain becomes possible due to the following mechanism . let us take two identical chains @xmath138 which allow the hpst between their end nodes . let @xmath139 be the minimal coupling constant between the nearest neighbours in these chains . we connect these chains by the weak bond with the coupling constant @xmath140 , resulting in the chain @xmath141 of @xmath142 nodes , see fig.[fig : d2_4n ] . then @xmath143 these hpsts may be understood as follows . let spin @xmath144 be directed opposite to the external magnetic field while all other spins of the chain @xmath141 be directed along the field initially . due to the small coupling constant @xmath145 this excited state remains inside of the first chain @xmath146 for the long time with the high probability to be detected in either @xmath144 or @xmath147 ( since , by definition , the chain @xmath146 provides hpst between end nodes ) . however , due to the bond between two chains @xmath146 , the excited state will be transfered to the second chain @xmath138 after comparatively long time interval @xmath148 with the high probability . if this happens , than the excited state remains in this chain for the long time with high probability to be detected in either @xmath149 or @xmath150 . similarly , after one more time interval @xmath148 the excited state will return to the first chain @xmath138 with the high probability , and so on . parameter @xmath145 may be fixed by the two conditions ( [ condition1 ] ) and ( [ condition2 ] ) . hereafter in this section we take @xmath66 in the definition ( [ hpst ] ) . consider two simple examples . * example 1 : the hpsts in the chain @xmath151 . * the simplest example of the chain @xmath146 providing the ideal end - to - end state transfer is the chain of two nodes , i.e. @xmath152 . connecting two equivalent chains @xmath138 we obtain the chain @xmath151 with @xmath137 , see fig.[fig : d3_4n_ex1 ] . in this case the set @xmath26 consists of all nodes of the chain @xmath151 . we set @xmath153 for simplicity and vary @xmath154 in order to satisfy the conditions ( [ condition1 ] ) and ( [ condition2 ] ) . we have found that the optimal parameters of the hpst(4;1,2,3,4 ) correspond to @xmath155 . these parameters are represented in table 1 , see also fig.[fig : n4 ] . let us find the time evolution of the external magnetic field which satisfies conditions ( [ gamma_b ] ) , i.e such @xmath128 that @xmath156 we may write conditions ( [ gamma0_2 ] ) as the following system of five equations : @xmath157 which is satisfied by the following function @xmath128 : @xmath158 we see that the function @xmath128 constructed in this way is positive over the interval @xmath159 . * example 2 : the hpsts in the chain @xmath160 . * we take two homogeneous chains of three nodes @xmath161 . it is known that the ideal state transfer is possible between the end nodes of these chains @xcite . we connect them by the weak bond obtaining the chain shown in fig.[fig : d3_4n_ex2 ] . thus , @xmath162 . let @xmath163 , @xmath164 in this chain . we vary @xmath165 to obtain the best correspondence to the conditions ( [ condition1 ] ) and ( [ condition2 ] ) . the optimal value is @xmath166 . the appropriate parameters of the hpst(6;1,3,4,6 ) are represented in table 2 . [ section:4n_mod ] the chain shown in fig.[fig : d2_4n ] is convenient for the state transfer between two chains @xmath167 if only the distance between them ( i.e. between @xmath147 and @xmath149 ) is not too long . otherwise the time interval @xmath168 becomes very long . to decrease this time interval we suggest the following modification of the chain @xmath141 . let us take a symmetrical chain @xmath169 of @xmath170 nodes with maximal coupling constant between neighbours @xmath171 satisfying the following condition : @xmath172 . using the coupling constant @xmath173 we may construct the chain @xmath174 of @xmath175 nodes shown in fig.[fig : d5_4n_mod ] . here @xmath26 consists of the end nodes of chains @xmath167 and may not involve any node of @xmath176 . this statement is valid due to the fact , that the probability for the spin to be detected in the chain @xmath169 may not be high because of the small coupling constants both inside of this chain and @xmath177 . these coupling constants should be fixed by the conditions ( [ condition1 ] ) and ( [ condition2 ] ) . * example : the hpsts in the chain @xmath178 . * we consider the hpst(6;1,2,5,6 ) in the chain @xmath178 shown in fig.[fig : d6_4n_mod_ex ] . this is a chain of 6 nodes ( @xmath179 ) . we fix @xmath180 and vary parameters @xmath181 and @xmath164 with the purpose to decrease the parameter @xmath182 and satisfy the conditions ( [ condition1],[condition2 ] ) . we have found that the following values of the parameters @xmath183 yield a good result : @xmath184 and @xmath185 . the appropriate parameters of the hpst(6;1,2,5,6 ) are represented in table 3 . here we demonstrate that the intermediate chain @xmath75 in the chain @xmath178 speeds up the state transfer between @xmath147 and @xmath149 separated by the distance @xmath186 . for this purpose we compare the parameters @xmath187 for the chain @xmath151 ( see fig.[fig : d3_4n_ex1 ] ) with @xmath188 and parameter @xmath189 from the table 3 . numerical simulation shows that @xmath190 , i.e. @xmath191 . [ section:8n ] chains @xmath141 and @xmath174 considered in sec.[section:4n ] provide the hpsts between any two nodes out of the set @xmath26 consisting of four nodes . however , the number of such nodes may be increased using the following obvious generalization of these chains . let us take two chains @xmath192 and @xmath193 ( see fig.[fig : gen ] ) and the symmetrical chain @xmath194 . the maximal coupling constant @xmath195 between the nearest neighbours in @xmath194 must satisfy inequality @xmath196 , where @xmath139 is the minimal coupling constant between the nearest neighbours in @xmath197 , @xmath198 . using the coupling constant @xmath199 , @xmath200 , we construct the chain of @xmath201 nodes shown in fig.[fig:8n ] . here @xmath26 consists of the eight end nodes of chains @xmath167 and @xmath202 : @xmath203 , @xmath204 . it is obvious that this algorithm may be extended to construct chains with @xmath26 consisting of @xmath0 nodes @xmath52 .. consider the simplest example where we take @xmath67 ( see eq.([hpst ] ) ) . * example : the hpsts in the chain @xmath205 . * we consider the chain of eight nodes as one obtained by joining of two 4-nodes chains @xmath151 constructed in example 1 of sec.[section:4n ] , see fig.[fig:8n_ex1 ] where @xmath206 . to provide the hpsts among all nodes we must take a small coupling constant @xmath207 between these two chains . namely , it must be less than @xmath155 introduced in example 1 of sec.[section:4n ] . we take @xmath208 , @xmath209 , @xmath210 , @xmath211 . the disadvantages of this chain are big parameter @xmath212 and comparatively small values of @xmath213 , @xmath214 ( @xmath215 ) . after optimization we obtain @xmath216 . the parameters of the hpst(8;1,2,3,4,5,6,7,8 ) are represented in table 4 . it may be shown that the parameters of the hpsts may be improved varying all coupling constants in the chains @xmath174 and @xmath217 in a proper way , i.e. varying not only the coupling constants between chains @xmath138 ( like it was done in secs.[section : ch1 ] and [ section : ch3 ] ) but also the coupling constants inside of @xmath138 . this allows one to decrease significantly parameter @xmath80 . after such process the above chains will be reduced to the `` optimized '' chains which we call @xmath218 and @xmath219 respectively ( compare the parameters from tables 2 and 5 and from tables 4 and 6 ) . * example 1 : the hpsts in the chain @xmath220 , @xmath66 . * the purpose of this section is to decrease the parameter @xmath182 which have been found in example 2 of sec.[section:4n ] , see table 2 and fig.[fig : d3_4n_ex2 ] . we take @xmath180 , @xmath181 , @xmath164 and vary @xmath183 , @xmath198 keeping in mind the conditions ( [ condition1],[condition2 ] ) . we have found that the optimal parameters of the hpst(6;1,3,4,6 ) correspond to the @xmath221 and @xmath222 , see table 5 and fig.[fig : d3_4n_ex2 ] . * example 2 : the hpsts in the chain @xmath223 , @xmath67 . * the parameter @xmath212 obtained in example 1 of sec.[section:8n ] may be decreased varying the coupling constants in the chain , see fig.[fig:8n_ex1 ] . thus , the optimal parameters of the hpst(8;1,2,3,4,5,6,7,8 ) have been found for @xmath224 , @xmath225 , @xmath226 , see table 6 and fig.[fig:8n_ex1 ] . [ cols="<,<,<,<,<,<,<,<",options="header " , ] [ table : hpst(8;8hom ) ] using the numerical simulations we demonstrate that the @xmath1-nodes spin 1/2 chains with wide spread of the properly adjusted coupling constants allow the hpsts between different nodes . it is important that the whole hamiltonian ( [ hamiltonian ] ) rather then approximation by the nearest neighbour interaction must be used for the correct description of the hpsts in such chains . we have found that two spin 1/2 chains @xmath138 with the hpst between end nodes may be connected by a relatively weak bond to get a chain with the hpsts among four nodes , see sec.[section:4n ] . in turn , having two such chains we may connect them by another weak bond to get a chain with the hpsts among eight nodes ( see sec.[section:8n ] ) , and so on . formally , the number of the nodes allowing the hpsts among all of them may be @xmath0 , where @xmath52 . however , the disadvantage of such chains is a rapid increase of the time interval @xmath80 with the number of nodes involved in the hpsts . the mechanism decreasing @xmath80 would be important for the implementation of these chains . we also demonstrate that the speedup of the state transfer between the nodes @xmath28 and @xmath227 separated by the distance @xmath228 may be achieved using the intermediate chain with properly adjusted coupling constants , see sec.[section:4n_mod ] .

this paper concerns the problem of the high probability state transfer among @xmath0 symmetrically placed nodes of the @xmath1-nodes spins 1/2 chain with the @xmath2 hamiltonian . we consider examples with @xmath3 , @xmath4 and @xmath5 . , spin dynamics , quantum state transfer , spin chain , ideal state transfer , high - probability state transfer

the `` nodal defect '' of an eigenfunction of a schrdinger operator is closely related to the difference between the upper bound on the number of nodal domains given by courant s theorem and the number of nodal domains . in the recent paper @xcite , gregory berkolaiko proves a nice formula for the nodal defect of an eigenfunction of a schrdinger operator on a finite graph in terms of the morse index of the corresponding eigenvalue as a function of a magnetic deformation of the operator . his proof remains mysterious and rather indirect . in order to get a better understanding in view of possible generalizations , it is desirable to have a more direct approach . this is what we do here . after reviewing our notation , we summarize the main result and give an informal description of the proof in section [ sec : statement ] . the proof itself is implemented in sections [ sec : qf ] and [ sec : hess ] with an alternative view provided in appendix [ sec : pedestrian ] . the continuous schrdinger operator on a circle is considered in appendix [ sec : hills ] and various special cases and further ideas are explored in other appendices . let @xmath0 be a finite connected graph where @xmath1 is the set of vertices and @xmath2 the set of unoriented edges . we denote by @xmath3 the edge linking the vertices @xmath4 and @xmath5 . we denote by @xmath6 the set of oriented edges and by @xmath7 $ ] the edge from @xmath4 to @xmath5 ; the set @xmath6 is a 2-fold cover of @xmath2 . a 1-form @xmath8 on @xmath9 is a map @xmath10 such that @xmath11)=-\ga([x , y])$ ] for all @xmath12 . we denote by @xmath13 the vector space of dimension @xmath14 of 1-forms on @xmath9 . the operator @xmath15 is defined by @xmath16)=f(y)-f(x)$ ] . if @xmath17 is a non - degenerate , not necessarily positive , quadratic form on @xmath13 , we denote by @xmath18 the adjoint of @xmath19 where @xmath20 carries the canonical euclidean structure and @xmath13 is equipped with the symmetric inner product @xmath21 associated to @xmath17 . we have @xmath22 where @xmath23 is the dimension of the space of cycles of @xmath9 . we will show later that , in our context , we have the hodge decomposition @xmath24 where both spaces are @xmath21-orthogonal . following @xcite , we denote by @xmath25 the set of @xmath26 real symmetric matrices @xmath27 which satisfy @xmath28 if @xmath29 and @xmath30 if @xmath31 and @xmath32 . note that the diagonal entries of @xmath27 are arbitrary . an element @xmath27 of @xmath25 is called a _ schrdinger operator _ on the graph @xmath9 . it will be useful to write the quadratic form associated to @xmath27 as @xmath33 with @xmath34 . a _ magnetic field _ on @xmath9 is a map @xmath35 defined by @xmath36)= e^{i \ga _ { x , y}}$ ] where @xmath7 \mapsto \ga _ { x , y}$ ] is a 1-form on @xmath9 . we denote by @xmath37 the manifold of magnetic fields on @xmath9 . the magnetic schrdinger operator @xmath38 associated to @xmath39 and @xmath40 is defined by the quadratic form @xmath41\in \vec{e } } h_{x , y}| f(x)-e^{i\ga_{x , y}}f(y)|^2 + \sum_{x\in x } v_x |f(x)|^2\ ] ] associated to a hermitian form on @xmath42 . more explicitly , if @xmath43 , @xmath44 we fix @xmath27 and we denote by @xmath45 the eigenvalues of @xmath38 . it will be important to notice that @xmath46 . moreover , we have a gauge invariance : the operators @xmath38 and @xmath47 with @xmath48 for some @xmath49 are unitarily equivalent . hence they have the same eigenvalues . this implies that , if @xmath50 ( this is not always the case because @xmath17 is not positive ) , it is enough to consider @xmath51forms in the subspace @xmath52 of @xmath13 when studying the map @xmath53 . this holds in particular for investigations concerning the hessian and the morse index . before stating the main result , we recall the [ def : index ] the morse index @xmath54 of a quadratic form @xmath55 on a real vector space @xmath2 is defined by @xmath56 where @xmath57 is a subspace of @xmath2 so that @xmath58 is @xmath59 . the nullity of @xmath55 is the dimension of the kernel of @xmath55 . the morse index of a smooth real - valued function @xmath60 defined on a smooth manifold @xmath61 at a critical point @xmath62 ( i.e. a point satisfying @xmath63 ) is the morse index of the hessian of @xmath60 , which is a canonically defined quadratic form on the tangent space @xmath64 . the critical point @xmath65 is called non - degenerate if the previous hessian is non - degenerate . the nullity of the critical point @xmath65 of @xmath60 is the nullity of the hessian of @xmath60 at the point @xmath65 . the aim of this note is to prove the following nice results due to berkolaiko @xcite : [ theo : main ] let @xmath0 be a finite connected graph and @xmath66 the dimension of the space of cycles of @xmath9 . we suppose that the @xmath67-th eigenvalue @xmath68 of @xmath69 is simple . we assume moreover that an associated non - zero eigenfunction @xmath70 satisfies @xmath71 for all @xmath72 . then , the number @xmath73 of edges along which @xmath74 changes sign satisfies @xmath75 . moreover @xmath76 is smooth at @xmath77 which is a critical point of @xmath78 and the nodal defect , @xmath79 is the morse index of @xmath80 at that point . if @xmath61 is the manifold of dimension @xmath81 of magnetic fields on @xmath9 modulo the gauge transforms , the function @xmath82 \ra \lambda _ n ( b)$ ] has @xmath83 $ ] as a non - degenerate critical point . the previous results can be extended by replacing the critical point @xmath84 by @xmath85 for all edges @xmath86 . the number @xmath73 is then the number of edges @xmath87 satisfying @xmath88 where @xmath74 is the corresponding eigenfunction . the assumptions on @xmath27 are satisfied for @xmath27 in an open dense subset of @xmath25 . the upper bound of @xmath89 in the first part of theorem [ theo : main ] is related to courant nodal theorem ( see @xcite section vi.6 ) as follows : a nodal domain on a graph for the eigenfunction @xmath74 is a connected component of the sub - graph @xmath90 of @xmath9 obtained by removing the edges along which @xmath74 changes sign . denoting by @xmath91 the number of nodal domains of @xmath74 , the courant theorem for graphs ( see @xcite , theorem 2.4 ) asserts that @xmath92 ; using euler formula for the graph @xmath90 and because @xmath93 , the number of connected components of the graph @xmath90 , we get also a lower bound ( see @xcite ) : under the assumptions of theorem [ theo : main ] , we have @xmath94 * important warning : * _ without loss of generality , we can and will assume in the rest of this note that @xmath95 . this implies that the morse index of @xmath96 is @xmath97 . _ in the course of the proof we will use a special choice of gauge in which we can compute the hessian explicitly . more precisely , according to the classical perturbation formulae , @xmath98 where we assumed that @xmath99 is at a critical point : @xmath100 . the first term is easy to calculate explicitly ; for perturbation in the direction of the 1-form @xmath101 it is @xmath102)^2 ~{\rm with } ~ a_{x , y}=-h_{x , y}\phi_n(x)\phi_n(y)=a_{y , x}~.\ ] ] considered as a quadratic form in @xmath101 , @xmath17 is already in the diagonal form . its index is clearly the number of negative values among @xmath103 , or , in other words , the number @xmath89 of edges where @xmath74 changes sign ! we will present an explicit choice of gauge in which the second term vanishes . the condition for this is @xmath104 which , after explicit calculation , can be interpreted as @xmath105 , where @xmath106 is the conjugate of @xmath19 with respect to the inner product induced by . finally , we observe that the index of @xmath107 has been computed to be @xmath89 in the whole of @xmath108 , whereas we should be restricting ourselves to our chosen gauge , @xmath105 . we will show that this restriction reduces the index precisely by @xmath97 . indeed , the splitting @xmath109 is orthogonal with respect to the form @xmath17 , therefore @xmath110 we establish that @xmath111 by relating the form @xmath17 on @xmath112 to the quadratic form @xmath96 around the point @xmath74 . the set of forms @xmath113 where @xmath114 , the set of subsets with two elements of @xmath1 , and @xmath115 with @xmath116 is a basis of the set of quadratic forms on @xmath20 . a quadratic form @xmath55 on @xmath117 is said of laplace type if @xmath118 where @xmath119 is the symmetric bi - linear form associated to @xmath55 . [ lemm : lapl ] the set of forms @xmath120 is a basis of the space of quadratic forms of laplace type . the form @xmath121 , where @xmath122 is the point - wise product of @xmath74 and @xmath60 , is of laplace type because @xmath123 hence @xmath124 . moreover , @xmath125 . indeed , because of lemma [ lemm : lapl ] , it is enough to compare the coefficients of the basis forms @xmath126 . the form @xmath127 is already expanded in this basis . to find the coefficient for the form @xmath128 , we observe that ( because we know it is of laplace type ) the coefficient in question is minus the coefficient in front of the term @xmath129 , divided by two . this evaluates to @xmath130 ( see equation ( [ equ : axy ] ) ) . in fact , we will need to use @xmath131 . the morse index of @xmath132 is equal to @xmath97 . it is a general fact that the morse index of the quadratic form @xmath133 is the same as the morse index of the restriction of @xmath17 to the image of @xmath134 . hence , the morse index of @xmath132 is the morse index of @xmath135 on @xmath20 . because @xmath136 is a linear isomorphism , this index is equal to the index of @xmath96 by sylvester theorem . since @xmath137 , the index of @xmath96 is @xmath97 by elementary spectral theory . [ lemm : hodge ] let us denote by @xmath18 the adjoint of @xmath19 where @xmath20 is equipped with the canonical euclidean structure and @xmath138 with the inner product associated to @xmath17 . the space @xmath138 splits as @xmath139 ( hodge type splitting ) , and this decomposition is @xmath17-orthogonal . more explicitly @xmath18 is given by @xmath140)~.\ ] ] if @xmath141 satisfies @xmath142 , we have @xmath143 . hence @xmath144 for all @xmath145 and @xmath146 . because @xmath147 is of multiplicity @xmath148 , this implies that @xmath60 is constant and hence @xmath149 . so @xmath150 and the conclusions follow . at this point , we know that the nodal defect is the morse index of the restriction of @xmath17 to the space @xmath151 of dimension @xmath66 . the first part of the theorem follows . we need one more fact to complete the proof : to identify the hessian of @xmath78 on @xmath152 at @xmath84 with the restriction of @xmath17 to @xmath151 . let us denote by @xmath153 the set of unit vectors @xmath60 normalized so that @xmath154 is real and @xmath155 where @xmath65 is chosen in @xmath156 . the point @xmath84 is a critical point of @xmath78 . if @xmath157 is the eigenfunction of @xmath38 corresponding to the eigenvalue @xmath158 , the differential of @xmath159 vanishes at @xmath84 on @xmath151 . the first property comes from the fact that @xmath160 . we can compute , for any variation @xmath161 , @xmath162 close to @xmath163 , of @xmath84 , @xmath164 . the condition @xmath165 can be written as @xmath166 for all @xmath116 . from equation ( [ equ : magn ] ) , this is equivalent to @xmath167 . hence @xmath168 and @xmath169 since @xmath170 is simple . from the normalization @xmath171 , we get @xmath172 and , since @xmath173 , the number @xmath174 is real . we deduce that @xmath175 . [ lemm : hess ] the function @xmath176 defined by @xmath177 admits @xmath178 as a critical point and the hessian of @xmath179 at the point @xmath84 is the form @xmath17 . the differential of @xmath57 with respect to @xmath60 vanishes because @xmath60 is an eigenfunction of @xmath27 . the differential with respect to @xmath151 vanishes , because @xmath180 . the hessian of @xmath57 at @xmath181 is well defined . because the differential at @xmath182 of @xmath183 vanishes on @xmath152 , the hessians of @xmath184 and @xmath185 agree . a simple calculation of the hessian of @xmath186 gives the result : @xmath187\in\vec{e } } h_{x , y}| \phi_n ( x)-e^{i\ga _ { x , y}}\phi_n ( y)|^2 + \sum _ { x\in x}v_x |\phi_n(x)|^2=\ ] ] @xmath188\in { e}}h_{x , y } \left ( \phi_n(x)^2 + \phi_n(y)^2 -2\cos \ga _ { x , y}\phi_n(x)\phi_n(y)\right)+\sum _ { x\in x}v_x |\phi_n(x)|^2 .\ ] ] computing the second derivative with respect to @xmath189 at @xmath190 gives @xmath191 . we will derive a direct approach to the calculus of the second derivative of an eigenvalue which could be used directly in the proof of lemma [ lemm : hess ] . let @xmath192 be a @xmath193 curve defined near @xmath194 in the space of hermitian matrices on a finite dimensional hilbert space @xmath195 . let us assume that @xmath196 is an eigenvalue of @xmath197 of multiplicity one with a normalized eigenvector @xmath198 . then , for @xmath162 close to @xmath163 , @xmath199 has a simple eigenvalue @xmath200 of multiplicity one which is a @xmath193 function of @xmath162 . we can choose an associated eigenfunction @xmath201 which is @xmath193 with respect to @xmath162 . the following assertions give the values of the first and second derivatives of @xmath200 at @xmath194 : [ prop : deriv ] under the previous assumptions , we have @xmath202 if @xmath203 , we have @xmath204 where @xmath205 is any solution of @xmath206 . in particular , if @xmath207 , @xmath208 we start with @xmath209 where @xmath201 is an eigenfunction of @xmath199 which depends in a @xmath193 way of @xmath162 . taking the first derivative , we get @xmath210 putting @xmath194 and taking the scalar product with @xmath211 , we get the formula for @xmath212 . similarly , the @xmath162-derivative of equation ( [ equ : first ] ) is @xmath213 pouting @xmath194 , taking the scalar product with @xmath211 and using @xmath203 , we get the result . we can apply this to @xmath214 with @xmath215 in order to get the hessian of @xmath78 in section [ sec : hess ] . the condition @xmath216 is exactly @xmath165 ! in this appendix , we will describe the case of a schrdinger operator on the circle , also called the hill s operator . this is the simplest continuous case , but it may be useful to do it with some details in order to try to extend the method to higher dimensional manifolds . the hill s operator is @xmath217 where @xmath218 is a smooth , @xmath148-periodic , function . the spectral theory of hill s operators has been well studied ; in particular , the inverse spectral theory for this operator allows to solve non - linear evolution equations , like the korteweg - de vries one . a presentation of the properties of hill s operators is given in @xcite . the following facts are known : if we denote by @xmath219 the spectra of @xmath27 acting on periodic ( resp anti - periodic ) functions of period @xmath148 , we have the inequalities @xmath220 and the spectrum of @xmath27 on @xmath221 is then union of intervals , called the bands , @xmath222\cup [ \gl_2 ^ - , \gl_2^+]\cup [ \gl_3^+,\gl_3 ^ -]\cup \cdots ~.\ ] ] these statements are linked to the properties of the _ discriminant _ @xmath223 : if @xmath224 and @xmath225 are the _ normalized solutions _ of @xmath226 whose cauchy data are @xmath227 , the discriminant @xmath228 is the entire function given by @xmath229 . the spectrum of @xmath27 on @xmath221 is the set of real @xmath230 s so that @xmath231 . the periodic ( resp . anti - periodic ) spectra are given by @xmath232 ( resp . @xmath233 ) . the function @xmath234 is a regularization of @xmath235 in the spirit of @xcite . it is proved in @xcite , section ii , that , if @xmath236 is simple , @xmath237 and the sign of this derivative is that of @xmath238 . we will assume that @xmath239 is equal to @xmath163 and is a simple eigenvalue of @xmath27 acting on @xmath148-periodic functions . up to gauge transform , every magnetic potential on the circle is a constant @xmath240 . the bands are linked to the addition of a magnetic field as follows : the @xmath67-th band is the image of the circle @xmath241 by the map @xmath242 where @xmath243 is the @xmath67-th eigenvalue of @xmath244 which is @xmath27 acting on functions @xmath60 so that @xmath245 . in particular , if @xmath67 is even , @xmath236 is a maximum of @xmath242 while if @xmath67 is odd , @xmath236 is a minimum of @xmath242 . this fits with berkolaiko s formula because the ( even ! ) number of zeros of the corresponding periodic eigenfunction @xmath74 is @xmath246 if @xmath67 is even and @xmath247 if @xmath67 is odd ( see @xcite theorem 2.14 ) . in this appendix , we will use the general formula for the second derivative in order to reprove this result and to show that the critical points are non - degenerate . a direct computation of @xmath248 using the discriminant works as follows : the spectrum of @xmath244 is given by @xmath249 . near @xmath250 , we have @xmath251 . this gives @xmath252 , hence the morse index of @xmath242 at @xmath190 is @xmath163 if @xmath67 is odd and @xmath148 is @xmath67 is even . we will denote with a `` dot '' the derivatives w.r . to @xmath240 and by a `` prime '' the derivatives w.r . to @xmath4 . the operator @xmath244 is unitarily equivalent to @xmath253 acting on @xmath148-periodic functions . we have @xmath254 the derivatives of @xmath255 w.r . to @xmath8 at @xmath190 are @xmath256 and @xmath257 . applying proposition [ prop : deriv ] and denoting by @xmath258 a corresponding normalized eigenfunction , we get @xmath259 moreover @xmath260 . let us denote by @xmath261 the function @xmath262 . then , using the method of `` variation of parameters '' ( i.e. making the ansatz @xmath263 with @xmath264 ) , we get @xmath265 where the constant @xmath266 is chosen so that @xmath267 is periodic and @xmath268 is an arbitrary constant which can be fixed by a normalization of @xmath74 . we can always assume that @xmath269 by shifting the origin of @xmath270 to some zero of @xmath74 . using the wronskian , we see that @xmath271 . we have to check the derivatives : @xmath272 or @xmath273 . this gives , using equation ( [ equ : phin ] ) , @xmath274 we get @xmath275\phi_n'(x)dx ~.\ ] ] by integration by parts , we have @xmath276 . moreover , again by integration by parts , @xmath277 and , since the wronskian @xmath278 is constant and @xmath279 , @xmath280 . we get @xmath281 moreover , it follows from equation ( 2.13 ) , page 16 in @xcite and the fact that @xmath282 , that this is exactly @xmath283 . in this appendix , we take @xmath284 and assume that @xmath137 is non - degenerate eigenvalue of @xmath27 with a normalized eigenfunction @xmath285 . we have the [ prop : var ] let us assume that , for all vertices @xmath4 satisfying @xmath286 , there exists a vertex @xmath287 so that @xmath288 . then , for any @xmath289 orthogonal to @xmath285 , there exists a smooth deformation @xmath290 of @xmath27 so that @xmath291 . it is enough to check that the space of @xmath292 is @xmath20 and to use the first variation formulae given in appendix a. let us assume that the function @xmath293 vanishes at the unique vertex @xmath65 . then , the nullity of the hessian of the magnetic variation of @xmath27 is at least @xmath294 where @xmath295 is the number of vertices @xmath296 so that @xmath297 . choose a smooth variation @xmath298 of @xmath27 so that @xmath299 . let @xmath73 be the number of sign changes of @xmath293 away of @xmath65 . then , for @xmath300 small enough , the number of sign change of @xmath301 is @xmath302 while , for @xmath303 small enough , it is @xmath304 . we see from theorem [ theo : main ] that the magnetic morse index is @xmath305 for @xmath300 and @xmath306 . the discontinuity of the morse index at @xmath194 is @xmath294 . this gives the lower bond on the nullity . if @xmath307 , the eigenvalue @xmath163 is degenerate . let us remark that this lower bound is not always sharp . in the following example , we have @xmath308 , @xmath309 and the nullity of the hessian is @xmath310 . the graph @xmath9 is made of 2 cycles of length 3 with a common vertex . the matrix of @xmath27 is chosen as follows : @xmath311=-\left(\begin{array}{ccccc}1&1&1&0&0 \\ 1&1&2&0&0 \\ 1&2&1&1&2 \\ 0&0&1&1&1 \\ 0&0&2&1&1 \end{array}\right)\ ] ] using the fact that the graph has a symmetry of order 2 exchanging the 2 cycles , one can split @xmath117 and the matrix @xmath27 into the even and odd parts . this allows to check that @xmath312 is non - degenerate . in order to compute the magnetic hessian , we check that it is possible to build a decomposition @xmath313 which is @xmath17 orthogonal and with @xmath314 . it is then easy to check that the magnetic hessian evaluated on @xmath315 vanishes . let @xmath316 be a bipartite graph , with @xmath317 and all edges have a vertex in @xmath318 , the other in @xmath319 . let @xmath320 be the involution on @xmath321 given by @xmath322 if @xmath323 and @xmath324 if @xmath325 and let @xmath326 be a magnetic field . then @xmath327 with @xmath328 . so that @xmath329 . an hence it follows form the diamagnetic inequality that @xmath330 has a maximum at @xmath84 . and hence the morse index of the hessian of @xmath330 at @xmath84 is the dimension of the manifold of magnetic fields namely @xmath81 . on the other hand the first eigenfunction @xmath331 of @xmath332 is everywhere @xmath333 and the number of sign changes of @xmath334 is @xmath335 . so berkolaiko s formula for @xmath336 gives @xmath337 . this is the euler formula . let us assume that we are in the discrete case and the eigenvalue we consider is @xmath338 . then we have @xmath339 where @xmath340 is the product of the eigenvalues @xmath341 for @xmath342 . we have @xmath343 . hence the index of @xmath344 is the same as the index of @xmath345 . there is a formula for the characteristic polynomial of a magnetic laplacian on graphs due to robin forman @xcite and reproved by kenyon @xcite and burman @xcite . using the gauge change @xmath346 as in my paper gives a laplace type operator whose entries can be of any sign . forman s formula extends to that case and it would be nice to get berkolaiko s formula form forman s formula .

bluein this note , we present a natural proof of a recent and surprising result of gregory berkolaiko interpreting the courant nodal defect as a morse index . this proof is inspired by a nice paper of miroslav fiedler published in 1975 .

conjugated organic materials have a variety of applications . they are already being used in organic light emitting diodes ( oleds ) and organic thin - film transistors.@xcite they are also considered to be potential candidates for single - molecule electronic and spintronic devices . the issue of spin and charge transport in @xmath3-conjugated organic systems , therefore , is of prime importance . thus , it is both important and necessary to understand the role of chemical modifications ( substituents ) on the dynamics of spin and charge transport in these systems . electrons in these molecular materials experience strong electron correlations by virtue of reduced dimensionality and these interactions are also long - ranged since the systems are semiconducting . therefore , to study transport of charge and spin in these systems , it is necessary to have appropriate models as well as techniques . dynamics in interacting systems can be studied either in the frequency or in the real - time domain . the dynamics of strongly interacting systems in the frequency domain has been possible due to the lanczos and correction vector techniques @xcite using both exact diagonalization method for small systems , and the density matrix renormalization group ( dmrg ) method for large systems . studying dynamics in the real - time domain throws a lot more light into the transport mechanism . exact time dependent techniques are possible only for small interacting systems . the advent of the time - dependent density matrix renormalization group ( td - dmrg ) method has allowed the study of _ out - of - equilibrium _ dynamics in large low - dimensional strongly correlated systems.@xcite recently , we have investigated non - equilibrium dynamics of spin and charge transport in unsubstituted polyenes , which are a class of @xmath3-conjugated organic materials , using the _ double time window targeting _ ( dtwt ) td - dmrg technique developed by us.@xcite in the present paper we extend our studies to address non - equilibrium spin and charge dynamics in the _ push - pull _ polyenes , characterized by the presence of an electron donating ( push ) group , and an electron accepting ( pull ) group , schematically represented as @xmath0 . presence of the donor and acceptor groups polarizes the @xmath3-electron bridge such that there is partial charge transfer from the donor to the acceptor group . the electronic structure of a push - pull polyene can be described by two canonical valence bond ( vb ) structures , namely , a neutral polyene structure and a zwitterionic structure , also known as charge transfer(ct ) configuration , where an electron is transferred from _ a_.@xcite this leads to the naive expectation that these groups will have significant influence on spin - charge dynamics . the _ push - pull _ polyenes have been so far studied mostly in the context of nonlinear optical response.@xcite in this paper we report our time - dependent dmrg studies on the spin and charge transport in push - pull polyenes and compare these results with those on polymethine imines which are quasi one - dimensional systems with alternate c and n atoms in the conjugation backbone . the organization of the paper is as follows : in the next section we provide details about the model hamiltonian and the computational method used in this study . in sec . iii we present our results with discussions . section iv concludes the paper . the pariser - parr - pople ( ppp ) hamiltonian @xcite with dimerization and site energies is appropriate for describing the low energy physics of @xmath3-conjugated systems . the ppp model hamiltonian is given by @xmath4 ( \hat{c}^{\dagger}_{i,\sigma}\hat{c}_{i+1,\sigma } + \text{h.c.})\\ & + \sum_{i=1}^{n } \epsilon_{i}\hat{n}_{i } + \sum_{i=1}^{n } \frac{u_{i}}{2}\hat{n}_{i}(\hat{n}_{i}-1 ) \\ & + \sum_{i > j } v_{ij } ( \hat{n}_{i}-z_{i})(\hat{n_{j}}-z_{j } ) . \end{split}\ ] ] here , @xmath5 is the number of carbon atoms in the polyene chain , @xmath6 ( @xmath7 ) creates ( annihilates ) an electron with spin orientation @xmath8 in the @xmath2 orbital of the @xmath9 carbon atom , @xmath10 is the average transfer integral and , @xmath11 ( 0 @xmath12 @xmath11 @xmath12 1 ) is the bond alternation parameter . the orbital energy and on - site coulomb repulsion of the @xmath2 orbital on the @xmath9 carbon atom are given by @xmath13 and @xmath14 , respectively and @xmath15 is the number operator on the @xmath9 site . @xmath16 is the inter - site coulomb repulsion between sites @xmath17 and @xmath18 , and @xmath19 is the on - site chemical potential at the @xmath9 site . in case of unsubstituted polyene systems,@xcite @xmath14 = 11.26 ev , @xmath13 = 0 and @xmath19 = 1 , for all sites , @xmath10 = @xmath202.4 ev and @xmath11 = 0.07 . the intersite interaction between electrons , @xmath16 , is interpolated according to the ohno scheme,@xcite between _ u _ for @xmath21 = 0 and @xmath22 for @xmath23 as , @xmath24^{-1/2}.\ ] ] we have used single - bond length of 1.495 , double - bond length of 1.299 , and a bond angle of 120@xmath25 between successive bonds . these parameters have correctly predicted a whole range of properties of the low - lying states of conjugated systems in general and polyenes in particular.@xcite when push and pull groups are introduced , we assume that only the energies of those @xmath2 orbitals to which the push and pull groups are attached , change due to inductive effect and all other parameters of the model remain unchanged . the donor group raises the orbital energy of the first carbon atom to which it is attached by @xmath26 , while the acceptor group lowers the orbital energy of the last carbon atom to which it is attached by @xmath27 , where @xmath28 and @xmath29 @xmath30 0 . we assume that @xmath28 = @xmath27 ( symmetric push - pull polyenes ) . we have studied push - pull polyenes of 30 and 40 carbon atoms with terminal push and pull groups , and have varied the push - pull strength @xmath31 . although , presence of the push and pull groups destroys both the electron - hole and the inversion symmetry , the total spin invariance of the hamiltonian , remains preserved . for studying the dynamics of spin and charge transport , an initial wave packet @xmath32 is constructed by annihilating an up spin electron from the first site of a push - pull polyene of @xmath5 sites , in the ground state , @xmath33 , @xmath34 the wave packet @xmath32 evolves under the influence of the ppp hamiltonian [ eq . ( 1 ) ] following the time - dependent schr"odinger equation and temporal dependence of site charge density @xmath35 , and site spin density @xmath36 , of this wave packet can be computed as , @xmath37 here , @xmath38 is the wave packet at time @xmath39 . real - time dynamics of the initial wave packet is studied using the dtwt scheme,@xcite with the following specifications : number of density matrix eigenvectors ( dmevs ) retained , @xmath40 = 300 , time - step for evolution @xmath41 = 0.0066 fs , total evolution time @xmath42 = 33.0 fs , the number of time steps in each window is kept at 130 which corresponds to 0.858 fs , and the number of windows is 39 . although we compute charge ( spin ) density at all the sites , we focus only on the quantities @xmath43 and @xmath44 and , @xmath45 and @xmath46 at the terminal sites attached to the substituents as these are sufficient for our purpose . another class of donor - acceptor substituted systems examined are the polymethine imines ( pmi ) , with the molecular formula @xmath1.@xcite these systems have alternately donor ( c ) and acceptor ( n ) atoms in conjugation . the bonding in this polymer corresponds to @xmath47 and both carbon and nitrogen atoms are in @xmath48 hybridization . this system has been studied extensively for linear and non - linear optical properties . the nitrogen 2@xmath2 orbitals are lower in energy than the carbon 2@xmath2 orbitals and intra - orbital repulsion of the nitrogen 2@xmath2 orbitals are greater than that of the carbon 2@xmath2 orbitals , both these can be rationalized on the basis of the more compact 2@xmath49 orbitals in nitrogen compared to carbon . the transfer integrals for @xmath50 and @xmath51 are @xmath202.767 ev and @xmath202.317 ev , respectively ; @xmath50 and @xmath51 bond lengths are 1.273 and 1.425 ; @xmath52 = 11.26 ev , @xmath53 = 12.34 ev ; @xmath54 = 0.0 ev , @xmath55 = @xmath202.96 ev . the initial wave packet is constructed as before [ see eq . ( 3 ) ] and time evolved by the dtwt scheme using a smaller time - step of @xmath41 = 0.0033 fs , necessitated by the larger transfer integrals . + in the absence of source and sink , the charge and spin of the hole decouple ( spin - charge separation ) and oscillate between sites @xmath56 and @xmath57 , with time . hence , the time evolution profiles of @xmath35 and @xmath58 ( @xmath17 = @xmath56 and @xmath57 ) , consist of a series of maxima and minima . the time @xmath59 taken for the charge of the hole , and @xmath60 for the spin of the hole , to propagate from site @xmath56 to site @xmath57 , is represented by the time at which the first major minimum ( dip ) appears in the time evolution profile of @xmath44 and @xmath46 , respectively . time required by the charge degree of freedom , @xmath61 , and spin degree of freedom , @xmath62 , of the hole to travel the round trip from site @xmath56 @xmath63 @xmath57 @xmath63 @xmath56 , is represented by the time at which the second major minima appears in the time evolution profiles of @xmath43 and @xmath45 . from these times , the charge and spin velocities of the injected hole are calculated as , @xmath64 = @xmath65 and @xmath66 = @xmath67 . the dynamics of spin and charge transport in systems with reduced dimensionality and electron - electron interactions , are expected to be different from each other , owing to spin - charge separation . when electron correlations are absent , the charge and spin velocities , @xmath68 and @xmath69 , are equal to the fermi velocity , @xmath70 . from our earlier td - dmrg studies @xcite we have found that the fermi velocities of tight - binding chains of lengths @xmath71 and @xmath72 are 2.63 and 2.72 ( / fs ) respectively , for @xmath11 = 0.0 , and 2.43 and 2.49 ( / fs ) , respectively , for @xmath11 = 0.07 . table i presents a comparison of the ratio of @xmath73 and @xmath74 , for hubbard chains of 30 and 40 sites with different @xmath75 values and @xmath11 = 0.07 . these quantities are also computed for dimerized ppp chains with 30 and 40 sites , with @xmath31 = 0.0 , and compared with those of hubbard chains . as evident from table i , the values of the ratios of @xmath73 and @xmath74 are much higher in the ppp model compared to the hubbard model , due to the presence of long - range electron - electron interactions . l*9c & & & & + @xmath5 & 30 & 40 & 30 & 40 & 30 & 40 & 30 & 40 + & & & & & & & & + @xmath73 & 1.33 & 1.32 & 1.35 & 1.39 & 1.30 & 1.18 & 5.05 & 4.80 + & & & & & & & & + @xmath74 & 0.97 & 0.91 & 0.69 & 0.80 & 0.68 & 0.80 & 2.49 & 2.63 + & & & & & & & & + as with unsubstituted polyenes,@xcite in push - pull polyenes ( fig . 1 ) the spin propagates slower than the charge with @xmath76 @xmath30 @xmath77 ( @xmath78 @xmath79 @xmath80 ) . however , we find that the _ d _ and _ a _ groups have no significant effect on spin and charge velocities in the push - pull polyenes . consequently , the charge and spin velocities in push - pull polyenes remain almost equal to those in unsubstituted polyenes ( table ii ) . the push - pull substitution however , has an effect on the amount of charge transported , as seen from the depths of the minima in the time evolution profiles of @xmath43 and @xmath44 . however , depth of the minima in @xmath45 or @xmath46 do not change ( see fig . 1 ) , implying that the push - pull groups have no influence on the spin degree of freedom of the hole . [ cols="^,^,^,^,^,^,^",options="header " , ] this implies that even though the spin and charge velocities in @xmath0 polyenes remain largely unaffected by the donor - acceptor strengths , the amount of charge transferred in unit time decreases with increase in @xmath31 as seen from the decrease in depth of the minima , when a hole is doped at the donor site . we can expect the opposite of when we dope an electron at the donor site ; more charge will be transported from the donor in these cases as @xmath31 increases . our results indicate that push - pull substitutions do not alter the many - body character of the @xmath0 polyenes , and that spin and charge velocities are controlled by the correlation strength while the magnitude of charge transfer is controlled by the strength of the substituents . to understand in detail , the reason for the inability of the donor and acceptor groups to alter the many - body character of the @xmath0 polyenes , we compute the inverse of the charge gap ( @xmath81 ) and spin gap ( @xmath82 ) of the push - pull polyenes . if @xmath83 , @xmath84 and @xmath85 denote the ground state energies of the @xmath86-particle , @xmath87-particle and @xmath5-particle systems respectively , then the charge gap is given by , @xmath88 similarly , the spin gap is defined as , @xmath89 where , @xmath90 and @xmath91 are the lowest energy states of the @xmath0 polyenes with @xmath92 @xmath93 and @xmath94 , respectively . in fig . 2 , we have plotted the variation in @xmath95 and @xmath96 as a function of the strength of the push - pull groups . it is clearly observed that the donor - acceptor groups fail to alter either of the gaps , as a result of which , the velocities of the spin and charge remain unaffected . + comparing figs . 1 and 3 it is clear that the dynamics of spin and charge transport in the @xmath1 chains is very different from that in the push - pull polyenes , owing to the different @xmath97 values and site energies of carbon and nitrogen atoms . for example , unlike in push - pull polyenes ( where the charge and spin of the injected hole after decoupling , oscillate back and forth between sites @xmath56 and @xmath57 ) in the @xmath1 systems , the oscillations are damped and the spin and charge of the injected hole do not return after reaching the chain end . this is demonstrated by the observation that , although the first dip in the time evolution profiles of @xmath44 and @xmath46 are well defined , there are no well defined second ( major ) minima in the @xmath43 and @xmath45 curves as a function of time . this may be because the hopping of an electron between unlike atoms is not a degenerate resonant tunneling process and hence , is not strictly reversible . in order to investigate the fate of the injected hole , we study the temporal variation of total charge and spin densities in the left and right halves of pmi chains of 30 and 40 sites , for @xmath31 = 0.0 ( fig . the total charge and spin densities for the left ( right ) half are defined as , @xmath98 = @xmath99 @xmath100 , and @xmath101 = @xmath99 @xmath102 ( @xmath57 @xmath103 @xmath104 $ ] , @xmath105 @xmath103 @xmath106 $ ] ) , respectively . these quantities serve as simple but effective probes to understand the motion of charge and spin of the hole , as they travel from the half of the system where charge injection occured to the opposite half of the system . if the time evolution profiles of these observables show significant oscillatory behavior , it implies that the charge and spin propagate back and forth between the ends of the system . however , if the oscillatory behavior is not pronounced and both the observables attain some average value , it signifies that the charge ( spin ) moves in a such a manner that a `` quasi - static '' state is generated in which charge ( spin ) distribution of the system remains unchanged within the time of study . it is observed from fig . 4 that for both the chain lengths , @xmath107 goes from 0.0 to @xmath200.25 while , @xmath108 goes from an initial value of @xmath200.5 to @xmath200.25 . once these quantities have attained the value of @xmath200.25 , they start to oscillate with time about this average value . the charge densities for the left and right halves of the pmi chains are however , found to behave little differently . in the case of the 30 site chain it is observed that , @xmath109 and @xmath110 take the average values of @xmath11114.25 and @xmath11114.75 , while for the longer chain of 40 sites , these quantities oscillate about a mean value of @xmath11119.5 . these data suggest that in the @xmath112 systems , with increase in system size , within the computational time a `` quasi - static '' state is achieved more rapidly . to understand this `` quasi - static '' state in more detail , we plot the time evolution profiles of charge and spin densities on carbon and nitrogen atoms in both halves of unsubstituted pmi chains of 30 and 40 sites ( fig . 5 ) . these quantities are defined as , @xmath113 where @xmath57 @xmath103 @xmath104 $ ] and @xmath105 @xmath103 @xmath106 $ ] . it is observed from figs . 5(ii ) and 5(iv ) that the oscillations in time , of @xmath114 and @xmath115 , are opposite in phase , signifying that spin densities on the carbon and nitrogen atoms ( in both half ) are antiferromagnetically coupled to each other . however , the total spin density carried by carbon atoms in both halves is found to be less than that carried by the nitrogen atoms . furthermore it is seen that with time , both @xmath116 and @xmath117 decrease , while @xmath118 and @xmath119 increase in magnitude , keeping the total spin density in either half at @xmath200.25 . from figs . 5(i ) and 5(iii ) we see that the average value of @xmath120 is less than @xmath121 , and the total charge density carried by nitrogen atoms on the right half of pmi chains is more compared to that in the left half in line with the fact that the nitrogens are the `` acceptor '' atoms , being more electronegative . the opposite is seen for the carbon atoms . the time evolution profiles of total spin densities on carbon and nitrogen in both halves , compared to the total charge densities , are found to be more oscillatory . + all these observations suggest that the `` quasi - static '' state is characterized by charge and spin of the hole distributed between the left and right halves of the chain in almost equal amount , the positive or hole charge predominantly residing on carbon atoms while the spin , on nitrogen atoms . the @xmath5 atoms being more electronegative than the @xmath122 atoms ( @xmath55 = @xmath202.96 ev , @xmath54 = 0.0 ev ) , prefer being electron rich by accommodating more charge density . however , due to large on - site coulomb repulsion ( @xmath53 = 12.34 ev ) , the average site charge on @xmath5 atoms is slightly larger than 1.0 which is observed from the charge density distribution in the neutral ground state and the initial state . on the other hand , due to @xmath52 = 11.26 ev and @xmath54 = 0.0 , the carbon atoms are slightly electron deficient in the @xmath112 system . thus , propagation of the charge and spin degrees of freedom of the hole in this polarized background results in formation of the `` quasi - static '' state , which once formed , prevents both the charge and spin from returning to the site of injection as it requires reversal of polarization and is hence , energetically unfavorable . in unsubstituted pmi chains , @xmath59 and @xmath60 are 1.75 fs and 3.65 fs for @xmath5 = 30 , and 2.01 fs and 6.60 fs for @xmath5 = 40 . we find that the @xmath123 ratios for the @xmath5 = 30 is 2.08 and for @xmath5 = 40 , 3.29 . the @xmath123 ratio for a pmi chain of 30 sites agrees well with that of push - pull polyene with @xmath5 = 30 ( @xmath31 = 0 ) . however , this ratio for @xmath5 = 40 ( @xmath31 = 0 ) is much higher in the pmi chain compared to push - pull polyene of same size . this is opposite to the behavior exhibited by push - pull polyenes , where the @xmath123 ratio decreases with increasing chain length . in both polyenes and polymethine imines , the @xmath123 ratios do not saturate , indicating that the @xmath3-coherence lengths for transport are rather long . it also appears that alternate donor - acceptor sites along the chain enhance the velocities of both charge and spin . we have also studied substituents at the terminal sites in the @xmath112 systems and as with the polyenes , the strength of the push - pull groups do not affect the charge and spin velocities , or the nature of the `` quasi - static '' state . to conclude , our studies show that the velocities of charge and spin transport are not affected by push - pull substituents , although the magnitude of charge transport depends on the push - pull strength . in the related @xmath112 system we find that , due to the underlying polarized structure both the spin and charge transport are affected , and charge travels much faster than spin compared to push - pull systems . furthermore , the transport of charge and spin results in the formation of a `` quasi - static '' state in which the injected hole resides in both halves of the system , almost in equal amount . we also note that in the pmis the finite size effects are very large compared to the push - pull polyenes , suggesting much longer @xmath3-coherence lengths . a. dodabalapur , l. torsi , and h. e. katz , science , * 268 * , 270 ( 1995 ) . a. dodabalapur , h. e. katz , l. torsi , and r. c. haddon , science , * 269 * , 1560 ( 1995 ) . j. h. burroughes , d. d. c. bradeley , a. r. brown , r. n. marks , k. machey , r. h. friend , p. l. burns , and a. b. holmes , nature ( london ) , * 347 * , 539 ( 1990 ) . a. nitzan , and m. a. ratner , science , * 300 * , 1384 ( 2003 ) . k. hallberg , phys . b * 52 * , r9827 ( 1995 ) . s. ramasesha and z. g. soos , chem . 153 * , 171 ( 1988 ) ; s. ramasesha , s. k. pati , h. r. krishnamurthy , z. shuai and j. l. brdas , syn . met . * 85 * , 1019 ( 1997 ) . e. jeckelmann , f. gebhard , and f. h. l. essler , phys . lett . * 85 * , 3910 ( 2000 ) ; e. jeckelmann , phys . rev . b * 66 * , 045114 ( 2002 ) . t. d. khner and s. r. white , phys . b * 60 * , 335 ( 1999 ) . h. g. luo , t.xiang , and x. q. wang , phys . lett . * 91 * , 049701 ( 2003 ) . a. e. feiguin and s. r. white , phys . lett * 93 * , 076401 ( 2004 ) ; a. j. daley , c. kollath , u. schollwck , and g. vidal , j. stat . p04005 ( 2004 ) ; a. e. feiguin and s. r. white , phys . b * 72 * , 020404(r ) ( 2005 ) . u. schollwck , rev . 77 , 259 ( 2005 ) . t. dutta and s. ramasesha , phys . b * 82 * , 035115 ( 2010 ) . t. a. skothem , r. l. elsenvaumer , and j. r. reynolds ( eds ) : _ handbook of conducting polymers _ , ( marcel dekker , inc . , 1998 ) . ( see references therein ) l. t. cheng , w. tam , s. r. marder , a. e. stiegman , g. rikken , and c. w. sprangler , j. phys . , * 95 * , 10643 ( 1991 ) . m. blanchard - desce , j. m. lehn , m. barzoukas , i. ledoux , and j. zyss , chem . phys . , * 181 * , 281 ( 1994 ) ; s. r. marder , l.t . cheng , b. g. tiemann , a. c. friedli , m. blanchard - desce , j. w. perry , and j. skindhoj , science , * 263 * , 511 ( 1994 ) ; m. blanchard - desce , v. bloy , j. m. lehn , c. runser , m. barzoukas , a. fort , and j. zyss , proc . spie bf 2143 , 20 ( 1994 ) ; m. blanchard - besce , c. runser , a. fort , m. barzoukas , j. m. lehn , v. bloy , and v. alain , j. chem . * 199 * , 253 ( 1995 ) . r. pariser and r. g. parr , j. chem . phys . * 21 * , 466 ( 1953 ) ; j. a. pople , trans . . soc . * 49 * , 1375 ( 1953 ) . k. ohno , theor . * 2 * , 219 ( 1964 ) ; g. klopman , j. am . * 86 * , 4550(1964 ) . s. ramasesha , proc . indian acad . 96 * , 509 ( 1986 ) ; s. ramasesha , j. mol . struc . * 194 * , 149 ( 1989 ) . z. g. soos and s. ramasesha , phys . rev b * 29 * , 5410 ( 1984 ) . s. ramasesha and z. g. soos , j. chem . phys . * 80 * , 3278 ( 1984 ) . i. d. l. albert and s. ramasesha , j . 94 * , 6540 ( 1990 ) . a. karpfen , chem . 64 * , 299 ( 1979 ) . j. l. bredas , b. themans , and j. m. andre , j. chem . phys . * 78 * , 6137 ( 1983 ) ; d. jacquemin , b. champagne , and j. m. andr , synthetic metals * 80 * , 205 ( 1996 ) ; d. jacquemina , b. champagnea , j. m. andr , and b. kirtman , chem . phys . * 213 * , 217 ( 1996 ) ; d. jacquemin , e. a. perpte , and j. m. andr , int . chem . * 105 * , 553 ( 2005 ) . i. d. l. albert , p. k. das , and s. ramasesha , chem . * 176 * , 217 ( 1991 ) . g. chen , z. m. su , z. w. shen , and y. j. yan , j. chem . * 109 * , 2565 ( 1998 ) . t. dutta and s. ramasesha , comput . lett . * 3 * , 457 ( 2007 ) .

in this paper we investigate the effect of terminal substituents on the dynamics of spin and charge transport in donor - acceptor substituted polyenes ( @xmath0 ) chains , also known as push - pull polyenes . we employ a long - range correlated model hamiltonian for the @xmath0 system , and time - dependent density matrix renormalization group technique for time propagating the wave packet obtained by injecting a hole at a terminal site , in the ground state of the system . our studies reveal that the end groups do not affect spin and charge velocities in any significant way , but change the amount of charge transported . we have compared these push - pull systems with donor - acceptor substituted polymethine imine ( pmi ) , @xmath1 , systems in which besides electron affinities , the nature of @xmath2 orbitals in conjugation also alternate from site to site . we note that spin and charge dynamics in the pmis are very different from that observed in the case of push - pull polyenes , and within the time scale of our studies , transport of spin and charge leads to the formation of a `` quasi - static '' state .

the bulk of the chemical elements heavier than iron are thought to be produced by neutron induced reactions in the astrophysical s- and r - processes @xcite . a tiny fraction of the heavy elements , however , can not be synthesized by these processes since some of the most proton rich isotopes ( the so - called p nuclei ) lie outside the path of both the s- and r - processes . it is generally accepted that the main mechanism contributing to the synthesis of the p isotopes is the so - called @xmath0-process @xcite which takes place in high temperature stellar environments . the @xmath0-process converts the isotopes at the bottom of the valley of stability to proton rich species by consecutive ( @xmath0,n ) photodisintegration reactions . charged particle emitting ( @xmath6 ) and ( @xmath0,p ) reactions play also an important role in a @xmath0-process reaction flow model . the presently available @xmath0-process models fail to reproduce the p isotope abundances observed in the solar system . this can in part be explained by the ambiguous astrophysical conditions under which the process may take place . on the other hand deficiencies of the nuclear physics input of the models can also be responsible for the failure . in lack of experimental data @xcite the reaction rates are typically taken from hauser - feshbach statistical model calculations . at the astrophysically relevant low energy range the few available experimental datasets show that the statistical model does not give a satisfactory estimation of the cross sections @xcite . especially severe deficiencies are encountered in the case of reactions involving charged particles . therefore , the measurement of reaction cross sections relevant for the @xmath0-process is important in order to test the statistical model predictions and to put the @xmath0-process models on a more reliable ground . the ( @xmath6 ) and ( @xmath0,p ) reactions are preferably studied by measuring the cross section of the inverse capture reactions @xcite . one of the first reactions that were studied for the purpose of the @xmath0-process have been the proton capture reactions on several stable mo isotopes by sauter and kppeler @xcite . this mass region deserves special attention owing to the exceptionally large natural abundances of @xmath2mo and @xmath7mo p isotopes which are notoriously and severely underproduced in @xmath0-process models ( * ? ? ? * ; * ? ? ? * and references therein ) . the results of the work of sauter and kppeler @xcite on the @xmath2mo(p,@xmath0)@xmath3tc reaction show a strong fluctuation of the cross section as a function of the energy ( see table 4 . and fig . 5 in @xcite this feature is explained by the low level density of the neutron magic @xmath3tc isotope . however , one of the basic assumptions of the statistical model is that level density is high enough so that a statistical treatment is possible . the statistical model is thus not able to reproduce the observed strong fluctuations . therefore , the careful study of this reaction is very important in order to provide reliable reaction rates for @xmath0-process networks . recently the @xmath2mo(p,@xmath0)@xmath3tc reaction was studied again by the cologne group with both activation and in - beam @xmath0-spectroscopy methods @xcite . no final results of these measurements have been published yet , but the preliminary results show rather big deviation from the data of ref.@xcite ( see fig . 2 . in @xcite ) . in order to investigate the apparent disagreement of the two available datasets and to study the fluctuations of the excitation function , the aim of the present work is to measure the @xmath2mo(p,@xmath0)@xmath3tc cross section using a different experimental technique . one possible source of uncertainty in cross section measurements utilizing thin targets is the determination of the number of target atoms , the uniformity of the target layers and the stability of the targets under beam bombardment . this uncertainty can be avoided if an infinitely thick target which stops completely the proton beam is used instead of a thin target layer with known thickness . in such a measurement the thick target yield is determined instead of the cross section itself . if the thick target yield as a function of energy is measured with small enough energy steps , the cross section can be obtained by differentiating the yield curve ( see section[sec : results ] for further details of the cross section determination ) . the only information which is needed about the target is the stopping power value which is known to a precision of 3.2% in the case of protons in metallic mo @xcite . radiative proton capture on @xmath2mo can populate either the ground state or the isomeric state of @xmath3tc . both of these states decay by electron capture or positron emission to @xmath3mo ( the isomeric state also decays to the ground state of @xmath3tc by internal transition ) . these decays are followed by characteristic gamma - ray emissions which can be used to determine the partial cross sections leading to the ground and isomeric states separately by the activation method . table [ tab : decay ] summarizes the relevant decay parameters of the studied isotopes . [ cols="<,^,^,^,^ " , ] the obtained reaction rate values are listed in table [ tab : rate ] . the columns labeled `` rate contribution '' indicate how much the different energy regions contribute to the rate . the column labeled `` exp . '' shows the contribution of the 17003100kev experimentally studied energy range to the rate . similarly , columns `` low '' and `` high '' , respectively , show the contribution of low and high energy yield extrapolation to the rate . the table clearly shows that in the temperature range between about 1.5 and 3.5 gk the experimentally determined yields give the dominant contribution to the reaction rate . this energy range is relevant for the @xmath0-process nucleosynthesis , therefore almost purely experimental reaction rate can be provided for this reaction . the uncertainty of the reaction rate is calculated by the following procedure : in the measured 17003100kev energy range the experimental yield uncertainty of typically 7% is used . outside this region , both to high and low energies , a 100% uncertainty is assigned to the extrapolated yield . the uncertainty values listed in table [ tab : rate ] reflects this procedure : where the rate is determined by the experimental energy range the uncertainty is relatively low , while it is increasing towards lower and higher temperatures . this is the reason why the rate values are shown only in the 1.04.5gk temperature interval . the last columns of the table show the rates obtained by sauter and kppeler @xcite , roughton _ _ @xcite and calculated with the non - smoker code @xcite . the results of roughton _ are in good agreement with our data while the rates of sauter and kppeler are higher by about a factor of 1.4 to 2.2 . the comparison with the statistical calculations carried out with the non - smoker code shows that the model overestimates the reaction rate by about a factor of 2 . this is in agreement with the s - factor comparison which is shown in fig.[fig:92totmo_res ] . the model overestimates the experimental cross section and the deviation increases with increasing energy . this explains the increasing difference between the measured and calculated rates towards higher temperatures . the @xmath2mo(p,@xmath0)@xmath3tc and @xmath4mo(p,@xmath0)@xmath8tc reaction were studied by measuring thick target yields in the proton energy interval between 1700 and 3100 kev . by differentiating the thick target yield curve , cross sections of the studied reactions were calculated . in the case of the @xmath4mo(p,@xmath0)@xmath8tc reaction the obtained cross sections are in good agreement with previous data and model calculations . the previously observed strong fluctuations in the @xmath2mo(p,@xmath0)@xmath3tc reaction cross section are confirmed in the present work . for this reaction partial cross sections leading to the ground and isomeric state in @xmath3tc were measured separately . the isomeric state cross section is in good agreement with the results of sauter and kppeler @xcite , while the obtained ground state cross sections are lower than the sauter and kppeler @xcite data and are in better agreement with the preliminary results of the cologne group @xcite . we have argued that in the case of a reaction with strongly fluctuating cross section it is more useful for astrophysical applications to measure thick target yield than differential cross section with thin targets . without the intermediate step of cross section determination we have derived astrophysical reaction rate for the @xmath2mo(p,@xmath0)@xmath3tc reaction . in the temperature range relevant for the @xmath0-process the reaction rate is predominantly determined by our experimental data . our new recommended reaction rates are about a factor of 2 lower than the rates provided by the non - smoker code which is widely used in astrophysical network calculations . this work was supported by erc grant 203175 , and otka nn83261 ( eurogenesis ) , k108459 , k101328 , pd104664 . g.g.k . is a bolyai fellow . f. kppeler , r. gallino , s. bisterzo , w. aoki , rev . 83 ( 2011 ) 157 . m. arnould , s. goriely , k. takahashi , phys 450 ( 2007 ) 97 . t. rauscher , n. dauphas , i. dillmann , c. frhlich , zs . flp , gy . gyrky , rep . 76 ( 2013 ) 066201 . t. szcs , i. dillmann , r. plag , zs . flp , j. phys . series 337 ( 2012 ) 012033 . t. rauscher , j. mod . e 20 ( 2011 ) 1071 . t. sauter , f. kppeler , phys . c 55 ( 1997 ) 3127 . m. arnould , s. goriely , phys . 384 ( 2003 ) 1 . j. hasper , m. bssing , m. elvers , j. endres , a. zilges , j. phys . ( 2010 ) 012005 . a. sauerwein , m. elvers , j. endres , j. hasper , a. hennig , l. netterdon , a. zilges , aip conf . proc . 1377 ( 2011 ) 423 . a. sauerwein , m. elvers , j. endres , j. hasper , a. hennig , l. netterdon , a. zilges , proceedings of science , pos(nic xi ) ( 2011 ) 244 . http://www.srim.org/srim/srimpics/stop01/stop0142.gif c.m . baglin , nucl . data sheets 112 ( 2011 ) 1163 . e. browne , j.k . tuli , nucl . data sheets 112 ( 2011 ) 275 . t. rauscher , phys . c 81 ( 2010 ) 045807 . z. halsz et al . , phys . c 85 , ( 2012 ) 025804 . roughton , m.r . fritts , r.j . peterson , c.s . zaidins , c.j . hansen , at . data nucl . data tables 23 ( 1979 ) 177 . a.j . et al . _ , aip conf . 769 ( 2005 ) 1154 . a.j . koning and j.p . delaroche , nucl . a713 ( 2003 ) 231 . brink , nucl . phys . 4 ( 1957 ) 215 . ; p. axel , phys . rev . 126 ( 1962 ) 671 . a. gilbert and a.g.w . cameron , can . 43 ( 1965 ) 1446 . c. iliadis , nuclear physics of stars , wiley - vch verlag gmbh & co. kgaa , 2007 . t. rauscher , f - k . thielemann , at . data nucl . data tables 79 2001 47 . , www.nucastro.org j.p . jeukenne , a. lejeune , c. mahaux , phys . c 16 ( 1977 ) 80 . t. rauscher , f - k . thielemann , k - l . kratz , phys . c 56 ( 1997 ) 1613 . h. sievers , nucl . data sheets 54 ( 1988 ) 99 . roughton , m.j . fritts , r.j . peterson , c.s . zaidins , c.j . hansen , astrophysical journal 205 , ( 1976 ) 302 .

for the better understanding of the astrophysical @xmath0-process the experimental determination of low energy proton- and @xmath1-capture cross sections on heavy isotopes is required . the existing data for the @xmath2mo(p,@xmath0)@xmath3tc reaction are contradictory and strong fluctuation of the cross section is observed which can not be explained by the statistical model . in this paper a new determination of the @xmath2mo(p,@xmath0)@xmath3tc and @xmath4mo(p,@xmath0)@xmath5tc cross sections based on thick target yield measurements are presented and the results are compared with existing data and model calculations . reaction rates of @xmath2mo(p,@xmath0)@xmath3tc at temperatures relevant for the @xmath0-process are derived directly from the measured thick target yields . the obtained rates are a factor of 2 lower than the ones used in astrophysical network calculations . it is argued that in the case of fluctuating cross sections the thick target yield measurement can be more suited for a reliable reaction rate determination . astrophysical gamma - process , reaction cross section , thick target yield , gamma - spectroscopy 24.60.dr , 25.40.lw , 26.30.-k , 27.60.+j

we consider a gas of electrons with @xmath3 spin states confined to a one dimensional wire , experiencing an external potential @xmath4 , where @xmath5 is the coordinate along the wire . the electrons interact through pairwise three dimensional coulomb interactions , @xmath6 , where @xmath7 is the dielectric constant of the medium and @xmath8 is the electrons effective mass . using parameters for n - doped gaas ( @xmath9 , where @xmath10 is the free - space electron mass ) , the effective bohr radius is quite large , @xmath11 m . in this wire , the electronic wavefunction has a transverse size @xmath12 nm . integrating out these transverse dimensions , the electrons feel a regularized coulomb interaction @xmath13 . two useful regularizations are @xmath14 and @xmath15 , where @xmath16 is the smaller of @xmath5 and @xmath17 . as @xmath18 the exact form of the regularization becomes unimportant . here we mostly rely on @xmath19 . one estimates the importance of correlations caused by the coulomb interactions by comparing the interaction energy between two neighboring particles @xmath20 to the fermi ( kinetic ) energy @xmath21 , where @xmath22 is the one dimensional density of particles , and we have assumed @xmath23 . kinetic energy dominates at high densities , when @xmath24 . in that limit one expects to find a liquid state , where electrons are delocalized . conversely , when @xmath25 , the coulomb interaction dominates , and a wigner crystal should be formed . as a compromise between experimental relevance and simplicity we concentrate on power law potentials @xmath26 , where @xmath27 , which parameterizes the potential strength , roughly coincides with the size of the single particle ground state in this potential . figure [ fig : tunnel](b ) illustrates the structure of this potential for different values of @xmath28 . more complicated potentials are briefly discussed in appendix [ doublewell ] . for the relatively flat ( large @xmath28 ) potentials used in experiments , @xmath27 also roughly coincides with the ` lithographic length , ' @xmath29 , which is the physical distance between the gates which generate the potential barriers . the limit @xmath30 corresponds to an infinite square well . analysis of experimental data by tserkovnyak et al . @xcite show that @xmath31 . taking the density of electrons to be @xmath32 , one expects crystallization when @xmath33 , with @xmath34 . the experiments of auslaender et al.@xcite measure the tunneling current between the short wire described in section [ model ] , and a parallel long wire , which we will take to be infinite and uniform . momentum is conserved in the tunneling . we will make the following approximations to analyze this experiment . ( 1 ) we will treat the tunneling within perturbation theory . this is an important limit to study , though , as will be shown below , the experiments are in a regime where higher order processes may be playing a role . ( 2 ) we will neglect correlations between electrons in the two wires . one expects that this approximation will break down in the extremely low density limt , when the separation between electrons in one of the wires becomes comparable to the interwire spacing . ( 3 ) in the bulk of this paper we use an independent electron picture to study the electrons in the short wire . the consequences and validity of this approximation are discussed in section [ indep ] and are explored in more detail in appendix [ two ] . we expect that in the limit where the splitting between different spin states is small compared to the temperature or linewidths that the independent electron picture correctly predicts the magnetic field dependence of the tunneling current . however , this approximation results in a vast overestimation of the overall amplitude of the current . we consider the geometry sketched in figure [ fig : tunnel ] . the wires are separated by @xmath35 nm , and the barrier separating the wires has height @xmath36mev , arising from the conduction band offset between the gaas wires , and the intervening algaas . using gate electrodes , the experimentalists control the chemical potential difference @xmath37 between the wires.@xcite a magnetic field , perpendicular to the wires , gives a momentum kick @xmath38 to a tunneling electron@xcite . we therefore consider the tunneling hamiltonian , @xmath39 where @xmath40 and @xmath41 are respectively the operators which annihilate particles with momentum @xmath42 and spin @xmath43 in the short and long wire . the magnitude of the tunneling matrix element @xmath44 is estimated by examining the energy states in the double square well geometry sketched in figure [ fig : tunnel ] . this one - dimensional potential is a crude model of the physics transverse to the wires . we can identify @xmath45 , where @xmath46 is the energy splitting between the two lowest energy single particle states . in the limit of a deep , wide barrier ( @xmath47 ) , one finds @xmath48 where @xmath49 ( so numerically @xmath50 ) . which gives @xmath51mev , which should be compared to the spacing of states transverse to the wire , @xmath52mev . due to this separation of scales , we treat @xmath53 perturbatively . note , however , that the ratio @xmath54 is not extremely small , so that quantitative comparison with experiment would require going beyond lowest order in perturbation theory . for @xmath55 . ( c ) potential transverse to the two wires [ dotted red line in ( b ) ] . ] the current operator , defined by @xmath56 , where @xmath57 is the number of particles in the short wire , is given by @xmath58.\ ] ] to lowest order in @xmath44 , the tunneling current is @xmath59\rangle\\ & = & t^2 a_i(q,\omega=\delta v),\end{aligned}\ ] ] where the spectral density of current fluctuations is related to the retarded current response function by @xmath60 where @xmath61,$ ] with @xmath62 and @xmath63 to simplify this expression we assume that in the electrons in the two wires are independent . thus we neglect correlations between the two wires caused by the coulomb interaction . such correlations are likely to be present in the extreme wigner crystal limit , where the spacing between electrons in one of the wires is large compared to the wire separation . quantitative comparison with experiments in this limit would require including these correlations , which will partially screen the long - range part of the coulomb interaction . within our approximation the expectation values can be factored , leading to @xmath64\\ & = & \int\frac{dk}{2\pi}\frac{dk^\prime}{2\pi}\sum_{\sigma}\int\frac{d\nu}{2\pi } \left[f(\nu)-f(\nu-\omega)\right]a_{\psi\sigma}(k , k^\prime;\nu ) a_{\phi\sigma}(k+q , k^\prime+q;\nu-\omega),\end{aligned}\ ] ] where the single particle greens functions @xmath65 are given by @xmath66 and equivalent expressions hold for @xmath67 . these greens functions are related to the appropriate single particle spectral density by @xmath68a(\omega)$ ] and @xmath69 , where @xmath70^{-1}$ ] is the fermi function . if we use the free electron spectral density for the long wire , @xmath71 one arrives at the simple result that the tunneling current is a direct measure of the single particle spectral density in the short wire , @xmath72 \nonumber\\&&\qquad\times a_{\psi\,\sigma}(k - q,\omega+k^2/2m^*)\label{specdens } \\ & \approx&\nonumber \beta\omega \int\frac{dk}{2\pi}\sum_{\sigma } f_k(1-f_k)a_{\psi\,\sigma}(k - q , k^2/2m^*),\end{aligned}\ ] ] where the last line neglects terms of order @xmath73 , and uses @xmath74 . at zero temperature , the fermi functions become step functions and to lowest order in @xmath75 , equation ( [ specdens ] ) becomes @xmath76.\end{aligned}\ ] ] from the definition @xmath77 , the single particle spectral density can be written as @xmath78\end{aligned}\ ] ] where @xmath79 represents a many body state containing @xmath80 electrons and possessing energy @xmath81 . normalization is given by the grand partition function @xmath82 . at zero temperature , the sum over @xmath83 is omitted , and @xmath79 is replace by the ground state . one thus sees that current flows whenever momentum and energy can be conserved in the tunneling process . the magnitude of the current is set by the degree of overlap between states with different particle number . when the electrons become localized in a wigner crystal , the overlap between states with different particle number becomes extremely small ( because removing a particle results in rearrangement of the entire crystal a form of orthogonality catastrophe ) . consequently , the average tunneling current should drop . this drop in average current is accompanied by a sharpening of line - widths and does not lead to a complete loss of the experimental signal . instead , it leads to coulomb - blockade " type peaks , where current only flows for discrete values of @xmath37 . one will see discrete peaks if the line width is small compared to the level spacing in the short wire , @xmath84mev . assuming that the lifetime is only due to coupling between the two wires , then the line width is given by an expression very similar to that for the tunneling current . since @xmath44 is large on the scale of @xmath46 , one will only see these discrete peaks once the overlaps in ( [ sd ] ) become extremely small . in an independent particle picture ( such as hartree - fock ) , @xmath85 , where @xmath86 is the momentum space wavefunction of the @xmath87 s single electron orbital of spin @xmath43 particles , with energy @xmath88 . in general @xmath89 , and the orbitals are chosen self - consistently . within such an approximation the tunneling current is @xmath90 where , @xmath91 , @xmath92 , and we have assumed that @xmath93 . the tunneling current therefore measures the momentum space density of the particles at the fermi surface . if there are no particles at the fermi surface , then no current flows . the delta functions in ( [ indi ] ) are broadened by finite lifetime and finite temperature . in particular if there are two different states @xmath94 with energies @xmath95 , and inverse lifetimes @xmath96 , which satisfy @xmath97 or @xmath98 , then the current is proportional to the incoherent sum @xmath99 this feature will be important in high symmetry situations where there is a near degeneracy ( see , for example , section [ fringe ] ) . in the remaining sections of this paper , we calculate @xmath100 . in its conventional form , the independent particle approximation does not include the effects of orthogonality between many - body states of different particle number , as described in section [ orthog ] . in appendix [ two ] we explore a simple example which lets us systematically study these effects . we find that the omitted physics predominately modifies the overall scale of the tunneling current , and can be partially accounted for by renormalizing @xmath101 in equation ( [ indi ] ) . hartree - fock also misses fine - detail fringe structure in the magnetic field dependance of the tunneling current . the prediction of the envelope of the tunneling current appears to be qualitatively correct . of spin-1/2 electron gas confined to a one - dimensional potential @xmath102 . interaction parameters are @xmath103 and @xmath104 . solid line : equation ( [ tf2 ] ) using regularization @xmath19 ; dashed line : approximation described above eq . ( [ trans ] ) ; dotted line : noninteracting profile . inset : renormalized dimensionless cloud radius @xmath105 as a function of @xmath106 from solving eq . ( [ trans ] ) . ] the simplest picture of a gas of electrons in a confined geometry comes from the thomas - fermi ( local density ) approximation , where the system is described by a local chemical potential @xmath107 . the global chemical potential is @xmath108 , and the effective potential @xmath109 includes both the external field and the interactions between the particles . within the hartree approximation , which is valid at high densities @xmath110 , the effective potential is @xmath111 the chemical potential is then related to the density by the relationship for a homogeneous gas , @xmath112 . self - consistency requires that the density obeys a nonlinear integral equation , @xmath113 this local density approximation only makes sense if the density changes slowly compared to the interparticle spacing [ @xmath114 . we write ( [ tf ] ) in dimensionless form by introducing @xmath115 , @xmath116 , @xmath117 , @xmath118 , @xmath119 , @xmath120 , @xmath121 at fixed chemical potential the interactions can only reduce the density . therefore the density is always bounded by the noninteracting result , @xmath122 . in particular , the density always vanishes for @xmath123 . for very steep potentials , @xmath30 , the noninteracting thomas - fermi radius @xmath124 approaches the trap length , @xmath125 . for @xmath126 , equation ( [ tf2 ] ) can be solved iteratively @xcite . in this limit , the interactions predominately renormalize the chemical potential . approximating the integral in ( [ tf2 ] ) by its value at @xmath127 , one self - consistently finds that @xmath128 where @xmath105 solves the transcendental equation @xmath129 the parameter @xmath130 depends logarithmically on the cut - off @xmath131 . for the regularization @xmath132 , @xmath133 . figure [ showt ] compares this approximation to the numerical solution of ( [ tf2 ] ) , and the inset shows @xmath105 as a function of @xmath134 . as previously explained , within an independent particle picture the tunneling current between two parallel wires is proportional to the momentum space wavefunction of the highest occupied single particle state . this momentum space wavefunction will be strongly peaked about @xmath135 , which can be approximated as @xmath136 , where @xmath137 , and @xmath105 is given by eq . ( [ trans ] ) . figure [ kplot ] shows @xmath138 as a function of @xmath139 for reasonable parameters . of wavefunction at fermi surface as a function of chemical potential @xmath108 within the thomas - fermi approximation . dotted line shows noninteracting result , @xmath140 . solid line includes coulomb interactions with @xmath141 , @xmath142 , in a power law trap with @xmath143 and @xmath144 m . this approximation will break down when @xmath145 . ] to go beyond this single wavevector approximation , we use semiclassical means to calculate the wavefunction of the state at the fermi surface . we find the modulus by imagining that we reduce the chemical potential by a very small amount , so that this last state goes from filled to empty . the density change must coincide with the density @xmath146 of that last single particle state , implying that @xmath147 where @xmath148 is a normalization constant determined by setting @xmath149 . this wavefunction should have a phase whose derivative gives the local fermi momentum . @xmath150 the @xmath151 and @xmath152 solutions correspond to left - moving and right moving waves . in principle , one must combine these two solutions to form a standing wave . within the hartree approximation , this procedure is equivalent to finding the wkb wavefunction in the self - consistent potential . we do not discuss the details as when we compute the momentum space wavefunction , only the structure near @xmath153 will be affected by how we superimpose these two solutions . in the case where the coulomb interaction can be neglected , equation ( [ dens ] ) reduces to @xmath154 , and equation ( [ phase ] ) becomes @xmath155 . not surprisingly , in this limit @xmath156 is _ exactly _ the wkb wavefunction for an electron with energy @xmath139 in the potential @xmath4 . our arguments therefore reduce to the semiclassical arguments of tserkovnyak et al . @xcite for the noninteracting gas . the analysis is very similar for our approximation where @xmath157 . since interactions only renormalize the chemical potential , the derivative @xmath158 is a constant , independent of space , and we can use the noninteracting result with a renormalized chemical potential . that is , we take @xmath159 . following tserkovnyak et al . @xcite , we calculate the momentum space wavefunction , @xmath160 by laplace s method . the fourier integral is dominated by regions of space near where @xmath161 . moreover , since @xmath162 is peaked around @xmath163 , we can expand @xmath164 near @xmath165 . since @xmath166 is symmetric , there is no loss of generality in taking @xmath167 . we introduce @xmath168 , with @xmath169 . changing variables to @xmath170 , and assuming @xmath171 , we find @xmath172,\ ] ] where @xmath173 is a normalization constant . in the case @xmath174 , the integral in ( [ airy ] ) is an airy function , though for other values of @xmath28 it is not a familiar special function . the argument of the exponential , @xmath175 , is stationary at @xmath176 . the curvature is @xmath177 . this is an isolated saddle point when @xmath178 in which case @xmath179 \right ] . \ ] ] by deforming the contour of integration , this equation holds for both @xmath180 and @xmath181 ( in the latter case one should take the principle branch , giving @xmath182 a small negative imaginary part ) . from ( [ ai2 ] ) , one sees that if @xmath183 , the wave function falls off exponentially as @xmath184 . for @xmath185 , instead of expanding about the saddle point , we expand about @xmath186 , finding as @xmath187 , @xmath188\\\nonumber \theta&= & p\,(\beta+1)^{1/(\beta+1 ) } \frac{\gamma\left(\frac{2}{\beta+1}\right ) } { \gamma\left(\frac{1}{\beta+1}\right ) } i^{1/(\beta+1)}\\\nonumber&&-i(\pi/2)(1/(\beta+1)).\end{aligned}\ ] ] in section [ sec : tf ] , we neglected the fact that electrons are discrete entities and that an electron does not interact with itself . including this discreteness , the interaction energy can be reduced by localizing each electron to a small volume maximizing the distance between the charge distribution of neighboring particles . since such localization costs kinetic energy , it is only favorable in the low density gas where interaction energy dominates over kinetic energy . to discuss this phase , we imagine that we have @xmath189 electrons , localized at positions @xmath190 where @xmath191 . neglecting kinetic terms , the energy of a given configuration is @xmath192 as with the profile in the dense limit ( fig . [ showt ] ) , we numerically find that the resulting density is quite flat , and is well approximated by evenly spaced electrons , @xmath193 , with uniform density @xmath22 . minimizing ( [ wen ] ) with respect to @xmath22 yields @xmath194^{1/(\beta+1)}\\\nonumber & \approx & \frac{2 w}{n } \left [ 1+\frac{1}{\beta+1}\log\left ( \frac{w n^2\log n}{2 a } \right)+{\cal o}(\beta^{-2 } ) \right]\end{aligned}\ ] ] in this wigner crystal state , each electron is confined to a length @xmath195 . to calculate this length , and the wavefunction of the localized electron , we expand the potential in eq . ( [ wen ] ) for small fluctuations in the position of one of the particles , @xmath196 we approximate the sum over a finite number of particles by that of an infinite chain , taking @xmath197 . numerically , the riemann zeta function is @xmath198 . the curvature of the external potential is generically negligible in this limit , hence each electron is trapped by a harmonic oscillator potential with @xmath199 and has a wavefunction @xmath200 with @xmath201 . this implies that the momentum space wavefunction has modulus @xmath202 . comparing this momentum space wavefunction to the high density predictions [ equation ( [ airy ] ) through ( [ sp ] ) ] , we see that in the wigner crystal , @xmath203 is more spread out in momentum space , and does not contain any oscillations . . the horizontal axis shows different interaction strengths parameterized by the ratio of the bohr radius @xmath204 to the wire length @xmath27 ( note the logarithmic scale ) . the potential has exponent @xmath143 , and a cutoff , @xmath205 . ] parameters coincide with fig . [ fig : hf ] . solid , blue dotted , and red dashed lines correspond to total density and the density of up and down spin electrons [ in ( a ) , the latter two curves coincide ] . panels ( a ) , ( b ) , ( c ) correspond to @xmath206=0.5 , 0.15 , 0.05 . ] we investigate the crossover between the high density electron gas and the low density wigner crystal within the hartree - fock approximation . in the extreme limits , hartree - fock reduces to our previous approximations . it also provides semi - quantitative understanding of the intermediate regime . in uniform higher dimensional systems , hartree - fock overestimates the stability of the wigner crystal @xcite , and one therefore expects some systematic errors as compared to an exact many - body calculation . unlike bosonization and luttinger liquid techniques , hartree - fock correctly accounts for band curvature . band curvature plays a significant role at low densities . appendix [ two ] compares hartree - fock and exact results for two particles . hartree - fock is most simply thought of as a variational method , where one searches for the slater determinant which minimizes the energy @xcite . this procedure is highly nontrivial as the energy landscape in the space of slater determinants is quite complex with many local minima . we start by considering the limit of vanishing interactions @xmath207 , where we find the exact solution by discretizing space and solving the single - particle schrodinger equation . with fixed numbers of up and down spin electrons ( @xmath208 and @xmath209 ) , we then gradually decrease @xmath210 . for each value of @xmath210 we iteratively solve the hartree - fock equations in discretized space . although there is no guarantee that the states we find in this manner are absolute energy minima , they consistently have lower energies than all other states that we have found by iterating the hartree - fock equations from different starting points . an advantage of the hartree - fock approximation is that since it involves an independent electron approximation , one has direct access to single - particle observables , such as the wavefunction of the last bound state . figure [ fig : hf ] shows the momentum - space wavefunction of the last bound state for a system of eighteen particles ( nine in each spin state ) . for large @xmath206 , this last single - particle state is delocalized in real space , resulting in a series of momentum - space peaks which are well described by equation ( [ airy ] ) . for small @xmath206 , this last state is localized , resulting in a broad spread of momenta . illustrative real - space density profiles are shown in figures [ fig : hf2 ] a , b , c . in ( a ) we see that even for arbitrarily weak interactions the real space density is corrugated . the corrugations are caused by the free electron response to an inhomogeneous potential and are analogous to the friedel @xcite oscillations seen in electron density near an impurity . the asymptotic behavior of these oscillations has been analytically calculated for harmonic confinement@xcite . there are nine peaks corresponding to the nine doubly occupied single particle states . in ( b ) we see that as we increase the interaction strength each of the density peaks split into two , so there is one peak per particle . a spin - density wave appears . in ( c ) we see a wigner crystal with an antiferromagnetic spin profile . the density varies smoothly as one increases the interactions . when an odd number of particles form a wigner crystal the highest occupied single particle state , with wavefunction @xmath211 , is in a symmetric superposition of two locations , as illustrated in figure [ symbreak ] , and has fringes in its momentum space wavefunction , @xmath212 . the next highest energy state , @xmath213 , is in the antisymmetric superposition of these same two locations , and has momentum space fringes which are 90 degrees out of phase with those for @xmath212 . since the splitting between these two states is exponentially small , the tunneling current is proportional to @xmath214 ( see section [ indep ] ) , which , as illustrated in figure [ symbreak]c , contains no fringes . going beyond hartree - fock , one generically expects the many - body wavefunction to include amplitudes for each particle to be in a superposition of more than one place . thus , in an exact treatment one would always expect that tunneling from any one particular state to any other particular state would result in fringes . however , as with the hartree - fock result , if one incoherently adds the fringes from several symmetry related states then the fringes will be washed out . in particular , if the temperature or lifetime is greater than the splitting between these states then no fringes will be observable . , see caption of fig [ fig : hf2 ] for key . ( b ) wavefunction @xmath215 ( solid ) and @xmath216 ( dashed ) of the two highest highest energy occupied states . ( c ) momentum space wavefunctions : red dashed , @xmath217 ; blue dotted , @xmath218 ; solid , @xmath219 . ] we have presented simple analytic theories of very high and very low density electrons in an inhomogeneous one dimensional potential . at high density an electron liquid is formed where the density displays friedel oscillations from response to the potential [ fig . [ fig : hf2](a ) ] . at low densities a crystal - like state is formed where the electrons are well - separated [ fig . [ fig : hf2](a ) ] . using the hartree - fock approximation we numerically studied the cross - over between these limits . intermediate between the two extremes we found a spin - density - wave state [ fig . [ fig : hf2](b ) ] . in the extreme high density limit our approach is exact . in the extreme low density limit , our approach will give the exact electronic density . our mean - field approach , however , can not describe the quantum spin correlations which occur in the wigner crystal state . instead , it predicts a classical nel order . one expects that our description of the crossover is qualitatively correct , but that the stability of the crystalline state is overestimated . within the independent electron picture we have calculated the tunneling current between electrons in a short and long wire . in the liquid phase , the zero bias momentum resolved tunneling is dominated by sharp peaks at @xmath220 , and there is large overlap between states with different particle number . as interactions increase , these two peaks broaden and merge . in the crystaline phase , the tunneling displays a large feature centered at @xmath153 . this feature is attributed to electron localization . going beyond our mean - field picture , it is clear that in this limit the overlap between states with different particle number becomes extremely small , which leads to coulomb - blockade structures . for these tunneling predictions we work in linear response , and neglect correlations between the electrons in the two wires . for quantitative comparison with experiments , one needs to include higher order terms in the tunneling and coulomb induced interwire correlations . although simple mean - field analysis ( sec . [ fringe ] ) and more general symmetry arguments@xcite suggest that the large peak at @xmath153 will show a fringe structure , we have demonstrated that if either the temperature or the inverse state lifetime @xmath96 is larger that the splitting between spin states then this structure is washed out , and may therefore be extremely hard to observe experimentally . we should emphasize that the hartree - fock theory used here is unable to accurately describe these fringes . from our studies of the two - particle system , we believe that hartree - fock does describe the envelope of the peak , though this needs to be confirmed by studying larger systems . i would like to thank amir yacoby for providing experimental details@xcite , bertrand halperin for stimulating correspondence , and paul mceuen for discussing related experiments on carbon nanotubes . this work was partially performed at the kavli institute for theoretical physics . it was supported in part by the national science foundation under grant no . phy99 - 07949 . the physics of interaction induced electron localization is very simply understood by studying two particles in a parabolic confinement . the two particle system is particularly appealing because it is equivalent to a problem in one dimensional single particle quantum mechanics and is therefore approachable through elementary means . the two and three dimensional version of this problem has been extensively studied @xcite . in section [ exact ] we show this exact solution . in section [ hf2 ] we solve this problem through the hartree - fock method used in section [ hf ] . by comparing these two results one quantifies the accuracy of our approximations . in coordinates scaled by the oscillator length @xmath27 , the two particle schrodinger equation is @xmath221 \psi(x_1,x_2)\\\nonumber = e \psi(x_1,x_2),\end{aligned}\ ] ] where the energy @xmath222 is scaled by the oscillator energy @xmath223 . this equation separates into center of mass an relative coordinates @xmath224 , @xmath225 by writing @xmath226 , which obey @xmath227 \psi(x)&= & e_x \psi(x)\\ \left [ -\partial_x^2+x^2/4+f_d(x)/a \right ] \psi(x)&= & e_x \psi(x ) , \label{spse}\end{aligned}\ ] ] with @xmath228 . up to normalization , the lowest energy center of mass wavefunction is @xmath229 and @xmath230 . our task therefore reduces to solving the single particle schrodinger equation ( [ spse ] ) . the electron spin enters in that for a spin singlet state @xmath231 must be even , while for a spin triplet , @xmath231 must be odd . ; ( b ) @xmath232 ; ( c ) @xmath233 . all lengths are measured in units of @xmath234 . solid line shows exact results , while dashed lines show variational approximation . ] figure [ exactpsie ] shows the energy @xmath222 of the singlet and triplet states as a function of @xmath204 , while figure @xmath235 shows @xmath231 for several representative values of @xmath204 . qualitative features of the exact solution are most readily understood through a variational approach . we take @xmath236 where @xmath105 and @xmath237 are free parameters . the expectation value of the energy is then @xmath238 for @xmath239 we extract the leading @xmath240 dependence of @xmath241 to arrive at @xmath242+{\cal o}(d^2/s^2),\end{aligned}\ ] ] where @xmath243 , @xmath244 is euler s gamma and @xmath245 is a constant which depends on the exact form of the regularization . for regularization @xmath246 , @xmath247 and for regularization @xmath19 , @xmath248 . the energy is only logarithmically sensitive to this constant . the separation between the two peaks in the wavefunction is given by @xmath105 , and the width of each peak is given by @xmath237 . these play the role that @xmath249 and @xmath250 played in section [ classical ] . the lowest energy variational wavefunctions are compared to the exact results in figure [ exactpsiwf ] . by expanding for small @xmath105 , one sees that for any @xmath251 , the lowest energy variational state always has @xmath252 . when appropriately biased , the tunneling current is proportional to @xmath253 , where @xmath254 , @xmath255 , and @xmath256 , @xmath257 represent the ground state with @xmath258 and @xmath259 particles . explicitly , @xmath260 figure [ comparef ] shows this quantity for two illustrative values of @xmath204 . as with the hartree - fock theory in section [ fringe ] , one sees fringes from the fact that each electron is in a quantum superposition of two places . the period ( in k - space ) of these fringes becomes smaller as interactions increase . if one expands the potential in ( [ spse ] ) about its minimum , one sees that the size of each peak in the wavefunction becomes independent of @xmath204 for sufficiently small @xmath204 . this is born out by analyzing the variational energy ( [ evare ] ) in the limit of large @xmath105 . consequently the envelope of the finges in @xmath261 becomes independent of @xmath204 for strong interactions . in a flatter potential ( @xmath262 with @xmath263 , the peaks in @xmath231 will become narrower as @xmath204 increases , resulting in a broadening of @xmath261 . for two harmonically trapped particles in a singlet state : exact @xmath264 ( solid ) , @xmath265 ( dashed ) ; hartree fock @xmath264 ( red dotted ) , @xmath265 ( green dotted , scaled by 0.32 ) . all lengths are measured in units of @xmath234 . inset shows triplet states . , title="fig : " ] we compare our exact results with the hartree - fock approximation . for two particles in the same / different spin states , hartree - fock is equivalent to the variational ansatz @xmath266,\end{aligned}\ ] ] where @xmath215 represent single particle orbitals and @xmath267 corresponds to having the @xmath87th particle in the spin up state . this ansatz clearly leads to a spin triplet state when the two particles have the same spin , but does not lead to any particular symmetry when the particles have dissimilar spins . we focus on the case of dissimilar spins . as might be expected from symmetry , we numerically find that the minimum energy configuration always has @xmath268 [ which when @xmath269 is even is equivalent to @xmath270 . we then find that the minimal energy hartree - fock state obeys the self - consistent equation @xmath271 which is readily solved numerically . as with the exact state , we gain extra insight by a variational solution , taking @xmath272 , for which @xmath273.\ ] ] not surprisingly , the interaction term in the energy is a linear combination of the interaction we previously found for triplet and singlet states . unlike when we applied the variational approximation to the exact equations ( [ evare ] ) , here we find that for large @xmath204 , the separation @xmath105 is strictly zero . as @xmath204 decreases there is a transition where @xmath105 discontinuously jumps to a nonzero value , corresponding to the formation of a spin density wave . this discontinuity is also seen in the full numerical solution to ( [ fullhf ] ) , and for @xmath274 occurs near @xmath275 . the variational approximation to the hartree - fock energy ( [ hfvar ] ) is almost indistinguishable from the exact hartree - fock energy ( [ fullhf ] ) . figure [ exactpsie ] compares this approximate result to the exact energy from solving ( [ spse ] ) and its variational approximation ( [ evare ] ) . as we anticipated , the hartree - fock result is nearly exact in the limits of very strong and very weak interaction . for intermediate coupling however the hartree - fock energy deviates from the exact result . within hartree - fock , the tunneling current is proportional to @xmath276 , where @xmath277 . figure [ comparef ] compares @xmath276 to the exact result for @xmath253 . since the hartree - fock result does not capture the orthogonality catastrophe physics , the two predictions differ by a significant scale factor [ 0.32 at @xmath265 ] . furthermore , the exact result contains fringes which are not seen in hartree - fock . these fringes occur because the exact wavefunction places each electron in a superposition of two positions . once rescaled , the hartree - fock prediction gives the envelope of the exact result . we expect these observations to be generic . even in the system with more particles , the hartree - fock prediction should match the scaled envelope of the exact result . if finite temperature / lifetime effects wash out the fringes then the hartree fock should correctly predict the dependance of the tunelling current on momentum ( magnetic field ) . the power law potentials used in the body of this paper capture the slow variation of electrostatic fields felt by electrons in typical experiments . some features of these power - law potentials are overly - simplistic . most importantly , due to experimental constraints on the placement of gate electrodes , a multiple - well structure may be formed along the wire , as illustrated in figure [ multiwell ] . qualitative understanding of the electronic state is found by noting that the local chemical potential is small in the central region but large near the edges , resulting in a low density crystaline state surrounded by a high density liquid state . a hartree - fock calculation , illustrated in figure [ multiwell ] , verifies this intuitive picture . the localized states which are formed _ above _ the barrier are a purely many - body effect . similar localized state are believed to occur above quantum point contacts@xcite and may give rise to the so called 0.7 anomaly@xcite . for key ] . as illustrated , the characteristic length scale of the potential is @xmath27 . the interactions are characterized by @xmath279 and @xmath280 , yielding a low density crystal - like central region surrounded by a high density liquid - like region . ] 99 a. yacoby , h. l. stormer , n. s. wingreen , l. n. pfeiffer , k. w. baldwin , and k. w. west , phys . lett . * 77 * , 4612 ( 1996 ) ; solid state communications * 101 * , 77 ( 1997 ) ; l. n. pfeiffer , a. jacoby , h. l. stormer , k. w. baldwin , j. hasen , a. pinczuk , w. wegscheider , and k. w. west , microelectron . j. * 28 * , 817 ( 1997 ) . o. m. auslaender , a. yacoby , r. de picciotto , k. w. baldwin , l. n. pfeiffer , and k. w. west , science * 295 * , 825 ( 2002 ) . h. steinberg , o. m. auslaender , a. yacoby , j. qian , g. a. fiete , y. tserkovnyak , b. i. halperin , r. de picciotto , k. w. baldwin , l. n. pfeiffer and k. w. west , _ localization transition in ballistic quantum wires _ , unpublished ( 2004 ) . also see d. yacoby , _ measurement of spin charge separation and localization in 1d _ , lecture at kitp , 5/20/2004 . http://online.kitp.ucsb.edu/online/exotic04/yacoby/ o. m. auslaender , h. steinber , a. yacoby , y. tserkovnyak , b. i. halperin , k. w. baldwin , l. n. pfeiffer , and k. w. west , science * 308 * , 88 ( 2005 ) . e. wigner , phys . rev . * 46 * , 1002 ( 1934 ) ; trans . faraday soc . * 34 * , 678 ( 1938 ) . the mermin - wagner theorem [ n. d. mermin , and h. wagner , phys . * 17 * , 1133 ( 1966 ) ] prohibits long - range crystalline order in one dimension , but does not restrict the short - range order or localization described here . y. tserkovnyak , b. i. halperin , o. m. auslaender , and a. yacoby , phys . * 89 * , 136805 ( 2002 ) ; y. tserkovnyak , b. i. halperin , o. m. auslaender , and a. yacoby , phys . rev . b * 68 * , 125312 ( 2003 ) . k. a. matveev , phys . rev . lett . * 92 * , 106801 ( 2004 ) , b. tanatar , i. al - hayek , and m. tomak , phys . rev . b * 58 * , 9886 ( 1998 ) , l. i. glazman , i. m. ruzin , and b. i. shklovskii , phys . rev . b * 45 * , 8454 ( 1992 ) . h. j. schulz , phys . lett . * 71 * , 1864 ( 1993 ) . q. yuan , h. chen , y. zhang , and y. chen , phys . b * 58 * , 1084 ( 1998 ) . g. fano , f. ortolani , a. parola , and l. ziosi , phys . b * 60 * , 15654 ( 1999 ) . w. husler and b. kramer , phys . b * 47 * , 16353 ( 1992 ) . b. szafran , f. m. peeters , s. bednarek , t. chwiej , j. adamowski , and a. dabrowska , phys . b * 70 * , 035401 ( 2004 ) . e. rsnen , h. saarikoski , v. n. stavrou , a. harju , m. j. puska , and r. m. nieminen , phys . b * 67 * , 235307 ( 2003 ) . f. anfuso , and s. eggert , phys . b * 68 * , 241301(r ) ( 2003 ) . g. a. fiete , j. qian , y. tserkovnyak , and b. i. halperin , cond - mat/0501684 . as the chemical potential difference between the wires , @xmath37 depends not only on the applied potential , but also on the number of electrons in the short wire . for example , the first correction involves setting @xmath281 on the right hand side of ( [ tf2 ] ) . using regularization @xmath132 this correction is @xmath282 where @xmath283 , and @xmath284/ [ ( n-4)\gamma(1/2 - 2/\beta)]$ ] . see j. r. trail , m. d. towler , and r. j. needs , phys . b * 68 * , 045107 ( 2003 ) , and references therein . n. w. ashcroft and n. d. mermin , solid state physics ( hartcourt brace , fort worth 1976 ) . j. friedel , nuovo cimento suppl . * 7 * , 287 ( 1958 ) . e. j. mueller , phys . lett . * 93 * , 190404 ( 2004 ) . m. taut , phys . a , * 48 * , 3561 ( 1993 ) . u. merkt , j. huser , and m. wagner , phys . b * 43 * , 7320 ( 1991 ) . o. p. sushkov , phys . b * 67 * , 195318 ( 2003 ) , k. hirose , y. meir , and n. s. wingreen , phys . * 90 * , 026804 ( 2003 ) , y. meir , k. hirose , and n. s. wingreen , phys . lett 89 , 196802 ( 2002 ) . r. de picciotto , l. n. pfeiffer , k. w. baldwin , and k. w. west , phys . lett . * 92 * , 036805 ( 2004 ) ; d. j. reilly , t. m. buehler , j. l. obrien , a. r. hamilton , a. s. dzurak , r. g. clark , b. e. kane , l. n. pfeiffer , and k. w. west , phys . lett . * 89 * , 246801 ( 2002 ) ; s. m. cronenwett , h. j. lynch , d. goldhaber - gordon , l. p. kouwenhoven , c. m. marcus , k. hirose , n. s. wingreen , and v. umansky , phys . rev 88 * , 226805 ( 2002 ) ; k. j. thomas , j. t. nicholls , m. y. simmons , m. pepper , d. r. mace , and d. a. ritchie , phys . lett . * 77 * , 135 ( 1996 ) .

we explore the theory of electrons confined by one dimensional power law potentials . we calculate the density profile in the high density electron gas , the low density wigner crystal , and the intermediate regime . we extract the momentum space wavefunction of the electron at the fermi surface , which can be measured in experiments on tunneling between parallel wires . the onset of localization leads to a dramatic broadening of the momentum space wavefunction together with pronounced sharpening ( in energy ) of the tunneling spectrum . advances in microprocessing have enabled scientists to construct ultra - high mobility one dimensional wires with transverse size @xmath0 nm , which is sufficiently small to freeze out all transverse electronic motion @xcite . by allowing tunneling between two such parallel wires , auslaender et al . @xcite have been able to map out the spectrum of elementary excitations , providing good evidence of spin - charge separation . further experiments have found evidence of electronic localization in these wires when a gate electrode is used to deplete the density of electrons @xcite . this localization is possibly caused by an interaction driven transition into a wigner crystal state @xcite , where each electron is localized by repulsion from its neighbors . here we investigate the theory of such crystallization in one dimensional wires . the wires used in these experiments are not infinitely long . their finite length , which is controlled by electrostatic potentials , leads to striking fringes in the tunneling spectrum @xcite . these fringes can not be explained by particles bounded by ` infinite walls ' , but require a more accurate modeling of the confinement . tserkovnyak et al . @xcite found that the experimental results were consistent with a power - law potential @xmath1 with @xmath2 . in discussing experimental signatures of crystallization in these systems , we must take into account this same physics , and explore the interplay between the external confining potential and the interparticle coulomb repulsion . due to the influence of the gate electrode , the electrons in the localization experiment@xcite feel a complicated potential , with two minima separated by a barrier . for most of our analysis we use the simpler power law potential . we are interested in gaining an intuitive understanding the crossover between electron liquids and solids in these inhomogeneous wires . we therefore choose to work with the simplest techniques which can capture the qualitative features of these experiments . we use a thomas - fermi description of the high density electron gas ( sec . [ sec : tf ] ) ; a semiclassical description of the wigner crystal ( sec . [ classical ] ) ; and a hartree - fock description of the cross - over ( sec . [ hf ] ) . these models provide direct insight into the experimental signatures of this transition . they do not , however , give direct information about excitations , transport properties , or fractionalization of the electrons into spin and charge degrees of freedom . our approach complements prior theoretical work , which predominantly treated homogeneous infinite systems @xcite , small numbers of electrons @xcite , or infinite square well potentials @xcite . particular note should be taken of the recent work by fiete et al.@xcite which analyzes the experiments discussed here@xcite in the limits of extremely small and extremely large particle numbers . our main accomplishment is that we are able to follow the evolution of the quantum state from the weakly interacting high density regime to the strongly interacting low density regime . we describe the electronic density profile , and see how friedel oscillations evolve into wigner - crystal correlations . in all of these regimes , we predict the tunnelling spectrum between parallel wires , giving qualitative explanation of the striking features in recent experiments@xcite .

although conspicous inss such as the crab and vela pulsars have been observed from the very beginning of the mission , hst started to play a key role on the study of the optical behaviour of these faint targets after the first refurbishing mission in 1993 . the study did not proceed systematically , e.g. from the brighter to the dimmer , but rather following a random walk dictated by the allocation of observing time . table 1 lists all the inss ( be they bona fide pulsars or radio - silent neutron stars ) observed so far by the hst . ccccc + i d & log(yr ) & log(de / dt ) & d(kpc ) & mag + + crab & 3.1 & 38.6 & 2.0 & 16.6 + b0540 - 69 & 3.2 & 38.2 & 55 & 22.5 + vela & 4.1 & 36.8 & 0.5 & 23.6 + b0656 + 14 & 5.0 & 34.6 & 0.76 & 25.0 + geminga & 5.5 & 34.5 & 0.16 ( ) & 25.5 + b1055 - 52 & 5.7 & 34.5 & 1.5 & 24.9(u ) + b1929 + 10 & 6.5 & 33.6 & 0.17 ( ) & 25.7(u ) + b0950 + 08 & 7.2 & 32.7 & 0.28 ( ) & 27.1(u ) + + rxj 1856 - 3754 & & & @xmath4 & 25.6 + although their number is limited , the objects in table 1 sample 10 magnitude in brightness and 4 decades in age , going from the youngest pulsars , such as the crab and psr b0540 - 69 , to rather old ones , such as psr b0950 + 08 . all inss , but the crab , are faint . all challenging , sometimes plainly impossible to observe from the ground . this was the case of psr b1055 - 52 ( mignani et al . 1997@xcite ) which , together with psr b1929 + 10 and psr b0950 + 08 ( pavlov et al . 1996@xcite ) have been seen only with the hst using the foc and the u filter . to the score of hst identifications we can add the ins candidate rxj 1856 - 3754 ( walter & matthews 1997@xcite ) . over the years , hst has collected light curves , for the crab ( percival et al . 1993@xcite ) and psr b0540 - 69 ( boyd et al . 1995@xcite ) , spectra , for the same two objects ( gull et al . 1998@xcite ; hill et al . 1997@xcite ) , and images in different filters for all of them . the major results obtained by hst in pulsar astronomy have been reviewed by mignani et al . ( 2000)@xcite . the observational efforts pursued by different groups with the imaging instruments on board hst are summarized in table 2 , where , for sake of clarity , the spectral coverage provided by hst has been roughly divided in two infrared channels ( ir and i ) , four optical ones ( r , v , b , u)- plus narrow bands ( nb)- and one ultraviolet . in table 2 , nicmos , wfpc2 , and foc observations are indicated . if an observation has been done more than once , the number in parenthesis gives the number of repetitions . lcccccccc + i d & ir & i & r & v & b & u & uv & nb + crab & & & & & & & & 547 m ( several ) + b0540 - 69 & & & & wfpc2 & & & & 656n , 658n + vela & & wfpc2 & wfpc2 & wfpc2(5 ) & & & & + b0656 + 14 & nicmos & & & wfpc2(2 ) & foc & foc & foc & + geminga & nicmos & & wfpc2 & wfpc2(4 ) & foc & foc & foc & + _ b1055 - 52 _ & & & & & & _ foc _ & & + & & & & & & _ foc _ & _ foc _ & + _ b0950 + 08 _ & & & & & & & _ foc _ & + & & & & _ wfpc2(2 ) _ & _ wfpc2 _ & _ wfpc2(2 ) _ & _ wfpc2 _ + table 2 shows quite eloquently that not all the entries in table 1 received the same amount of observing time : it is worth noticing that , apart from the `` dancing crab '' , the objects with the highest number of observations is the rather dim geminga , followed by psrb0656 + 14 , to show that objects fainter than v=25 were not discriminated in this study . the amount of information contained in this comprehensive list has been used : * to measure pulsars proper motions and parallactic displacements , * to study plerion phenomenology * to assess the spectral distribution of objects too faint for spectroscopy the major achievements are summarized in the next sections . for all the pulsars observed more than once , namely the crab , vela , psr b0656 + 14 and geminga , a proper motion has been measured , yielding also new and independent measurements of the objects transverse velocities . this topic is reviewed in these proceedings by mignani et al . sometimes , the accurate determination of the proper motion has been a by - product of a sequence of observations aimed at the measurement of the object s parallactic displacement and hence its distance ( see also de luca et al . , these proceedings ) . this has been done for geminga ( caraveo et al . 1996@xcite ) and is currently underway for the vela pulsar . determining the distance to a pulsar allows the assessment of the absolute optical luminosity which , compared with the overall energy loss de / dt , yields the efficiency to convert rotational energy loss into optical emission , an important parameter in pulsar modelling . hst imaging of crab , vela and psr b0540 - 69 allows one to trace the relativistic pulsar wind and to better study the plerion phenomenology . moreover , with the proper motion vectors clearly aligned with the axes of symmetry of the crab and vela plerions , proper motions , or rather the mechanisms responsible for them , seem to play a role in shaping the inner remnants ( see mignani et al . , these proceedings , and pavlov et al . 2000@xcite ) . comparisons between hst frames and recently obtained chandra high resolution images open new avenues to study the multiwavelength behaviour of young energetic plerions . the case of psr b0540 - 69 is discussed in an accompanying paper by caraveo et al . . hst multicolor imaging appears to be the next best thing to a spectrum for studying the spectral shape of faint objects and discriminating between thermal emission from the ins surface and non thermal magnetospheric one . two classical examples are * psr b0656 + 14 , where pavlov et al . ( 1997)@xcite have shown a composite spectral shape featuring both a thermal and non - thermal components ( see fig.1 , left panel ) * geminga , for which bignami et al . ( 1996)@xcite and mignani et al . ( 1998)@xcite have provided the evidence of a cyclotron spectral feature on top of a thermal continuum ( see fig.1 , right panel ) . if correct , the cyclotron identification ( discussed also by jacchia et al . 1999@xcite ) of the feature would provide the first in situ measurement of the magnetic field of an isolated neutron star . all in all , the study of inss , in spite of their faintness , has yielded a wealth of interesting results definitely worth the time and efforts devoted to them . new identifications have been secured while new insights have been achieved for pulsars already identified . of course , a lot remains to be done . young pulsars are definitely promising targets , thus we should concentrate on newly discovered young objects , such as the 16 msec one in the lmc . here the timing capability of the stis , so far poorly exploited , should be fully used . bignami , g.f . 1996 , apj 456 , l111 boyd , p.t . 1995 , apj 448 , 365 caraveo , p.a . , bignami , g.f , mignani , r. & taff , l.g . 1996 , apj 461 , l91 gull , t.r . et al . 1998 , apj 495 , l51 hill , r.j . et al . 1997 , apj 486 , l99 jacchia , a. et al . 1999 , a&a 347 , 494 koptsevich , a.b . , pavlov , g.g . , zharikov , s.v . , sokolov , v.v . , shibanov , yu.a . , & kurt , v.g . 2000 , a&a , submitted , astro - ph/0009064 mignani , r. , caraveo , p.a . & bignami , g.f . 1997 , apj 474,l51 mignani , r. , caraveo , p.a . & bignami , g.f . 1998 , a&a 332 , l37 mignani , r.p . , caraveo , p.a . & bignami , g.f . 2000 , stsci newsletter 17 , no 1 , 3 pavlov , g.g . , stringfellow , g.s . & cordova , f.a . 1996 , apj 467,370 pavlov , g.g . , welty , a.d . & cordova , f.a . 1997 apj 489 , l75 pavlov , g.g . , sanwal , d. , garmire , g.p . , zavlin , v.e . , burwitz , v. & dodson , r.g . 2000 , aas meeting 196 , @xmath5 percival , j.w . 1993 , apj 407 , 276 walter , f.m . & matthews l.d . 1997 , nature 389 , 358

hst observations have contributed significantly to our knowledge on the behaviour of isolated neutron stars ( inss ) as optical emitters . first , hst has been instrumental both to discover new optical counterparts ( psr b1055 - 52 , psr b1929 + 10 , psr b0950 + 08 ) and to confirm proposed identifications ( psr b0656 + 14 ) . second , hst multicolor photometry provided useful information to characterize the optical emission mechanism(s ) at work in middle - aged inss like psr b0656 + 14 and geminga . last , but not least , the superior angular resolution of the hst allowed both to perform a very accurate morphological study of the plerionic environments of young pulsars ( e.g. the crab and psr b0540 - 69 ) and to perform very accurate astrometric measurements yielding proper motions ( crab , vela , geminga , psr b0656 + 14 ) and parallaxes ( geminga ) . psfig.sty @xmath0 ifc - cnr , via bassini 15 , i-20133 milan , italy @xmath1 st - ecf , karl schwarzschild str.2 , d8574o garching b. munchen , germany @xmath2 pennsylvania state univ . , 525 davey lab , university park , pa 16802 , usa @xmath3 asi , via liegi 26 , i-00198 rome , italy

models of particle physics including weak scale supersymmetry ( susy ) are amongst the most promising candidates@xcite for new physics at the tev scale . of this class of models , the minimal supergravity ( msugra ) model stands out as providing one of the most economic explanations for the diversity of soft supersymmetry breaking terms in the susy lagrangian@xcite . in this model@xcite , supersymmetry is communicated from a hidden sector ( whose dynamics leads to the breaking of supersymmetry ) to the observable sector ( consisting of the fields of the minimal supersymmetric standard model or mssm ) via gravitational interactions . with the assumption of canonical kinetic terms for scalars in the lagrangian , this leads to a universal mass @xmath13 for all scalar particles at some high scale @xmath16 , usually taken to be @xmath17 . at @xmath17 , gaugino masses and trilinear terms are assumed to unify at @xmath14 and @xmath18 , respectively . these parameters , along with the bilinear soft term @xmath19 , provide boundary conditions for the renormalization group evolution of the various soft terms from @xmath17 to @xmath20 . requiring in addition radiative electroweak symmetry breaking leaves a rather small parameter set @xmath21 from which the entire susy particle mass spectrum and mixing parameters may be derived . the flavor changing neutral current decay of the bottom quark @xmath4 is well known to be particularly sensitive to new physics effects . new weak scale particles ( _ e.g. _ , a chargino @xmath22 and top squark @xmath23 ) could give loop contributions which would be comparable to the standard model ( sm ) @xmath24 loop amplitude . measurements from the cleo experiment@xcite restrict the inclusive @xmath25 branching ratio to be @xmath26 , where @xmath27 at 95% cl . many analyses have been performed@xcite which compare theoretical predictions of susy models to experimental results . in a previous report@xcite , predictions of the @xmath4 decay rate were made as functions of the msugra model parameter space . in this study , a number of qcd improvements were incorporated into the calculation which reduced the inherent uncertainty of the @xmath4 decay rate predictions due to the qcd scale choice from @xmath28 down to @xmath29 . susy contributions to the @xmath4 decay amplitude included @xmath5 , @xmath30 and @xmath31 loops . results were presented for @xmath32 and @xmath33 , and for both signs of @xmath34 . for @xmath10 , large regions of parameter space were excluded , especially for @xmath35 . for @xmath12 , all the parameter space scanned was allowed by cleo data : in fact , for some ranges of parameters , the model predicts values of @xmath36 close to the central value measured by the cleo collaboration . recently , sparticle mass spectra and sparticle decay branching ratios in the msugra model have been reanalysed for large values of the parameter @xmath0@xcite . in the msugra model , the range of @xmath0 is typically @xmath37 , where the lower limit depends somewhat on the precise value of @xmath38 . for @xmath8 , @xmath39 and @xmath40 yukawa couplings become non - negligible and can affect the sparticle mass spectrum and decay branching fractions . the upper and lower limits on @xmath0 are set by a combination of requiring a valid solution to radiative electroweak symmetry breaking , and requiring perturbativity of third generation yukawa couplings between the scales @xmath20 and @xmath17 . some optimization of scale choice at which the one - loop effective potential is minimized was found to be needed in ref . @xcite in order to gain stable sparticle and higgs boson mass contributions . this scale optimization effectively includes some portion of two - loop corrections to the effective potential@xcite . it was shown that the mass of the pseudoscalar higgs boson @xmath41 , and the related masses of @xmath42 and @xmath43 , suffer a sharp decrease as @xmath0 increases . in addition , the masses of the lighter tau slepton @xmath44 and bottom squark @xmath45 also decrease , although less radically . naively , one might expect corresponding increases in the loop contributions to @xmath4 decay involving @xmath45 and @xmath43 . indeed , borzumati has shown in ref . @xcite that as @xmath46 decreases , the charged higgs contribution to @xmath4 decay does increase . however , for large values of @xmath0 , the chargino loop contributions increase even more dramatically , and dominate the decay amplitude . she further notes that at intermediate to large @xmath0 values , there is a non - negligible contribution from @xmath6 loops . in this paper , we re - examine constraints on the msugra model from @xmath4 decay at large @xmath0 . in doing so , we incorporate several improvements over previous analyses . * we present our analysis using updated msugra mass predictions for large @xmath0 , using a renormalization group improved one - loop effective potential with optimized scale choice @xmath47 . we use an updated value of top mass @xmath48 gev . * we include in this analysis contributions from @xmath49 and @xmath50 loops . these contributions require knowledge of the full squark mixing matrices , and hence an improved calculation of renormalization group evolution of soft susy breaking parameters . * as in ref . @xcite , we include the dominant next - to - leading order ( nlo ) virtual and bremsstrahlung corrections to the operators mediating @xmath4 decay at scale @xmath51 . in addition , we include nlo rg evolution of wilson coefficients between scales @xmath52 and @xmath53 . we also include appropriate renormalization group evolution of wilson coefficients at high scales @xmath54 for @xmath5 , @xmath30 and @xmath31 loops following the procedure of anlauf@xcite . the corresponding rg evolution of wilson coefficients for @xmath49 and @xmath50 loops is not yet available . * we compare our results to recent calculations at large @xmath0 of the neutralino relic density and direct dark matter detection rates for the msugra model . in sec . ii of this paper , we present some details of our calculations , especially those regarding the inclusion of @xmath49 and @xmath50 loops . in sec . iii , we present qcd improved results for the @xmath4 branching fraction in msugra parameter space for large @xmath0 . in this section , we also make comparisons with cosmological and collider search expectations . in sec . iv , we relax some of the assumptions of the msugra framework to see whether @xmath49 loops can become large or even dominant . this question is important when considering the model dependence of our results . the calculation of the width for @xmath4 decay proceeds by calculating the loop interaction for @xmath4 within a given model framework , _ e.g. _ , msugra , at some high mass scale @xmath55 , and then matching to an effective theory hamiltonian given by @xmath56 where the @xmath57 are wilson coefficients evaluated at scale @xmath16 , and the @xmath58 are a complete set of operators relevant for the process @xmath59 , given , for example , in ref . all orders approximate qcd corrections are included via renormalization group resummation of leading logs ( ll ) which arise due to a disparity between the scale at which new physics enters the @xmath60 loop corrections ( usually taken to be @xmath55 ) , and the scale at which the @xmath60 decay rate is evaluated ( @xmath51 ) . resummation then occurs when we solve the renormalization group equations ( rge s ) for the wilson coefficients @xmath61 where @xmath62 is the @xmath63 anomalous dimension matrix ( adm ) , and @xmath64 the matrix elements of the operators @xmath58 are finally calculated at a scale @xmath51 and multiplied by the appropriately evolved wilson coefficients to gain the final decay amplitude . the dominant uncertainty in this leading - log theoretical calculation arises from an uncertainty in the scale choice @xmath16 at which effective theory decay matrix elements are evaluated . varying @xmath16 between @xmath65 to @xmath66 leads to a theoretical uncertainty of @xmath67% . recently , next - to - leading order qcd corrections have been completed for @xmath4 decay . these include _ i _ ) complete virtual corrections@xcite to the relevant operators @xmath68 and @xmath69 which , when combined with bremsstrahlung corrections@xcite results in cancellation of associated soft and collinear singularities ; _ ii _ ) calculation of @xmath70 contributions to the adm elements @xmath71 for @xmath72 ( by ciuchini _ et al._@xcite ) , for @xmath73 by misiak and mnz@xcite , and for @xmath74 by chetyrkin , misiak and mnz@xcite . in addition , if two significantly different masses contribute to the loop amplitude , then there can already exist significant corrections to the wilson coefficients at scale @xmath52 . in this case , the procedure is to create a tower of effective theories with which to correctly implement the rg running between the multiple scales involved in the problem . the relevant operator bases , wilson coefficients and rge s are given by cho and grinstein@xcite for the sm and by anlauf@xcite for the mssm . the latter analysis includes contributions from just the @xmath5 , @xmath75 and @xmath76 loops ( which are the most important ones ) . we include the above set of qcd improvements ( with the exception of @xmath74 , which has been shown to be small@xcite ) into our calculations of the @xmath4 decay rate for the msugra model . the contributions to @xmath77 and @xmath78 from @xmath6 and @xmath7 loops ( susy contributions to @xmath79 are suppressed by additional factors of @xmath80 ) have been presented in ref . @xcite , although some defining conventions must be matched between ref . @xcite and ref . @xcite and ref . the only complication is that the squark mixing matrix @xmath81 which enters the couplings must be derived . to accomplish this , we incorporate the following procedure into our program for renormalization group running . * we first calculate the values of all running fermion masses in the sm at the mass scale @xmath82 . from these , we derive the corresponding yukawa couplings @xmath83 , @xmath84 and @xmath85 for each generation , and construct the corresponding yukawa matrices @xmath86 , @xmath87 and @xmath88 , where @xmath89 runs over the 3 generations . we choose a basis that yields flavor diagonal matrices for @xmath87 and @xmath88 , whereas the ckm mixing matrix creates a non - diagonal matrix @xmath86@xcite . * the three yukawa matrices are evolved within the mssm from @xmath90 up to @xmath91 and the values are stored . we use 1-loop rges except for the evolution of gauge couplings . * at @xmath91 , the matrices @xmath92 , @xmath93 and @xmath94 are constructed ( assuming @xmath95 ) . the squark and slepton mass squared matrices @xmath96 are also constructed , where @xmath97 and @xmath98 . these matrices are assumed diagonal at @xmath91 with entries @xmath99 . * the @xmath100 and @xmath96 matrices are evolved along with the rest of the gauge / yukawa couplings and soft susy breaking terms between @xmath82 and @xmath17 iteratively via runge - kutta method until a stable solution is found . the entire solution requires the simultaneous solution of 134 coupled rge s ( with some slight redundancy ) . we use 1-loop rges except for the evolution of gauge couplings . * at @xmath90 , the @xmath101 @xmath102-squark mass squared matrix is constructed . numerical diagonalization of this matrix yields the squark mass mixing matrix @xmath81 which is needed for computation of the @xmath6 and @xmath7 loop contributions . at this point , the wilson coefficients @xmath77 and @xmath78 can be calculated and evolved to @xmath51 as described above , so that the @xmath4 decay rate can be calculated @xcite . as an example , we show in fig . [ nfig1 ] the calculated contributions to the wilson coefficient @xmath77 versus @xmath0 for the msugra point @xmath103 gev , @xmath104 and @xmath12 . in frame _ a _ ) , we show contributions from @xmath105 loops , as well as from @xmath5 and @xmath75 . the @xmath5 contribution is of course constant , while the @xmath75 contribution is of the same sign , and increasing slightly in magnitude . the various contributions from chargino loops increase roughly linearly with @xmath0 at a much faster rate , and thus form the dominant components of the @xmath4 decay amplitude . in the case shown , there are several large negative as well as positive contributions , so that significant cancellations take place . the sum of all chargino loop contributions is shown by the dotted curve . in frame _ b _ ) we show the contributions to @xmath77 from different @xmath6 loops . these contributions vary with @xmath0 as well and are comparable to corresponding contributions from chargino loops . the sum of all gluino loop contributions is shown by the dotted curve ; in this case , however , the cancellation amongst the various loop contributions is nearly complete . in frame _ c _ ) , we show the individual and summed contributions from @xmath7 loops . these also increase with @xmath0 but , as expected , are tiny compared to the @xmath105 and @xmath6 loop contributions . the sum is again shown by the dotted curve . here , the cancellations are not as complete as in the gluino loop case due to the higgsino interactions of the neutralinos which increase with @xmath0 . our first numerical results for @xmath11 decay are shown in fig . [ nfig2 ] , where we plot the branching fraction versus @xmath0 for the msugra point @xmath103 gev , @xmath104 and for _ a _ ) @xmath10 and _ b _ ) @xmath12 . the sm value , after qcd corrections , is @xmath106 , where the error comes from varying the scale choice @xmath107@xcite . the sm result is denoted by the dot - dashed line , and of course does not vary with @xmath0 . if we include in addition the contribution from the @xmath75 loop , then we obtain the dotted curves , which always increase the value of @xmath11 . for this parameter space point , including the @xmath75 loop always places the value of @xmath11 above the cleo 95% cl excluded region of @xmath108 . if we include the full contribution of supersymmetric particles to the computation of @xmath11 , then we arrive at the solid curves in fig . [ nfig2 ] . for the @xmath10 case in frame _ a _ ) , the susy loops increase the branching fraction , which increases with @xmath0 , so that the cleo restriction on @xmath11 severely constrains the msugra model for large @xmath0 . for the frame _ b _ ) case with @xmath12 , the susy loop contributions generally act to decrease the branching fraction , so that much of the parameter space is allowed for moderate values of @xmath0 . ultimately , as @xmath0 increases , the decrease in @xmath11 becomes so severe that the msugra model becomes in conflict with the cleo lower 95% cl bound that @xmath109 , so that for this particular msugra point , all values of @xmath110 are excluded for the particular choice of msugra parameters . in figure [ nfig3 ] , we show the main result of this paper : the contours of constant @xmath11 in the @xmath111 parameter plane for large @xmath112 , for @xmath104 and for _ a _ ) @xmath10 and _ b _ ) @xmath12 . the contours are evaluated at a renormalization scale choice @xmath113 . the region marked by th is disallowed by theoretical considerations : either electroweak symmetry is not properly broken ( the large @xmath13 , small @xmath14 region ) or the lightest neutralino @xmath114 is not the lightest susy particle ( lsp ) . for small @xmath13 , the light tau slepton @xmath44becomes so light that in the th region , @xmath115 . the region denoted by ex is excluded by lep2 constraints which require that the light chargino mass @xmath116 gev . in frame _ a _ ) , we see the value of @xmath11 is large throughout the entire parameter space plane . the region with small values of @xmath13 and @xmath14 which is most favored by fine - tuning considerations@xcite is in the most severe violation of the cleo constraint . the region below the dotted contour is in violation of the cleo 95% cl bound for _ all _ choices of renormalization scale @xmath107 . thus , susy models allowed by cleo for @xmath112 and @xmath10 would be required to have @xmath117 gev ( at @xmath118 gev ) and @xmath119 gev ( for @xmath120 gev ) . in frame _ b _ ) for @xmath12 , we see that the values of @xmath11 are uniformly _ below _ the sm value , and so usually in better agreement with the cleo measured value of @xmath121 ( where errors have been combined in quadrature ) . in fact , we note that the region with @xmath122 gev agrees with the cleo central value for @xmath11 ! this region corresponds to parameter space points with @xmath123 gev , and @xmath124 gev . in this frame , the entire plane shown , except the region below the dotted contour , is _ allowed _ by the cleo constraint . the region below the dotted contour falls below the cleo 95% cl value of @xmath109 for all values of scale choice @xmath107 . this is again the region most favored by fine - tuning . in this plane , @xmath125 gev and @xmath126 gev . up to this point , we have only shown results for a constant value of @xmath104 . in figure [ nfig4 ] , we show contours of constant @xmath11 in the @xmath127 plane for @xmath128 gev , for @xmath112 and _ a _ ) @xmath10 and _ b _ ) @xmath12 . for frame _ a _ ) , we see that the branching fraction can change by typically a factor of 2 over the parameter range shown , with most of the variation occuring for changes in @xmath13 , instead of with @xmath18 . the entire plane shown in frame _ a _ ) is excluded by the cleo bound . in frame _ b _ ) , for @xmath12 , the branching fraction can change by up to a factor of @xmath129 over the plane shown , again with most of the variation coming due to changes in @xmath13 . in this case , the region to the left of the dotted contour is excluded by the cleo bound for all choices of renormalization parameter @xmath130 . an important constraint on the msugra model comes from implications for the relic density of dark matter in the universe . the idea here is that in the very early universe , the lsp ( the lightest neutralino ) was a constituent of the matter and radiation assumed to be in thermal equilibrium at some very high temperature . as the universe expanded and cooled , the lsp s could no longer be produced , although they could still annihilate with one another . upon further expansion , the neutralino flux dropped to such low levels that further annihilations would rarely occur , and a relic abundance of neutralinos was locked in . these relic lsp s could make up the bulk of dark matter in the universe today . the neutralino relic density is calculable as a function of msugra model parameter space@xcite . the relic density is usually parametrized in terms of @xmath131 , where @xmath132 , @xmath133 is the relic density , @xmath134 is the critical closure density of the universe ( @xmath135 ) and @xmath136 km / sec / mpc is the scaled hubble constant with @xmath137 . a value of @xmath138 implies a universe with age less than 10 billion years , in conflict with the ages of the oldest stars . if @xmath139 , then the relic density of neutralinos can not even account for the dark matter required by galactic rotation curves . some popular cosmological models that account for the cobe cosmic microwave background measurements as well as structure formation in the universe actually prefer a mixed dark matter ( mdm ) universe@xcite , with a matter density ratio of 0.3/0.6/0.1 for a hot dark matter / cold dark matter / baryonic matter mix . in this case , values of @xmath140 are preferred . in fig . [ nfig5 ] , we show contours of constant relic density @xmath131 . the region to the right of the solid @xmath141 contour is excluded by the age of the universe constraint , while the region below the dotted contour has @xmath139 . the region betwen the dashed - dotted contours is favored by a mdm universe . the region excluded by cleo data is below the solid contour labelled @xmath4 . in frame _ a _ ) , we see that combining the two constraints allows only a small patch of allowed parameter space with @xmath142 gev and @xmath143 gev . over almost all of this region , the relic density @xmath144 . in frame _ b _ ) , the @xmath4 excluded region hardly intersects with the mdm region , so that large regions of parameter space are favorable for cosmology as well as for cleo constraints ! in fig . [ nfig5 ] we also plot one contour for expected rates for direct detection of neutralino dark matter via cryogenic dark matter detectors@xcite . the calculations have been performed for neutralino scattering from a @xmath145 detector . current experiments are sensitive to detection rates of 1 - 10 events / day / kg of detector . the goal of such experiments is to achieve a sensitivity of @xmath146 events / kg / day by about the year 2000 . towards this end , we show the 0.01 event / kg / day contour in the @xmath111 plane ; below the contour the event rates exceed the 0.01/kg / day benchmark . in frame _ a _ ) , we see that the region accessible to direct neutralino detection coincides with the region with very large @xmath11 rates well beyond the cleo 95% cl limit . however , in frame _ b _ ) , for @xmath12 , there exists a significant region with large direct detection rates , which is allowed by the cleo constraint , and is also in the favorable cosmological region ! the lep2 @xmath147 collider is expected to reach a peak cm energy of @xmath148 gev , which should allow sm higgs bosons of mass @xmath149 gev to be explored . for the @xmath112 value shown in fig . [ nfig5 ] , the light higgs scalar @xmath150 gev over the entire @xmath111 plane shown . hence , for these values of @xmath0 , we would expect no higgs signals to be seen at lep2 . the reach of lep2 via @xmath151 and @xmath152 searches is shown by the dashed contour just above the region marked ex . this contour is defined by requiring @xmath153 gev and @xmath154 gev . for both cases of @xmath10 and @xmath12 shown in fig . [ nfig5 ] , the lep2 sparticle reach falls below both the @xmath11 excluded region , and below the @xmath155 contours . if we accept the msugra model literally , then the prediction is that lep2 should see no evidence for either a higgs or susy if @xmath156 . the fermilab tevatron @xmath15 collider is expected to operate at @xmath157 tev in run 2 , and to amass @xmath158 fb@xmath159 of integrated luminosity by use of the main injector ( mi ) . ultimately , experiments hope to acquire @xmath67fb@xmath159 of integrated luminosity under the tev33 program . recently completed calculations of the reach of the tevatron mi for msugra at large @xmath156 show a maximal reach in @xmath14 to @xmath160 gev in the @xmath161jets channel@xcite . a similar reach has been calculated for tev33 , and finds points with @xmath162 accessible . comparing these regions to fig . [ nfig5 ] shows that , like lep2 , the reach of tevatron mi and tev33 are below both the @xmath11 excluded contour and the @xmath155 contour , making discovery of susy particles highly unlikely for msugra if @xmath0 is large . over much of the parameter space plane in frame _ b _ ) of fig . [ nfig5 ] , however , @xmath163 gev , which ( optimistically ) corresponds to the maximal reach for @xmath164 at tev33 . hence , if msugra is correct and @xmath165 , then tev33 experiments may see a hint of the higgs boson in their data sample . of course , the entire large @xmath0 parameter space shown should be easily visible in at least the jets@xmath166 channel at the cern lhc , even with modest integrated luminosity . it is well known that within the msugra framework the chargino loop gives the dominant susy contribution to the amplitude for the decay @xmath59 . this can also be seen from fig . [ nfig1 ] where we see that while the contributions from the gluino loops are individually comparable ( or even larger ! ) than those from chargino loops , these cancel out almost completely leaving only a small residual contribution . in contrast , while there is indeed considerable cancellation amongst the various chargino contributions , there is nonetheless a sizeable residue that remains . we may understand the large cancellations among the gluino contributions in analogy with the familiar gim cancellation in the sm : indeed such a cancellation would be exact if squarks were precisely degenerate ( _ i.e. _ , the squark mass matrix is proportional to the unit matrix ) because we can then , by a unitary transformation , align the squark and quark mass matrices , so that the gluino - squark - quark vertex is exactly flavour diagonal . within the msugra framework with universal soft breaking squark mass matrices at the unification scale , squarks are indeed ( approximately ) degenerate , and gluino loop contributions to the flavour violating @xmath59 decay are suppressed . the gim - like cancellation that we have described above does not occur when yukawa couplings enter the calculation as occurs , for instance , via the higgsino components of chargino and neutralino loops . in this case , as can be seen from fig . [ nfig1 ] , the cancellation is indeed incomplete ( particularly for the chargino where the large top quark yukawa coupling enters ) . these considerations lead us to examine whether the breaking of the degeneracy of the soft susy - breaking squark masses at the unification scale so strongly upsets the delicate cancellations that it results in large gluino contributions to the amplitude for @xmath59 decay . of course , by allowing soft - breaking mass squared matrices with arbitrary off - diagonal entries , it should be possible to get very large flavour violating gluino interactions . the issue that we address , however is whether large gluino contributions are possible even if we choose these soft squark matrices to be diagonal at the unification scale . as we will soon see , the physics of this ansatz is basis - dependent . to parametrize the breaking of the squark degeneracy we begin by noting that we may always choose a quark basis so that _ either _ the down or the up type yukawa couplings are diagonal at the weak scale . we will call these the @xmath102- and @xmath167- cases , respectively . next , these couplings are evolved to @xmath17 , where both up and down yukawa matrices have off - diagonal components . in the @xmath102-case , the down type yukawa matrices get off - diagonal contributions just from the rge , while the up type yukawas start off off - diagonal right at the weak scale ; in the @xmath167-case , the situation is reversed . up to now , the choice to work in the @xmath167 or @xmath102 cases is purely a matter of convention , and indeed in previous sections we have used the @xmath102 case . this is , however , no longer the case if we further assume that the soft breaking squark mass squared matrix is diagonal ( but not a multiple of the identity ) at the gut scale . this is because the transformation that takes us from the @xmath102-case to the @xmath167-case does not leave the squark mass squared matrix diagonal ( except in the case when this matrix is @xmath168 ) . the @xmath167- and @xmath102- cases are thus physically distinct . we have , therefore , studied these two cases separately . to keep things simple , we split only the @xmath39-squarks ( @xmath169 splits with @xmath170 , of course ) keeping the others degenerate at @xmath13 . the splitting is given by a single parameter @xmath171 where @xmath53 and @xmath13 are soft squark masses at the gut scale . thus @xmath172 corresponds to the msugra case . we further consider three possibilities where ( _ i _ ) just @xmath173 , ( _ ii _ ) just @xmath174 and ( _ iii _ ) both @xmath173 and @xmath174 masses are split from those of other squarks . the results of our calculation of the gluino contribution to @xmath77 where the squark degeneracy is broken as described above is shown in fig . [ nfig6 ] for the @xmath102-case labelled @xmath102-diagonal , and for the @xmath167-case , labelled @xmath167-diagonal . in each frame , we have three curves labelled @xmath175 , @xmath176 and @xmath177 for the cases where just left , just right , and both left and right sbottom soft masses are different from @xmath13 . in our calculation , we have chosen @xmath178 gev , @xmath128 gev , @xmath179 , @xmath104 and @xmath180 . we choose a large value of @xmath13 so that the squark masses are not dominated by @xmath181 ( in which case splitting due to non - universal soft mass term would be unimportant ) . also , for the large value of @xmath0 the bottom yukawa coupling is significant . the following features are worth noting . * for @xmath172 the value of @xmath182 is the same for the up and down cases for reasons that we have already explained . * for non - degenerate squarks , gluino loop effects are significantly larger in the @xmath167-case . this may be understood if we recall that in the @xmath102-case , the mixing of down type yukawas at @xmath17 arises _ only _ due to rge . * non - degeneracy effects when just the right squarks are split show only small variation with the non - degeneracy parameter @xmath183 because flavour mixing in the right squark sector is suppressed . * somewhat surprising is the fact that despite the large degree of non - degeneracy , the gluino contribution to @xmath77 , which increases by up to an order of magnitude relative to that in the msugra case , never becomes really large . for our choice of parameters , this may be partly due to the fact that gluinos and squarks are significantly heavier than @xmath184 and @xmath185 . this is not the complete reason though . in our computation we find that the three main contributions ( from the two sbottoms and the @xmath186 ) cancel one another leaving a remainder that is typically smaller than 10 - 15% of the largest contribution . while this cancellation is much less complete than in the msugra case , we are unable to give a simple argument for why such a cancellation occurs . finally , we comment on the neutralino contributions for the case of non - degenerate squarks . we have already noted that cancellations among various chargino contributions are incomplete because of the effect of large higgsino couplings to the @xmath187 system . for large values of @xmath0 , we may expect a similar effect for neutralinos . we may further guess that this effect is largest in the @xmath167-case where flavor mixing does not originate solely in the rge . we have checked that for values of parameters in fig . [ nfig6 ] above a small value of @xmath188 , the neutralino contribution to @xmath77 is indeed enhanced by a factor of 6 - 7 above its msugra value , and further , that this enhancement is largely due to incompleteness in the cancellation between various contributions . for very large values of @xmath183 , @xmath77 is about 1.5 times its msugra value , but opposite in sign . we thus conclude that while the neutralino contribution is somewhat sensitive to the splitting of squark masses , it never appears to become very dominant . to summarize the results of this section , we see that with our assumptions , gluino contributions to the amplitude for the @xmath189 decay never dominate the susy contribution . this contribution may nonetheless be non - negligible even if gluinos and squarks are well beyond the reach of the tevatron ( and its proposed upgrades ) as seen in fig . [ nfig6 ] . we emphasize though that our conclusion is special to models where all the flavour violation in the gluino - squark - quark vertex at the gut scale comes from non - diagonal yukawa interactions . larger contributions from gluino loops may be possible in other models . _ note added : _ after completion of this manuscript , a related paper by blazek and raby appeared on the topic of @xmath4 constraints on @xmath190 susy models at large @xmath0@xcite . since ref . @xcite adopts a particular @xmath190 framework and does not include the radiative electroweak symmetry breaking constraint , comparison of results between the two papers is not straightforward . also , we have not included @xmath191 mixing included in ref . @xcite in our evaluation of the chargino loop . we are grateful to m. drees and t. ter veldhuis for helpful conversations and comments . we thank the aspen center for physics for hospitality while a portion of this work was completed . this research was supported in part by the u. s. department of energy under grant number de - fg-05 - 87er40319 . e.g. _ , h. haber in _ woodlands superworld _ , hep - ph/9308209 ( 1993 ) . e.g. _ , m. drees and s. martin in _ electroweak symmetry breaking and new physics at the tev scale _ , edited by t. barklow , s. dawson , h . haber and j. seigrist , ( world scientific ) 1995 ; see also j. amundson _ _ in _ new directions for high energy physics _ , edited by d. g. cassel , l. trindle gennari and r. h. siemann , ( stanford linear accelerator center , 1996 ) , hep - ph/9609374 . a. chamseddine , r. arnowitt and p. nath , phys . lett . * 49 * , 970 ( 1982 ) ; r. barbieri , s. ferrara and c. savoy , phys . lett . * b119 * , 343 ( 1982 ) ; l.j . hall , j. lykken and s. weinberg , phys . 2359 ( 1983 ) . m. s. alam _ et al . _ , ( cleo collaboration ) , phys . rev . lett . * 74 * , 2885 ( 1995 ) . note that recently the aleph collaboration ( at the international europhysics conference on high energy physics , jerusalem , israel , aug . 19 - 26 , 1997 ) has made a preliminary announcement of a measurement of @xmath192 . we do not include this preliminary result in our analysis , and caution the reader that some of our conclusions may change if the @xmath193 decay rate turns out to be closer to this aleph value . s. bertolini , f. borzumati , a. masiero and g. ridolfi , nucl . b*353 * , 591 ( 1991 ) . r. barbieri and g. f. giudice , phys . b*309 * , 86 ( 1993 ) ; j. lopez , d. nanopoulos and g. park , phys . rev . d*48 * , 974 ( 1993 ) ; n. oshimo , nucl . b*404 * , 20 ( 1993 ) ; r. garisto and j. ng , phys . b*315 * , 372 ( 1993 ) ; m. diaz , phys . b*322 * , 207 ( 1994 ) ; y. okada , phys . b*315 * , 119 ( 1993 ) ; f. borzumati , zeit . fr physik c*63 * , 291 ( 1994 ) ; p. nath and r. arnowitt , phys . b*336 * , 395 ( 1994 ) ; g. kane , c. kolda , l. roszkowski and j. wells , phys . rev . d*49 * , 6173 ( 1994 ) ; f. borzumati , m. drees and m. nojiri , phys . rev . d*51 * , 341 ( 1995 ) ; v. barger , m. berger , p. ohmann and r. phillips , phys . rev . d*51 * , 2438 ( 1995 ) ; f. bertolini and f. vissani , zeit . fr physik c*67 * , 513 ( 1995 ) ; j. lopez , d. nanopoulos , x. wang and a. zichichi , phys . rev . d*51 * , 147 ( 1995 ) ; j. wu , r. arnowitt and p. nath , phys . rev . d*51 * , 1371 ( 1995 ) ; b. de carlos and j. a. casas , phys . b*349 * , 300 ( 1995 ) and erratum-_ibid _ b*351 * , 604 ( 1995 ) . h. baer and m. brhlik , phys . d*55 * , 3201 ( 1997 ) . h. baer , c. h. chen , m. drees , f. paige and x. tata , phys . * 79 * , 986 ( 1997 ) . m. carena , j. espinosa , m. quiros and c. wagner , phys . b355 * , 209 ( 1995 ) ; m. carena , m. quiros , c.e.m . wagner , nucl . * b461 * , 407 ( 1996 ) ; h. haber , r. hempfling and a. hoang , z. phys . * c*75 , 539 ( 1997 ) . see f. borzumati , ref . h. anlauf , nucl . b*430 * , 245 ( 1994 ) . e.g. _ , b. grinstein , m. j. savage and m. wise , nucl . b*319 * , 271 ( 1989 ) ; a. ali , in _ 20th international nathiagali summer college on physics and contemporary needs _ , bhurban , pakistan , 1995 , hep - ph/9606324 . c. greub , t. hurth and d. wyler , phys . b*380 * , 385 ( 1996 ) and phys . rev . d*54 * , 3350 ( 1996 ) . a. ali and c. greub , z. phys . c*49 * , 431 ( 1991 ) , phys . b*259 * , 182 ( 1991 ) , * 287 * , 191 ( 1992 ) and * 361 * , 146 ( 1995 ) ; z. phys . c*60 * , 433 ( 1993 ) ; n. pott , phys . * , 938 ( 1996 ) ; m. ciuchini , e. franco , g. martinelli and l. reina , nucl . b*415 * , 403 ( 1994 ) . m. misiak and m. mnz , phys . b*344 * , 308 ( 1995 ) . k. g. chetyrkin , m. misiak and m. mnz , phys . b400 * , 206 ( 1997 ) . p. cho and b. grinstein , nucl . b*365 * , 279 ( 1991 ) . h. arason _ et al . _ , rev . d*46 * , 3945 ( 1992 ) . @xcite , a. buras , a. kwiatkowski and n. pott , hep - ph/9707482 ( 1997 ) and c. grueb and t. hurth , hep - ph9708214 ( 1997 ) quote a somewhat higher sm value of @xmath11 than is presented here . the above works include a 3% increase due to non - perturbative effects calculated by m. b. voloshin , phys . lett . * b*397 , 275 ( 1997 ) plus the corrected value of @xmath194 in their calculations , which conspire to increase the sm decay rate by about 10% . these effects would similarly increase our estimate of the susy contributions . g. anderson and d. castao , phys . b347 * , 300 ( 1995 ) and phys . rev . d*52 * , 1693 ( 1995 ) ; see also k. l. chan , u. chattopadhyay and p. nath , hep - ph/9710473 ( 1997 ) . for a recent review , see g. jungman , m. kamionkowski and k. griest , phys . rep . * 267 * , 195 ( 1996 ) . see also m. drees and m. nojiri , phys . rev . * d47 * , 376 ( 1993 ) . the result shown here are from h. baer and m. brhlik , phys . * d53 * , 597 ( 1996 ) . further references are included in these reports . e.g. _ , j. primack , astro - ph/9707285 ( 1997 ) . h. baer and m. brhlik , phys . d , in press , hep - ph/9706509 . h. baer , c. h. chen , m. drees , f. paige and x. tata , manuscript in preparation . t. blazek and s. raby , hep - ph/9712257 ( 1997 ) .

in the minimal supergravity model ( msugra ) , as the parameter @xmath0 increases , the charged higgs boson and light bottom squark masses decrease , which can potentially increase contributions from @xmath1 , @xmath2 and @xmath3 loops in the decay @xmath4 . we update a previous qcd improved @xmath4 decay calculation to include in addition the effects of gluino and neutralino loops . we find that in the msugra model , loops involving charginos also increase , and dominate over @xmath5 , @xmath1 , @xmath6 and @xmath7 contributions for @xmath8 . we find for large values of @xmath9 that most of the parameter space of the msugra model for @xmath10 is ruled out due to too large a value of branching ratio @xmath11 . for @xmath12 and large @xmath0 , most of parameter space is allowed , although the regions with the least fine - tuning ( low @xmath13 and @xmath14 ) are ruled out due to too _ low _ a value of @xmath11 . we compare the constraints from @xmath4 to constraints from the neutralino relic density , and to expectations for sparticle discovery at lep2 and the fermilab tevatron @xmath15 colliders . finally , we show that non - universal gut scale soft breaking squark mass terms can enhance gluino loop contributions to @xmath4 decay rate even if these are diagonal .

cosmological observations coming from type ia supernovae @xcite , cosmic microwave background radiation @xcite and the large scale structure @xcite , provide evidences that the universe is currently in an accelerating phase . this result is , in general , ascribed to the existence of a sort of dark energy ( de ) sector in the universe , an exotic energy source characterized by a negative pressure . at late times , the dark - energy sector eventually dominates over the cold dark matter ( cdm ) , and drives the universe to the observed accelerating expansion . the simplest candidate for de is the cosmological constant @xmath2 , which has an equation - of - state parameter @xmath3 . although this model is in agreement with current observations , it is plagued by some difficulties related to the small observational value of de density with respect to the expected one arising from quantum field theories ( the well known cosmological constant problem @xcite ) . moreover , the @xmath2cdm paradigm , where cold dark matter ( cdm ) is considered into the game , may also suffer from the age problem , as it was shown in @xcite , while the present data seem to slightly favor an evolving de with the equation - of - state parameter crossing @xmath3 from above to below in the near cosmological past @xcite . over the past decade several de models have been proposed , such as quintessence @xcite , phantom @xcite , k - essence @xcite , tachyon @xcite , quintom @xcite , chaplygin gas @xcite , generalized chaplygin gas ( gcg ) @xcite , holographic de @xcite , new agegraphic de @xcite , ricci de @xcite etc . on the other hand , there are also numerous models that induce an effective dark energy which arises from modifications of the gravitational sector itself , such as @xmath4 gravity @xcite ( this class is very efficient in verifying observational and theoretical constraints and explain the universe acceleration and phantom crossing @xcite ) , or gravity with higher curvature invariants @xcite , by coupling the ricci scalar to a scalar field @xcite , by introducing a vector field contribution @xcite , or by using properties of gravity in higher dimensional spacetimes @xcite ( for a review see @xcite ) . a possibility that can be explored to explain the accelerated phase of the universe is to consider a theory of gravity based on the weitzenbck connection , instead of the levi - civita one , which deduces that the gravitational field is described by the torsion instead of the curvature tensor . in such theories , the torsion tensor is achieved from products of first derivatives of tetrad fields , and hence no second derivatives appear . this _ teleparallel _ approach @xcite , is closely related to general relativity , except for `` boundary terms '' @xcite that involve total derivatives in the action , and thus one can construct the teleparallel equivalent of general relativity ( tegr ) , which is completely equivalent with general relativity at the level of equations but is based on torsion instead of curvature . teleparallel gravity possesses a number of attractive features related to geometrical and physical aspects @xcite . hence , one can start from tegr and construct various gravitational modifications based on torsion , with @xmath1 gravity being the most studied one @xcite . in particular , it may represent an alternative to inflationary models without the use of the inflaton , as well as to effective de models , in which the universe acceleration is driven by the extra torsion terms @xcite ( for a detailed review , see @xcite ) . the main advantage of @xmath1 gravity is that the field equations are 2nd - order ones , a property that makes these theories simpler if compared to the dynamical equations of other extended theories of gravity , such as @xmath4 gravity . the aim of this paper is to explore the implications of @xmath1 gravity to the formation of light elements in the early universe , i.e. to the big bang nucleosynthesis ( bbn ) . on the other hand , we want to explore the possibility to constrain @xmath1 gravity by bbn observatio nal data . bbn has occurred between the first fractions of second after the big bang , around @xmath5 sec , and a few hundreds of seconds after it , when the universe was hot and dense ( indeed bbn , together with cosmic microwave background radiation , provides the strong evidence about the high temperatures characterizing the primordial universe ) . it describes the sequence of nuclear reactions that yielded the synthesis of light elements @xcite , and therefore drove the observed universe . in general , from bbn physics , one may infer stringent constraints on a given cosmological model . hence , in this work , we shall confront various @xmath1 gravity models with bbn calculations based on current observational data on primordial abundance of @xmath0 , and we shall extract constraints on their free parameters . the layout of the paper is as follows . in section [ revmodel ] we review @xmath1 gravity and the related cosmological models . in section [ bbnanal ] we use bbn calculations in order to impose constraints on the free parameters of specific @xmath1 gravity models . conclusions are reported in section [ conclusions ] . finally , in the appendix we summarize the main notions of bbn physics . let us briefly review @xmath1 gravity , and apply it in a cosmological framework . in this formulation , the dynamical variable is the vierbein field @xmath6 , @xmath7 , which forms an orthonormal basis in the tangent space at each point @xmath8 of the manifold , i.e. @xmath9 , with @xmath10 the minkowsky metric with signature @xmath11 : @xmath12 . denoting with @xmath13 , @xmath14 the components of the vectors @xmath15 in a coordinate basis @xmath16 , one can write @xmath17 . as a convection , here we use the latin indices for the tangent space , and the greek indices for the coordinates on the manifold . the dual vierbein allows to obtain the metric tensor of the manifold , namely @xmath18 . in teleparallel gravity , one adopts the curvatureless weitzenbck connection ( contrarily to general relativity which is based on the torsion - less levi - civita connection ) , which gives rise to the non - null torsion tensor : @xmath19 remarkably , the torsion tensor ( [ torsion ] ) encompasses all the information about the gravitational field . the lagrangian density is built using its contractions , and hence the teleparallel action is given by @xmath20 with @xmath21 , and where the torsion scalar @xmath22 reads as @xmath23 here , it is @xmath24 with @xmath25 the contorsion tensor which gives the difference between weitzenbck and levi - civita connections . finally , the variation of action ( [ action0 ] ) in terms of the vierbiens gives rise to the field equations , which coincide with those of general relativity . that is why the above theory is called the teleparallel equivalent of general relativity ( tegr ) . one can now start from tegr , and generalize action ( [ action0 ] ) in order to construct gravitational modifications based on torsion . the simplest scenario is to consider a lagrangian density that is a function of @xmath22 , namely @xmath26},\ ] ] that reduces to tegr as soon as @xmath27 . considering additionally a matter lagrangian @xmath28 , variation with respect to the vierbein gives the field equations @xcite @xmath29- e_i^\lambda { t^\rho}_{\mu\lambda}{s_\rho}^{\nu\mu}[1+f']\nonumber\\ & & + e^\rho_i { s_\rho}^{\,\,\mu\nu}(\partial_\mu t)f '' + \frac{1}{4}e^\nu_i [ t+f]=4\pi g\,{e_i}^\rho\ , { \theta_\rho}^\nu\,,\end{aligned}\ ] ] where @xmath30 , @xmath31 and @xmath32 is the energy - momentum tensor for the matter sector . in order to explore the cosmological implications of @xmath1 gravity , we focus on homogeneous and isotropic geometry , considering the usual choice for the vierbiens , namely @xmath33 which corresponds to a flat friedmann - robertson - walker ( frw ) background metric of the form @xmath34 where @xmath35 is the scale factor . equations ( [ torsion ] ) , ( [ lagrangian ] ) , ( [ s ] ) and ( [ contorsion ] ) allow to derive a relation between the torsion @xmath22 and the hubble parameter @xmath36 , namely @xmath37 hence , in the case of frw geometry , and assuming that the matter sector corresponds to a perfect fluid with energy density @xmath38 and pressure @xmath39 , the @xmath40 component of ( [ equations ] ) yields @xmath41+[t+f]=16\pi g\rho,\ ] ] while the @xmath42 component gives @xmath43-(t - f)=16\pi gp.\ ] ] the equations close by considering the equation of continuity for the matter sector , namely @xmath44 . one can rewrite ( [ friedmann ] ) and ( [ acceleration ] ) in the usual form @xmath45 @xmath46 where @xmath47,\\ p_t&=&\frac{1}{16\pi g}\ , \frac{f - t f'+ 2t^2f''}{1+f'+ 2{t}f '' } \ , , \label{pt}\end{aligned}\ ] ] are the effective energy density and pressure arising from torsional contributions . one can therefore define the effective torsional equation - of - state parameter as @xmath48 in these classes of theories , the above effective torsional terms are responsible for the accelerated phases of the early or / and late universe @xcite . let us present now three specific @xmath1 forms , which are the viable ones amongst the variety of @xmath1 models with two parameters out of which one is independent , i.e which pass the basic observational tests @xcite . 1 . the power - law model by bengochea and ferraro ( hereafter @xmath49cdm ) @xcite is characterized by the form @xmath50 where @xmath51 and @xmath52 are the two model parameters . inserting this @xmath1 form into friedmann equation ( [ friedmann ] ) at present , we acquire @xmath53 where @xmath54 is the matter density parameter at present , and @xmath55 is the current hubble parameter value . the best fit on the parameter @xmath52 is obtained taking the @xmath56 observational data , and it reads @xcite @xmath57 clearly , for @xmath58 the present scenario reduces to @xmath2cdm cosmology , namely @xmath59 , with @xmath60 . 2 . the linder model ( hereafter @xmath61cdm ) @xcite arises from @xmath62 with @xmath63 and @xmath39 ( @xmath64 ) the two model parameters . in this case ( [ friedmann ] ) gives that @xmath65 the @xmath56 observational data imply that the best fit of @xmath64 is @xcite @xmath66 as we can see , for @xmath67 the present scenario reduces to @xmath2cdm cosmology . 3 . motivated by exponential @xmath4 gravity @xcite , bamba et . al . introduced the following @xmath1 model ( hereafter @xmath68cdm ) @xcite : @xmath69 with @xmath63 and @xmath39 ( @xmath64 ) the two model parameters . in this case we acquire @xmath70 for this model , and using @xmath56 observational data , the best fit is found to be @xcite @xmath71 similarly to the previous case we can immediately see that @xmath68cdm model tends to @xmath2cdm cosmology for @xmath67 . the above @xmath1 models are considered viable in literature because pass the basic observational tests @xcite . they are characterized by two free parameters . actually there are two more models with two free parameters , namely the logarithmic model @xcite , @xmath72 and the hyperbolic - tangent model @xcite , @xmath73 nevertheless since these two models do not possess @xmath2cdm cosmology as a limiting case and since they are in tension with observational data @xcite , in this work we do not consider them . finally , let us note that one could also construct @xmath1 models with more than two parameters , for example , combining the above scenarios . however , considering many free parameters would be a significant disadvantage concerning the corresponding values of the information criteria . in the section , we examine the bbn in the framework of @xmath1 cosmology . as it is well known , bbn occurs during the radiation dominated era . the energy density of relativistic particles filling up the universe is given by @xmath74 , where @xmath75 is the effective number of degrees of freedom and @xmath76 the temperature ( in the appendix we review the main features related to the bbn physics ) . the neutron abundance is computed via the conversion rate of protons into neutrons , namely @xmath77 and its inverse @xmath78 . the relevant quantity is the total rate given by @xmath79 explicit calculations of eq . ( [ lambda ] ) lead to ( see ( [ lambdafinapp ] ) in the appendix ) @xmath80 where @xmath81 is the mass difference of neutron and proton , and @xmath82gev@xmath83 . the primordial mass fraction of @xmath84 can be estimated by making use of the relation @xcite @xmath85 here @xmath86 , with @xmath87 the time of the freeze - out of the weak interactions , @xmath88 the time of the freeze - out of the nucleosynthesis , @xmath89 the neutron mean lifetime given in ( [ rateproc3 ] ) , and @xmath90 is the neutron - to - proton equilibrium ratio . the function @xmath91 is interpreted as the fraction of neutrons that decay into protons during the interval @xmath92 $ ] . deviations from the fractional mass @xmath93 due to the variation of the freezing temperature @xmath94 are given by @xmath95 \frac{\delta { \cal t}_f}{{\cal t}_f}\,,\ ] ] where we have set @xmath96 since @xmath97 is fixed by the deuterium binding energy @xcite . the experimental estimations of the mass fraction @xmath93 of baryon converted to @xmath98 during the big bang nucleosynthesis are @xcite @xmath99 inserting these into ( [ deltayp ] ) one infers the upper bound on @xmath100 , namely @xmath101 during the bbn , at the radiation dominated era , the scale factor evolves as @xmath102 , where @xmath103 is cosmic time . the torsional energy density @xmath104 is treated as a perturbation to the radiation energy density @xmath38 . the relation between the cosmic time and the temperature is given by @xmath105 ( or @xmath106mev ) . furthermore , we use the entropy conservation @xmath107 . the expansion rate of the universe is derived from ( [ modfri ] ) , and can be rewritten in the form @xmath108 where @xmath109 ( @xmath110 is the expansion rate of the universe in general relativity ) . thus , from the relation @xmath111 , one derives the freeze - out temperature @xmath112 , with @xmath113 mev ( which follows from @xmath114 ) and @xmath115 from which , in the regime @xmath116 , one obtains : @xmath117 with @xmath118gev@xmath83 . in what follows we shall investigate the bounds that arise from the bbn constraints , on the free parameters of the three @xmath1 models presented in the previous section . these constraint will be determined using eqs . ( [ deltat / tboundg ] ) and ( [ rhot ] ) . moreover , we shall use the numerical values @xmath119 where @xmath120 is the present value of cmb temperature . 1 . @xmath121cdm model . + for the @xmath121cdm model of ( [ eq : ftmyrzabis ] ) relation ( [ rhot ] ) gives @xmath122 \nonumber \\ & = & \frac{3h_0 ^ 2}{8\pi g}\ , \omega_{m0}\left(\frac{{\cal t}}{{\cal t}_0}\right)^{4n } \label{rhoth } \ , , \end{aligned}\ ] ] and then ( [ deltat / tboundg ] ) yields @xmath123 in fig . [ deltatf1 ] we depict @xmath124 from ( [ deltat / tboundgfin ] ) vs @xmath52 , as well as the upper bound from ( [ deltat / tbound ] ) . as we can see , constraints from bbn require @xmath125 . remarkably , this bound is in agreement with the best fit for @xmath52 of ( [ bestfit1 ] ) , namely @xmath126 , that was obtained using @xmath56 observational data in @xcite . + _ @xmath124 from ( [ deltat / tboundgfin ] ) vs @xmath52 ( thick line ) for the @xmath121cdm model of ( [ eq : ftmyrzabis ] ) , and the upper bound for @xmath124 from ( [ deltat / tbound ] ) ( dashed line ) . as we can see , constraints from bbn require @xmath125._,width=278,height=216 ] 2 . @xmath127cdm model . + in the case of @xmath128cdm model of ( [ modf2 ] ) and @xmath129cdm model of ( [ modf3 ] ) , and for the purpose of this analysis , we can unified their investigation parameterizing them as @xmath130\,,\ ] ] with @xmath131 where @xmath132 for model @xmath128cdm and @xmath133 for model @xmath129cdm . inserting ( [ f12insieme ] ) into ( [ deltat / tboundg ] ) we acquire @xmath134e^{-p({\cal t}_f/{\cal t}_0)^{4 m } } \ ! -\!\frac{1}{2}\right\}.\end{aligned}\ ] ] hence , using this relation we can calculate the value of @xmath135 for various values of @xmath136 that span the order of magnitude of the best fit values ( [ bestfit1 ] ) and ( [ bestfit2 ] ) that were obtained using @xmath56 observational data in @xcite , and we present our results in table [ tab1 ] . as we can see , in all cases the value of @xmath135 is well below the bbn bound ( [ deltat / tbound ] ) . hence , bbn can not impose constraints on the parameter values of @xmath128cdm and @xmath129cdm models . .@xmath135 from ( [ delta12insieme ] ) for different values of @xmath136 , for @xmath137 ( @xmath128cdm model ) and @xmath138 ( @xmath129cdm model ) . [ cols="^,^,^,^,^",options="header " , ] in this work we have investigated the implications of @xmath1 gravity to the formation of light elements in the early universe , i.e. to the bbn . in particular , we have examined the three most used and well studied viable @xmath1 models , namely the power law , the exponential and the square - root exponential , and we have confronted them with bbn calculations based on current observational data on primordial abundance of @xmath0 . hence , we were able to extract constraints on their free parameters . concerning the power - law @xmath1 model , the obtained constraint on the exponent @xmath52 , is @xmath125 . remarkably , this bound is in agreement with the constraints obtained using @xmath56 observational data @xcite . concerning the exponential and the square - root exponential , we showed that , for realistic regions of free parameters , they always satisfy the bbn bounds . this means that , in these cases , bbn can not impose strict constraints on the values of free parameters . in summary , we showed that viable @xmath1 models , namely those that pass the basic observational tests , can also satisfy the bbn constraints . this feature acts as an additional advantage of @xmath1 gravity , which might be a successful candidate for describing the gravitational interaction . as discussed in @xcite , this kind of constraints could contribute in the debate of fixing the most realistic picture that can be based on curvature or torsion . this article is based upon work from cost action ca15117 `` cosmology and astrophysics network for theoretical advances and training actions '' ( cantata ) , supported by cost ( european cooperation in science and technology ) . in this appendix we briefly review the main features of big bang nucleosynthesis following @xcite . in the early universe , the primordial @xmath0 was formed at temperature @xmath139 mev . the energy and number density were formed by relativistic leptons ( electron , positron and neutrinos ) and photons . the rapid collisions maintain all these particles in thermal equilibrium . interactions of protons and neutrons were kept in thermal equilibrium by means of their interactions with leptons @xmath140 the neutron abundance is estimated by computing the conversion rate of protons into neutrons , i.e. @xmath141 , and its inverse @xmath78 . thus , the weak interaction rates ( at suitably high temperature ) are given by @xmath142 the rate @xmath143 is the sum of the rates associated to the processes ( [ proc1])-([proc3 ] ) , namely @xmath144 finally , the rate @xmath143 is related to the rate @xmath145 as @xmath146 , with @xmath147 the mass difference of neutron and proton . during the freeze - out stage , one can use the following approximations @xcite : ( i ) the temperatures of particles are the same , i.e. @xmath148 . ( ii ) the temperature @xmath76 is lower than the typical energies @xmath149 that contribute to the integrals entering the definition of the rates ( one can therefore replace the fermi - dirac distribution with the boltzmann one , namely @xmath150 ) . ( iii ) the electron mass @xmath151 can be neglected with respect to the electron and neutrino energies ( @xmath152 ) . having these in mind , the interaction rate corresponding to the process ( [ proc1 ] ) is given by @xmath153 where @xmath154\ , , \label{wa}\\ { \cal p } & \equiv & p_n+p_{\nu_e}-p_p - p_e\ , , \\ { \cal m } & = & \left(\frac{g_w}{8m_w}\right)^2 [ { \bar u}_p\omega^\mu u_n][{\bar u}_e\sigma_\mu v_{\nu_e}]\ , , \label{m } \\ \omega^\mu & \equiv & \gamma^\mu(c_v - c_a \gamma^5)\ , , \\ \sigma^\mu & \equiv & \gamma^\mu(1-\gamma^5 ) \ , . \end{aligned}\ ] ] in ( [ m ] ) we have used the condition @xmath155 , where @xmath156 is the mass of the vector gauge boson @xmath157 , with @xmath158 the transferred momentum . from eq . ( [ rateproc1 ] ) it follows that @xmath159 where @xmath160 and where @xmath161d\epsilon,\ ] ] with @xmath162 lastly , for the neutron decay ( [ proc3 ] ) one obtains @xmath166 hence , in the calculation of ( [ sumprocess ] ) we can safely neglect the above interaction rate of the neutron decay , i.e. during the bbn the neutron can be considered as a stable particle . the above approximations ( i)-(iii ) lead to @xcite @xmath167 thus , inserting ( [ auxilirel ] ) into ( [ sumprocess ] ) , and then into ( [ lambdaa ] ) , allows to derive the expression for @xmath168 , namely @xmath169 which using ( [ ne - pnu - fin ] ) leads to @xmath170 99 a. g. riess , et al . , astron . j. * 116 * , 1009 ( 1998 ) . s. perlmutter , et al . , astrophys . j. * 517 * , 565 ( 1999 ) . d. n. spergel , et al . , apjs . * 148 * , 175 ( 2003 ) . d. n. spergel , et al . , apjs . * 170 * , 377 ( 2007 ) . m. tegmark , et al . d * 69 * , 103501 ( 2004 ) . d. j. eisenstein , et al . , astrophys . j. * 633 * , 560 ( 2005 ) . s. m. carroll , living rev . rel . * 4 * , 1 ( 2001 ) . e. j. copeland , m. sami and s. tsujikawa , int . d. * 15 * , 1753 ( 2006 ) . r. j. yang and s. n. zhang , mon . not . . soc . * 407 * , 1835 ( 2010 ) . b. feng , x. l. wang and x. m. zhang , phys . b. * 607 * , 35 ( 2005 ) . r. r. caldwell , r. dave and r. j. steinhardt , phys . 80 * , 1582 ( 1998 ) . r. r. caldwell , phys . b. * 545 * , 23 ( 2002 ) . c. armendariz - picon , v. mukhanov and p. j. steinhardt , phys . d * 63 * , 103510 ( 2001 ) . t. padmanabhan , phys . rev . d. * 66 * , 021301 ( 2002 ) . a. sen , phys . t. * 117 * , 70 ( 2005 ) . y. f. cai , e. n. saridakis , m. r. setare and j. q. xia , phys . rept . * 493 * , 1 ( 2010 ) . e. elizalde , s. nojiri and s. d. odintsov , phys . d. * 70 * , 043539 ( 2004 ) . a. kamenshchik , u. moschella and v. pasquier , phys . b. * 511 * , 265 ( 2001 ) . m. c. bento , o. bertolami and a. a. sen , phys . d. * 66 * , 043507 ( 2002 ) . s. capozziello and m. de laurentis , phys . * 509 * , 167 ( 2011 ) . a. einstein , sitz . p. 217 ; ibid a. einstein ( 2005 ) , translations of einstein papers by a. unzicker and t. case , ( arxiv : physics/0503046 ) . s. bahamonde and c. g. bhmer , eur.phys.j . c * 76 * , 578 ( 2016 ) .

we use bbn observational data on primordial abundance of @xmath0 to constrain @xmath1 gravity . the three most studied viable @xmath1 models , namely the power law , the exponential and the square - root exponential are considered , and the bbn bounds are adopted in order to extract constraints on their free parameters . for the power - law model , we find that the constraints are in agreement with those acquired using late - time cosmological data . for the exponential and the square - root exponential models , we show that for realiable regions of parameters space they always satisfy the bbn bounds . we conclude that viable @xmath1 models can successfully satisfy the bbn constraints .

in a bidirectional relay network , two users exchange information via a relay node @xcite . several protocols have been proposed for such a network under the practical half - duplex constraint , i.e. , a node can not transmit and receive at the same time and in the same frequency band . the simplest protocol is the traditional two - way relaying protocol in which the transmission is accomplished in four successive point - to - point phases : user 1-to - relay , relay - to - user 2 , user 2-to - relay , and relay - to - user 1 . in contrast , the time division broadcast ( tdbc ) protocol exploits the broadcast capability of the wireless medium and combines the relay - to - user 1 and relay - to - user 2 phases into one phase , the broadcast phase @xcite . thereby , the relay broadcasts a superimposed codeword , carrying information for both user 1 and user 2 , such that each user is able to recover its intended information by self - interference cancellation . another existing protocol is the multiple access broadcast ( mabc ) protocol in which the user 1-to - relay and user 2-to - relay phases are also combined into one phase , the multiple - access phase @xcite . in the multiple - access phase , both user 1 and user 2 simultaneously transmit to the relay which is able to decode both messages . generally , for the bidirectional relay network without a direct link between user 1 and user 2 , six transmission modes are possible : four point - to - point modes ( user 1-to - relay , user 2-to - relay , relay - to - user 1 , relay - to - user 2 ) , a multiple access mode ( both users to the relay ) , and a broadcast mode ( the relay to both users ) , where the capacity region of each transmission mode is known @xcite , @xcite . using this knowledge , a significant research effort has been dedicated to obtaining the achievable rate region of the bidirectional relay network @xcite-@xcite . specifically , the achievable rates of most existing protocols for two - hop relay transmission are limited by the instantaneous capacity of the weakest link associated with the relay . the reason for this is the fixed schedule of using the transmission modes which is adopted in all existing protocols , and does not exploit the instantaneous channel state information ( csi ) of the involved links . for one - way relaying , an adaptive link selection protocol was proposed in @xcite where based on the instantaneous csi , in each time slot , either the source - relay or relay - destination links are selected for transmission . to this end , the relay has to have a buffer for data storage . this strategy was shown to achieve the capacity of the one - way relay channel with fading @xcite . moreover , in fading awgn channels , power control is necessary for rate maximization . the highest degree of freedom that is offered by power control is obtained for a joint average power constraint for all nodes . any other power constraint with the same total power budget is more restrictive than the joint power constraint and results in a lower sum rate . therefore , motivated by the protocols in @xcite and @xcite , our goal is to utilize all available degrees of freedom of the three - node half - duplex bidirectional relay network with fading , via an adaptive mode selection and power allocation policy . in particular , given a joint power budget for all nodes , we find a policy which in each time slot selects the optimal transmission mode from the six possible modes and allocates the optimal powers to the nodes transmitting in the selected mode , such that the sum rate is maximized . adaptive mode selection for bidirectional relaying was also considered in @xcite and @xcite . however , the selection policy in @xcite does not use all possible modes , i.e. , it only selects from two point - to - point modes and the broadcast mode , and assumes that the transmit powers of all three nodes are fixed and identical . although the selection policy in @xcite considers all possible transmission modes for adaptive mode selection , the transmit powers of the nodes are assumed to be fixed , i.e. , power allocation is not possible . interestingly , mode selection and power allocation are mutually coupled and the modes selected with the protocol in @xcite for a given channel are different from the modes selected with the proposed protocol . power allocation can considerably improve the sum rate by optimally allocating the powers to the nodes based on the instantaneous csi especially when the total power budget in the network is low . moreover , the proposed protocol achieves the maximum sum rate in the considered bidirectional network . hence , the sum rate achieved with the proposed protocol can be used as a reference for other low complexity suboptimal protocols . simulation results confirm that the proposed protocol outperforms existing protocols . finally , we note that the advantages of buffering come at the expense of an increased end - to - end delay . however , with some modifications to the optimal protocol , the average delay can be bounded , as shown in @xcite , which causes only a small loss in the achieved rate . the delay analysis of the proposed protocol is beyond the scope of the current work and is left for future research . [ c][c][0.75]@xmath0 [ c][c][0.75]@xmath1 [ c][c][0.75]@xmath2 [ c][c][0.75]@xmath3 [ c][c][0.75]@xmath4 [ c][c][0.75]@xmath5 [ c][c][0.75]@xmath6 [ c][c][0.75]@xmath7 [ c][c][0.75]@xmath8 [ c][c][0.75]@xmath9 [ c][c][0.5]@xmath0 [ c][c][0.5]@xmath1 [ c][c][0.5]@xmath2 [ c][c][0.75]@xmath10 [ c][c][0.75]@xmath11 [ c][c][0.75]@xmath12 [ c][c][0.75]@xmath13 [ c][c][0.75]@xmath14 [ c][c][0.75]@xmath15 in this section , we first describe the channel model . then , we provide the achievable rates for the six possible transmission modes . we consider a simple network in which user 1 and user 2 exchange information with the help of a relay node as shown in fig . we assume that there is no direct link between user 1 and user 2 , and thus , user 1 and user 2 communicate with each other only through the relay node . we assume that all three nodes in the network are half - duplex . furthermore , we assume that time is divided into slots of equal length and that each node transmits codewords which span one time slot or a fraction of a time slot as will be explained later . we assume that the user - to - relay and relay - to - user channels are impaired by awgn with unit variance and block fading , i.e. , the channel coefficients are constant during one time slot and change from one time slot to the next . moreover , in each time slot , the channel coefficients are assumed to be reciprocal such that the user 1-to - relay and the user 2-to - relay channels are identical to the relay - to - user 1 and relay - to - user 2 channels , respectively . let @xmath3 and @xmath4 denote the channel coefficients between user 1 and the relay and between user 2 and the relay in the @xmath16-th time slot , respectively . furthermore , let @xmath17 and @xmath18 denote the squares of the channel coefficient amplitudes in the @xmath16-th time slot . @xmath19 and @xmath20 are assumed to be ergodic and stationary random processes with means @xmath21 and @xmath22 in expectations for notational simplicity . ] , respectively , where @xmath23 denotes expectation . since the noise is awgn , in order to achieve the capacity of each mode , nodes have to transmit gaussian distributed codewords . therefore , the transmitted codewords of user 1 , user 2 , and the relay are comprised of symbols which are gaussian distributed random variables with variances @xmath24 , and @xmath7 , respectively , where @xmath25 is the transmit power of node @xmath26 in the @xmath16-th time slot . for ease of notation , we define @xmath27 . in the following , we describe the transmission modes and their achievable rates . in the considered bidirectional relay network only six transmission modes are possible , cf . fig . [ figmodes ] . the six possible transmission modes are denoted by @xmath28 , and @xmath29 , denotes the transmission rate from node @xmath30 to node @xmath31 in the @xmath16-th time slot . let @xmath8 and @xmath9 denote two infinite - size buffers at the relay in which the received information from user 1 and user 2 is stored , respectively . moreover , @xmath32 , denotes the amount of normalized information in bits / symbol available in buffer @xmath33 in the @xmath16-th time slot . using this notation , the transmission modes and their respective rates are presented in the following : @xmath34 : user 1 transmits to the relay and user 2 is silent . in this mode , the maximum rate from user 1 to the relay in the @xmath16-th time slot is given by @xmath35 , where @xmath36 . the relay decodes this information and stores it in buffer @xmath8 . therefore , the amount of information in buffer @xmath8 increases to @xmath37 . @xmath38 : user 2 transmits to the relay and user 1 is silent . in this mode , the maximum rate from user 2 to the relay in the @xmath16-th time slot is given by @xmath39 , where @xmath40 . the relay decodes this information and stores it in buffer @xmath9 . therefore , the amount of information in buffer @xmath9 increases to @xmath41 . @xmath42 : both users 1 and 2 transmit to the relay simultaneously . for this mode , we assume that multiple access transmission is used , see @xcite . thereby , the maximum achievable sum rate in the @xmath16-th time slot is given by @xmath43 , where @xmath44 . since user 1 and user 2 transmit independent messages , the sum rate , @xmath45 , can be decomposed into two rates , one from user 1 to the relay and the other one from user 2 to the relay . moreover , these two capacity rates can be achieved via time sharing and successive interference cancelation . thereby , in the first @xmath46 fraction of the @xmath16-th time slot , the relay first decodes the codeword received from user 2 and considers the signal from user 1 as noise . then , the relay subtracts the signal received from user 2 from the received signal and decodes the codeword received from user 1 . a similar procedure is performed in the remaining @xmath47 fraction of the @xmath16-th time slot but now the relay first decodes the codeword received from user 1 and treats the signal of user 2 as noise , and then decodes the codeword received from user 2 . therefore , for a given @xmath48 , we decompose @xmath45 as @xmath49 and the maximum rates from users 1 and 2 to the relay in the @xmath16-th time slot are @xmath50 and @xmath51 , respectively . @xmath52 and @xmath53 are given by @xmath54 the relay decodes the information received from user 1 and user 2 and stores it in its buffers @xmath8 and @xmath9 , respectively . therefore , the amounts of information in buffers @xmath8 and @xmath9 increase to @xmath37 and @xmath41 , respectively . @xmath55 : the relay transmits the information received from user 2 to user 1 . specifically , the relay extracts the information from buffer @xmath9 , encodes it into a codeword , and transmits it to user 1 . therefore , the transmission rate from the relay to user 1 in the @xmath16-th time slot is limited by both the capacity of the relay - to - user 1 channel and the amount of information stored in buffer @xmath9 . thus , the maximum transmission rate from the relay to user 1 is given by @xmath56 , where @xmath57 . therefore , the amount of information in buffer @xmath9 decreases to @xmath58 . @xmath59 : this mode is identical to @xmath55 with user 1 and 2 switching places . the maximum transmission rate from the relay to user 2 is given by @xmath60 , where @xmath61 and the amount of information in buffer @xmath8 decreases to @xmath62 . @xmath63 : the relay broadcasts to both user 1 and user 2 the information received from user 2 and user 1 , respectively . specifically , the relay extracts the information intended for user 2 from buffer @xmath8 and the information intended for user 1 from buffer @xmath9 . then , based on the scheme in @xcite , it constructs a superimposed codeword which contains the information from both users and broadcasts it to both users . thus , in the @xmath16-th time slot , the maximum rates from the relay to users 1 and 2 are given by @xmath64 and @xmath65 , respectively . therefore , the amounts of information in buffers @xmath8 and @xmath9 decrease to @xmath62 and @xmath58 , respectively . our aim is to develop an optimal mode selection and power allocation policy which in each time slot selects one of the six transmission modes , @xmath28 , and allocates the optimal powers to the transmitting nodes of the selected mode such that the average sum rate of both users is maximized . to this end , we introduce six binary variables , @xmath66 , where @xmath67 indicates whether or not transmission mode @xmath68 is selected in the @xmath16-th time slot . in particular , @xmath69 if mode @xmath68 is selected and @xmath70 if it is not selected in the @xmath16-th time slot . furthermore , since in each time slot only one of the six transmission modes can be selected , only one of the mode selection variables is equal to one and the others are zero , i.e. , @xmath71 holds . in the proposed framework , we assume that all nodes have full knowledge of the csi of both links . thus , based on the csi and the proposed protocol , cf . theorem [ adaptprot ] , each node is able to individually decide which transmission mode is selected and adapt its transmission strategy accordingly . in this section , we first investigate the achievable average sum rate of the network . then , we formulate a maximization problem whose solution is the sum rate maximizing protocol . we assume that user 1 and user 2 always have enough information to send in all time slots and that the number of time slots , @xmath72 , satisfies @xmath73 . therefore , using @xmath67 , the user 1-to - relay , user 2-to - relay , relay - to - user 1 , and relay - to - user 2 average transmission rates , denoted by @xmath74 , @xmath75 , @xmath76 , and @xmath77 , respectively , are obtained as lll[ratreg123 ] & & |r_1r = _ i = 1^n + & & |r_2r = _ i = 1^n + & & |r_r1 = _ i = 1^n \{c_r1(i),q_2(i-1 ) } + & & |r_r2 = _ i = 1^n \{c_r2(i),q_1(i-1)}. the average rate from user 1 to user 2 is the average rate that user 2 receives from the relay , i.e. , @xmath77 . similarly , the average rate from user 2 to user 1 is the average rate that user 1 receives from the relay , i.e. , @xmath76 . in the following theorem , we introduce a useful condition for the queues in the buffers of the relay leading to the optimal mode selection and power allocation policy . [ queue ] the maximum average sum rate , @xmath78 , for the considered bidirectional relay network is obtained when the queues in the buffers @xmath8 and @xmath9 at the relay are at the edge of non - absorbtion . more precisely , the following conditions must hold for the maximum sum rate lll[ratregapp456-buffer ] |r_1r=|r_r2 = _ i = 1^n c_r2(i ) + where @xmath74 and @xmath75 are given by ( [ ratreg123]a ) and ( [ ratreg123]b ) , respectively . please refer to ( * ? ? ? * appendix a ) . using this theorem , in the following , we derive the optimal transmission mode selection and power allocation policy . the available degrees of freedom in the considered network in each time slot are the mode selection variables , the transmit powers of the nodes , and the time sharing variable for multiple access . herein , we formulate an optimization problem which gives the optimal values of @xmath67 , @xmath25 , and @xmath48 , for @xmath79 , @xmath80 , and @xmath81 , such that the average sum rate of the users is maximized . the optimization problem is as follows cll[adaptprob ] & |r_1r+|r_2r + & : |r_1r=|r_r2 + & : |r_2r=|r_r1 + & : |p_1+|p_2+|p_r p_t + & : _ k = 1 ^ 6 q_k ( i ) = 1 , i + & : q_k(i ) [ 1-q_k(i ) ] = 0 , i , k + & : p_j(i)0 , i , j + & : 0t(i)1 , i where @xmath82 is the total average power constraint of the nodes and @xmath83 , and @xmath84 denote the average powers consumed by user 1 , user 2 , and the relay , respectively , and are given by lll |p_1= _ i = 1^n ( q_1(i)+q_3(i))p_1 ( i ) + in the optimization problem given in ( [ adaptprob ] ) , constraints @xmath85 and @xmath86 are the conditions for sum rate maximization introduced in theorem [ queue ] . constraints @xmath87 and @xmath88 are the average total transmit power constraint and the power non - negativity constraint , respectively . moreover , constraints @xmath89 and @xmath90 guarantee that only one of the transmission modes is selected in each time slot , and constraint @xmath91 specifies the acceptable interval for the time sharing variable @xmath48 . furthermore , we maximize @xmath92 since , according to theorem 1 ( and constraints @xmath85 and @xmath86 ) , @xmath93 and @xmath94 hold . in the following theorem , we introduce a protocol which achieves the maximum sum rate . [ adaptprot ] assuming @xmath73 , the optimal mode selection and power allocation policy which maximizes the sum rate of the considered three - node half - duplex bidirectional relay network with awgn and block fading is given by lll[seleccrit ] q_k^*(i)= 1 , & k^*= \{_k(i ) } + 0 , & where @xmath95 is referred to as _ selection metric _ and is given by lll[selecmet ] _ 1(i ) = ( 1-_1)c_1r(i ) - p_1(i ) |_p_1(i)=p_1^_1(i ) + _ 2(i ) = ( 1-_2)c_2r(i ) - p_2(i)|_p_2(i)=p_2^_2(i ) + _ 3(i ) = ( 1-_1)c_12r(i)+(1-_2)c_21r(i ) + - ( p_1(i)+p_2(i))|_p_2(i)=p_2^_3(i)^p_1(i)=p_1^_3(i ) + _ 6(i ) = _ 1 c_r2(i)+_2 c_r1(i ) - p_r(i)|_p_r(i)=p_r^_6(i ) where @xmath96 denotes the optimal transmit power of node @xmath30 for transmission mode @xmath68 in the @xmath16-th time slot and is given by lll[optpower ] p_1^_1 ( i ) = ^+ + p_2^_2 ( i ) = ^+ + p_1^_3 ( i ) = ^+ , _ 1_2 + ^+ , + p_2^_3 ( i ) = ^+ , _ 1 _ 2 + ^+ , + p_r^_6 ( i ) = ^+ where @xmath97^+=\max\{x,0\}$ ] , @xmath98 , and @xmath99 . the thresholds @xmath100 and @xmath101 are chosen such that constraints @xmath85 and @xmath86 in ( [ adaptprob ] ) hold and threshold @xmath102 is chosen such that the total average transmit power satisfies @xmath87 in ( [ adaptprob ] ) . the optimal value of @xmath48 in @xmath52 and @xmath53 is given by lll t^*(i ) = 0 , & _ 1 _ 2 + 1 , & _ 1 < _ 2 please refer to appendix [ appkkt ] . we note that the optimal solution utilizes neither modes @xmath13 and @xmath14 nor time sharing for any channel statistics and channel realizations . the mode selection metric @xmath95 introduced in ( [ selecmet ] ) has two parts . the first part is the instantaneous capacity of mode @xmath68 , and the second part is the allocated power with negative sign . the capacity and the power terms are linked via thresholds @xmath100 and/or @xmath101 and @xmath102 . we note that thresholds @xmath100 , @xmath101 , and @xmath102 depend only on the long term statistics of the channels . hence , these thresholds can be obtained offline and used as long as the channel statistics remain unchanged . to find the optimal values for the thresholds @xmath100 , @xmath101 , and @xmath102 , we need a three - dimensional search , where @xmath103 and @xmath104 . adaptive mode selection for bidirectional relay networks under the assumption that the powers of the nodes are fixed is considered in @xcite . based on the average and instantaneous qualities of the links , all of the six possible transmission modes are selected in the protocol in @xcite . however , in the proposed protocol , modes @xmath13 and @xmath14 are not selected at all . moreover , the protocol in @xcite utilizes a coin flip for implementation . therefore , a central node must decide which transmission mode is selected in the next time slot . however , in the proposed protocol , all nodes can find the optimal mode and powers based on the full csi . in this section , we evaluate the average sum rate achievable with the proposed protocol in the considered bidirectional relay network in rayleigh fading . thus , channel gains @xmath19 and @xmath20 follow exponential distributions with means @xmath105 and @xmath106 , respectively . all of the presented results were obtained for @xmath107 and @xmath108 time slots . in fig . [ mixed ] , we illustrate the maximum achievable sum rate obtained with the proposed protocol as a function of the total average transmit power @xmath82 . in this figure , to have a better resolution for the sum rate at low and high @xmath82 , we show the sum rate for both log scale and linear scale @xmath109-axes , respectively . the lines without markers in fig . [ mixed ] represent the achieved sum rates with the proposed protocol for @xmath110 . we observe that as the quality of the user 1-to - relay link increases ( i.e. , @xmath105 increases ) , the sum rate increases too . however , for large @xmath105 , the bottleneck link is the relay - to - user 2 link , and since it is fixed , the sum rate saturates . [ c][t][0.75]@xmath111 [ c][b][0.75]@xmath112 [ l][c][0.45]@xmath113 [ l][c][0.45]@xmath114 [ l][c][0.45]@xmath115 [ l][c][0.45]@xmath116}\,(\omega_1=1)$ ] [ l][c][0.45]@xmath117}\,(\omega_1=1)$ ] [ l][c][0.45]@xmath118 [ l][c][0.45]@xmath119 [ c][c][0.6]@xmath120 [ c][c][0.6]@xmath121 [ c][c][0.6]@xmath122 [ c][c][0.6]@xmath123 [ c][c][0.6]@xmath124 [ c][c][0.6]@xmath125 [ c][c][0.6]@xmath126 [ c][c][0.6]@xmath127 [ c][c][0.6]@xmath128 [ c][c][0.6]@xmath129 [ c][c][0.6]@xmath130 [ c][c][0.6]@xmath131 [ c][c][0.6]@xmath132 [ c][c][0.6]@xmath133 [ c][c][0.30]@xmath124 [ c][c][0.30]@xmath134 [ c][c][0.30]@xmath135 [ c][c][0.30]@xmath136 [ c][c][0.30]@xmath137 [ c][c][0.30]@xmath138 [ c][c][0.30]@xmath139 [ c][c][0.30]@xmath140 [ c][c][0.30]@xmath141 [ c][c][0.30]@xmath142 [ c][c][0.30]@xmath125 [ c][c][0.30]@xmath120 [ c][c][0.30]@xmath121 [ c][c][0.30]@xmath122 [ c][c][0.30]@xmath123 [ c][c][0.30]@xmath124 [ c][c][0.30]@xmath125 [ c][c][0.30]@xmath126 [ c][c][0.30]@xmath127 [ c][c][0.30]@xmath128 for different protocols.,title="fig:",width=3 ] as performance benchmarks , we consider in fig . [ mixed ] the sum rates of the tdbc protocol with and without power allocation @xcite and the buffer - aided protocols presented in @xcite and @xcite , respectively . for clarity , for the benchmark schemes , we only show the sum rates for @xmath143 . for the tdbc protocol without power allocation and the protocol in @xcite , all nodes transmit with equal powers , i.e. , @xmath144 . for the buffer - aided protocol in @xcite , we adopt @xmath145 and @xmath146 is chosen such that the average total power consumed by all nodes is @xmath82 . we note that since @xmath143 and @xmath147 , the protocol in @xcite only selects modes @xmath12 and @xmath15 . moreover , since @xmath143 , we obtain @xmath148 in the proposed protocol . thus , considering the optimal power allocation in ( [ optpower]c ) and ( [ optpower]d ) , we obtain that either @xmath149 or @xmath150 is zero . therefore , for the chosen parameters , only modes @xmath151 , and @xmath15 are selected , i.e. , the same modes as used in @xcite , the protocol in @xcite is optimal for given fixed node transmit powers . ] . hence , we can see how much gain we obtain due to the adaptive power allocation by comparing our result with the results for the protocol in @xcite . on the other hand , the gain due to the adaptive mode selection can be evaluated by comparing the sum rate of the proposed protocol with the result for the tdbc protocol with power allocation . from the comparison in fig . [ mixed ] , we observe that for high @xmath82 , a considerable gain is obtained by the protocols with adaptive mode selection ( ours and that in @xcite ) compared to the tdbc protocol which does not apply adaptive mode selection ( around @xmath152 db gain ) . however , for high @xmath82 , power allocation is less beneficial , and therefore , the sum rates obtained with the proposed protocol and that in @xcite converge . on the other hand , for low @xmath82 , optimal power allocation is crucial and , therefore , a considerable gain is achieved by the protocols with adaptive power allocation ( ours and tdbc with power allocation ) . we have derived the maximum sum rate of the three - node half - duplex bidirectional buffer - aided relay network with fading links . the protocol which achieves the maximum sum rate jointly optimizes the selection of the transmission mode and the transmit powers of the nodes . the proposed optimal mode selection and power allocation protocol requires the instantaneous csi of the involved links in each time slot and their long - term statistics . simulation results confirmed that the proposed selection policy outperforms existing protocols in terms of average sum rate . in this appendix , we solve the optimization problem given in ( [ adaptprob ] ) . we first relax the binary condition for @xmath67 , i.e. , @xmath153=0 $ ] , to @xmath154 , and later in appendix [ appbinrelax ] , we prove that the binary relaxation does not affect the maximum average sum rate . in the following , we investigate the karush - kuhn - tucker ( kkt ) necessary conditions @xcite for the relaxed optimization problem and show that the necessary conditions result in a unique sum rate and thus the solution is optimal . to simplify the usage of the kkt conditions , we formulate a minimization problem equivalent to the relaxed maximization problem in ( [ adaptprob ] ) as follows cll[adaptprobmin ] & -(|r_1r+|r_2r ) + & : |r_1r-|r_r2=0 + & : |r_2r-|r_r1=0 + & : |p_1+|p_2+|p_r - p_t0 + & : _ k = 1 ^ 6 q_k ( i ) - 1 = 0 , i + & : q_k(i)-1 0 , i , k + & : -q_k(i ) 0 , i , k + & : -p_j(i ) 0 , i , k + & : t(i)-1 0 , i + & : -t(i ) 0 , i. the lagrangian function for the above optimization problem is provided in ( [ kkt function ] ) at the top of the next page where @xmath155 , and @xmath156 are the lagrange multipliers corresponding to constraints @xmath157 , and @xmath158 , respectively . the kkt conditions include the following : l[kkt function ] = + - ( |r_1r+|r_2r ) + _ 1(|r_1r-|r_r2 ) + _ 2(|r_2r-|r_r1 ) + ( |p_1+|p_2+|p_r - p_t ) + + _ i = 1^n ( i ) ( _ k = 1 ^ 6 q_k ( i ) - 1 ) + _ i = 1^n _ k = 1 ^ 6 _ k ( i ) ( q_k ( i ) - 1 ) - _ i = 1^n _ k = 1 ^ 6 _ k ( i ) q_k ( i ) + -_i = 1^n + _ i = 1^n _ 1(i ) ( t(i)-1 ) - _ i = 1^n _ 0(i ) t(i ) * 1 ) * stationary condition : the differentiation of the lagrangian function with respect to the primal variables , @xmath159 , and @xmath160 , is zero for the optimal solution , i.e. , cccl[stationary condition ] & = & 0 , & i , k + & = & 0 , & i , j + & = & 0 , & i. * 2 ) * primal feasibility condition : the optimal solution has to satisfy the constraints of the primal problem in ( [ adaptprobmin ] ) . * 3 ) * dual feasibility condition : the lagrange multipliers for the inequality constraints have to be non - negative , i.e. , lll[dual feasibility condition ] _ k(i)0 , & i , k + _ k(i)0,&i , k + 0 , & + _ j(i ) 0 , & i , j + _ l(i ) 0 , & i , l . * 4 ) * complementary slackness : if an inequality is inactive , i.e. , the optimal solution is in the interior of the corresponding set , the corresponding lagrange multiplier is zero . thus , we obtain lll[complementary slackness ] _ k ( i ) ( q_k ( i ) - 1 ) = 0,&i , k + _ k ( i ) q_k ( i ) = 0 , & i , k + ( |p_1+|p_2+|p_r - p_t ) = 0 & + _ j(i ) p_j(i ) = 0 , & i , j + _ 1(i ) ( t(i)-1 ) = 0 , & i + _ 0(i ) t(i ) = 0 , & i. a common approach to find a set of primal variables , i.e. , @xmath161 and lagrange multipliers , i.e. , @xmath162 , which satisfy the kkt conditions is to start with the complementary slackness conditions and see if the inequalities are active or not . combining these results with the primal feasibility and dual feasibility conditions , we obtain various possibilities . then , from these possibilities , we obtain one or more candidate solutions from the stationary conditions and the optimal solution is surely one of these candidates . in the following subsections , with this approach , we find the optimal values of @xmath163 and @xmath164 . in order to determine the optimal selection policy , @xmath165 , we must calculate the derivatives in ( [ stationary condition]a ) . this leads to lll[stationary mode ] = - ( 1-_1)c_1r(i)+(i)+_1(i)-_1(i ) + p_1(i)=0 + = -(1-_2)c_2r(i)+(i)+_2(i)-_2(i ) + p_2(i)=0 + = -[(1-_1)c_12r(i)+(1-_2)c_21r(i)]+(i ) + _ 3(i)-_3(i)+(p_1(i)+p_2(i))=0 + = - _ 2 c_r1(i)+(i)+_4(i)-_4(i)+p_r(i)=0 + = -_1 c_r2(i)+(i)+_5(i)-_5(i)+p_r(i)=0 + = -[_1 c_r2(i)+_2 c_r1(i)]+(i)+_6(i)-_6(i)+p_r(i)=0 . without loss of generality , we first obtain the necessary condition for @xmath166 and then generalize the result to @xmath167 . if @xmath168 , from constraint @xmath89 in ( [ adaptprobmin ] ) , the other selection variables are zero , i.e. , @xmath169 . furthermore , from ( [ complementary slackness ] ) , we obtain @xmath170 and @xmath171 . by substituting these values into ( [ stationary mode ] ) , we obtain lll[met ] ( i)+_1(i ) = ( 1-_1)c_1r(i ) -p_1(i ) _ 1(i ) + ( i)-_2(i ) = ( 1-_2)c_2r(i ) -p_2(i ) _ 2(i ) + ( i)-_3(i ) = ( 1-_1)c_12r(i)+(1-_2)c_21r(i ) -(p_1(i)+p_2(i ) ) _ 3(i ) + ( i)-_4(i ) = _ 2 c_r1(i ) -p_r(i ) _ 4(i ) + ( i)-_5(i ) = _ 1 c_r2(i ) -p_r(i ) _ 5(i ) + ( i)-_6(i ) = _ 1 c_r2(i)+_2 c_r1(i ) -p_r(i ) _ 6(i ) , where @xmath95 is referred to as selection metric . by subtracting ( [ met]a ) from the rest of the equations in ( [ met ] ) , we obtain rcl[eq_2_1 ] _ 1(i ) - _ k(i ) = _ 1(i)+_k(i ) , k=2,3,4,5,6 . from the dual feasibility conditions given in ( [ dual feasibility condition]a ) and ( [ dual feasibility condition]b ) , we have @xmath172 . by inserting @xmath172 in ( [ eq_2_1 ] ) , we obtain the necessary condition for @xmath166 as lll _ 1(i ) \ { _ 2(i ) , _ 3(i ) , _ 4(i ) , _ 5(i ) , _ 6(i ) } . repeating the same procedure for @xmath167 , we obtain a necessary condition for selecting transmission mode @xmath173 in the @xmath16-th time slot as follows lll[optmet ] _ k^*(i ) \{_k(i ) } , where the lagrange multipliers @xmath174 , and @xmath102 are chosen such that @xmath175 , and @xmath87 in ( [ adaptprobmin ] ) hold and the optimal value of @xmath48 in @xmath52 and @xmath53 is obtained in the next subsection . we note that if the selection metrics are not equal in the @xmath16-th time slot , only one of the modes satisfies ( [ optmet ] ) . therefore , the necessary conditions for the mode selection in ( [ optmet ] ) is sufficient . moreover , in appendix [ appbinrelax ] , we prove that the probability that two selection metrics are equal is zero due to the randomness of the time- continuous channel gains . therefore , the necessary condition for selecting transmission mode @xmath68 in ( [ optmet ] ) is in fact sufficient and is the optimal selection policy . in order to determine the optimal @xmath25 , we have to calculate the derivatives in ( [ stationary condition]b ) . this leads to [ stationary power ] @xmath177 \nonumber \\ & + \gamma \frac{1}{n } ( q_1(i)+q_3(i ) ) - \nu_1(i ) = 0 \quad \tag{\stepcounter{equation}\theequation}\\ \frac{\partial\mathcal{l}}{\partial p_2(i ) } { \hspace{-0.7mm}=\hspace{-0.7mm}}&-\frac{1}{n\mathrm{ln}2 } \big [ \big \ { ( 1{\hspace{-0.7mm}-\hspace{-0.7mm}}\mu_2)q_2(i ) { \hspace{-0.7mm}+\hspace{-0.7mm}}(1{\hspace{-0.7mm}-\hspace{-0.7mm}}t(i))(\mu_1{\hspace{-0.7mm}-\hspace{-0.7mm}}\mu_2)q_3(i ) \big\ } \nonumber \\ & \times\frac{s_2(i)}{1{\hspace{-0.7mm}+\hspace{-0.7mm}}p_2(i)s_2(i ) } { \hspace{-0.7mm}+\hspace{-0.7mm}}\big\{t(i)(\mu_1{\hspace{-0.7mm}-\hspace{-0.7mm}}\mu_2){\hspace{-0.7mm}+\hspace{-0.7mm}}1{\hspace{-0.7mm}-\hspace{-0.7mm}}\mu_1 \big\}q_3(i ) \nonumber \\ & \times\frac{s_2(i)}{1+p_1(i)s_1(i)+p_2(i)s_2(i ) } \big]\nonumber \\ & + \gamma \frac{1}{n } ( q_2(i)+q_3(i ) ) - \nu_2(i)=0 \quad\tag{\stepcounter{equation}\theequation}\\ \frac{\partial\mathcal{l}}{\partial p_r(i ) } = & -\frac{1}{n\mathrm{ln}2 } \big [ \mu_2\left(q_4(i)+q_6(i)\right)\frac{s_1(i)}{1+p_r(i)s_1(i)}\nonumber \\ & + \mu_1\left(q_5(i)+q_6(i)\right)\frac{s_2(i)}{1+p_r(i)s_2(i ) } \big]\nonumber \\ & + \gamma \frac{1}{n } ( q_4(i)+q_5(i)+q_6(i ) ) - \nu_r(i)=0 \tag{\stepcounter{equation}\theequation}\end{aligned}\ ] ] the above conditions allow the derivation of the optimal powers for each transmission mode in each time slot . for instance , in order to determine the transmit power of user 1 in transmission mode @xmath10 , we assume @xmath166 . from constraint @xmath89 in ( [ adaptprobmin ] ) , we obtained that the other selection variables are zero and therefore @xmath178 . moreover , if @xmath10 is selected then @xmath179 and thus from ( [ complementary slackness]d ) , we obtain @xmath180 . substituting these results in ( [ stationary power]a ) , we obtain lll [ eq_11 ] p_1^_1 ( i ) = ^+ , where @xmath97^+=\max\{0,x\}$ ] . in a similar manner , we obtain the optimal powers for user 2 in mode @xmath11 , and the optimal powers of the relay in modes @xmath13 and @xmath14 as follows : lll[p245 ] p_2^_2 ( i ) = ^+ + p_r^_4 ( i ) = ^+ + p_r^_5 ( i ) = ^+ in order to obtain the optimal powers of user 1 and user 2 in mode @xmath12 , we assume @xmath181 . from @xmath89 in ( [ adaptprobmin ] ) , we obtain that the other selection variables are zero , and therefore @xmath182 and @xmath183 . we note that if one of the powers of user 2 and user 1 is zero mode @xmath12 is identical to modes @xmath10 and @xmath11 , respectively , and for that case the optimal powers are already given by ( [ eq_11 ] ) and ( [ p245]a ) , respectively . for the case when @xmath179 and @xmath184 , we obtain @xmath180 and @xmath185 from ( [ complementary slackness]d ) . furthermore , for @xmath181 , we will show in appendix [ appkkt].c that @xmath48 can only take the boundary values , i.e. , zero or one , and can not be in between . hence , if we assume @xmath186 , from ( [ stationary power]a ) and ( [ stationary power]b ) , we obtain lll[powerm3 ] & - + = 0 + & - + = 0 by substituting ( [ powerm3]a ) in ( [ powerm3]b ) , we obtain @xmath187 and then we can derive @xmath188 from ( [ powerm3]a ) . this leads to lll [ pm3-t0 ] p_1^_3 ( i ) = p_1^_1 ( i ) , s_2 + ^+ , + p_2^_3 ( i ) = p_2^_2 ( i ) , s_2 s_1 + ^+ , similarly , if we assume @xmath189 , we obtain lll [ pm3-t1 ] p_1^_3 ( i ) = p_1^_1 ( i ) , s_2 s_1 + ^+ , + p_2^_3 ( i ) = p_2^_2 ( i ) , s_2 + ^+ , we note that when @xmath190 , we obtain @xmath191 which means that mode @xmath12 is identical to mode @xmath10 . thus , there is no difference between both modes so we select @xmath10 . in figs [ figsregion ] a ) and [ figsregion ] b ) , the comparison of @xmath192 , and @xmath193 is illustrated in the space of @xmath194 . moreover , the shaded area represents the region in which the powers of users 1 and 2 are zero for @xmath151 , and @xmath12 . [ c][c][0.6]@xmath19 [ c][c][0.6]@xmath20 [ c][c][0.6]@xmath195 [ c][c][0.6]@xmath196 [ c][c][0.6]@xmath195 [ c][c][0.6]@xmath196 [ c][c][0.6]@xmath197 [ c][c][0.6]@xmath198 [ c][c][0.75]@xmath199 [ c][b][0.7]@xmath200 [ c][c][0.7]@xmath201 [ c][c][0.75]@xmath202 [ c][c][0.75]@xmath203 [ c][c][0.75]@xmath204 [ c][c][0.65]@xmath200 [ c][t][0.65]@xmath201 [ c][c][0.75]@xmath202 [ c][c][0.75]@xmath205 [ c][c][0.75]@xmath206 [ c][c][0.75]a ) @xmath207 [ c][c][0.75]b ) @xmath208 [ c][c][0.75]c ) for mode @xmath15 , we assume @xmath209 . from constraint @xmath89 in ( [ adaptprobmin ] ) , we obtain that the other selection variables are zero and therefore @xmath210 and @xmath211 . moreover , if @xmath209 then @xmath212 and thus from ( [ complementary slackness]d ) , we obtain @xmath213 . using these results in ( [ stationary power]c ) , we obtain lll[powerm6 ] _ 2 + _ 1 = 2 the above equation is a quadratic equation and has two solutions for @xmath7 . however , since we have @xmath214 , we can conclude that the left hand side of ( [ powerm6 ] ) is monotonically decreasing in @xmath7 . thus , if @xmath215 , we have a unique positive solution for @xmath7 which is the maximum of the two roots of ( [ powerm6 ] ) . thus , we obtain lll[pm6 ] p_r^_6 ( i ) = ^+ , where @xmath216 , and @xmath99 . in fig . [ figsregion ] c ) , the comparison between selection metrics @xmath217 and @xmath218 is illustrated in the space of @xmath194 . we note that @xmath219 and @xmath220 hold and the inequalities hold with equality if @xmath221 and @xmath222 , respectively , which happen with zero probability for time - continuous fading . to prove @xmath219 , from ( [ met ] ) , we obtain rcl _ 6(i ) & = & _ 1c_r2(i ) + _ 2c_r1(i ) -p_r(i ) |_p_r(i)=p_r^_6(i ) + & & _ 1c_r2(i ) + _ 2c_r1(i ) -p_r(i)|_p_r(i)=p_r^_4(i ) + & & _ 2c_r1(i ) -p_r(i ) |_p_r(i)=p_r^_4(i ) = _ 4(i ) , where @xmath223 follows from the fact that @xmath224 maximizes @xmath218 and @xmath225 follows from @xmath226 . the two inequalities @xmath223 and @xmath225 hold with equality only if @xmath221 which happens with zero probability in time - continuous fading or if @xmath227 . however , in appendix [ appmuregion ] , @xmath227 is shown to lead to a contradiction . therefore , the optimal policy does not select @xmath13 and @xmath14 and selects only modes @xmath228 , and @xmath15 . to find the optimal @xmath48 , we assume @xmath230 and calculate the stationary condition in ( [ stationary condition]c ) . this leads to lll [ stationary t ] = & -(_1-_2 ) + & + _ 1(i)-_0(i)=0 now , we investigate the following possible cases for @xmath229 : * case 1 : * if @xmath231 then from ( [ complementary slackness]e ) and ( [ complementary slackness]f ) , we have @xmath232 . therefore , from ( [ stationary t ] ) and @xmath233 , we obtain @xmath148 . then , from ( [ stationary power]a ) and ( [ stationary power]b ) , we obtain rcl[contradict ] - ( 1-_1 ) + = 0 + - ( 1-_1 ) + = 0 in appendix [ appmuregion ] , we show that @xmath234 , therefore , the above conditions can be satisfied simultaneously only if @xmath235 , which , considering the randomness of the time - continuous channel gains , occurs with zero probability . hence , the optimal @xmath48 takes the boundary values , i.e. , zero or one , and not values in between . * case 2 : * if @xmath236 , then from ( [ complementary slackness]e ) , we obtain @xmath237 and from ( [ dual feasibility condition]e ) , we obtain @xmath238 . combining these results into ( [ stationary t ] ) , the necessary condition for @xmath186 is obtained as @xmath239 . * case 3 : * if @xmath240 , then from ( [ complementary slackness]f ) , we obtain @xmath241 and from ( [ dual feasibility condition]e ) , we obtain @xmath242 . combining these results into ( [ stationary t ] ) , the necessary condition for @xmath189 is obtained as @xmath243 . we note that if @xmath148 , we obtain either @xmath244 or @xmath245 . therefore , mode @xmath12 is not selected and the value of @xmath48 does not affect the sum rate . moreover , from the selection metrics in ( [ met ] ) , we can conclude that @xmath207 and @xmath208 correspond to @xmath246 and @xmath247 , respectively . therefore , the optimal value of @xmath48 is given by lll t^*(i ) = 0 , & _ 1 _ 2 + 1 , & _ 1 < _ 2 now , the optimal values of @xmath159 , and @xmath160 are derived based on which theorem [ adaptprot ] can be constructed . this completes the proof . in this appendix , we prove that the optimal solution of the problem with the relaxed constraint , @xmath248 , selects the boundary values of @xmath67 , i.e. , zero or one . therefore , the binary relaxation does not change the solution of the problem . if one of the @xmath249 , adopts a non - binary value in the optimal solution , then in order to satisfy constraint @xmath89 in ( [ adaptprob ] ) , there has to be at least one other non - binary selection variable in that time slot . assuming that the mode indices of the non - binary selection variables are @xmath250 and @xmath251 in the @xmath16-th time slot , we obtain @xmath252 from ( [ complementary slackness]a ) , and @xmath253 and @xmath254 from ( [ complementary slackness]b ) . then , by substituting these values into ( [ stationary mode ] ) , we obtain lll[binrelax ] ( i ) = _ k(i ) + ( i)= _ k(i ) + ( i)-_k(i ) = _ k(i ) , kk , k . from ( [ binrelax]a ) and ( [ binrelax]b ) , we obtain @xmath255 and by subtracting ( [ binrelax]a ) and ( [ binrelax]b ) from ( [ binrelax]c ) , we obtain rcl _ k(i ) - _ k(i ) & = & _ k(i ) , kk , k + _ k(i ) - _ k(i ) & = & _ k(i ) , kk , k . from the dual feasibility condition given in ( [ dual feasibility condition]b ) , we have @xmath256 which leads to @xmath257 . however , as a result of the randomness of the time - continuous channel gains , @xmath258 holds for some transmission modes @xmath259 and @xmath260 , if and only if we obtain @xmath261 or @xmath262 which leads to a contradiction as shown in appendix [ appmuregion ] . this completes the proof . in this appendix , we find the intervals which contain the optimal value of @xmath100 and @xmath101 . we note that for different values of @xmath100 and @xmath101 , some of the optimal powers derived in ( [ eq_11 ] ) , ( [ p245 ] ) , ( [ pm3-t0 ] ) , ( [ pm3-t1 ] ) , and ( [ pm6 ] ) are zero for all channel realizations . for example , if @xmath263 , we obtain @xmath264 from ( [ eq_11 ] ) . [ figmuregion ] illustrates the set of modes that can take positive powers with non - zero probability in the space of ( @xmath174 ) . in the following , we show that any values of @xmath100 and @xmath101 except @xmath265 and @xmath266 can not lead to the optimal sum rate or violate constraints @xmath85 or @xmath86 in ( [ adaptprobmin ] ) . * case 1 : * sets @xmath267 and @xmath268 lead to selection of either the transmission from the users to the relay or the transmission from the relay to the users , respectively , for all time slots . this leads to violation of constraints @xmath85 and @xmath86 in ( [ adaptprobmin ] ) and thus the optimal values of @xmath100 and @xmath101 are not in this region . * case 2 : * in set @xmath269 , both modes @xmath13 and @xmath15 need the transmission from user 2 to the relay which can not be realized in this set . thus , this set leads to violation of constraint @xmath86 in ( [ adaptprobmin ] ) . similarly , in set @xmath270 , both modes @xmath14 and @xmath15 require the transmission from user 1 to the relay which can not be selected in this set . thus , this region of @xmath100 and @xmath101 leads to violation of constraint @xmath85 in ( [ adaptprobmin ] ) . * case 3 : * in set @xmath271 , there is no transmission from user 2 to the relay . therefore , the optimal values of @xmath100 and @xmath101 have to guarantee that modes @xmath13 and @xmath15 are not selected for any channel realization . however , from ( [ met ] ) , we obtain where @xmath223 follows from the fact that @xmath224 maximizes @xmath218 and @xmath225 follows from @xmath272 . the two inequalities @xmath223 and @xmath225 hold with equality only if @xmath222 which happens with zero probability for time - continuous fading , or @xmath273 which is not included in this region . therefore , mode @xmath15 is selected in this region which leads to violation of constraint @xmath86 in ( [ adaptprobmin ] ) . a similar statement is true for set @xmath274 . thus , the optimal values of @xmath100 and @xmath101 can not be in these two regions . where inequality @xmath223 comes from the fact that @xmath276 and the equality holds when @xmath221 which happens with zero probability , or @xmath227 . inequality @xmath225 holds since @xmath277 maximizes @xmath278 and holds with equality only if @xmath279 and consequently @xmath227 . if @xmath280 , mode @xmath15 is not selected and there is no transmission from the relay to user 2 . therefore , the optimal values of @xmath100 and @xmath101 have to guarantee that modes @xmath10 and @xmath12 are not selected for any channel realization . thus , we obtain @xmath263 which is not contained in this region . if @xmath227 , from ( [ met ] ) , we obtain where both inequalities @xmath223 and @xmath225 hold with equality only if @xmath281 . if @xmath282 , modes @xmath13 and @xmath15 are not selected thus , there is no transmission from the relay to the users which leads to violation of @xmath85 and @xmath86 in ( [ adaptprobmin ] ) . if @xmath281 , we obtain @xmath283 , thus mode @xmath11 can not be selected and either @xmath244 or @xmath284 , thus mode @xmath12 can not be selected either . since both modes @xmath13 and @xmath15 require the transmission from user 2 to the relay , and both modes @xmath11 and @xmath12 are not selected , constraint @xmath86 in ( [ adaptprobmin ] ) is violated and @xmath227 and @xmath281 can not be optimal . a similar statement is true for set @xmath285 . therefore , the optimal values of @xmath100 and @xmath101 are not in this region . s. j. kim , n. devroye , p. mitran , and v. tarokh , `` achievable rate regions and performance comparison of half duplex bi - directional relaying protocols , '' _ ieee trans . inf . theory _ , 57 , no . 6405 6418 , oct . 2011 . n. zlatanov and r. schober , `` capacity of the state - dependent half - duplex relay channel without source - destination link , '' _ submitted ieee transactions on information theory _ , 2013 . [ online ] . available : http://arxiv.org/abs/1302.3777 v. jamali , n. zlatanov , a. ikhlef , and r. schober , `` adaptive mode selection in bidirectional buffer - aided relay networks with fixed transmit powers , '' _ submitted in part to eusipco13 _ , 2013 . [ online ] . available : http://arxiv.org/abs/1303.3732

in this paper , we consider the problem of sum rate maximization in a bidirectional relay network with fading . hereby , user 1 and user 2 communicate with each other only through a relay , i.e. , a direct link between user 1 and user 2 is not present . in this network , there exist six possible transmission modes : four point - to - point modes ( user 1-to - relay , user 2-to - relay , relay - to - user 1 , relay - to - user 2 ) , a multiple access mode ( both users to the relay ) , and a broadcast mode ( the relay to both users ) . most existing protocols assume a fixed schedule of using a subset of the aforementioned transmission modes , as a result , the sum rate is limited by the capacity of the weakest link associated with the relay in each time slot . motivated by this limitation , we develop a protocol which is not restricted to adhere to a predefined schedule for using the transmission modes . therefore , all transmission modes of the bidirectional relay network can be used adaptively based on the instantaneous channel state information ( csi ) of the involved links . to this end , the relay has to be equipped with two buffers for the storage of the information received from users 1 and 2 , respectively . for the considered network , given a total average power budget for all nodes , we jointly optimize the transmission mode selection and power allocation based on the instantaneous csi in each time slot for sum rate maximization . simulation results show that the proposed protocol outperforms existing protocols for all signal - to - noise ratios ( snrs ) . specifically , we obtain a considerable gain at low snrs due to the adaptive power allocation and at high snrs due to the adaptive mode selection .

einstein s general theory of relativity ( gtr ) has proved to be one of the most successful physical theories ever formulated . as a theory of the gravitational interaction , it has a number of predictions definitely corroborated by experiments or astronomical observations , which makes it the correct gravity theory here - to - fore . after readdressing an earlier study , einstein ( 1936 ) @xcite came up with the prediction of the gravitational lensing effect . the idea is that the gravitational field produced by a massive astrophysical object , the sun for instance , can act as a convergent lens able to deviate light rays flying - by . observations of far away quasars and galaxies and more familiar total solar eclipses confirm the reality of this phenomenon . more than half of the star s surface may be seen by a distant observer . this means that a photon emitted at a given colatitude on the star s surface , reaching the observer at infinity , must be emitted at a smaller angle with respect to the normal at that point . because of this effect , a distant compact star must then appear to any observer larger then it actually is . the relation between the apparent radius @xmath1 of a spherical ( non - rotating ) star , as seen by a distant observer , and its physical radius @xmath2 is @xmath3 this relation is obtained from the photon trajectory in polar coordinates ( @xmath4 ) by using a schwarzschild metric @xmath5 here @xmath6 is the schwarzschild radius of the star of mass @xmath7 . as is clear from this line - element no effects from physical fields other than the gravitational have been taken into account in prescribing the star apparent size . paper _ shows that the phenomenon changes in a fundamental fashion if nled is called into play . the analysis below , based on well - known nonlinear lagrangeans , as the heisenberg - euler ( h - e , @xcite ) and the exact born & infeld ( 1934 ) ( b - i @xcite ) , proves that when nled is included to describe the photon dynamics , the trajectory depends on the background @xmath0-field pervading the star . we stress from the very beginning that the results below are obtained upon an idealization of the @xmath0-field structure . in a real star , the @xmath0-field is neither purely radial nor constant , so that the apparent radius hardly goes to zero at high @xmath0 . we stress in passing that this theory _ is not a new theory of gravitation_. in our formalism einstein s general relativity remains definitively unaltered . only the photons propagating out of the ultramagnetized neutron stars see " the effective metric that appears once one incorporates nled into a self - consistent physical description of light propagation from those stars . that nled is a theory already tested in a number of experiments is demonstrated by the laboratory experiment described by burke , field , horton - smith , et al . ( 1997 ) @xcite , as quoted below . our theory attempts to call to the attention of the astrophysics community the fundamental changes that should appear in determining physical properties as radius , mass , and more crucial : the equation of state , of neutron star pulsars endowed with extremely strong magnetic fields , since a property as the star radius ( which can be estimated from the x - ray variability of very short period of the star emission ( light curve ) , or by measuring the gravitational redshift of absorption or emission lines received from the star surface , or etc . ) can say too much about the stellar structure and its equation of state . a fundamental property that encodes basic information of the physical constituents of the neutron star matter . according to quantum electrodynamics ( qed : see delphenich 2003 @xcite for a complete review on nled and qed ) a vacuum has nonlinear properties @xcite , which affect the photon propagation . a noticeable advance in the realization of this theoretical prediction has been provided by burke , field , horton - smith , et al . ( 1997 ) @xcite , who demonstrated experimentally that the inelastic scattering of laser photons by gamma - rays is definitely a nonlinear phenomenon . the propagation of photons in nled has been examined by several authors ( bialynicka - birula & bialynicka - birula 1970 ; salazar et al . 1989 ; dietrich & gies 1998 ; de lorenci 2000 ; denisov & svertilov 2003 ; alarcon 1981 ; garcia & plebanski 1989 ; boillat 1970 ; vazquez et al . 1969 ; torrence 1984 ; garcia 1984 ) @xcite . in the geometric optics approximation , it was shown by novello et al . ( 2000 ) @xcite ; and novello & salim ( 2001 ) @xcite , that when the photon propagation is identified with the propagation of discontinuities of the em field in a nonlinear regime , a remarkable feature affloats : the discontinuities propagate along null geodesics of an _ effective _ geometry which depends on the em field on the background . this means that the nled interaction can be geometrized , in analogy to gravity in gtr . a key outcome of this formalism is introduced in this _ paper_. the h - e @xcite lagrangean @xmath8 where @xmath9 is a quantum parameter , is a gauge invariance description of the photon propagation in nled that uses two invariants @xmath10 constructed upon the maxwell em tensor and its dual .- fields smaller than @xmath11 g ( schwinger s critical @xmath0-field ) . ] the method of characteristic surfaces or shock waves followed here ( introduced by hadamard ( 1903 ) @xcite ) can be applied to any field theory having hyperbolic field equations , including electrodynamics . hence , the result in eq.([fg - metric ] ) ( see appendix ) , implies two possible paths of propagation or polarization modes , according to the double solution @xmath12 . using eq.([h - e ] ) to compute the lagrangean derivatives @xmath13 , one arrives to the couple of effective metrics ( first order in @xmath14 ) @xmath15 in terms of electric ( @xmath16 ) and magnetic ( @xmath0 ) fields the tensor @xmath17 can be written as @xmath18 where @xmath19 is the normalized @xmath20 , eq.([schw ] ) ) velocity of the reference frame where the fields are measured . our main concern here is the behavior of photons , in the nled context , emitted from a highly magnetized neutron star . corotating charges in the pulsar magnetosphere or a rotating magnetic dipole lead to induced @xmath16-fields in the star surface . we consider here slowly rotating neutron stars in order that the @xmath16-field contribution could be neglected . since @xmath21 , the above expression simplifies to : @xmath22 . its self - product reads @xmath23 where @xmath24 . after discarding a nonphysical conformal factor the effective metrics become @xmath25 the inverse ( covariant ) metric of eq.([contrav - metric ] ) is obtained from the relation @xmath26 , which then reads @xmath27 by @xmath28 we mean @xmath29 and @xmath30 . in order to pursue our calculations we assume a radial magnetic field . this is clearly not realistic , but is an approximate description of the field geometry near the polar caps of a magnetized neutron star . thus @xmath31 and consequently one arrives to the effective metric components @xmath32 notice the structural similarity of both metric components . from eq.([l - dependent ] ) is straightforward to rewrite the expression for the apparent radius of the spherical star in eq.([r - max ] ) in the putative background metric given by eq.([schw ] ) . after doing so , one can compute the ratio of the star area with h - e nled effects included to the area without ( einstein s derivation ) , as seen from the distant observer . this quantity is represented here by @xmath33 . is a solution of the nled field equations , to compute the ratio `` @xmath34 '' is only needed to know the local ( on the star surface ) magnitude or modulus of the field . in the particular case of the computation of the null geodesics discussed below , is indeed necessary to know the analytic expression for the field as a function of the coordinates , in particular of the radial @xmath35-coordinate . unfortunately , such a solution for the magnetic field structure ( or dependence on the coordinates ) is only known for the linear ( maxwell s ) case . ] = 6.6 cm as stated above , `` @xmath34 '' is intended to be a quantity describing the ratio between the area of the star as seen by an observer at infinity that receives the actual traveling luminous ray , and the actual physical area of the neutron star , both projected on the plane of the sky . this ratio of areas can be consistently figured out by the following qualitative reasoning . in a general relativistic framework , the local neutron star luminosity is defined as l_= 4 r^2__sb t_^4 , where @xmath36 is the neutron star radius and @xmath37 is its temperature . both measured locally at the star surface . the luminosity measured by the observer at infinity , which carries the information on the spacetime metric around the neutron star , is defined @xmath38 where @xmath39 is the @xmath40-metric component of a schwarzschild spacetime , i.e. , @xmath41 notice that @xmath42 . on the other hand , the luminosity at infinity can be written as @xmath43 thus , the temperature measured by an observer at infinity relates to the actual star temperature via @xmath44 therefore , after manipulating all these relations one arrives to the relation between both the star radii ( local and at infinity ) @xmath45 in an equivalent approach , once one realizes that the star luminosity is modified by the spacetime geometry induced by the matter distribution , one can readily verify that in the case of nled its geometric effects on the photon propagation from the ultra magnetized neutron star leads to an expression that is similar to the one above . hence , one can write , taking into account the general relativistic effects superposed to the nled effects , the local neutron star luminosity as given by @xmath46 the luminosity measured by the observer at infinity , which carries the information on the spacetime metric as modified by nled , as in the heisenberg - euler ( h - e ) nled , now reads @xmath47\ ] ] or equivalently , @xmath48 where the term @xmath49 $ ] now substitutes the @xmath40-metric component of the pure schwarzschild spacetime presented above . notice that in this case the effective @xmath40-metric component reads @xmath50~ .\ ] ] this implies that @xmath51 } = \frac { \left [ 1 - { \tilde{\beta}}^2_{\pm } b^2 \right ] } { g_{tt } } .\ ] ] thus , the temperature measured by an observer at infinity now reads @xmath52^{1/2 } ~ .\ ] ] once again , after manipulating these relations one obtains @xmath53 from these relations one can write the area ratio `` @xmath34 '' that estimates the magnitude of the change in the apparent visible area of the star as seen from infinity compared to the actual star area , as measured by a local observer : = = . hence , by substituting the corresponding terms in these last equations one arrives to = ( ) = . as one can notice , in this new expression for the area ratio there is no any functional of the @xmath35 coordinate but rather only a function of the magnetic field strength at the surface of the neutron star . this is the reason of why we took @xmath0 as a constant in the numerical calculation leading to the plots in figures 1 , and 2 . the propagation of light from hypermagnetized neutron stars can also be viewed within the framework of the born - infeld lagrangean @xmath54 where @xmath55 as is well known , this is an exceptional lagrangean . one of its remarkable properties is that it does not exhibit birefringence @xcite . in this case , the deduction we present in the appendix fails since the quantities @xmath56 , with @xmath57 ; vanish identically . we need hence to go back to eq.([eq25],[eq26 ] ) ; and substituting in these equations the form for @xmath58 of eq.([exc ] ) to obtain a unique _ characteristic equation _ , which then leads to the effective metric @xmath59 using the self - product of the tensor @xmath17 given by ( [ eq80 ] ) , and noting that in our case @xmath60 , the effective metric then reads ( see reference @xcite for details ) @xmath61 by computing the inverse metric via @xmath62 , the covariant form of this effective metric is obtained as @xmath63 after recalling the relations for @xmath64 and @xmath65 given in eq.([4-vectors ] ) , one can verify that the covariant @xmath66 effective metric component is then written as @xmath67~ g_{r r } \ ; .\ ] ] = 6.6 cm = 6.6 cm by following steps similar to the h - e case , the area ratio is presented in fig.2 . although our hypothesis about the field geometry maximizes the nled effect on the observed properties of the star , they should properly be taken into account when analyzing the x - ray emission . ( this point will be considered in a forthcoming paper ) . it should also be relevant in pondering the effects on supernova dynamics of photons radiated either by highly magnetized proto - neutron stars ( miralles et al . ( 2002 ) @xcite ) or stellar - mass black holes enshrouded by super strong @xmath0-fields ( see van putten et al . ( 2004 ) @xcite ) . a self - consistent treatment of this problem does require the study of the light propagation in the neutron star spacetime geometry modified by nled . in nled the lagrangean defined by eq.([schw ] ) becomes @xmath68 where @xmath69 . the resulting euler - lagrange equations read ( with @xmath70 ; @xmath71 ) : ( - 2 g^eff_tt u^t ) = 0 g^eff_tt u^t = e_0 , [ tt - component ] ( 2 g^eff _ u^ ) = 0 g _ u^= l_0 , [ thetatheta - component ] ( 2 g^eff _ u^ ) = 0 g^eff _ u^= [ phiphi - component ] , where @xmath72 is defined as the _ impact parameter _ of the photon propagating from the ns surface to an observer at @xmath73 , and @xmath74 a constant , which may be put equal to zero . as the photons travel along null geodesics , the euler - lagrange equation for the @xmath66-component of the metric can be obtained from the ( first integral ) relation @xmath75 in analogy with the planetary system one can chose a particular `` orbital '' plane at @xmath76 , so that @xmath77 and @xmath78 , and thus @xmath79 . by making use of the standard change of variables ( @xmath80 , @xmath81 , and defining @xmath82 ) one can write the photon propagation equation as @xmath83 which one can in turn write as ( @xmath84 , @xmath72 constants ) @xmath85 after performing the derivative of equation ( [ photon - traject ] ) with respect to the angular coordinate @xmath86 , and taking into account the fact that the solution @xmath87 can be discarted by an analogous ( physical ) reason as it is done in the case of a pure photon propagation in a schwarzschild geometry , one arrives to @xmath88 \frac{da}{dr } . \label{nonlinear}\ ] ] this equation is highly nonlinear and quite hard to solve by analytical means . ( notice that eq.([nonlinear ] ) is quite similar to that one obtains in the case of the photon propagation in a schwarzschild spacetime . nonetheless , we stress that in order to find such a solution ; an analogous procedure as the one usually pursued in general relativity ( _ the perturbation approach _ ) does not work in the present case , since for the extremely large magnetic field we are considering the induced nled effects are of the same order of magnitude as those purely obtained from gravitation . ) besides , one needs to know the function @xmath89 , which is a solution of the born - infeld equations for this problem . perhaps a numerical solution can be found , but that would require extra work that we think would not add too much to the new results we are presenting in this version of the paper . ( a full detailed solution will be given elsewhere ) . eqs.([photon - traject],[nonlinear ] ) shows very clearly that nled does modify the standard propagation of photons as compared to that one in a pure schwarzschild gravitational field @xcite , as described in most textbooks of gtr . a first look at eq.([photon - traject ] ) suggests that very large @xmath0-fields do reduce the effective star s visible area . the nonlinear electrodynamics is a well stablished theory of the electromagnetic interaction , which overcomes the failure of maxwell s theory in describing the stability of atomic structures . since its introduction in physics by euler and heisenberg @xcite it has gained the status of a theory with a strong experimental support ( for a recent complete review see delphenich ( 2003 ) @xcite ) . despite being known since long , its very key features have become recognized only in the late years ( see references above ) . it had been shown in the last few years ( novello et al2000 ; novello & salim 2001 ) that the force acting along the photons trajectory can be geometrized in such a way that in an effective metric @xmath90 the photons follow geodesic paths in that effective geometry ( novello et al2000 ; novello & salim 2001 ) @xcite . in particular , the effects of nonlinear electrodynamics in the physics of strongly magnetized neutron stars have been recently studied by mosquera cuesta & salim ( 2003 ; 2004 ) @xcite . it was shown there that for extremely supercritical magnetic fields nonlinear electrodynamics effects force photons to propagate along accelerated curves , in such a way that the surface gravitational redshift of emission lines from hypermagnetized neutron stars , as measured by a distant observer , is significantly modified . such a new effect turns out to be of the same order of magnitude of the one produced by the pure gravitational field . in this paper , we present a new astrophysical application of the peculiar effects brought by nled , which is intimately related to the one previously discussed by ( mosquera cuesta & salim 2003 ; 2004 ) @xcite . we show that the famous gravitational lensing effect , originally introduced by einstein , is also modified when the electromagnetic ( nonlinear ) effects are taken into the depiction of the physics around strongly magnetized neutron stars . such a dynamics is described by using both the euler - heisenberg and the born - infeld approaches to nled . incidentally , we show , by comparing eqs.([l - dependent ] ) and ( [ effective - metric ] ) , that eq.([l - dependent ] ) clearly exhibits a divergence . such an unphysical behavior , in turn , invalidates the results presented by denisov et al . in ref . @xcite obtained within the h - e approach , since they extended to the case of magnetars ( neutron stars having @xmath0-fields @xmath91 g ) their formula to estimate the light - ray bending angle beyond the qed @xmath0-field limit . that is inconsistent . besides , both approaches : the h - e and the b - i nled induce critical changes in the @xmath66 metric component that modifies the quantification of the lensing effect . more relevant , yet , the area of a putative star may appear to a distant observer nonphysically diminished in h - e nled , whereas it is physically largely reduced , and may even `` disappear '' , for fields @xmath92 g in the b - i approach . the impact of this on the star flux may be dramatic . thence , the effect introduced by ( mosquera cuesta & salim 2003 ; 2004 ) @xcite and this new here crucially alters the dynamics of photons from very high @xmath0-field stars , while entangles ( makes it difficult ) the inference of their physical properties . to the very end , what kind of observations can be performed in a search for this peculiar effect ? as light is the prime messenger from the stars , one can look for variability of very short period in the x - ray ( or @xmath93-ray ) light curves from most hypermagnetized pulsars . a comparative analysis of the characteristics of those light curves with the ones from canonical pulsars may render the clues on the presence of nled effects . a similar analysis can be pursued by using the measurements of the surface gravitational redshift of that kind of hypermagnetic stars . following hadamard ( 1903 ) @xcite , the surface of discontinuity of the em field is denoted by @xmath94 . the field is continuous when crossing @xmath94 , while its first derivative presents a finite discontinuity . these properties are specified as follows : @xmath95_{\sigma } = 0\ ; , $ ] 0.3 truecm @xmath96_{\sigma } = f_{\mu\nu}k_\lambda \ ; \protect \label{eq14 } \;$ ] , where the symbol @xmath95_{\sigma } = \lim_{\delta \to 0^+ } \left(j|_{\sigma + \delta}-j|_{\sigma - \delta}\right)$ ] represents the discontinuity of the arbitrary function @xmath97 through the surface @xmath94 . the tensor @xmath98 is called the discontinuity of the field , @xmath99 is the propagation vector , and the symbol `` @xmath100 '' stands for partial derivative . here - after we investigate the effects of nonlinearities of very strong @xmath0-fields in the evolution of em waves ; described onwards as the surface of discontinuity of the em field ( represented here - to - fore by @xmath17 ) . extremizing the lagrangian with respect to the potentials @xmath101 yields the following field equation : we seek thus for a master relation representing the propagation of field discontinuities , which should be independent of the quantities @xmath112 , that is , independent of @xmath113 and @xmath114 there is a simple way to achieve such a goal . we firstly isolate the common term @xmath115 which appears in both eqs.([63],[64 ] ) . then , by assuming that @xmath116 , the difference of these equations can be put in the form of an algebraic linear relation between @xmath113 and @xmath114 given as : @xmath117 where we define : @xmath118 and @xmath119 and @xmath120 . solving the quadratic equation for @xmath121 we obtain : @xmath122 , with : @xmath123 . using this solution in eqs.([63 ] ) and ( [ 64 ] ) , and assuming @xmath124 , after some algebra one gets the dispersion relation v. i. denisov , i. p. denisova , s. i. svertilov , dokl . 46 , 705 ( 2001a ) ; v. i. denisov , i. p. denisova , s. i. svertilov , dokl . nauk serv . 380 , 435 ( 2001b ) ; v. i. denisov , s. i. svertilov , a & a. 399 , l39 ( 2003 )

einstein ( 1936 ) predicted the phenomenon presently known as gravitational lensing ( gl ) . a prime feature of gl is the magnification , because of the gravitational field , of the star visible surface as seen from a distant observer . we show here that nonlinear electrodynamics ( nled ) modifies in a fundamental basis einstein s general relativistic ( gr ) original derivation . the effect becomes apparent by studying the light propagation from a strongly magnetic ( @xmath0 ) pulsar ( smp ) . unlike its gr counterpart , the photon dynamics in nled leads to a new effective gl , which depends also on the @xmath0-field permeating the pulsar . the apparent radius of a smp appears then unexpectedly diminished , by a large factor , as compared to the classical einstein s prediction . this may prove very crucial in determining physical properties of high @xmath0-field stars from their x - ray emission . 0.5 truecm received(07/04/2004 ) revised(16/05/2005 )

a study of disorder in condensed matter systems is useful from two perspectives . disorder is inherent in most condensed matter systems and often has profound effects on their properties . incorporation of small amounts of paramagnetic impurities in a metallic host can result in the kondo effect which gives the well - known logarithmic temperature dependence of the resistivity upon cooling , and eventually crosses over to a fermi - liquid regime with a characteristic low energy scale , the kondo temperature . conversely , impurities at low concentrations can act as a probe providing specific signatures of the environment they exist in . from the latter perspective , the kondo effect is a set of signatures of certain low - energy excitations of the host lacking long - range magnetic order . for instance , exotic kondo effects are known to arise in itinerant electron magnets near criticality @xcite and in insulating quantum spin - liquid systems @xcite owing to the paramagnons and spinonic excitations respectively . a study of impurity effects in the spin-@xmath0 honeycomb kitaev model @xcite is very appealing in this context . this kitaev model is integrable and the ground state can be either a gapless or gapped quantum ( @xmath3 ) spin - liquid with extremely short - ranged spin correlations . @xcite the elementary excitations are not spin-@xmath1 bosons that one typically expects for magnetic systems in two dimensions and higher , but emergent dispersing majorana fermions ( spinons ) and @xmath3 vortices ( @xmath4flux excitations associated with spins at the vertices of the hexagonal plaquettes ) which in the gapless phase are known to be non - abelian anyons . @xcite experimental realization looks increasingly imminent with several interesting proposals to realize kitaev physics in two - dimensional quantum - compass materials such as the alkali iridates @xcite and ruthenium trichloride , @xcite and independently in cold - atom optical lattices . @xcite introducing a paramagnetic impurity into the model through local exchange - coupling of the impurity spin with a host ( kitaev ) spin results in a highly unusual kondo effect @xcite owing to the peculiar elementary excitations in the host . for an @xmath5 kitaev model with an energy scale @xmath6 coupled locally to a spin-@xmath7 impurity , the perturbative scaling equations for the impurity coupling @xmath8 turn out to be independent of its sign , with an intermediate coupling unstable fixed point @xmath9 separating weak and strong coupling regimes . such scaling differs qualitatively from the kondo effect in metals ( and graphene , @xcite ) where a nontrivial effect is seen only for antiferromagnetic coupling , but is similar to the kondo scaling reported for paramagnetic impurities in certain pseudogapped bosonic spin - liquids . @xcite the distinguishing feature of the kitaev - kondo problem is that the weak and strong coupling limits correspond to different topologies of the ground state . @xcite despite the insensitivity of the scaling equations to the sign of impurity coupling , the strong coupling limits for @xmath10 and @xmath11 are very different physically . in the antiferromagnetic case ( @xmath10 ) , the strong - coupling limit for an @xmath5 impurity spin corresponds to a spin singlet at the impurity site - equivalent to the kitaev model with a missing site , which is an integrable model . in the ferromagnetic case , the strong - coupling limit corresponds to a non - integrable problem where one of the sites has @xmath12 while the rest have @xmath13 the problem of missing sites ( spinless vacancies ) in the kitaev model has received much attention in recent times . it was independently reported in ref . and ref . that the ground state of the kitaev model with a missing site is associated with a finite @xmath3 flux through the defect . that would not be the case , for example in graphene , where although the dirac fermions have the same dispersion as the emergent majorana fermions in the kitaev model , the phases of the intersite hopping matrix elements in graphene are identical for every bond and do not change upon the creation of defects . in contrast , the phases of the intersite hopping elements of the majorana fermions in the kitaev model are a degree of freedom and can take values @xmath14 or @xmath15 for the missing site problem , the magnetic susceptibility is predicted@xcite to have logarithmic singularities both as a function of the magnetic field as well as the temperature . some of the singularities in magnetic susceptibility reported in refs . are reminiscent of the two - channel kondo problem , and we shall later discuss a connection between such singularities and the presence of bound , zero energy majorana fermions in the kitaev model with a vacancy @xcite as well as in the two - channel kondo model . @xcite vacancy induced spin textures have also been studied by exact diagonalization@xcite of finite clusters of up to 24 spins described by more general ( and nonintegrable ) kitaev - heisenberg models in the presence of a small magnetic field . in ref . , it was reported that a vacancy induces longer ranged spin - spin correlations that extend beyond the single bond correlations one has in the defect - free kitaev model . @xcite the low - energy properties of the kitaev model with a random and dilute concentration of vacancies are also quite interesting . such rare but locally singular perturbations are predicted to result in qualitatively different low - energy properties compared to that expected for gaussian white noise type disorder.@xcite in contrast to the understanding we currently have on the effects of single and multiple vacancies in the kitaev model , much less is known about the effect of spinful defects where the defect site has a different spin from the host sites . part of the reason is that while the vacancy problem is integrable and affords a simplification of a difficult problem where interactions and disorder are both present , the problem with an @xmath16 defect can not be reduced to a noninteracting one . some progress was made in ref . where it was explicitly demostrated that the ground states of both the vacancy as well as @xmath16 defect problems have a two - fold degeneracy . however , it could not be established whether the @xmath16 defect was also associated with a finite @xmath3 flux that is the case when a vacancy is present . it was also not demonstrated whether the strong coupling fixed points were indeed stable . the stability is an important issue , for otherwise one would expect new intermediate coupling stable fixed points and not only would our understanding of the kondo effect in the kitaev model be incomplete , but also the paramagnetic impurity route for generating vacancies and spinful defects would no longer be appropriate . the latter issue is of interest from a practical point of view too since it would make it possible to generate nonabelian anyons conveniently using a spin - polarized stm tip to bind a kitaev spin ferromagnetically or antiferromagnetically . it is also not clear whether the topological transition and the magnetic transitions are coincident or occur at the same value of the impurity coupling strength . in this paper we address these open questions and make the following new findings . we demonstrate the stability of the strong coupling limit demonstrated for the ferromagnetic and antiferromagnetic cases which implies stability of the spin-@xmath14 vacancies and spin-@xmath1 defects created through this route . we perform exact diagonalization calculations for a finite fragment of the kitaev model coupled to a paramagnetic impurity and show that while for weak impurity coupling , the ground state corresponds to zero @xmath3 flux at the impurity site , for strong coupling , the ground state has a finite flux irrespective of the sign of impurity coupling . for the value of impurity coupling that corresponds to this topological transition , we also observe the total spin at the impurity site going to zero or one depending on the sign of coupling - this establishes that the the topological as well as the weak coupling to strong coupling transitions occur very close to each other and possibly at the same point . as a corollary , we find that the ground state with a spin-@xmath1 defect corresponds to a finite flux at the defect site . finally , we report an intriguing connection between the susceptibilities of a vacancy in the kitaev model and of the magnetic impurity in a two - channel kondo model . the rest of the paper is organized as follows . sec.[sec : model ] provides a brief introduction to the honeycomb spin-@xmath0 kitaev model . in sec.[sec : kondo ] we study the effect of coupling an external paramagnetic impurity to the kitaev host through a local exchange coupling . a poor man s scaling analysis of the impurity coupling reveals an unstable fixed point separating the weak and strong - coupling regimes . the stability if the strong coupling fixed point is demonstrated for both ferromagnetic and antiferromagnetic impurity couplings . it is also shown that the strong and weak coupling limits correspond to different topology of the ground state . new conserved quantities are identified which are composite operators of an impurity spin component and two kitaev flux operators . sec.[sec : numerical ] contains the result of exact diagonalization calculations of finite kitaev fragments coupled to external spins . the key findings in this section are ( a)establishing that a finite @xmath2 flux is associated with the ground state of the spin@xmath17 kitaev model with a spin@xmath18 defect - just as in the case of a vacancy , and ( b ) the magnetic ( kondo ) and topological transitions occur very close to each other and are possibly coincident . in sec.[sec : two - channel ] we discuss the intriguing parallels between the low - temperature magnetic response of the kitaev model with a missing site and the two - channel kondo model . sec.[sec : discussion ] contains a discussion of the results and possible future directions . kitaev s spin-@xmath0 honeycomb lattice model for a quantum spin liquid is a model of direction dependent nearest neighbour exhchage interactions on a honeycomb lattice . @xcite the hamiltonian for this model is given by @xmath20 where the three bonds at each site ( see fig.[fig : basis ] ) are labeled as @xmath21 , @xmath22 and @xmath23 the model is exactly solavable . @xcite as was shown by kitaev , the flux operators @xmath24 defined for each elementary plaquette @xmath25 are conserved ( fig.[fig : basis ] ) , with eigenvalues @xmath26 and form a set of commuting observables . the kitaev spins can be represented in terms of majorana fermions @xmath27 as @xmath28 this representation spans a larger fock space , and we restrict to the physical hilbert space of the spins by choosing the gauge@xcite @xmath29 ( a ) schematic of a fragment of the kitaev lattice showing the @xmath30 and @xmath31 sites and the @xmath32 @xmath22 and @xmath33 types of bonds . ( b ) figure showing the reciprocal lattice vectors for the @xmath30 sublattice . the dirac point for the massless majorana fermions is denoted by @xmath34 and momentum summations are over the ( shaded ) half brillouin zone . ] for each @xmath35type bond , @xmath36 is also conserved and the flux operators can be written as a product of @xmath37 s on the plaquette @xmath38 the ground state manifold corresponds to a vortex - free state where all @xmath39 are equal . in the vortex - free state , we can fix all @xmath40 ( corresponds to @xmath41 ) and the hamiltonian can be written as a tight - binding model of noninteracting majorana fermions . the reduced hamiltonian for this ground state manifold is given by @xmath42 where @xmath43 if @xmath44 are neighboring sites on an @xmath35bond and zero otherwise . the excited states ( with finite vorticity ) are separated from the ground state manifolds by a gap of order @xmath45 the free majorana fermion hopping hamiltonian can be diagonalized in momentum space by defining the bravais lattice with a two - point basis ( fig.[fig : basis ] ) . in momentum space , @xmath46 with @xmath47 , where @xmath48 is the nearest neighbor spin distance . the eigenstates are @xmath49 and @xmath50 with @xmath51 being the phase of @xmath52 the kitaev model has gapless excitations for a region of parameter space where @xmath53 s satisfy the triangle inequalities @xmath54 etc . and a gapped spectrum outside this parameter regime . the gapless phase has a point fermi surface where @xmath55 and @xmath56 has a linear dispersion around @xmath57 ( fig.[fig : basis ] ) . for simplicity , we will assume @xmath58 for further analysis . for this case , the fermi points are at @xmath59 the ground state of the kitaev model is a quantum spin liquid with only nearest neighbor spin - spin correlations . @xcite on an @xmath35bond , only @xmath60 is non zero and other two spin correlations are zero . four spin bond - bond correlations are long - ranged with power - law decay in the gapless phase of the kitaev model . consider a spin @xmath7 impurity locally exchange - coupled to a host ( kitaev ) spin at an @xmath30 site ( @xmath61 ) : @xmath62 we perform a poor man s scaling analysis@xcite for the kondo coupling @xmath8 to study the screening of the impurity spin by the host excitations . to study the system properties at low temperatures , we can compute the effective hamiltonian for a reduced bandwidth for the fermionic excitations ( @xmath63 ) by integrating out the excitations in the band edges ( ( @xmath64 ) . this process is successively repeated to get a scaling law for the coupling constants in the hamiltonian . we consider the lippmann - schwinger expansion for the @xmath65matrix element , @xmath66 making a perturbation expansion @xmath67 in increasing powers of @xmath8 and following its variation as a function of the decrease of the bandwidth @xmath68 we find that the first correction to the bare @xmath65matrix comes from two _ third _ order terms ( see fig . [ fig : kondo - on - site ] ) . the contribution from on - site scattering ( fig.[fig : kondo - on - site]a ) is @xmath69 + here @xmath70 is the density of states at the band edge , @xmath71 is the lattice constant and @xmath72 similarly , the contribution from fig . [ fig : kondo - on - site](b ) is @xmath73 .,title="fig : " ] + .,title="fig : " ] + [ fig : kondo - on - site ] adding the two contributions ( taking @xmath74 ) , @xmath75 here we have taken @xmath76 and neglected them . if either the impurity is a @xmath77 spin , or the kondo interaction is rotationally symmetric , the above contribution renormalizes the kondo coupling constant . however for @xmath78 with anisotropic coupling , new terms are generated and one needs to go to higher order diagrams to obtain the scaling of these new coupling terms . for @xmath5 or for symmetric impurity coupling we thus have @xmath79 just as for the kondo effect in graphene@xcite , owing to the change in the density of states with bandwidth ( here @xmath80 ) , we also need to consider the change in @xmath8 due to the rescaling done in order to keep the total number of states fixed . this gives a contribution @xmath81 in addition , as we shall scale the bandwidth @xmath82 to smaller values , the second term in eq . [ eq : delta - k ] may be dropped . thus @xmath83 thus , as we decrease the bandwidth by integrating out the high energy excitations , the effective coupling @xmath8 has an unstable fixed point at @xmath84};$ ] or in other words , @xmath85 here we used @xmath86 and @xmath87 clearly for @xmath88 the coupling flows to infinity independent of the nature of coupling ( ferromagnetic or antiferromagnetic ) , while for @xmath89 the coupling flows to zero . for anisotropic kondo coupling we can show @xmath90.\label { eq : anis - kondo}\end{aligned}\ ] ] the two - parameter kondo flow is therefore given by @xmath91 a comparison of the kondo effect in graphene @xcite , a bosonic spin bath @xcite and the kitaev model are shown in table [ tab : comparison ] . bosonic spin bath with pseudogap density of states @xmath92 & kitaev , honeycomb lattice kondo scaling & unstable intermediate coupling fixed pt . only for afm coupling . only afm flows to strong coupling above unstable fixed pt . & flow direction is independent of the sign of magnetic impurity coupling . unstable intermediate coupling fixed pt . for both fm and afm . & scaling same as @xmath3 bosonic spin bath case . however a topological transition is associated with the unstable fixed point . the poor man s scaling analysis is only valid for small kondo couplings as the perturbation theory breaks down much before the critical value of the coupling . while we have shown that the coupling flows to larger values above the critical value @xmath93 it remains to be seen whether there is any other fixed point beyond @xmath94 but less than the @xmath95 below we study the model in the strong coupling limit and see if it is a stable fixed point . in the strong coupling limit , @xmath8 is the largest energy scale and the impurity spins forms a singlet / triplet with the kitaev spin at origin . we consider the hamiltonian such that the kondo term and the kitaev model with one spin missing ( @xmath96 ) constitute the unperturbed hamiltonian and kitaev coupling to the site at origin is the perturbation : @xmath97 for antiferromagnetic kondo coupling ( @xmath10 ) , the ground state consists of a kondo singlet of @xmath98 and the kitaev model with one spin missing . the perturbation term causes transitions from singlet to triplet states of the kondo singlet . we use effective hamiltonian scheme@xcite to include the effects of the perturbation terms within the projected ground state subspace . @xmath99 where @xmath100 is chosen such that the terms which take us out of the reduced hilbert space are canceled order by order . this gives the reduced hamiltonian as @xmath101 where @xmath102 belong to the ground state manifold and @xmath103 belongs to excited state manifold . the eigenstates of the kondo term are singlet @xmath104 and triplet states @xmath105 : @xmath106 here @xmath107 refers to the kitaev spin and @xmath108 refers to the impurity spin state . for the antiferromagnetic kondo coupling case , ground state is the singlet state . as @xmath109 , @xmath110 and @xmath111 here , @xmath112 denotes the eigenstates of the kitaev model with the spin at origin missing . since change in energy of the kitaev state is @xmath113 , we ignore their contribution in the energy denominators of the perturbation term . the matrix elements of @xmath114 are then @xmath115 here , in @xmath116 the subscript @xmath117 refers to a neighoring site of the origin in the direction of the @xmath35bond . thus , in the antiferromagnetic coupling case , the kondo singlet decouples from the rest of the kitaev model and a small interaction ( @xmath118 ) is generated between the kitaev spin at the origin and the sins at the three neighboring sites in the second order perturbation . the strong coupling fixed point is thus a stable fixed point and is equivalent to the kitaev model with one site missing . in the ferromagnetic kondo coupling case , the triplet states form the ground state manifold . we perform degenerate perturbation theory to get the effective hamiltonian : @xmath119 if we calculate the matrix elements of @xmath120 these matrices are just the spin@xmath18 matrices : @xmath121 and the hamiltonian in the reduced subspace becomes @xmath122 where @xmath123 represents the spin-1 at the origin . thus for ferromagnetic impurity coupling , the new terms which couple the triplet and the rest of the kitaev model are similar to the original kitaev coupling and of the same strength . we get a kitaev - like model with a spin@xmath18 at the origin and spin@xmath124 elsewhere . here the kondo triplet does not decouple from rest of the kitaev model in the strong coupling limit and does not lend itself to a simple treatment , unlike the corresponding antiferromagnetic case . a remarkable property of the kondo effect in kitaev model is that the unstable fixed point is associated with a topological transition from the zero flux state to a finite flux state . the strong antiferromagnetic coupling limit amounts to studying the kitaev model with a missing site or cutting the three bonds linking this site to the neighbors . it was shown in kitaev s original paper@xcite that such states with an odd number of cuts are associated with a finite flux , and also that these vortices are associated with unpaired majorana fermions and have non - abelian statistics under exchange . it has also been shown numerically for the gapless phase@xcite that the ground state of the kitaev model with one spin missing has a finite flux pinned to the defect site . we argue the existence of a localized zero energy majorana mode from the degeneracy of the ground state in presence of impurity spin and elucidate on the nature of this zero mode . for the hamiltonian @xmath125 the three plaquettes @xmath126 @xmath127 and @xmath128 ( fig . [ fig : unpaired ] ) that touch the impurity site are no longer associated with conserved flux operators , while the flux operators that do not include the origin remain conserved . the three plaquette operator @xmath129 is still conserved and @xmath130 in the ground state of the unperturbed kitaev model . we now define composite operators @xmath131 @xmath132 and @xmath133 ( @xmath134 are the pauli spin matrices corresponding to the impurity ) . remarkably , these composite operators represent conserved quantities for arbitrary values of the impurity coupling . the @xmath135 s do not commute with each other and instead obey an @xmath136 algebra , @xmath137=2i\epsilon_{\alpha\beta\gamma}\tau^{\gamma}.$ ] this @xmath136 symmetry , which is exact for all couplings is realized in the spin-1/2 representation @xmath138 . clearly , all eigenstates , including the ground state are doubly degenerate ( corresponding to @xmath139 ) , and this applies also to the strong coupling limit . schematic of the three unpaired b majorana fermions formed as a result of cutting the links to the kitaev spin at the origin . any two of the three can be given an expectation value ( dotted bond ) . ] in the strong antiferromagnetic coupling limit @xmath140 the low energy states will be the ones in which the spin at the origin forms a singlet @xmath141 with the impurity spin , @xmath142 here @xmath143 represents the low energy states of the kitaev model with the spin at the origin removed . to see the action of the @xmath136 symmetry generators on these states , we note that they can be written as @xmath144 and @xmath145 do not involve the components of the spin at the origin , @xmath146 we then have @xmath147 . so , in the strong coupling limit , the symmetry generators act non - trivially only in the kitaev model sector , implying that the low energy states of the kitaev model with one spin removed are all doubly degenerate , with the double degeneracy emerging from the kitaev sector . this is also true for the zero - energy mode in the single particle spectrum : the two degenerate states correspond to the zero mode being occupied or unoccupied . let us examine the structure of the zero mode . removing a kitaev spin creates three unpaired @xmath148majorana fermions at the neighboring sites , say , @xmath149 @xmath150 and @xmath151 ( fig . [ fig : unpaired ] ) . now @xmath152 is conserved and commutes with all the conserved flux operators @xmath39 but not with the two other combinations @xmath153 and @xmath154 so , we can choose a gauge where the expectation value of @xmath152 is equal to + 1 such that these two @xmath148modes drop out of the problem and we equivalently have one unpaired @xmath148majorana fermion . the unpaired @xmath155 majorana has dimension @xmath156 and therefore , there must be an unpaired majorana mode in the @xmath157sector ( again of dimension @xmath156 ) so that together these two give the full ( doubly degenerate ) zero energy mode . also , while the @xmath155 mode is sharply localized , the wave function of the @xmath158 mode can be spread out in the lattice . for the ferromagnetic case , while the strong coupling limit also leads to a model with doubly - degenerate levels , we are unable to explicitly identify a zero energy unpaired majorana fermion and not address the question as to whether a nontrivial @xmath2 flux can be associated with closed paths enclosing the @xmath16 defect . for this purpoose , we perform a numerical exact diagonalization analysis below . we have used a modified lanczos algorithm to calculate the ground state properties of a finite kitaev fragment exchange - coupled to an external @xmath5 impurity spin as discussed in the previous sections . for the antiferromagnetic case , where we already know that the strong - coupling limit @xmath159 is associated with a nontrivial flux @xmath160 at the defect site , an exact - diagonalization calculation with a fragment as small as three hexagons ( open boundary conditions , impurity spin coupled to central site ) is sufficient to confirm @xmath161 figure [ fig_imp_flux ] shows the expectation value of the flux operator @xmath162 and total spin @xmath163 as a function of the impurity coupling @xmath164 it is seen that the @xmath165 for @xmath166 ( i.e in the pure kitaev case ) as it should be , since the ground state of kitaev model is flux free , whereas it changes to -1 as @xmath8 is increased , implying a finite @xmath2 flux at the origin . within numerical accuracy , it also appears that the topological transition from @xmath165 to @xmath160 practically occurs at the same value of @xmath8 at which a bound singlet state is formed between the impurity spin and the kitaev host . for negative values of @xmath8 for this three - hexagon fragment , @xmath167 appears to stay close to one implying a flux - free state even as the total spin at the defect site begins approaching @xmath168 we suspect this anomalous result is not generally true for larger fragments and may have originated from the presence of a large number of boundary spins that interact with the central spin . we thus performed exact - diagonalization calculations with a larger fragment with six hexagons ( open boundary conditions , impurity spin coupled to central site ) . as we increase the ferromagnetic impurity coupling , we clearly observe the @xmath169 to @xmath160 topological transition . once again , the topological phase transition and the magnetic transition ( at which the total spin at the defect site becomes @xmath12 are practically coincident . ) and the total spin at the impurity site with @xmath8 evaluated on a three hexagon fragment . the impurity spin is coupled to the central spin and the flux is calculated over the boundary . the inset shows the slight decrease of @xmath167 from 1 for ferromagnetic coupling whereas for antiferromagnetic coupling @xmath167 changes to -1 . _ ( bottom ) _ @xmath167 and total spin at the impurity site evaluated for a six hexagon fragment showing the transition to @xmath167 = -1 state for ferromagnetic coupling . ] in ref . , the temperature and magnetic field dependences of the magnetic susceptibility of the kitaev model with a missing site were obtained as @xmath170 and @xmath171 respectively , which bears striking resemblance to the low - temperature impurity susceptibility in the two - channel kondo model @xcite . this is not a mere coincidence . in the two - channel kondo problem , it is long known @xcite that the low energy physics is described by a model of a localized , zero energy majorana fermion interacting with a band of dispersing majorana fermions with a finite density of states at the fermi energy . likewise , in the kitaev model with a missing site , we can clearly identify a localized zero energy majorana @xmath148fermion coexisting with dispersing majorana @xmath157fermions with a _ nonvanishing _ density of states @xcite at zero energy . a nonvanishing density of states for the dispersing majoranas is a very unusual result for a honeycomb lattice , and is associated with the fact that the missing site is associated with a finite @xmath2 flux . if one instead estimates the density of states for the dispersing fermions in the absence of a @xmath4flux , the density of states would vanish @xcite at the fermi energy ; indeed , this would be the case in graphene . to compute the magnetic susceptibility , we choose a gauge where @xmath172 is the zero energy localized majorana fermion . using @xmath173 the magnetic susceptibility of the defect can be expressed as @xmath174 where @xmath175 and @xmath176 are respectively the green functions of the dispersing and the ( zero - energy ) localized majorana fermions and we used the fact that the density of states of the dispersing majorana fermions does not vanish at zero energy . for finite fields @xmath177 in the low temperature limit , the logarithmic divergence of eq.[chi_imp1 ] gets cut off by the field , and one obtains @xmath178 for the ground state entropy , one notes that the dispersing majorana fermions have zero entropy at @xmath179 while the localized zero energy majorana fermion has a finite entropy @xmath180 this again agrees with the two - channel kondo result.@xcite in summary , we showed that the problem of spinless and spinful defects in the honeycomb @xmath5 kitaev model can be approached from a more general `` kondo perspective '' of local exchange coupling of external paramagnetic impurities with a host spin . on one hand , such an approach gives us a new class of kondo effects where the magnetic binding - unbinding transition is accompanied by a change of topology of the ground state . on the other hand , some intriguing recent observations , such as logarithmic singularities in the magnetic response of kitaev models with vacancies , are now recognizable as familiar kondo stories - in this case , we note a remarkable similarity with the two - channel kondo problem . it would be interesting to study a lattice of vacancies in the kitaev model from the perspective of a two - channel kondo lattice . one would like to better understand the kitaev model with a @xmath16 defect . this nonintegrable nature of this problem prevents us from repeating the kind of analysis one could make for the vacancy case where similarity with the two - channel kondo problem was observed . our numerical approach , based on exact diagonalization calculations of relatively small fragments , can not answer questions such as the density of states of low energy excitations . another direction for future study would be to consider a more general kitaev - heisenberg model and track the kondo effect as a function of the relative strengths of kitaev and kondo interactions . this should give insights into magnetic impurity response in kitaev candidate materials where kitaev and heisenberg interactions are believed to compete with each other . we are grateful for useful discussions with k. damle . v.t . acknowledges financial support from argonne natl . lab . , the university of chicago center in delhi , and dst ( india ) swarnajayanti grant ( no . dst / sjf / psa-0212012 - 13 ) . s.d.d acknowledges the financial support provided by cambridge commonwealth trust(cct ) and hospitality provided by dtp ( tifr ) . a. banerjee , c.a . bridges , j - q yan , a.a . aczel , l.li , m.b.stone , g.e . granroth , m.d . lumsden , y.yui , j.knolle , d. l. kovrizhin , s.bhattacharjee , r. moessner , d.a.tennant,d.g.mandrus,and s.e.nagler , arxiv 1504.08037,(2015 ) .

paramagnetic impurities in a quantum spin - liquid can result in kondo effects with highly unusual properties . we have studied the effect of locally exchange - coupling a paramagnetic impurity with the spin-@xmath0 honeycomb kitaev model in its gapless spin - liquid phase . the ( impurity ) scaling equations are found to be insensitive to the sign of the coupling . the weak and strong coupling fixed points are stable , with the latter corresponding to a noninteracting vacancy and an interacting , spin-@xmath1 defect for the antiferromagnetic and ferromagnetic cases respectively . the ground state in the strong coupling limit in both cases has a nontrivial topology associated with a finite @xmath2 flux at the impurity site . for the antiferromagnetic case , this result can be obtained straightforwardly owing to the integrability of the kitaev model with a vacancy . the strong - coupling limit of the ferromagnetic case is however nonintegrable , and we address this problem through exact - diagonalization calculations with finite kitaev fragments . our exact diagonalization calculations indicate that that the weak to strong coupling transition and the topological phase transition occur rather close to each other and are possibly coincident . we also find an intriguing similarity between the magnetic response of the defect and the impurity susceptibility in the two - channel kondo problem .

we study packings of @xmath1-joins in plane graphs . let @xmath0 be a graph and @xmath1 an even - size subset @xmath1 of its vertices . @xmath1-join _ is a subgraph @xmath7 of @xmath0 such that the odd - degree vertices of @xmath7 are precisely those in @xmath1 . a _ cut _ is a partition of the vertex set of a graph @xmath0 into two sets @xmath8 and @xmath9 , which we refer to as _ sides _ ; the _ size _ of the cut is the number of edges with one end - vertex in @xmath8 and the other end - vertex in @xmath9 . a cut is _ trivial _ if one its sides consists of a single vertex and a cut is _ odd _ if the size of @xmath8 is odd . finally , a _ @xmath1-cut _ is a cut such that @xmath10 is odd . clearly , if @xmath0 has a @xmath1-cut of size @xmath2 , it can not have more than @xmath2 edge - disjoint @xmath1-joins . we are interested when the converse is also true . seymour @xcite ( also see problem 12.18 in @xcite ) conjectured the following for @xmath4 . [ conj ] let @xmath0 be a plane graph and @xmath1 an even - size subset of its vertices . if the sizes of all @xmath1-cuts in @xmath0 have the same parity and the size of every @xmath1-cut is at least @xmath2 , then @xmath0 contains @xmath2 edge - disjoint @xmath1-joins . the case @xmath3 is equivalent to the four color theorem . the cases @xmath4 and @xmath5 were proved by guenin @xcite . we remark that the case @xmath4 implies the four color theorem , as pointed out by seymour @xcite . here , we prove the next open case . our main result is the following . [ thm - main ] let @xmath0 be a plane multigraph and @xmath1 an even subset of its vertices . if every @xmath1-cut of @xmath0 has the same parity and the size of every @xmath1-cut is at least six , then @xmath0 contains six edge - disjoint @xmath1-joins . guenin @xcite argued that it suffices to prove conjecture [ conj ] for plane graphs @xmath0 with @xmath11 that are @xmath2-regular , i.e. , every vertex has degree @xmath2 . in such case , the existence of @xmath2 edge - disjoint @xmath1-joins is equivalent to the existence of a @xmath2-edge - coloring ; a _ @xmath2-edge - coloring _ is an assignment of colors to the edges such that no vertex is incident with two edges of the same color . hence , theorem [ thm - main ] for @xmath6 is equivalent to the next theorem which we prove in the following sections of the paper . [ thm - edge - coloring ] let @xmath0 be a @xmath12-regular plane multigraph . if @xmath0 has no odd cut of size less than @xmath12 , then @xmath0 has a @xmath12-edge - coloring . note that the condition that @xmath0 has no odd cut of size less than six implies that the number of vertices of @xmath0 is even ( otherwise , consider a cut with one of the sides empty ) . let us remark that conjecture [ conj ] would be implied by the following more general conjecture of seymour ( replacing the condition of not containing petersen by a stronger condition of being planar yields a statement equivalent to conjecture [ conj ] ) . [ conj - petersen ] let @xmath0 be a @xmath2-regular graph with no petersen minor . the graph @xmath0 is @xmath2-edge - colorable if and only if every odd cut of @xmath0 has size at least @xmath2 . the case @xmath3 is the well - known tutte s three edge - coloring conjecture , whose solution is announced by robertson , sanders , seymour and thomas ( see @xcite ) . indeed , the case @xmath3 is a special case of another well known conjecture by tutte , which is known as the tutte s four flow conjecture . conjecture [ conj - petersen ] would also imply the following conjecture of conforti and johnson @xcite , also see @xcite . [ conj - postman ] let @xmath0 be a graph with no petersen minor and @xmath1 a set of its odd - degree vertices . then , the maximum number of edge - disjoint @xmath1-joins is equal to the size of the smallest @xmath1-cut . conjectures [ conj ] , [ conj - petersen ] and [ conj - postman ] are paid attention by many researchers , because they are connected not only to @xmath1-joins , @xmath1-cuts and edge - coloring but also to cycle covers and flows . for more details , we refer the reader to the books by cornujols @xcite and by schrijber @xcite , respectively . we now introduce notation used throughout the paper . a vertex of degree @xmath13 is called a _ @xmath13-vertex_. an _ @xmath14-vertex _ is a vertex of degree at least @xmath13 and an _ @xmath15-vertex _ is a vertex of degree at most @xmath13 . in a @xmath16-connected plane graph , a @xmath13-face is a face incident with exactly @xmath13 edges . analogously to vertices , we use an @xmath15-face and an @xmath14-face . two faces in a plane graph are _ adjacent _ if they share a common edge . we say that a face is _ @xmath2-big _ if it is adjacent to @xmath2 @xmath17-faces . a _ bigon _ is a @xmath16-face and a series of consecutively adjacent bigons is called _ multigon_. the _ order _ of a multigon is the number of edges forming it , i.e. , the number of bigons forming it increased by one . multigons of order three are called _ trigons _ and those of order four _ quadragons_. two multigons are _ incident _ if they share a vertex . if @xmath18 is a face , then two multigons are _ @xmath18-incident _ if they contain edges consecutive on the boundary of @xmath18 . we extend this notion to a multigon and a face and to two faces in the natural way . a @xmath19-face is _ dangerous _ if either it is adjacent to two trigons and at least one bigon or it is adjacent to a trigon and at least three bigons . a @xmath20-face is _ dangerous _ if it is adjacent to three trigons and three bigons . finally , a multigon @xmath21 is _ dangerous _ if * @xmath21 is a quadragon , * @xmath21 is a trigon adjacent to a dangerous @xmath19-face @xmath18 such that @xmath21 is @xmath18-incident with a multigon , or * @xmath21 is a trigon adjacent to a dangerous @xmath20-face @xmath18 such that @xmath21 is @xmath18-incident with two multigons . the rest of the paper is devoted to proving theorem [ thm - edge - coloring ] . with respect to this proof , a plane graph @xmath0 is said to be a _ minimal counterexample _ if @xmath0 satisfies the assumptions of theorem [ thm - edge - coloring ] , i.e. , * @xmath0 is @xmath12-regular , * every odd cut of @xmath0 has size at least six , and * @xmath0 has no @xmath12-edge - coloring , and it also holds that * subject to the previous conditions , @xmath0 has the smallest order , * subject to the previous conditions , @xmath0 has as many quadragons as possible , * subject to the previous conditions , @xmath0 has as many trigons as possible , and * subject to the previous conditions , @xmath0 has as many bigons as possible . by lemma 2.2 from @xcite , in order to prove theorem [ thm - edge - coloring ] , it is enough to exclude the existence of a minimal counterexample . in our arguments , we will often need to transform an edge - coloring to another one . to simplify our arguments , we will use the letters @xmath22 , @xmath23 , @xmath24 , @xmath25 , @xmath26 and @xmath27 to denote the colors used on edges . if @xmath0 is a graph with maximum degree @xmath13 that is @xmath13-edge - colored , then an _ @xmath28-chain _ , where @xmath22 and @xmath23 are two colors used on the edges of @xmath0 , is a cycle or a maximal path formed by edges with the colors @xmath22 and @xmath23 only . _ swapping _ the colors the of edges on an @xmath28-chain means recoloring @xmath22-colored edges of the chain with @xmath23 and @xmath23-colored edges with @xmath22 . in this section , we analyze structure of a minimal counterexample ; in the next section , we then prove that there exists no minimal counterexample using the discharging method . we start with analyzing sizes and structure of odd cuts in a minimal counterexample . as the first step , we prove the following simple observation . [ lm - odd - cut ] every non - trivial odd cut in a minimal counterexample @xmath0 has size at least eight . since @xmath0 is @xmath12-regular , every cut in @xmath0 has even size . hence , if @xmath0 has a non - trivial odd cut of size less than eight , its size must be six . let @xmath8 and @xmath9 be the sides of such a non - trivial odd cut . let @xmath29 be the ( plane ) graph obtained from @xmath0 by replacing @xmath8 with a single vertex incident with the six edges of the cut @xmath30 . similarly , @xmath31 is the graph obtained from @xmath0 by replacing @xmath9 with a single vertex incident with the six edges of the cut @xmath30 . by the minimality of @xmath0 , both @xmath29 and @xmath31 have @xmath12-edge - colorings . these edge - colorings combine to a @xmath12-edge - coloring of @xmath0 ( the edges of the cut receive six distinct colors ) which contradicts that @xmath0 is a minimal counterexample . using lemma [ lm - odd - cut ] , we prove the following simple observation on the structure of multigons in a minimal counterexample . [ lm - multigon ] in a minimal counterexample @xmath0 , the order of every multigon is a most four and the sum of the orders of any two incident multigons is at most five . if @xmath0 contains a multigon of order five or two incident multigons with orders summing to six , then there is a vertex @xmath32 that has only two neighbors , say @xmath33 and @xmath34 . unless @xmath0 has exactly four vertices , the set @xmath35 forms a side of a non - trivial odd cut of size six which is impossible by lemma [ lm - odd - cut ] hence , @xmath0 has exactly four vertices and it is straightforward to show that it can be @xmath12-edge - colored . one of possible ways to to obtain an edge - coloring of minimal counterexample , which is assumed not to exist , is reducing a minimal counterexample to another @xmath12-regular graph of the same order but with multigons of larger order . such a reduction can only be possible if the resulting graph has no odd cuts of size less than six . one operation that will observe this property is the swapping operation that we now introduce . if @xmath0 is a plane graph such that it is possible to draw a closed curve in the plane that intersects @xmath0 only at vertices @xmath36 for @xmath2 even and @xmath0 contains edges @xmath37 , @xmath38 , @xmath39 , @xmath40 and @xmath41 ( such a @xmath2-tuple of vertices @xmath36 is called _ eligible _ ) , then the graph obtained from @xmath0 by the _ @xmath42-swap _ is the plane graph obtained by removing the edges @xmath37 , @xmath38 , @xmath39 , @xmath40 and @xmath41 and inserting the edges @xmath43 for @xmath44 . a crucial property of this operation is that the size of odd cuts can decrease by at most two if @xmath2 is four or six . we prove this in the next two lemmas . [ lm - swap ] let @xmath0 be a minimal counterexample and let @xmath2 be either four or six . any graph @xmath45 obtained from @xmath0 by a @xmath46-swap for eligible vertices @xmath36 is @xmath12-regular and has no odd cut of size less than six . clearly , the graph @xmath45 is @xmath12-regular . consider an odd cut with sides @xmath8 and @xmath9 of @xmath45 . observe that this cut is also odd in @xmath0 . by symmetry , we can assume that @xmath47 . let @xmath48 be those vertices @xmath49 of @xmath8 such that @xmath50 or @xmath51 is in @xmath8 . since @xmath47 , the set @xmath52 is either empty or formed by two or three consecutive vertices . if @xmath52 is empty or formed by three consecutive vertices , the number of edges leaving @xmath52 to @xmath9 is the same in @xmath0 and @xmath45 . if @xmath8 is formed by two consecutive vertices , then the number of such edges leaving @xmath52 to @xmath9 is either increased or decreased by two . since the number of edges between the vertices of @xmath53 and the vertices of @xmath9 is preserved , every odd cut of @xmath45 has size at least six . a crucial property of a minimal counterexample is the existence of an @xmath54-coloring . let us define this notion formally . if @xmath0 is a @xmath12-regular graph and @xmath54 an edge of @xmath0 , then an _ @xmath54-coloring _ of @xmath0 is a coloring of edges of @xmath0 with six colors such that every edge except for @xmath54 is assigned one color , @xmath54 is assigned three or more colors , and for every color , each vertex is incident with an odd number of edges assigned that color . observe that in an @xmath54-coloring , every vertex except the end - vertices of @xmath54 must be incident with edges of mutually different colors , i.e. , an @xmath54-coloring is proper at all vertices except for the end vertices of @xmath54 . also observe that the number of colors assigned to @xmath54 in an @xmath54-coloring must always be odd . the following lemma appears as lemma 2.5 in @xcite . [ lm - e - coloring ] let @xmath0 be a minimal counterexample . for every edge @xmath54 , there exists an @xmath54-coloring . we now strengthen lemma [ lm - e - coloring ] for the case when @xmath54 is contained in a multigon of order at least three . [ lm - e - trigon ] let @xmath0 be a minimal counterexample and @xmath55 an edge of @xmath0 contained in a multigon of order three or more . there exists an @xmath54-coloring such that @xmath54 is assigned precisely three colors , say @xmath22 , @xmath23 and @xmath24 , and one of these colors , say @xmath22 , is assigned to other two edges incident with @xmath32 as well as other two edges incident with @xmath33 . by lemma [ lm - e - coloring ] , there exists an @xmath54-coloring of @xmath0 . recall that @xmath54 must be assigned an odd number of colors assume that @xmath54 is assigned exactly three colors , say @xmath22 , @xmath23 and @xmath24 . if an edge contained in the multigon with @xmath54 is assigned one of these three colors , say @xmath24 , then we can obtain a @xmath12-edge - coloring of @xmath0 by assigning to this edge the color @xmath22 and assigning the edge @xmath54 the color @xmath23 only . hence , either @xmath54 is contained in a trigon and the other two edges of the trigon are assigned the colors @xmath25 and @xmath26 ( by symmetry ) or @xmath54 is contained in a quadragon and the other three edges are assigned the colors @xmath25 , @xmath26 and @xmath27 . by permuting the colors of the edges of the multigon and exchanging the roles of @xmath32 and @xmath33 , we can assume that one of the following cases apply : * * the vertex @xmath32 is incident with two edges colored with @xmath22 in addition to @xmath54 and the vertex @xmath33 is also incident with two edges colored with @xmath22 in addition to @xmath54 . * in this case , the edge - coloring is of the type described in the statement of the lemma . * * the vertex @xmath32 is incident with two edges colored with @xmath22 in addition to @xmath54 and the vertex @xmath33 is also incident with two edges colored with @xmath23 in addition to @xmath54 . * let @xmath56 be one of the edges incident with @xmath33 assigned the color @xmath23 . let @xmath57 be the end edge of the @xmath28-chain of @xmath58 starting at @xmath56 . the edge @xmath57 is either an edge colored with @xmath22 incident with @xmath32 or the other edge colored with @xmath23 incident with @xmath33 . if @xmath57 is incident with @xmath32 , we can obtain a proper @xmath12-edge - coloring of @xmath0 by swapping the colors on the chain and assigning @xmath54 the color @xmath24 only . hence , @xmath57 is incident with @xmath33 . in this case , we swap the colors on the chain and obtain an @xmath54-coloring of the type described in the statement of the lemma . * * the vertex @xmath32 is incident with two edges colored with @xmath22 in addition to @xmath54 and the vertex @xmath33 is incident with three edges colored with @xmath27 . * observe that the order of the multigon must be three in this case . + consider the @xmath59-chains of @xmath58 starting at edges incident with @xmath33 . since @xmath33 is incident with three edges colored with @xmath27 and no edge colored with @xmath22 in @xmath58 , at least one of these chains ends at an edge incident with @xmath32 . let @xmath60 be this edge . + if @xmath60 is colored with @xmath22 , swap the colors @xmath22 and @xmath27 on the @xmath59-chain and color @xmath54 with the colors @xmath23 , @xmath24 and @xmath27 . an @xmath54-coloring of the desired type can now be obtained by exchanging the colors @xmath22 and @xmath27 in the whole graph . + hence , we can assume that there is no @xmath59-chain starting an edge incident with @xmath33 and ending at an edge incident with @xmath32 with the color @xmath22 . consequently , there exists an @xmath59-chain starting and ending at an edge incident with @xmath33 . by swapping the colors @xmath22 and @xmath27 on this chain , we obtain an @xmath54-coloring of the type described in the statement of the lemma . * * each of the vertices @xmath32 and @xmath33 is incident with three edges colored with @xmath27 . * observe that the order of the multigon must again be three . + since @xmath32 is incident with three edges colored with @xmath27 and no edge colored with @xmath22 in @xmath58 , there exists a @xmath59-chain in @xmath58 that starts at an edge incident with @xmath32 and ends at an edge incident with @xmath33 . swap now the colors @xmath22 and @xmath27 on this chain and color @xmath54 with the colors @xmath23 , @xmath24 and @xmath27 . the desired edge coloring of @xmath0 can now be obtained by exchanging the colors @xmath22 and @xmath27 in the whole graph . assume now that @xmath54 is assigned five colors , say @xmath22 , @xmath23 , @xmath24 , @xmath25 and @xmath26 . if one of these colors , say @xmath26 , is assigned to another of edge of the multigon , then assign @xmath54 the colors @xmath22 , @xmath23 and @xmath24 only and assign @xmath25 to the edge colored @xmath26 . on the other hand , if other two edges in the multigon are assigned the color @xmath27 , then assign to @xmath54 the colors @xmath22 , @xmath23 and @xmath24 only and recolor the other two edges with @xmath25 and @xmath26 . in either of the two cases , we have found an @xmath54-coloring of @xmath0 assigning @xmath54 exactly three colors . since we have already analyzed such @xmath54-colorings , the proof is now completed . hereafter , we will sometimes refer to the assumption of lemma [ lm - e - trigon ] . in particular , the color @xmath22 will be the color assigned to other two edges incident with @xmath32 as well as other two edges incident with @xmath33 . the existence of @xmath54-colorings is related to the notion of mates which also appeared in @xcite . we use a slightly different but equivalent terminology to that in @xcite . if @xmath0 is a minimal counterexample , @xmath54 an edge of @xmath0 and @xmath61 one of the colors used in an @xmath54-coloring , then a _ @xmath61-mate _ @xmath62 is a set of edges of @xmath0 that form an odd cut containing @xmath54 such that , for every color @xmath63 , @xmath62 contains exactly one edge that is assigned the color @xmath64 . a @xmath61-mate is _ non - trivial _ if it forms a non - trivial cut of @xmath0 . lemma 2.6 in @xcite asserts the existence of mates in a minimal counterexample . [ lm - mate ] let @xmath0 be a minimal counterexample and @xmath54 an edge of @xmath0 . for every @xmath54-coloring and every color @xmath61 , there exists a @xmath61-mate . the following observation on the structure of mates is often used in our arguments . we state it as a proposition for future references . [ prop - mate ] let @xmath0 be a minimal counterexample and @xmath54 an edge of @xmath0 . in each @xmath54-coloring , every non - trivial @xmath61-mate @xmath62 contains at least five edges ( possibly including @xmath54 ) assigned the color @xmath61 . by lemma [ lm - odd - cut ] , the mate @xmath62 contains at least eight edges . since the edge @xmath54 is assigned at least three colors and @xmath62 contains exactly one edge assigned each of the colors @xmath63 , @xmath62 must include at least five edges assigned the color @xmath61 . we use the existence of a @xmath61-mate @xmath62 in the following way : if @xmath62 contains an edge adjacent to a face @xmath18 or an edge in a multigon adjacent to @xmath18 , then @xmath18 is @xmath17-face unless @xmath18 is incident with an end - vertex of @xmath54 , or @xmath18 is adjacent to an edge of @xmath62 colored with a color different from @xmath61 or @xmath18 is adjacent to a multigon containing edges with colors different from @xmath61 only . let us demonstrate the use of this approach in the following lemma . [ lm - face - quadragon ] every face @xmath18 adjacent to a quadragon is @xmath65-big . let @xmath54 be the edge of the quadragon incident with @xmath18 and consider an @xmath54-coloring as described in lemma [ lm - e - trigon ] . let @xmath62 be a @xmath61-mate for @xmath66 and @xmath67 another edge of @xmath18 contained in @xmath62 . let @xmath68 be the other face containing the edge @xmath67 . since the edges of the quadragon are assigned all six colors and the mate @xmath62 contains at least five edges with the color @xmath61 by proposition [ prop - mate ] , the face @xmath68 contains another edge with the color @xmath61 . hence , @xmath68 is @xmath17-face . since the faces @xmath68 differ for different choices of @xmath61 , we conclude that @xmath18 is @xmath65-big . another application of the mates is the following . [ lm - face - trigon ] every face @xmath18 adjacent to a trigon is adjacent to at least one edge not contained in a multigon . let @xmath54 be an edge of the trigon and consider an @xmath54-coloring as described in lemma [ lm - e - trigon ] . let @xmath69 be a @xmath27-mate . since the edges of the trigon have all the five colors different from @xmath27 , all the other edges contained in @xmath69 have the color @xmath27 . since the only vertices incident with two edges of the same color are the end vertices of @xmath54 , no edge of @xmath69 except for those contained in the trigon is contained in a multigon . in particular , the edge of @xmath18 contained in @xmath69 and not in the trigon is not contained in a multigon . in the rest of this section , we will focus in more detail on multigons adjacent to faces of various sizes . in this subsection , we focus on @xmath70-faces . we start with @xmath70-faces adjacent to a trigon . [ lm-3-face - trigon ] if a @xmath70-face @xmath18 in a minimal counterexample is adjacent to a trigon , then @xmath18 is adjacent to no other multigon , both the faces adjacent to @xmath18 are @xmath65-big and the other face adjacent to the trigon is also @xmath65-big . in particular , it is @xmath71-face . let @xmath72 be an edge of the trigon adjacent to @xmath18 and consider an @xmath54-coloring as in lemma [ lm - e - trigon ] . since the edges incident with @xmath73 and @xmath74 not contained in the trigon have the colors @xmath22 and @xmath27 only and no vertex except for @xmath73 and @xmath74 is incident with two edges of the same color , the face @xmath18 can not be adjacent to another multigon and the other two edges adjacent to @xmath18 have the colors @xmath22 and @xmath27 . let @xmath75 be the remaining vertex of the face @xmath18 , @xmath76 the other face adjacent to the trigon , @xmath77 the face incident with the edge @xmath78 and @xmath79 a neighbor of @xmath75 on the face @xmath77 ( see figure [ fig-3-face - trigon ] ) . the graph @xmath45 obtained from @xmath0 by the @xmath80-swap has a @xmath12-coloring by the minimality of @xmath0 ( there is a quadragon between @xmath73 and @xmath74 after the swap ) and lemma [ lm - swap ] . by symmetry , we can assume that the colors of the edges of the quadragon between @xmath73 and @xmath74 are @xmath23 , @xmath24 , @xmath25 and @xmath26 , and the color of the edge @xmath78 is @xmath27 . if the edge @xmath81 has a color different from @xmath22 , then we can obtain a proper @xmath12-edge - coloring of @xmath0 by assigning the edges @xmath37 and @xmath82 the color of the edge @xmath81 . hence , the color of @xmath81 is @xmath22 . consider now the following @xmath54-coloring : the edge @xmath54 is colored with @xmath22 , @xmath23 , @xmath24 , the other edges of the trigon with @xmath25 and @xmath26 , the edge @xmath37 is colored with @xmath22 , the edge @xmath83 is also colored with @xmath22 and the remaining edges of @xmath0 have the same colors as in @xmath45 . let @xmath62 be a @xmath61-mate for @xmath66 . observe that @xmath62 contains the three edges of the trigon and the edge @xmath78 which is colored with @xmath27 ( it can not contain the edge @xmath37 because its color is @xmath22 ) . we show that both @xmath76 and @xmath77 are @xmath65-big . since @xmath0 is @xmath16-connected ( it is @xmath12-regular and @xmath12-edge - connected ) , the faces @xmath76 and @xmath77 are different . we now argue that @xmath76 is @xmath65-big . let @xmath67 be the edge incident with @xmath76 that is contained in @xmath62 and that is not contained in the trigon . clearly , the color of @xmath67 is @xmath61 . let @xmath68 be the face adjacent to @xmath67 that is distinct from @xmath76 . since the mate @xmath62 contains at least five edges with the color @xmath61 by proposition [ prop - mate ] , @xmath68 is also distinct from @xmath77 . since @xmath68 contains another edge colored with @xmath61 by proposition [ prop - mate ] , @xmath68 must be @xmath17-face . since the faces @xmath68 are different for different choices of @xmath66 , the face @xmath76 is @xmath65-big . the argument that @xmath77 is @xmath65-big follows the same lines . switching the roles of @xmath73 and @xmath74 yields that the face adjacent to @xmath18 distinct from @xmath77 is also @xmath65-big . we now focus on @xmath70-faces adjacent to bigons . [ lm-3-face-23-bigon ] if a @xmath70-face of a minimal counterexample is adjacent to at least two bigons , then it is adjacent exactly to two bigons and the other faces adjacent to these bigons are @xmath65-big . let @xmath73 , @xmath74 and @xmath75 be the vertices of @xmath18 in such an order that there is a bigon between @xmath73 and @xmath74 and between @xmath73 and @xmath75 . let @xmath76 be the other face adjacent to the bigon between @xmath73 and @xmath74 and @xmath77 the other face adjacent to the bigon between @xmath73 and @xmath75 . finally , let @xmath79 be the neighbor of @xmath73 on @xmath77 different from @xmath75 . also see figure [ fig-3-face-23-bigon ] . by the minimality of @xmath0 and lemma [ lm - swap ] , the graph @xmath45 obtained from @xmath0 by the @xmath80-swap has a @xmath12-edge - coloring . let @xmath23 , @xmath24 and @xmath25 be the colors assigned to the edges between @xmath73 and @xmath74 and @xmath26 and @xmath27 the colors of the edges between @xmath73 and @xmath75 . since the edge @xmath81 can not have the color @xmath26 or @xmath27 , we can assume that its color is @xmath22 or @xmath23 by symmetry . in the latter case , coloring the edges of the bigon between @xmath73 and @xmath74 with the colors @xmath24 and @xmath25 , the edges @xmath83 and @xmath37 with @xmath23 and the other edges with their colors in @xmath45 yields a @xmath12-edge - coloring of @xmath0 . hence , the color of @xmath81 is @xmath22 . observe that this implies that the edge @xmath37 is not contained in a multigon . consider now the following @xmath54-coloring where @xmath54 is one of the edges of the bigon between @xmath73 and @xmath74 : the edge @xmath54 is assigned the colors @xmath22 , @xmath23 and @xmath24 , the other edge of the bigon the color @xmath25 , the edges @xmath83 and @xmath37 the color @xmath22 and the remaining edges preserve their colors . consider a @xmath61-mate @xmath62 for @xmath66 . this mate must contain all the edges of the bigons between @xmath73 and @xmath74 and between @xmath73 and @xmath75 . let @xmath67 be the edge of @xmath76 contained in @xmath62 . the color of @xmath67 must be @xmath61 and the other face containing @xmath67 must contain another edge with the color @xmath61 . consequently , it is a @xmath17-face . we conclude ( by considering all choices of @xmath61 ) that @xmath76 is @xmath65-big face . a symmetric argument applies to the face adjacent to the bigon between @xmath73 and @xmath75 . before considering @xmath70-faces adjacent to a single bigon , we have to prove the following lemma : [ lm - trigon-2big ] in a minimal counterexample , any trigon adjacent to an @xmath84-big face is also adjacent to an @xmath17-big face . let @xmath54 be an edge of the trigon , @xmath18 the @xmath84-big face adjacent to the trigon and @xmath76 the other face adjacent to the trigon . consider an @xmath54-coloring described in lemma [ lm - e - trigon ] and let @xmath62 be a @xmath61-mate for @xmath66 . the mate @xmath69 contains at least five edges colored with @xmath27 by proposition [ prop - mate ] and no edges of other colors except for those contained in the trigon . hence , one of the ( at most ) two @xmath17-faces adjacent to @xmath18 shares with @xmath18 an edge colored @xmath27 . by symmetry with respect to the colors @xmath23 , @xmath24 , @xmath25 and @xmath26 , we can assume that the other @xmath17-face adjacent to @xmath18 ( if it exists ) shares with @xmath18 an edge with color different from @xmath23 , @xmath24 and @xmath25 . consider now a mate @xmath62 , @xmath85 . on the face @xmath18 , the mate @xmath62 either contains an edge colored with @xmath27 or ( if @xmath86 ) an edge colored with the color @xmath61 that lies on a @xmath87-face ( which forces @xmath62 to contain an edge incident with this face that is colored with @xmath27 ) . in both cases , since the mate @xmath62 contains at least five edges colored with @xmath61 by proposition [ prop - mate ] , one of these edges ( which are colored with @xmath61 ) must lie on the face @xmath76 and the face containing this edge different from @xmath76 is @xmath17-face . since these faces are different for different values of @xmath61 in @xmath88 , the face @xmath76 is @xmath17-big . we are now ready to consider @xmath70-faces adjacent to a single bigon . [ lm-3-face-1-bigon ] if a @xmath70-face @xmath18 in a minimal counterexample is adjacent to a single bigon and the the other face adjacent to this bigon @xmath84-big , then the other two faces adjacent to @xmath18 are @xmath89-big . let @xmath73 , @xmath74 and @xmath75 be the vertices of @xmath18 in such an order that the bigon is between @xmath73 and @xmath74 . let @xmath76 be the face incident with the edge @xmath78 , @xmath77 the other face adjacent to the bigon and let @xmath79 be the neighbor of @xmath73 on @xmath76 different from @xmath75 . also see figure [ fig-3-face-1-bigon ] . if @xmath0 contains a trigon between the vertices @xmath73 and @xmath79 , the claim follows from lemma [ lm - trigon-2big ] . otherwise , the minimality of @xmath0 and lemma [ lm - swap ] implies that the graph @xmath45 obtained from @xmath0 by the @xmath80-swap has a @xmath12-edge - coloring . by symmetry , we can assume that the colors assigned to the edges between @xmath73 and @xmath74 are @xmath23 , @xmath24 and @xmath25 , and @xmath26 is the color of the edge @xmath78 . if the color of the edge @xmath81 is one of the colors @xmath23 , @xmath24 and @xmath25 , we can obtain a @xmath12-coloring of @xmath0 as in the proof of lemma [ lm-3-face - trigon ] . hence , we can assume that the color of @xmath81 is @xmath22 . let @xmath54 be one of the edges of the bigon and consider the following @xmath54-coloring : the edge @xmath54 is assigned the colors @xmath22 , @xmath23 and @xmath24 , the other edge of the bigon the colors @xmath25 , the edges @xmath83 and @xmath37 are assigned the color @xmath22 and the remaining edges preserve their colors . let @xmath62 be a @xmath61-mate for @xmath66 . observe that each mate @xmath62 , @xmath66 , contains the two edges of the bigon as well as the edge @xmath78 . by proposition [ prop - mate ] , the mate @xmath69 contains at least five edges colored with @xmath27 . the construction of the @xmath54-coloring yields that all these edges must lie on @xmath17-faces ( since their end - vertices are distinct ) . hence , one of the ( at most ) two @xmath17-faces adjacent to @xmath77 shares an edge with color @xmath27 with @xmath77 . the argument that we now present is symmetric with respect to the colors @xmath23 , @xmath24 , @xmath25 and @xmath26 . hence , we can assume that the other @xmath17-face adjacent to @xmath77 ( if it exists ) shares with @xmath77 an edge with color different from @xmath26 . on the face @xmath77 , each of the mates @xmath62 , @xmath90 , either contains an edge colored with @xmath27 or an edge colored with the color @xmath61 that lies in a @xmath87-face . in the latter case , the other edge of that face contained in @xmath62 must have the color @xmath27 . hence , @xmath62 contains an edge @xmath27 in both cases . since the mate @xmath62 contains at least five edges colored with @xmath61 by proposition [ prop - mate ] , it contains at least three additional edges colored with @xmath61 . one of these edges must lie on the face @xmath76 and the face containing this edge different from @xmath76 must be a @xmath17-face . since these faces are different for different values of @xmath90 , the face @xmath76 is @xmath89-big . we finish this subsection with an observation on faces around @xmath70-faces adjacent to trigons . [ lm-35-face - bigon - trigon ] a minimal counterexample @xmath0 does not contain a vertex @xmath73 incident with mutually adjacent @xmath70-face @xmath91 and dangerous a @xmath19-face @xmath92 , a bigon contained in @xmath91 but not in @xmath92 , and a trigon @xmath21 contained in @xmath92 such that @xmath21 is @xmath92-incident to a bigon . let @xmath74 be the other vertex contained in @xmath21 , let @xmath75 , @xmath79 and @xmath93 be the other vertices of @xmath92 ( in this order ) and let @xmath94 be the remaining vertex of @xmath91 . also see figure [ fig-35-face - bigon - trigon ] . assume first that @xmath92 is adjacent to a trigon between the vertices @xmath75 and @xmath79 . let @xmath45 be the graph obtained from @xmath0 by the @xmath95-swap . by the minimality of @xmath0 ( none of the affected multigons has order four in @xmath0 by lemma [ lm - face - quadragon ] ) and lemma [ lm - swap ] , the graph @xmath45 has a @xmath12-edge - coloring . let @xmath96 be the set of colors assigned to the four edges between @xmath73 and @xmath74 in @xmath45 . both the colors of the edges between @xmath93 and @xmath94 are in @xmath96 : the two colors not contained in @xmath96 are assigned to the edges @xmath97 and @xmath98 and thus the two edges between @xmath93 and @xmath94 can not have either of these two colors . at least three of the colors from @xmath96 are used on the edges of the quadragon between @xmath75 and @xmath79 in @xmath45 . hence , there is a color @xmath61 assigned to an edge between @xmath73 and @xmath74 , an edge between @xmath75 and @xmath79 , and an edge between @xmath93 and @xmath94 . remove the three edges colored with @xmath61 between these pairs of vertices and insert the edges of @xmath0 missing in @xmath45 . coloring the new edges with @xmath61 yields a @xmath12-edge - coloring of @xmath0 . assume next that @xmath0 contains a trigon between @xmath79 and @xmath93 . let @xmath99 be the graph obtained by the @xmath100-swap and consider its @xmath12-edge - coloring which exists by the minimality of @xmath0 and lemma [ lm - swap ] . let @xmath101 be the colors of the edges in the trigon between @xmath73 and @xmath94 in @xmath99 , @xmath102 the colors of the edges in the trigon between @xmath74 and @xmath75 , and @xmath103 the colors of the edges in the quadragon between @xmath79 and @xmath93 . since @xmath99 contains a bigon between @xmath73 and @xmath74 , it holds that @xmath104 . similarly , because the vertices @xmath73 and @xmath93 are joined by an edge , it holds that @xmath105 . consequently , there exists a color @xmath106 . removing edges of this color from the quadragon and the two trigons and coloring the edges of @xmath0 missing in @xmath99 with @xmath61 yields a @xmath12-edge - coloring of @xmath0 . the final case to consider is that @xmath92 is adjacent to three bigons but the only trigon adjacent to @xmath92 is the one formed by the edges between @xmath73 and @xmath74 . let @xmath99 be again the graph obtained by the @xmath100-swap and consider its @xmath12-edge - coloring which exists by the minimality of @xmath0 and lemma [ lm - swap ] . let @xmath101 be the colors of the edges in the trigon between @xmath73 and @xmath94 in @xmath99 , @xmath102 the colors of the edges in the trigon between @xmath74 and @xmath75 , and @xmath103 the colors of the edges in the quadragon between @xmath79 and @xmath93 . since @xmath99 contains a bigon between @xmath73 and @xmath74 , it holds that @xmath104 . if @xmath103 does not contain a color from @xmath107 , then the edges @xmath98 and @xmath81 have the same color that is the unique color not contained in @xmath108 . this however implies that the sets @xmath101 and @xmath102 must be the same : they do not contain the colors of the two edges between @xmath73 and @xmath74 and the common color of the edges @xmath98 and @xmath81 . since the sets @xmath102 and @xmath103 have a common color , we conclude that there exists a color @xmath61 contained in all the sets @xmath101 , @xmath102 and @xmath103 . removing the edges colored with @xmath61 from the three trigons and coloring the edges of @xmath0 missing in @xmath99 with @xmath61 yields a @xmath12-edge - coloring of @xmath0 . in this subsection , we just prove two simple lemmas on @xmath109-faces . [ lm-4-face - trigon ] if a trigon in a minimal counterexample @xmath0 is adjacent to a @xmath109-face @xmath18 , then @xmath18 is adjacent to no other multigon . let @xmath110 be the vertices of @xmath18 in such an order that the trigon adjacent to @xmath18 is between @xmath73 and @xmath74 . apply the @xmath80-swap . since the resulting graph @xmath45 contains a new quadragon ( and @xmath18 is not adjacent to a quadragon in @xmath0 by lemma [ lm - face - quadragon ] ) , @xmath45 has a @xmath12-edge - coloring by the minimality of @xmath0 and lemma [ lm - swap ] . let @xmath23 , @xmath24 , @xmath25 and @xmath26 be the colors of the edges of the quadragon in @xmath45 . if one of these colors appears on the edges between @xmath75 and @xmath79 , we can use this color for the edges @xmath83 and @xmath37 and obtain a proper coloring of @xmath0 . hence , @xmath45 contains a bigon between the vertices @xmath75 and @xmath79 , the colors of its two edges are @xmath22 and @xmath27 , and @xmath45 contains neither an edge @xmath83 nor @xmath37 . in particular , in @xmath0 , @xmath18 is adjacent to a single multigon which is the considered trigon . [ lm-4-face-3-bigon ] no @xmath109-face in a minimal counterexample is adjacent to three or four bigons . let @xmath110 be the vertices of @xmath18 in such an order that the bigons are ( at least ) between the pairs @xmath73 and @xmath74 , @xmath74 and @xmath75 , and @xmath75 and @xmath79 . by the minimality of @xmath0 and lemma [ lm - swap ] , the graph @xmath45 obtained by the @xmath80-swap has a @xmath12-edge - coloring . since @xmath45 contains an edge @xmath37 , the trigons between the pairs @xmath73 and @xmath74 , and @xmath75 and @xmath79 must have two edges with the same color . remove edges of this color from the trigons and insert them as edges between @xmath74 and @xmath75 , and @xmath73 and @xmath79 . this yields a @xmath12-edge - coloring of @xmath0 . we start this subsection with a lemma on @xmath19-faces . [ lm-5-face-5-multi ] in a minimal counterexample , if a @xmath19-face @xmath18 is adjacent to @xmath19 multigons , then all these multigons are bigons and each such bigon @xmath111 is adjacent to a @xmath17-big face . by lemmas [ lm - face - quadragon ] and [ lm - face - trigon ] , @xmath18 is adjacent to five bigons . let @xmath54 be an edge of the bigon @xmath111 and @xmath76 the other face adjacent to it . consider an @xmath54-coloring and @xmath61-mates @xmath62 for different colors @xmath61 . if @xmath54 is assigned five colors , then each mate @xmath62 includes all six colors on the edges of @xmath111 . since a mate @xmath62 can not include additional edges with a color different from @xmath61 , such a mate can not include any other bigons which is impossible . hence , we can assume that @xmath54 is assigned three colors , say @xmath22 , @xmath23 and @xmath24 . by symmetry , we can assign that the other edge of @xmath111 is colored with @xmath25 . the mate @xmath112 includes in addition to the bigon @xmath111 another bigon adjacent to @xmath18 . the two edges of this bigon must be colored with the colors @xmath26 and @xmath27 ( they can not have the same color since they are incident with a vertex that is not an end - vertex of @xmath54 ) . let @xmath113 be this bigon . similarly , the mate @xmath69 must include a bigon with one edge colored @xmath26 and the other @xmath27 . we claim that the bigons @xmath111 and @xmath113 are @xmath18-incident . if they were not , there would a bigon @xmath114 @xmath18-incident with both @xmath111 and @xmath113 . let @xmath32 be the vertex shared by @xmath114 and @xmath111 . since @xmath32 is incident with three edges assigned one of the colors and the remaining five colors are assigned to a single edge each , one of the edges @xmath113 has the color @xmath26 or the color @xmath27 . this is however not possible , since the edges of @xmath114 have the colors @xmath26 and @xmath27 . observe that the face @xmath18 is adjacent to at most two bigons containing an edge colored @xmath26 , one of these bigons being @xmath113 . let @xmath115 be such a bigon different from @xmath113 if it exists . similarly , let @xmath116 be the bigon different from @xmath113 containing an edge colored @xmath27 ( if it exists ) . let @xmath117 and @xmath118 be the colors of the other edges contained in @xmath115 and @xmath116 ( if the bigons do not exist , choose @xmath117 and @xmath118 arbitrarily ) . let @xmath61 be a color different from @xmath117 and @xmath118 . the mate @xmath62 includes the two edges assigned colors @xmath22 , @xmath23 , @xmath24 and @xmath25 which appear on the bigon @xmath111 . since it contains exactly one edge of each color different from @xmath61 , @xmath62 must include a bigon with edges colored @xmath26 and @xmath27 ( which can be @xmath113 or a different bigon of this type ) . consequently , @xmath62 includes an edge colored with @xmath61 on a face @xmath76 and this edge must be contained in a @xmath119-face since @xmath62 includes more edges colored with @xmath61 ( see proposition [ prop - mate ] ) . since there are four choices of @xmath61 different from @xmath117 and @xmath118 , the face @xmath76 is @xmath17-big . in the next two lemmas , we apply the six - vertex swap operation to analyze structure of @xmath12-faces . [ lm-6-face-23 ] in a minimal counterexample @xmath0 , no @xmath12-face @xmath18 is adjacent to one or two bigons and three trigons . by lemma [ lm - multigon ] , the trigons adjacent to @xmath18 can not be @xmath18-incident . hence , we can assume that the trigons adjacent to @xmath18 are between the pairs of the vertices @xmath73 and @xmath74 , @xmath75 and @xmath79 , and @xmath93 and @xmath94 . by symmetry , there is a bigon between the vertices @xmath74 and @xmath75 . consider the graph @xmath45 obtained from @xmath0 by the @xmath95-swap . by the minimality of @xmath0 and lemma [ lm - swap ] , @xmath45 has a @xmath12-edge - coloring . let @xmath120 , @xmath121 and @xmath122 be the sets of colors of the three quadragons . since the vertices @xmath74 and @xmath75 are joined by an edge in @xmath45 , we have @xmath123 . since @xmath124 , there exists a color @xmath125 . removing the edges of the three quadragons colored with @xmath61 and inserting the edges @xmath37 , @xmath38 and @xmath126 colored with @xmath61 yields a @xmath12-edge - coloring of @xmath0 . this contradicts the fact that @xmath0 is a minimal counterexample . [ lm-6-face-32 ] in a minimal counterexample @xmath0 , no @xmath12-face @xmath18 is adjacent to three bigons and two trigons . let @xmath127 be the vertices of the face @xmath18 . by lemma [ lm - face - trigon ] , @xmath18 is incident with an edge not contained in a multigon ; let @xmath97 be such an edge . consider the graph @xmath45 obtained from @xmath0 by the @xmath95-swap . let @xmath128 , @xmath129 , be the set of colors assigned to the edges of the multigon @xmath130 between @xmath131 and @xmath132 . since the order of each multigon between @xmath131 and @xmath132 , @xmath129 , is at least three , the size of each @xmath128 is at least three . assume first that @xmath133 is a trigon , i.e. , @xmath134 . if @xmath135 is a quadragon , then @xmath136 since the multigons @xmath135 and @xmath133 are joined by an edge . if @xmath135 is a trigon , then it also holds @xmath136 since the multigons @xmath135 and @xmath133 are joined by two edges ( there must be a trigon between @xmath74 and @xmath75 in @xmath0 ) . a symmetric argument yields that @xmath137 . consequently , there exists a color @xmath125 . removing the edges with the color @xmath61 from the multigons @xmath135 , @xmath133 and @xmath138 , and coloring the edges of @xmath0 not present in @xmath45 with @xmath61 yields a @xmath12-edge - coloring of @xmath0 . we now consider the case that @xmath133 is a quadragon , i.e. , @xmath139 . since no trigons in @xmath0 can share a vertex , we can assume by symmetry that that @xmath135 is a quadragon and @xmath138 is a trigon , i.e. , @xmath140 and @xmath141 . since the quadragons @xmath135 and @xmath133 are joined by an edge in @xmath45 , it follows @xmath123 . similarly , @xmath133 and @xmath138 are joined by an edge and thus @xmath137 . since the size of @xmath121 is four , it follows that there exists a color @xmath125 . we can now obtain a @xmath12-edge - coloring of @xmath0 as in the case that @xmath133 is a trigon . we now prove a lemma on bigons and trigons adjacent to @xmath20-faces . [ lm-7-face - trigon+ ] in a minimal counterexample , every trigon @xmath21 adjacent to a dangerous @xmath20-face @xmath18 is adjacent to @xmath65-big face ( which is different from @xmath18 ) . let @xmath54 be an edge of the trigon and consider an @xmath54-coloring as described in lemma [ lm - e - trigon ] . let @xmath62 be a @xmath61-mate . the mate @xmath69 can not contain any edge with color @xmath86 except for those contained in the trigon . hence , @xmath69 must include the only edge of @xmath18 not contained in a multigon and this edge must be colored with @xmath27 . consequently , every mate @xmath62 , @xmath86 , includes either this single edge , which has the color @xmath27 , or a multigon which has one edge of color @xmath61 and the other of color @xmath27 . we conclude that every mate @xmath62 contains the edge with the color @xmath27 either incident to @xmath18 or contained in a multigon adjacent to @xmath18 . let @xmath76 be the face adjacent to @xmath21 different from @xmath18 . the mate @xmath62 , @xmath66 , can only contain an edge colored with @xmath61 on @xmath76 and the face @xmath68 containing this edge must contain another edge colored with @xmath61 by proposition [ prop - mate ] . since no vertex is incident with two edges with the color @xmath61 , the face @xmath68 is @xmath17-face . we conclude that @xmath76 is @xmath65-big . in the next lemma , we consider @xmath142-faces . [ lm-8-face ] in a minimal counterexample , no @xmath142-face @xmath18 is adjacent to three bigons and four trigons . let @xmath143 be the vertices of @xmath18 in the cyclic order around @xmath18 . by lemma [ lm - multigon ] and the symmetry , we can assume that the trigons are between the vertices @xmath49 and @xmath51 for @xmath144 and the bigons are between the vertices @xmath49 and @xmath51 for @xmath145 . hence , there is either a single edge or a bigon between the vertices @xmath73 and @xmath146 . consider now the graph @xmath45 obtained from @xmath0 by the @xmath147-swap . we claim that @xmath45 has no odd cut of size less than six . consider an odd cut with sides @xmath8 and @xmath9 that has the smallest size in @xmath45 . by symmetry , we can assume that @xmath148 . let @xmath48 be those vertices @xmath49 of @xmath8 such that @xmath50 or @xmath51 is contained in @xmath8 . observe that @xmath149 . if @xmath52 is empty or @xmath52 is formed by a single sequence of an odd number of vertices , then the size of the cut with sides @xmath8 and @xmath9 is the same in @xmath0 and @xmath45 . if @xmath52 is formed by a single sequence of an even number of vertices , then the size is either increased or decreased by two , and thus the size of the cut is at least six by lemma [ lm - odd - cut ] . otherwise , @xmath52 is formed by at least two sequences . since @xmath149 , @xmath52 is formed by two sequences of consecutive vertices , each of length two . suppose that the size of the cut with the sides @xmath8 and @xmath9 is less than six in @xmath45 . since its size in @xmath0 is at least eight and the size of the cut can not decrease by more than two unless @xmath150 or @xmath151 , @xmath52 must be equal to one of these two sets . by symmetry , @xmath152 and thus @xmath153 . note that the roles of @xmath8 and @xmath9 are now completely symmetric . since the sizes of the cut with the sides @xmath8 and @xmath9 in @xmath0 and @xmath45 can differ by at most four , it follows that its size in @xmath0 is eight and in @xmath45 is four . since @xmath45 is a plane graph , then @xmath73 and @xmath93 are not joined by a path formed by vertices of @xmath8 only or @xmath75 and @xmath154 are not joined by a path formed by vertices of @xmath9 only . by symmetry , we can assume the former to be the case : let @xmath155 be the component of the side @xmath8 containing the vertices @xmath73 and @xmath75 , and set @xmath156 . since the size of the cut with sides @xmath8 and @xmath9 in @xmath45 is four , the size of the cut with sides @xmath155 and @xmath157 is at most two or the size of the cut with sides @xmath158 and @xmath159 is at most two . this contradicts the choice of @xmath8 and @xmath9 . we have shown that @xmath45 has no odd cuts of size less than six which implies that @xmath45 has a @xmath12-edge - coloring by the minimality of @xmath0 . let @xmath128 be the set of the four colors assigned to the edges between @xmath131 and @xmath132 , @xmath160 , in @xmath45 . since the vertices @xmath132 and @xmath161 for @xmath129 are joined by an edge , it holds that @xmath162 for @xmath129 . hence , @xmath163 . this combines with @xmath164 to the fact that @xmath165 . let @xmath61 be the color contained in all the sets @xmath128 , @xmath160 . we can now obtain a @xmath12-edge - coloring of @xmath0 by removing the edges in the quadragons of @xmath45 colored with @xmath61 and coloring the edges of @xmath0 not contained in @xmath45 with @xmath61 . we finish this section with two lemmas on dangerous trigons which we will apply in the proof of lemma [ lm - final - face ] . [ lm - danger - a ] let @xmath0 be a minimal counterexample and @xmath18 a @xmath65-big face of @xmath0 . if @xmath18 is adjacent to a dangerous multigon @xmath21 , then @xmath18 is adjacent to at least four @xmath17-faces that are not @xmath18-incident with @xmath21 . consider an @xmath54-coloring for an edge @xmath54 contained in @xmath21 of the type described in lemma [ lm - e - trigon ] and let @xmath62 be a @xmath61-mate . if @xmath21 is a quadragon , then @xmath62 contains only edges colored with @xmath61 except for the edges contained in @xmath21 . hence , each @xmath62 , @xmath66 , contains an edge on @xmath18 colored with @xmath61 contained in @xmath17-face . since the edges @xmath18-incident with @xmath21 are colored with @xmath22 , we conclude that @xmath18 is adjacent to at least five @xmath17-faces that are not @xmath18-incident with @xmath21 . in the rest , we assume that @xmath21 is a trigon . let @xmath76 be the dangerous face adjacent to @xmath21 . if @xmath76 is a @xmath20-face , then @xmath21 is @xmath18-incident with two edges not contained in a bigon . the colors of these edges are @xmath22 . the mate @xmath69 must contain an edge of @xmath76 not contained in a multigon and thus the only such edge incident with @xmath76 has the color @xmath27 . every mate @xmath62 , @xmath66 , either contains the edge of @xmath76 not contained in a multigon , which is colored with @xmath27 , or a multigon adjacent to @xmath76 . in either of the two cases , it includes an edge colored with @xmath27 . consequently , it must include an edge of @xmath18 that is colored with @xmath61 ( and thus not @xmath18-incident with @xmath21 ) and that is contained in @xmath17-face . we conclude that @xmath18 is adjacent to at least five @xmath17-faces not @xmath18-incident with @xmath21 . it remains to consider the case that @xmath76 is a @xmath19-face . if @xmath21 is @xmath76-incident with two multigons , the same argument as in the case of a @xmath20-face applies . hence , we can assume that @xmath21 is @xmath76-incident with a single multigon and let @xmath166 be the edge incident with @xmath76 that is @xmath76-incident to @xmath21 and not contained in a multigon . if the color of @xmath166 is @xmath22 , then @xmath76 must contain an edge with the color @xmath27 not contained in a multigon . each mate @xmath62 , @xmath167 , either contains the edge colored with @xmath27 or a multigon adjacent to @xmath18 ( which has one of its edges colored with @xmath27 ) . hence , @xmath62 must contain an edge of @xmath18 colored with @xmath61 that is contained in a @xmath17-face . such a @xmath17-face can not be @xmath18-incident with @xmath21 and the claim follows . assume now that the color of @xmath166 is @xmath27 . hence , the two edges @xmath18-incident with @xmath21 have the color @xmath22 . let @xmath168 be the color of the other edge of @xmath76 not contained in a multigon ( if it exists ) . each mate @xmath62 , @xmath169 , contains an edge colored with @xmath27 contained in a multigon adjacent to @xmath76 and thus it contains an edge colored with @xmath61 incident with @xmath18 that is contained in a @xmath17-face different from @xmath18 . again , @xmath18 is adjacent to at least @xmath17-faces that are not @xmath18-incident with @xmath21 . [ lm - danger - b ] let @xmath0 be a minimal counterexample and @xmath18 a @xmath65-big face of @xmath0 . if @xmath18 is a adjacent to a dangerous multigon @xmath21 that is @xmath18-incident with a bigon , then @xmath18 is adjacent to at least five @xmath17-faces that are not @xmath18-incident with @xmath21 . lemma [ lm - multigon ] and the fact that @xmath0 is @xmath12-regular implies that the face @xmath76 adjacent to @xmath21 that is different from @xmath18 is a dangerous @xmath19-face . let @xmath54 be an edge of @xmath21 and consider an @xmath54-coloring as described in lemma [ lm - e - trigon ] . let @xmath166 be the edge of @xmath76 that is @xmath76-incident with @xmath21 and not contained in the bigon @xmath76-incident with @xmath21 . observe that the color of @xmath166 must be @xmath22 ( the bigon @xmath18-incident with @xmath21 must have edges with the colors @xmath22 and @xmath27 ) . let @xmath62 be a @xmath61-mate for @xmath66 . since the bigon @xmath18-incident with @xmath21 contains an edge colored with @xmath22 and the edge @xmath18-incident with @xmath21 ( which is on the other side of @xmath21 than the bigon ) is colored with @xmath22 , the mate @xmath62 , @xmath66 contains neither this bigon nor this edge . let us focus on the mate @xmath69 . since the trigon @xmath21 contains edges of all the colors different from @xmath27 , the mate @xmath69 can not include any multigon adjacent to @xmath76 . consequently , @xmath76 is adjacent to an edge @xmath170 different from @xmath166 that is not contained in a multigon and the color of this edge is @xmath27 . since @xmath76 is dangerous , we conclude that @xmath76 is adjacent to the trigon @xmath21 , another trigon , a bigon and the edges @xmath166 and @xmath170 . the mate @xmath69 contains the trigon @xmath21 , the edge @xmath170 and some edges colored with @xmath27 . since no two edges colored with @xmath27 share the same vertex , we conclude that @xmath18 is incident with an edge colored with @xmath27 that is not @xmath18-incident with @xmath21 and that is contained in a @xmath17-face different from @xmath18 . the mate @xmath62 , @xmath167 , contains the trigon @xmath21 and the edge @xmath170 ( note that the edges of the bigon @xmath76-incident with @xmath21 are @xmath22 and @xmath27 ) . hence , @xmath62 must contain an edge of @xmath18 colored with @xmath61 that is not @xmath18-incident with @xmath21 and that is contained in a @xmath17-face different from @xmath18 . it follows that @xmath18 is adjacent to at least five @xmath17-faces that are not @xmath18-incident with @xmath21 . we consider a minimal counterexample and assign every @xmath13-face , @xmath171 , @xmath172 units of charge , every bigon @xmath173 unit of charge , every trigon @xmath174 units of charge and every quadragon @xmath175 units of charge . note that the minimal counterexample can contain multigons of order at most four ( see lemma [ lm - multigon ] ) . vertices are assigned no charge . since the minimal counterexample is @xmath12-regular , the euler formula implies that the total sum of charge assigned to faces is negative . next , charge gets redistributed among @xmath89-faces and multigons using the following rules ( also see figure [ fig - rules ] ) . we attempt to name the rules mnemotechnically : the names start with r , followed by a character b , t , m and 3 to denote the type of faces it involves ( bigons , trigons , multigons and @xmath70-faces ) and sometimes by another character to distinguish the rules further ( e.g. , `` b '' stands for big , `` d '' for dangerous , `` t '' for a trigon , etc . ) . rule rmb : : every dangerous multigon @xmath21 adjacent to a @xmath17-big face @xmath18 receives @xmath176 units of charge from @xmath18 . rule rmd : : every dangerous multigon @xmath21 adjacent to a dangerous face @xmath18 receives @xmath177 unit of charge from @xmath18 . rule rt : : every trigon such that neither rule rmb nor rule rmd applies to it receives @xmath178 unit of charge from each adjacent face . rule rb3 : : a bigon adjacent to a @xmath70-face and a @xmath89-big face @xmath18 receives @xmath178 unit of charge from @xmath18 . rule rb5 : : a bigon adjacent to a @xmath17-big face @xmath18 and to a @xmath19-face adjacent to @xmath19-multigons receives @xmath178 unit of charge from @xmath18 . rule rb : : a bigon such that neither rule rb3 nor rule rb5 applies to it receives @xmath177 unit of charge from each adjacent face . rule r3 : : a @xmath70-face @xmath18 that is adjacent to a bigon and two @xmath89-big faces receives @xmath179 unit of charge from each adjacent @xmath89-big face . t : : a @xmath70-face @xmath18 that is adjacent to a trigon and a @xmath65-big face @xmath76 receives @xmath177 unit of charge from @xmath76 . charge of @xmath89-faces and multigons after these rules are applied is referred to as _ final _ charge . in the remainder of this section , we show that final charge of every face and every multigon is non - negative . recall that the initial amount of charge of a multigon of order @xmath180 is @xmath181 . by lemma [ lm - multigon ] , @xmath0 can contain only bigons , trigons and quadragons . every bigon receives either @xmath178 unit of charge by rule rb3 or rb5 from an adjacent face or @xmath177 unit of charge by rule rb from each adjacent face . hence , every bigon receives at least @xmath178 unit of charge in total . let us consider trigons . if a trigon is not dangerous , then rule rt applies twice . if a trigon is dangerous , then one of the faces adjacent to it is @xmath17-big by lemma [ lm - trigon-2big ] ( note that a dangerous face must be @xmath84-big ) and rules rmb and rmd apply ( each once ) . hence , every trigon also receives two units of charge in total . finally , every quadragon receives @xmath176 units of charge by rule rmb from each adjacent face which is @xmath17-big by lemma [ lm - face - quadragon ] . faces of a minimal counterexample send charge to adjacent multigons and @xmath70-faces only . hence , if @xmath18 sends out any charge , it is @xmath84-big . in particular , @xmath18 can send out some charge by rules rt and rb only . consequently , if @xmath18 is adjacent to no multigon , @xmath18 sends out no charge and its final charge is non - negative . if @xmath18 is not adjacent to multigons of order three or more , then it is adjacent to at most two bigons by lemma [ lm-3-face-23-bigon ] . if it is adjacent to two bigons , each of these bigons is adjacent to a @xmath65-big face by lemma [ lm-3-face-23-bigon ] , and thus rule rb3 applies to it . consequently , rule rb does not apply to @xmath18 . if @xmath18 is adjacent to a single bigon and the other face adjacent to the bigon is @xmath89-big , again , rule rb3 applies and @xmath18 sends out no charge . if the other face adjacent to the bigon is @xmath84-big , then @xmath18 is adjacent to two @xmath89-faces by lemma [ lm-3-face-1-bigon ] . in this case , @xmath18 sends a half of unit of charge to the bigon by rule rb and receives twice a quarter of unit of charge by rule r3 . if @xmath18 is adjacent to a trigon , then lemma [ lm-3-face - trigon ] yields that @xmath18 is adjacent to no other multigon and the other two faces adjacent to @xmath18 are @xmath65-big . hence , @xmath18 sends to the trigon @xmath178 unit of charge by rule rt and receives @xmath177 from each of the adjacent @xmath65-big face by rule r3 t . the final charge is again non - negative . in the rest of the proof , we assume that @xmath18 is @xmath84-big which implies that only rules rt and rb can apply . if @xmath18 is adjacent to a trigon , then @xmath18 is adjacent to no other multigon by lemma [ lm-4-face - trigon ] . hence , rule rt applies once and no other rule can apply to @xmath18 . consequently , @xmath18 sends out one unit of charge and its final charge is zero . if @xmath18 is adjacent to no multigons of order three or more , then , by lemma [ lm-4-face-3-bigon ] , @xmath18 is adjacent to at most two bigons . hence , rule rb can apply at most twice to @xmath18 and thus the final amount of charge of @xmath18 is non - negative . if @xmath18 is @xmath70-big , then either at most two rules apply to @xmath18 each once or the same rule applies to @xmath18 twice . since @xmath18 can not send out charge by rules rmb , @xmath18 sends out at most two units of charge in total and its final charge is non - negative . if @xmath18 is @xmath84-big , then only rules rmd , rt and rb can apply to @xmath18 . since no two trigons are incident by lemma [ lm - multigon ] , @xmath18 is adjacent to at most two trigons . if @xmath18 is adjacent to two trigons and no other multigons , rule rt applies twice and the final charge of @xmath18 is zero . if @xmath18 is adjacent to two trigons and one bigon , then at least one of the trigons is @xmath18-incident with the bigon . observe that @xmath18 is dangerous in this case . hence , rule rmd applies once ( with respect to the trigon @xmath18-incident with the bigon ) , rule rmd or rt applies once and rule rb applies . we conclude that the final amount of charge of @xmath18 is non - negative . if @xmath18 is adjacent to two trigons and two bigons , then both trigons are @xmath18-adjacent to a bigon . again , @xmath18 is dangerous . hence , rule rmd applies twice and rule rb also applies twice . consequently , final charge of @xmath18 is zero . if @xmath18 is adjacent to a single trigon and at most two bigons , it sends out at most two units of charge ( rule rt applies once , rule rb at most twice ) . if @xmath18 is adjacent to a single trigon and three bigons , then rule rmd applies once ( note that @xmath18 is dangerous ) and rule rb three times . again , @xmath18 sends out at most two units of charge . finally , @xmath18 can not be adjacent to a single trigon and four bigons by lemma [ lm-5-face-5-multi ] . in all these case , @xmath18 sends out at most two units of charge and its final amount of charge is non - negative . if @xmath18 is adjacent to no trigon and at most four bigons , @xmath18 sends out at most two units of charge in total by rule rb . on the other hand , if @xmath18 is adjacent to five bigons , lemma [ lm-5-face-5-multi ] yields that rule rb5 will apply to each of the adjacent bigons and @xmath18 sends out no charge . again , the final amount of charge of @xmath18 is non - negative . before proving the final lemma ( lemma [ lm - final - face ] ) of this section which deals with @xmath71-faces , we have to state two auxiliary lemmas that will be useful in analyzing final charge of @xmath109-big and @xmath19-big faces . the next two lemmas use the same notation which we will later use in the proof of lemma [ lm - final - face ] . [ lm - final - next - a ] let @xmath0 be a minimal counterexample and @xmath18 an @xmath182-face , @xmath183 . let @xmath184 be the vertices incident with the face @xmath18 in the cyclic order around @xmath18 , and let @xmath185 be the face or the multigon adjacent to @xmath18 through the edge @xmath186 ( indices taken modulo @xmath182 ) . finally , let @xmath187 be the amount of charge sent by @xmath18 to @xmath188 , @xmath189 . it holds that @xmath190 for every @xmath189 ( indices again taken modulo @xmath182 ) . let @xmath191 be the vertex shared by @xmath188 and @xmath192 . assume that @xmath193 . since the charge sent out by @xmath18 following the discharging rules can only be equal to @xmath194 , @xmath179 , @xmath177 , @xmath178 and @xmath176 , it follows that @xmath187 or @xmath195 is equal to @xmath176 . by symmetry , assume that @xmath196 . consequently , rule rmb applies to @xmath188 and @xmath188 is either a trigon or a quadragon . by lemma [ lm - multigon ] , @xmath192 can not be a trigon or a quadragon . consequently , @xmath197 and one of rules rb3 or rb5 must apply . since @xmath0 is @xmath12-regular , rule rb5 can not apply . consequently , @xmath192 is a bigon adjacent to a @xmath70-face and @xmath188 is a trigon . the trigon @xmath188 can not be adjacent to a dangerous @xmath19-face @xmath76 and @xmath76-incident with a bigon by lemma [ lm-35-face - bigon - trigon ] . on the other , the trigon @xmath188 can not be adjacent to a dangerous @xmath20-face @xmath76 and @xmath76-incident with two bigons since @xmath0 is @xmath12-regular . hence , rule rmb can not apply . the inequality ( [ eq - final - next - a ] ) now follows . [ lm - final - next - b ] let @xmath0 be a minimal counterexample and @xmath18 a @xmath89-big @xmath182-face , @xmath183 . let @xmath184 be the vertices incident with the face @xmath18 in the cyclic order around @xmath18 , and let @xmath185 be the face or the multigon adjacent to @xmath18 through the edge @xmath186 ( indices taken modulo @xmath182 ) . finally , let @xmath187 be the amount of charge sent by @xmath18 to @xmath188 , @xmath189 . it holds that @xmath198 for every @xmath189 ( indices again taken modulo @xmath182 ) and the equality can hold if and only if @xmath199 , @xmath200 and @xmath192 is a bigon . the values of @xmath187 , @xmath195 and @xmath201 are among @xmath194 , @xmath179 , @xmath177 , @xmath178 and @xmath176 . hence , @xmath202 . on the other hand , lemma [ lm - final - next - a ] yields that @xmath203 . we conclude that ( [ eq - final - next - b ] ) holds and the equality is attained only if @xmath199 and @xmath200 . assume that @xmath199 and @xmath200 . observe that @xmath188 and @xmath204 are multigons of order three or four . let @xmath191 and @xmath205 be the vertices shared by @xmath188 and @xmath192 and @xmath192 and @xmath204 , respectively . if @xmath192 is not a bigon and @xmath200 , then rule r3 t must apply . hence , @xmath192 is a @xmath70-face adjacent to a trigon . since both @xmath188 and @xmath204 are multigons of order three or more , this is impossible by lemma [ lm - multigon ] . the lemma now follows . let @xmath182 be the size of @xmath18 . assume first that @xmath18 is @xmath84-big . hence , only rules rmd , rt and rb can apply to @xmath18 . if @xmath18 is adjacent to no trigon , then @xmath18 sends out at most @xmath206 units of charge . since the initial amount of charge of @xmath18 is @xmath207 ( recall @xmath183 ) , the final charge of @xmath18 is non - negative . if @xmath18 is adjacent to a trigon , @xmath18 is adjacent to at most @xmath208 multigons by lemma [ lm - face - trigon ] . moreover , lemma [ lm - multigon ] implies that no two trigons are incident . hence , rule rt applies at most @xmath209 times and rules rmd , rt and rb together apply at most @xmath208 times . consequently , @xmath18 sends out at most @xmath210 units of charge . the value of ( [ eq - final - face ] ) is at most @xmath211 unless @xmath212 . by considering the number of bigons and trigons adjacent to @xmath18 , we derive that @xmath18 sends out at most the amount of its initial charge unless one of the following cases applies ( recall that one of the edges incident to @xmath18 is not in a multigon and no two trigons can share a vertex ) : 1 . * the face @xmath18 is a @xmath12-face and @xmath18 is adjacent to two trigons and three bigons . * this case is excluded by lemma [ lm-6-face-32 ] . the face @xmath18 is a @xmath12-face and @xmath18 is adjacent to three trigons and at least one bigon . * this case is excluded by lemma [ lm-6-face-23 ] . the face @xmath18 is a @xmath20-face and @xmath18 is adjacent to three trigons and three bigons . * in this case , @xmath18 is dangerous and at least one of the trigons is @xmath18-incident with two bigons . hence , rule rb applies three times and either rule rmd applies twice and rule rt once or rule rmd applies once and rule rt twice . in both cases , the face @xmath18 sends out at most four units of charge and its final charge is non - negative . the face @xmath18 is a @xmath142-face and @xmath18 is adjacent to four trigons and three bigons . * this case is excluded by lemma [ lm-8-face ] . adopt now the notation from the statements of lemmas [ lm - final - next - a ] and [ lm - final - next - b ] . if @xmath18 is @xmath70-big , then rule rmb never applies . hence , it holds @xmath213 for every @xmath189 . since @xmath18 is @xmath70-big , @xmath18 sends no charge to at least three of @xmath185 , and thus @xmath18 sends out at most @xmath211 units of charge . let us assume that @xmath18 is @xmath2-big . further , let @xmath214 be the indices @xmath215 such that @xmath188 is a @xmath17-face and set @xmath216 ( indices modulo @xmath182 and @xmath2 where appropriate ) . if @xmath217 , then @xmath218 . we now prove the following claim : if @xmath222 , the estimate ( [ eq - final - face - toobig ] ) follows from the fact that no face or multigon receives more than @xmath176 units of charge from @xmath18 . if @xmath223 , the estimate ( [ eq - final - face - toobig ] ) directly follows from lemmas [ lm - final - next - a ] and [ lm - final - next - b ] . lemmas [ lm - final - next - a ] and [ lm - final - next - b ] also yield that the equality in ( [ eq - final - face - toobig ] ) holds only if @xmath224 is odd , @xmath225 and @xmath226 . we now establish the inequality ( [ eq - final - face - danger ] ) . if @xmath222 , then rule rmb applies to @xmath227 and the face @xmath18 is adjacent to at least four @xmath17-faces distinct from @xmath228 and @xmath229 by lemma [ lm - danger - a ] . this implies that @xmath18 is @xmath71-big , i.e. , @xmath230 which contradicts the assumptions . assume that @xmath231 . since @xmath232 , @xmath233 is a bigon by lemma [ lm - final - next - b ] . lemma [ lm - danger - b ] applied to @xmath227 and the bigon @xmath233 now implies that the face @xmath18 is adjacent to at least five @xmath17-faces besides @xmath228 which again implies that @xmath230 . the proof of the claim is now finished . by claim [ claim ] , if @xmath18 is @xmath109-big or @xmath19-big , then the sum @xmath234 is at most @xmath235 . in particular , the charge sent out by the face @xmath18 is at most @xmath236 and the final charge of @xmath18 is non - negative . if @xmath18 is @xmath71-big , then claim [ claim ] yields that the sum @xmath234 is at most @xmath237 . again , the charge sent out by @xmath18 is at most @xmath211 and its final charge is non - negative . in order to prove theorem [ thm - edge - coloring ] which implies theorem [ thm - main ] , we have to exclude the existence of a minimal counterexample . assume that @xmath0 is a minimal counterexample and assign charge to the multigons and @xmath89-faces of @xmath0 as described in subsection [ sub - rules ] and apply the rules as described . by lemmas [ lm - final - multigon][lm - final - face ] , the final amount of charge of every multigon and every face of @xmath0 is non - negative . since charge is preserved during the application of the rules and the sum of the amounts of initial charge is negative , a minimal counterexample can not exist . this establishes theorem [ thm - edge - coloring ] . 9 g. cornujols : packing and covering , siam press , 2001 . m. conforti , e. l. johnson : two min - max theorems for graphs noncontractible to a four wheel , technical repot , ibm thomas j. watson research center , yorktown heights , new york , 1987 . t. r. jensen , b. toft : graph coloring problems , wiley interscience publications , 1995 . b. guenin : packing @xmath1-joins and edge colouring in planar graphs , manuscript . n. robertson , p. d. seymour and r. thomas : tutte s edge colouring conjecture , j. combin . theory ser . b , 70 ( 1997 ) , 166183 . a. schrijver : combinatorial optimization : polyhedra and efficiency , ser . algorithm and combinatorics , vol . 24 , springer verlag , 2003 . d. seymour : on multi - colourings of cubic graphs , and conjectures of fulkerson and tutte , proc . london math . soc . 3 ( 1979 ) , 423450 .

let @xmath0 be a plane graph and @xmath1 an even subset of its vertices . it has been conjectured that if all @xmath1-cuts of @xmath0 have the same parity and the size of every @xmath1-cut is at least @xmath2 , then @xmath0 contains @xmath2 edge - disjoint @xmath1-joins . the case @xmath3 is equivalent to the four color theorem , and the cases @xmath4 , which was conjectured by seymour , and @xmath5 were proved by guenin . we settle the next open case @xmath6 .

a global economic crisis , such as the recent 2008 - 2009 crisis , is certainly due to a large number of factors . in today s global economy , with strong economic relations between countries , it is important to investigate how a crisis propagates from the country of origin to other countries in the world . indeed , several significant crises in the past few decades have been originated in a single country . however , it is still not clear how and to what extent domestic economies of other countries may be affected by this spreading , due to the inter - dependence of economies @xcite . here , we use a statistical physics approach to deal with the modern economy , as it has been done successfully in the recent years for the case of financial markets and currencies @xcite . more precisely , we view the global economy by means of a complex network @xcite , where the nodes of the network correspond to the countries and the links to their economic relations . for generating the economic network we use two databases , in order to avoid any bias due to the network selection . a global corporate ownership network ( con ) is extracted from a database of the 4000 world corporations with the highest turnover , obtained from the _ bureau van dijk _ . this database includes all the corporate ownership relations to their 616000 direct or indirect subsidiaries for the year 2007 . the trade conducted by these companies , in terms of import / export , is a large fraction of the total world trade . furthermore , the network of subsidiaries is a direct measure of the investment of large corporations in order to grow . foreign investment is a key factor for the development of global and local economies while , as recent economic crises suggest , the role of large corporations to the spreading of crisis in the global economy is yet not clearly understood . the second network , the international trade network ( itn ) , is extracted by the 2007 version of the chelem database obtained by _ bureau van dijk _ @xmath0 , which contains detailed information about international trade , and gdp values for 82 countries in million us dollars . this database provides us with an economic network based on import / export relations between countries . for both networks we are able to locate a nucleus of countries that are the most likely to start a global crisis , and to sort the remaining countries crisis spreading potential according to their `` centrality '' . initially , a crisis is triggered in a country and propagates from this country to others . the propagation probability depends on the strength of the economic ties between the countries involved and on the strength of the economy of the target country . our results show that , besides the large economies , even smaller countries have the potential to start a significant crisis outbreak . the con is a network that connects 206 countries around the globe , using as links the ownership relations within large companies . if companies listed in country a have subsidiary corporations in country b , there is a link connecting these two countries directed from country a to country b. the weight of the link , @xmath1 , equals the number of the subsidiary corporations in country b controlled by companies of country a. next , if companies from country b have subsidiary corporations in country c , then again there is a weighted link , @xmath2 , connecting these two countries directed from b to c , and so on . this way we obtain a network with total 2886 links among 206 nodes ( countries ) . of these links 685 are bi - directional , meaning that if there is a link from node @xmath3 to @xmath4 , as well as a link from node @xmath4 to @xmath3 , and the rest 1516 are one directional only . we assume that the total link weight between a pair of nodes ( countries ) @xmath5 is the sum of all links independently of their direction , @xmath6 . the total link weight represents the strength of the economic ties between two countries in the network . we quantify the total economic strength of a country @xmath3 by its total node weight , @xmath7 , i.e. , summing the weights of all links of node @xmath3 . the probability density distributions of the total node weights and of the total link weights is skewed and heavy tailed , as shown in fig . s1 in the supplementary information . we find an almost linear relation between @xmath8 and the gdp of country @xmath3 , ( as shown in supplementary fig . s2 ) which indicates that the total weight of a country in our network is strongly correlated to a traditional economic measure . the itn is calculated from the second database after we aggregate the trade actions between all pairs of countries . using the trading relations between each pair of countries e.g. , a and b , we can create a bi - directional network where @xmath9 represents the export of a to b , and @xmath10 represents the export of b to a. of course @xmath9 is equal to @xmath11 , which stands for the imports of b from a. in accordance to the above notations , the total link weight is given by @xmath12 , but the total node weight @xmath8 which quantifies the economic strength of a node equals to its gdp value . to identify the uneven roles of different countries in the global economic network , we use the @xmath13-shell decomposition and assign a shell index , @xmath14 , to each node . the @xmath13-shell is a method identifying how central is a node in the network , the higher its @xmath14 the more central role the node is considered to have as a spreader @xcite . the nodes in the highest shell , called the nucleus of the network , represent the most central countries . to determine the @xmath13-shell structure we start by removing all nodes having degree @xmath15 , and we repeat this procedure until we are left only with nodes having @xmath16 . these nodes constitute shell @xmath17 . in a similar way , we remove all nodes with @xmath18 until we are left only with nodes having degree @xmath19 . these nodes constitute @xmath20 . we apply this procedure until all nodes of the network have been assigned to one of the @xmath13-shells . with this approach we view the network as a layered structure . the outer layers ( small @xmath14 ) include the loosely connected nodes at the boundary of the network , while in the deeper layers we are able to locate nodes that are more central . an illustration of this structure is shown in fig . [ fig1 ] . to identify the nucleus of con we consider in the @xmath13-shell the network having only links with weights above a cut - off threshold @xmath21 . by using different threshold values we locate different nuclei of different sizes , as shown in fig . [ fig2](a ) . however , for the whole range of the threshold values used , namely for @xmath22 $ ] , we are able to identify twelve countries that are _ always _ present in the nucleus with @xmath23 . furthermore , for @xmath24 the nucleus is fully connected and always include only the same twelve countries . these countries are the usa ( us ) , united kingdom ( gb ) , france ( fr ) , germany ( de ) , netherlands ( nl ) , japan ( jp ) , sweden ( se ) , italy ( it ) , switzerland ( ch ) , spain ( es ) , belgium ( be ) , and luxembourg ( lu ) , sorted according to their total node weight @xmath25 . when performing the @xmath13-shell decomposition method on the itn we locate a nucleus at @xmath26 composed of 11 countries , that is stable and always the same for @xmath27 m as shown in fig . [ fig2](b ) . actually we may view it as a 12 country core , because the chelem database regards belgium and luxembourg as one trade zone , and therefore the core consists of the following twelve countries : china ( cn ) , russia ( ru ) , japan ( jp ) , spain ( es ) , united kingdom ( gb ) , netherlands ( nl ) , italy ( it ) , germany ( de ) , belgium ( be ) , luxembourg ( lu ) , usa ( us ) , and france ( fr ) . this nucleus is almost the same as the nucleus found for con . there are only two differences , which can be clearly understood due to the complimentary nature of the networks . the first difference is that sweden ( se ) and switzerland ( ch ) are now located at shell 9 in itn , which is one shell below of the core . the second difference is that china ( cn ) and russia ( ru ) are now part of the nucleus , while in the con they were located one shell and five shells away from the core , respectively . the presence of these countries in the nucleus of itn , and their absence in the nucleus of con can be explained considering the unusual structure of their economies . there are not many large corporations in the global corporate database having headquarters in china or in russia , because most of the goods exported by these countries ( especially by china ) are branded under western brand names . therefore most of their trade comes from subsidiaries of western corporations . on the other hand their total trade volume is significantly high in the total global trade . compared to the other shells , the countries in the nucleus are not only strongly interconnected among themselves , but also have many links to other nodes of the network . this is clearly demonstrated in fig . [ fig2](c ) for con and in fig . [ fig2](d ) for the itn . more specifically , for con the 12 countries of the nucleus have significantly large average degree @xmath28 compared to the countries in the shells below , like @xmath29 for shell @xmath26 . for itn the average degree of the nucleus is @xmath30 , while the average degree of shell @xmath31 is @xmath32 . surprisingly , not all 12 countries have the largest total weights or the largest gdp . nevertheless , our results suggest that they do play an important role in the global economic network . for con i.e. we find that six of the g8 members are part of the nucleus ( except russia and canada that belong to lower shells ) , while the other six are smaller countries in absolute size ( relative to the large countries ) . this is explained by the fact that these smaller countries do not support only their local economy , but they are a haven for foreign investments , as they attract funds from large countries for taxation purposes , safekeeping , etc . and a problem in such investments can easily lead to a chain reaction in other countries . countries such as lu and ch , which are headquarters for some of the world s largest companies and subsidiaries , interact very strongly with a very large number of countries . for example , about 95% of all pharmaceutical products of the swiss industry are not intended for local consumption , but for exporting . although , both con and itn are based on different databases , it is interesting that most of the countries are located in very close @xmath13-shells , supporting the robustness of our approach . this is shown in the scatter plot of fig . [ fig2](e ) which maps the change in the @xmath13-shell ranking for both networks . the ranking is done in units of distance from the nucleus . we find that most of the countries are located in almost the same distance from the nucleus for both networks , since 82% of the countries are inside the two lines ( fig . 2(e ) ) , representing distance @xmath33 . the few countries that are in very different @xmath13-shells are usually large countries ( in terms of population ) , which are very active in terms of trade but do not own many large corporations with global status , i.e , turkey ( tr ) and india ( in ) . it should be noted that when we apply the clique percolation method ( cpm ) @xcite , both con for @xmath24 and itn for @xmath27 m , we obtain that the strongest connected communities are the same as the nuclei found using the @xmath13-shell method , as shown in supplementary fig . next we study how an economic crisis can propagate on these networks . an economic crisis is a very complex phenomenon that can not be reduced to an `` all or nothing '' situation . in order to get some insight on the mechanisms of crisis spreading and on the role of the network topology in this spreading , we applied a susceptible - infected - recovered ( sir)-type model . the sir model has been used successfully to model spreading of epidemics in various networks @xcite . the basic characteristic of sir is that it usually assumes a fixed probability for neighbour - to - neighbour infection . in our case we assume a probability which depends on the economic weights of the links and the strength of the targeted country ( see eq . [ eq1 ] ) . initially all nodes are in the susceptible ( s ) state . we chose one node ( country ) and set it to be in a crisis ( infected ) state ( i ) . during the first time step it will infect all its neighbouring nodes with probability calculated from eq . [ eq1 ] , and the status of the infected nodes switches from s to i. this process is repeated with all infected nodes trying to infect during each time step their susceptible ( s ) neighbours . after each time step , the status of the original infected nodes changes to recovered ( r ) and can no longer infect or become infected . in our economic crisis epidemics the recovery means that a set of successful measures has been applied and the country overcomes the crisis . the simulation stops when there are no more infected nodes , or when all the nodes have been infected . we assume for both , con and itn , that the probability @xmath34 that node @xmath3 infects its neighbouring node @xmath4 depends on the total weight of the link , the total weight ( strength ) of the targeted node , and on the magnitude @xmath35 of the crisis , as follows @xmath36 the ratio @xmath37 represents the relative economic dependence of country @xmath4 on country @xmath3 , which we consider as a factor in the probability that country @xmath4 will be infected by country @xmath3 . the factor @xmath35 represents the strength of the crisis and can obtain any positive value . in economic relations directionality plays an important role . for con i.e. , a crisis in a subsidiary could create some problems in the mother company , but if there is a crisis in the mother company its effect on the subsidiaries is always significantly more severe . to account for directionality , we study the case when we keep only one link , directed or undirected , for each pair of connected nodes . to this end we calculate for all pairs of nodes the quantity @xmath38 if @xmath39 is smaller than a certain threshold value , which is a parameter , then the link will be regarded as undirected . otherwise , the link is directed pointing from @xmath3 to @xmath4 if @xmath40 , and from @xmath4 to @xmath3 if @xmath41 . in all cases the weight of the link is @xmath6 . in the current study we set the threshold value of @xmath39 to be zero . by increasing this value , we increase the percentage of undirected links in the resulting network , and if we set @xmath42 then the entire network is undirected . our results are not sensitive to the value of t. we find that for all values of @xmath43 the crisis epidemics is nearly the same as shown in supplementary fig . s4 . in fig . [ fig3 ] we present the results of the sir simulations . [ fig3](a ) and [ fig3](b ) show that a crisis propagates to larger parts of the world when it has larger magnitude . note also that the directed network significantly delays the actual propagation of the crisis , in comparison to the random case ( see legend ) and to the undirected case . this means that if the directionality were not present , the crisis spreading would have been more severe . figs . [ fig3](c ) and [ fig3](d ) show how a crisis spreading process in con depends on different origin countries according to their shell . we find that the countries in the nucleus can spread a crisis to larger parts of the world compared to countries in the outer shells , even if the crisis originates in a small country , such as be . applying the sir model to the itn yields the results of figs . [ fig3](e ) and [ fig3](f ) . it is clear that from both figures we draw the same conclusion , i.e. the inner the shell , the more severe a crisis outbreak originating to this shell will be . additionally , we show that in the itn a crisis could have more severe effects , since the outer shells are capable of larger global impact in comparison to con . it is interesting to note that the error bars in figs . [ fig3](c ) - [ fig3](f ) are small , which means that all the countries in the shell have similar behavior , and the shell determines the spreading instead of the country size . this indicates the predictability power of our method . in fig . [ fig4](a ) we show the world countries colored according to the shell they belong in con , and in fig . [ fig4](b ) we show the countries affected by the economic crisis that started in 2008 in usa , which has spread globally . in fig . [ fig4](c ) we show a prediction of the model for con . we simulate an infection of a crisis starting in the us , using @xmath44 , which leads to roughly the same percentage of infected countries as the actual crisis of 2008 - 2009 ( @xmath45 of the world countries reported economic slowdown due to the crisis ) . the percentage of correctly predicted countries infected by the crisis using the model is above the average percentage of the prediction using random selection . a quantitative agreement between the prediction power of our model and the actual infection strength is shown in fig . [ fig4](d ) . considering the example of be ( ranked 29th according to its total gdp ) , we find that a crisis originating in this country with magnitude @xmath44 is able to affect for con almost 60% of the world countries ( average result of 50 realizations ) , while the worst case scenario that is given by the maximum value of the fraction of infected countries ( out of the same 50 realizations ) is @xmath46 of global infection . for a crisis starting in the nucleus of both networks we find that this maximum fraction has a sharp transition ( see suplementary figs . s5 and s8 ) , verifying that even crises with small magnitude can propagate to the rest of the world , if they originate in the inner shell . this shows the crucial role that these 12 countries play in the world economy . but surprisingly even countries of intermediate shells may play a major role in crisis spreading , as well . such countries , as we show in figs . [ fig3](c ) and [ fig3](d ) , are able to spread a crisis with sufficient magnitude to large fractions of the world as well . this is similar to what happened in the case of the asian currency crisis , starting in thailand in 1997 . thailand belongs to the @xmath47 shell for con and to the @xmath31 shell for itn , but a crisis originating there was able to affect countries of higher shells , and eventually triggered a new crisis that started in 1998 in russia @xcite . we model the way an economic crisis could spread globally , depending to the country of origin . we first create two global networks , which describe strong interaction patterns of the world economy . the first network ( con ) is based on the world s largest companies and all their subsidiaries and links together 206 countries , and the second network ( itn ) is created using aggregated trade data linking together 82 countries . a crisis is triggered with a controlled magnitude and propagates from one country to another with a probability that depends on the strength of the economic ties between the countries involved and on the strength of the economy of the target country . furthermore , using the @xmath13-shell decomposition method we are able to identify the role of the different countries in a world crisis , according to the shell they belong , and we identify the 12 most effective countries for crisis spreading . we find that although a recent global economic crisis originated in the usa , this might not be always the case and even smaller countries have the potential to start similar crisis outbreaks . in retrospect , we know that this has happened several times in the past ( e.g. the crisis in indonesia ) , and is happening these days as well , with a crisis in greece threatening to cause an avalanche effect to other european economies . in part iii of the supplementary information we examine in more detail the case study of greece , where we show that the fears of contagion of other major european economies by the greek crisis are justified . we report almost 40% probability of infection for the major eu countries only through the contagion channels modeled using con and itn i.e. not taking into acount the loans greece received from other countries . this is in contrast to the common belief that smaller countries partake only in crises that are locally contained , but do not spread out to affect the larger countries . thus , our findings show the predicting power of the network approach using the @xmath13-shell ranking methodology . this analysis is important when establishing international commerce rules , policies and legislations , trade treaties and alliances , as they can have a serious impact on monetary policies and international affairs . krugman pr 1992 _ currencies and crises _ ( the mit press cambridge , massachusetts london , england ) . mantegna rn and stanley he 2000 _ an introduction to econophysics : correlations and complexity in finance _ ( cambridge univ press , cambridge , u.k . ) . gabaix x , gopikrishnan p , plerou v and stanley he 2003 _ nature _ * 423 * 267 . mantegna rn 1999 _ eur . j. b _ * 11 * 193 . onnela jp , kaski k and kertsz j 2004 _ eur . j. b _ * 38 * 353 . garas a and argyrakis p 2007 _ physica a _ * 380 * 399 . garas a , argyrakis p and havlin s 2008 _ eur . j. b _ * 63 * 265 . garas a and argyrakis p 2009 _ europhys . lett . _ * 86 * 28005 . coelho r , gilmore cg , lucey b , richmond p and hutzler s 2007 _ physica a _ * 376 * 455 . takayuki m , takayasu h and takayasu m 2006 _ physica a _ * 364 * 336 . albert r and barabsi al 2002 _ rev . phys . _ * 74 * 47 . dorogovtsev sn and mendes jff 2003 _ evolution of networks : from biological networks to the internet and www _ ( oxford university press , new york ) . hidalgo ca , klinger b , barabsi al and hausmann r 2007 _ science _ * 317 * 482 . pittel b , spencer j and wormald n 1996 _ j combin theory _ * 67 * 111 . carmi s , havlin s , kirkpatrick s , shavitt y and shir e 2007 _ proc natl acad sci usa _ * 104 * 11150 . seidman sb ( 1983 ) _ social networks _ * 5 * 269 . palla g , dernyi i , farkas i and vicsek t 2005 _ nature _ * 435 * 814 . newman mej 2002 _ phys rev e _ * 66 * 016128 . colizza v , barrat a , barthlemy m and vespignani a 2006 _ proc natl acad sci usa _ * 103 * 2015 . desai p 2000 _ the american economic review _ * 90 * 48 . -shell decomposition of the network . * ( a ) the size of the nucleus for different cut - off values , using con . ( b ) the size of the nucleus for different cut - off values , using the itn . ( c ) the average degree of the countries of shell @xmath14 for con.the higher the shell number , the closer we get to the nucleus of the network . this plot corresponds to the structure obtained for the cut - off value @xmath48 . ( d ) the average degree of the countries of shell @xmath14 for the itn . this plot corresponds to the structure obtained for the cut - off value @xmath49 m . ( e ) scatter plot showing the distance from the nucleus of different countries for both con and itn.,width=566 ] of the crisis . dir is the real topology of con , taking into account the directionality , und is a network similar to the real one but undirected , and ran is a simulated network with the same number of nodes , the same number of links per node , but with shuffled weights and directions . ( b ) zoom of the area showing the spreading for smaller crisis magnitudes @xmath35 . ( c ) fraction of nodes infected by a crisis originating from different shells of the network versus its magnitude @xmath35 for con . ( d ) zoom of the area showing the spreading for smaller crisis magnitudes @xmath35 . the dashed line shows the spreading of a crisis originated in belgium , which is one of the smaller countries that belong to the nucleus of the network . note that a crisis originating in be , as @xmath35 gets larger , becomes more severe in comparison to the average case for all the countries in shell 11 . ( e ) fraction of nodes infected by a crisis originating from different shells of the network versus its magnitude @xmath35 for the itn . ( f ) zoom of the area showing the spreading for smaller crisis magnitudes @xmath35 . the dashed line again shows the spreading of a crisis originated in belgium . the results are averages over 50 realizations for each node of the network , and the error bars are showing the standard deviation.,width=566 ] -shell they belong for con . ( b ) the countries affected by the economic crisis which started in the us in 2008 . the effect of the crisis is an economic slowdown , as it is reflected by the annual change of the gdp published by the international monetary fund . ( c ) model results for con , showing the crisis starting in the us and spreading to the rest of the world . we performed 1000 realizations always starting an infection in the us . for each realization , when the simulation ends , every country has a score = 1 if it is infected and 0 if it is not infected . a sum of the scores per country for all realizations leads to a number inside the interval @xmath50 $ ] , where 0 means that in all runs this country was not infected and 1000 means that in all runs it was infected . we set a threshold value at 80% , so if a country has score @xmath51 then it is considered as infected . if it has score @xmath52 then it is considered as not infected , and if it is in the range @xmath53 we say that we can not conclude about its status . we find that the average percentage of infected countries is 90.6% , while the worst case scenario which is given by the maximum number of infected countries in all our runs is 96.6% . for comparison purposes if we start a crisis of the same magnitude in countries of lower shells we find a much lower percentage . for example , if we start the crisis in russia ( ks = 6 ) the average fraction of infected countries is 3.34% , while the worst case scenario which is given by the maximum value of the fraction of infected countries is 18.9% . ( d ) probability of the correct prediction of the model as a function of the 2009 vs. 2008 change in the gdp percentage rate . the curve is smoothed using 6 point moving average . the trend that is shown by the linear fit ( red curve ) shows quantitatively that the more affected a country is by the real crisis of 2008 ( represented by larger change in gdp ) , the higher the probability of our model to yeld a correct prediction.,width=642 ]

* we model the spreading of a crisis by constructing a global economic network and applying the susceptible - infected - recovered ( sir ) epidemic model with a variable probability of infection . the probability of infection depends on the strength of economic relations between the pair of countries , and the strength of the target country . it is expected that a crisis which originates in a large country , such as the usa , has the potential to spread globally , like the recent crisis . surprisingly we show that also countries with much lower gdp , such as belgium , are able to initiate a global crisis . using the _ k_-shell decomposition method to quantify the spreading power ( of a node ) , we obtain a measure of `` centrality '' as a spreader of each country in the economic network . we thus rank the different countries according to the shell they belong to , and find the 12 most central countries . these countries are the most likely to spread a crisis globally . of these 12 only six are large economies , while the other six are medium / small ones , a result that could not have been otherwise anticipated . furthermore , we use our model to predict the crisis spreading potential of countries belonging to different shells according to the crisis magnitude . *

most of the galaxies of the local universe shine by effect of stellar and interstellar gas emissions , predominantly in the optical band . typically , the emitted spectra are characterized by stellar absorption lines and emission lines by hii regions . these sources are called _ normal galaxies_. a small fraction of galaxies , about @xmath0 , do not follow this behavior . in fact , they show strong and broad emission lines , consistent with velocity dispersion of the emitting gas attaining several thousand of kilometers per second . most remarkably , the non - thermal emission coming from a central nucleus , with size @xmath1 pc , dominates over the thermal one coming from stars and interstellar gas and extends well beyond the optical band , from radio to gamma rays ( fig . [ fig:1 ] ) . these sources are called _ active galaxies _ and host _ active galactic nuclei _ ( agns ) at their centers . [ cols="<,<",options="header " , ] + agns are not all the same , on the contrary they can be extremely different in their properties . it resulted convenient to classify agns according to their radio power , in fact they can be separated into two distinct classes : radio quiet and radio loud agns . typically , the luminosity in the ghz band of radio loud agns exceeds the radio quiet ones by about three orders of magnitudes . observationally , seyfert 1 show broad emission lines in the spectrum ( broad line region , blr , with velocity dispersion @xmath2 km s@xmath3 ) , while for seyfert 2 the spectral lines are narrower ( narrow line region , nlr , @xmath4 km s@xmath3 ) . table [ tab:1 ] summarizes the objects belonging to the two classes . this complex zoology " has been interpreted by @xcite within the so called unified model for agns . according to this picture ( fig . [ fig:2 ] ) , radio quiet agns have no jets ( or very weak ones ) while the radio loud ones have jets ; the angle between the line of sight and the plane of the obscuring torus determines the observed properties of agns . we will discuss the main properties of radio loud agns , in particular of the radio galaxies , since the presence of a jet may be crucial for accelerating cosmic ray at the highest energies . in section 2 we review the main properties of radio galaxies and in section 3 we will discuss the particle acceleration from agn jets . the main outcomes are summarized in section 4 . radio galaxies are seen in the radio band emitting power - law spectra , indicative of synchrotron emission with typical spectral index @xmath5 . moreover , jets and hot - spots of some bright sources are seen in the optical and x - ray bands as well . the emission mechanism at these higher frequencies is again continuum and it may be synchrotron or synchrotron - self - compton . direct observations of radio galaxies give us : i ) the radio luminosity @xmath6 ; ii ) the size , from a few kiloparsec to some megaparsec ; iii ) the morphological brightness distribution , and iv ) the polarization degree of the radio emission . we can then derive , by indirect means , the main physical parameters such as the life timescale , @xmath7 ys , the mean magnetic field , @xmath8 g , and the kinetic power , @xmath9 . the values of the jet main physical parameters , such as jet velocity , density and composition , are still under debate after many decades of investigations . the reason for these uncertainties in constraining the basic physical parameters is due to the very nature of the radiation emission which is typically non - thermal continuum , i.e. the absence of any lines in the radiation spectrum @xcite . therefore , the value of the magnetic field is derived by the minimum energy assumption , the jet kinetic power by the work done against the ambient to evacuate cavities for accommodating the radio lobes , the jet velocity by the radiative flux contrast of the approaching to receding jets and by proper motion observations , and the jet density by comparison between observed morphologies and the outcome of numerical simulations . historically , the extragalactic radio sources have been classified into two categories @xcite based upon their radio morphology : a first class of objects , preferentially found in rich clusters and hosted by weak - lined galaxies , shows jet - dominated emission and two - sided jets and was named fr i ; a second one , found isolated or in poor groups and hosted by strong emission - line galaxies , presents lobe - dominated emission and one - sided jets and was called fr ii ( or classical doubles " ) . besides morphology , fr i and fr ii radio sources were discriminated in power as well : objects below @xmath10 w hz@xmath3 str@xmath3 were typically referred as fr i sources . a perhaps more illuminating criterion has been found by @xcite who plotted the radio luminosity at @xmath11 ghz against the optical absolute magnitude of the host galaxy : they found the bordering line of fr i to fr ii regions correlating as @xmath12 ( fig . [ fig : ledlow ] ) , i.e. in a luminous galaxy more radio power is required to form a fr ii radio sources . this correlation is important since it can be interpreted as an indication that the environment may play a crucial role in determining the source structure . the above argument yields the basic question of the origin of fr i / fr ii dichotomy , whether intrinsic or ambient driven @xcite . we note as well that there are no sources below a luminosity of @xmath13 w hz@xmath3 ( at @xmath11 ghz ) . [ eq : lnu1]).,scaledwidth=50.0% ] the problem of the sites of cosmic ray acceleration , after many decades of experimental and theoretical investigations , is not yet solved . concerning the cosmic rays in the highest - energy region , above @xmath14 ev , there is a common , even though not unanimous , agreement about an extragalactic origin . several mechanisms and classes of astrophysical objects have been proposed in the literature @xcite . among the proposed candidates for the origin of the highest energy particles the are : i ) neutron stars and other compact object @xcite ; ii ) the sources of gamma - ray bursts @xcite ; iii ) large - scale shocks due to merging galaxies @xcite or in accretion shocks in clusters of galaxies @xcite ; iv ) the active galactic nuclei @xcite ; v ) the hot - spots of fr ii radio galaxies @xcite ; vi ) the agn jets by inductive particle acceleration @xcite . considering the hypothesis that uhecrs originates from agns , and in particular from the hot - spots of fr ii sources via fermi i acceleration , @xcite has shown that 15 fr iis only are located within 130 mpc from us , distance set by the gzk cutoff . moreover , @xcite argues that the fr ii sources ( ngc4261 , pks1343 - 60 ) , closest to the highest energy event detected by the pierre auger experiment , were @xmath15 away from its position . however , more statistics is required before ruling out this possibility . on the other hand , uhecrs can come from agn jets themselves , as pointed out by @xcite , by increasing the energy of protons from @xmath16 ev up to @xmath17 ev by electric fields induced in sheared , relativistic magnetized jets , whenever : @xmath18 where @xmath19 are the electric and magnetic fields and jet velocity , respectively , and @xmath20 the electric potential . the resulting particle spectrum becomes @xmath21 asymptotically . sheared jets , needed for this kind of acceleration , are actually invoked for interpreting the limb - brightening of radio jets at parsec scales . one must recall that , while fr i jets are non - relativistic at kiloparsec scales and fr ii ones are relativistic , vlbi observations of radio galaxies @xcite have shown that , albeit these two kind of sources have different radio power and kiloparsec scale morphology , they appear similar at parsec scales where the jet bulk lorentz factor is typically @xmath22 for both classes . moreover , as said before , about ten fr i sources , e.g. m87 @xcite and b2 1144 + 35 @xcite , observable in the radio band at the vlbi show limb - brightened emission . this limb brightening can be interpreted assuming the presence in the jet of a central spine with high lorentz factor , @xmath23 , and an outer layer with @xmath24 . the synchrotron emission from the central spine would be deboosted when viewed at angles larger than about @xmath25 . the problem of the interaction of relativistic , fluid jets with the ambient was studied numerically , in three spatial dimensions , by @xcite ; they followed the onset and nonlinear growth of unstable modes in the jets while propagating into a uniform medium . the unstable perturbations caused mixing and deceleration of the jet , processes controlled by the initial jet mach number @xmath26 and ambient - to - jet density ratio @xmath27 , with the jet lorentz factor kept fixed to 10 . they found that very light jets were substantially decelerated , and the jet deceleration was not uniform , but more effective at the outer layers leaving a central spine " traveling at higher lorentz factor . thus , fr i jets as well may meet the conditions to be source of uhecrs via inductive particle acceleration . one can then ask the same question of @xcite : how many fr i sources , potential accelerators of cosmic rays , can be found within @xmath28 mpc from earth ? the radio luminosity function ( rlf ) of a sample of radio - loud agns from the @xmath11 ghz nrao vla sky survey was obtained by @xcite . if one extracts from this rlf the number of objects , with luminosity exceeding @xmath13 w hz@xmath3 , expected within @xmath28 mpc from us one finds about @xmath28 sources , that should be identified and their positions possibly correlated to the arrival directions of pierre auger events . independently of the particular acceleration mechanism considered , not all radio - loud agns can effectively accelerate uhecrs . cosmic rays must be contained into the acceleration region in order to attain the maximum energies @xcite , in other words the acceleration timescale must not exceed the dynamical time available for the acceleration . this condition reads , in the frame comoving with the jet : @xmath29 with @xmath30 where @xmath31 is an efficiency parameter that is appropriate for the actual acceleration mechanism , @xmath32 the jet longitudinal coordinate , @xmath33 is the larmor time in the jet magnetic field , @xmath34 the jet lorentz factor and @xmath35 its velocity . applying condition ( [ eq : cond ] ) , one derives the maximum energy achievable , in the observer s frame @xcite : @xmath36 this quantity must be related with the jet magnetic field intensity . to the jet total power contribute both kinetic and magnetic ( poynting ) luminosities : @xmath37 with @xmath38 and @xmath39 where @xmath40 is the jet opening angle and @xmath41 the jet density in the comoving frame . since magnetic fields are likely to play a crucial role in the acceleration process , one may think that the main constituent of jets , close to its origin , is not mass but fields in form of poynting - dominated beams . however , @xcite argued that , even though jets could be poynting - dominated at the origin , observational data imply that they become kinetically dominated beyond about 1,000 gravitational radii from the central acceleration region , with at its center a supermassive black - hole of @xmath42 m@xmath43 . furthermore , @xcite discussed the role of kink instability in poynting - flux dominated jets and find that the poynting flux dissipates and the jet becomes kinetically - dominated , again at about 1,000 gravitational radii . thus it is therefore reasonable to assume that if close to the agn @xmath44 , at the kiloparsec scale @xmath45 . combining eq . [ eq : eobs ] with eq . [ eq : mag ] , one obtains @xcite : @xmath46 as discussed previously , the jet magnetic power is not directly observable , however this form of energy , probably dominant at the beginning , transforms into kinetic one via propagation - related processes . most importantly , jet kinetic power has been connected to the ( observable ) luminosity of the radio core by @xcite and @xcite who derived this quantity from the work done by the radio lobes to evacuate intracluster cavities , seen at the x - ray energy band . they showed a correlation between the kinetic luminosity and the luminosity of the radio core , in the limit of flat power - law spectra of the radio emission ( @xmath47 ) : @xmath48 with @xmath49 w hz@xmath3 and @xmath50 w. this correlation is shown in the log - log plot of fig . [ fig : heinz ] , where the core radio power @xmath51 , with @xmath52 ghz , and the luminosities are in cgs units . we have now all the elements for relating condition ( [ eq : lumb ] ) with quantities _ that can be actually observed_. in fact , recalling the above discussion about the transformation of a poynting - flux dominated jet into a kinetically dominated one , from eq . ( [ eq : lumb ] ) and eq . ( [ eq : kinradio ] ) we derive : @xmath53 solving for @xmath54 , and in case of cosmic rays of @xmath55 ev , we have : @xmath56 this limit can be checked against observation of @xmath54 . however , it can be more instructive to gain a visual impression of this limit , for fr is , by using fig . [ fig : ledlow ] . for doing this we must convert the radio core luminosity into the total luminosity . as shown by @xcite , this two quantities are correlated , once corrected for doppler effects on the radio flux . following @xcite , we set a mean value of @xmath57 for this correction , obtaining : @xmath58 we may set for the jet velocity @xmath59 , @xmath60 for protons , and a typical opening angle of radio jet is @xmath61 . one thus obtains : @xmath62 that should be compared with the data in fig . [ fig : ledlow ] , after having decided which is the favored particle acceleration mechanisms in order to set the parameter @xmath27 . setting @xmath63 , condition ( [ eq : lnu1 ] ) is represented in fig . [ fig : ledlow ] by the thick , horizonthal line . we notice that @xmath27 can not attain values much smaller that unity , otherwise we run out of fr i sources suitable for particle acceleration . we will be then able to look among the about 100 radio - loud agns within @xmath28 mpc the ones that fulfill condition ( [ eq : lnu1 ] ) for accelerating particles up to @xmath55 ev . we have reviewed the main properties of agns , their separation into radio - loud and radio - quiet sources and the implications of this classification on the problem of cosmic ray acceleration from these objects . we have then examined , on quite general bases , the conditions for uhecrs production and set observational limits to the radio luminosity of radio sources in order to be able to be sources of uhecrs . 27 s. w. allen , r. j. h. dunn , a. c. fabian , g. b. taylor and c. s. reynolds , mnras 372 ( 2006 ) , 21 . l. biermann and p. a. strittmatter , ap . j. 322 ( 1987 ) 643 . p. l. biermann , j. phys . g : nucl . part . 23 ( 1997 ) 1 . r. d. blandford , phst 85 ( 2000 ) , 191 . c. cesarsky and v. ptuskin , 23rd ircr ( university of calgary ) 2 ( 1993 ) 341 . b. l. fanaroff and j. m. riley , mnras 167 ( 1974 ) 31 . d. giannios and h. c. spruit , a&a 450 ( 2006 ) , 887 . g. giovannini , w. d. cotton , l. feretti , l. lara and t. venturi , ap . j. 552 ( 2001 ) , 508 . g. giovannini , m. giroletti and g.b . taylor , a&a 474 ( 2007 ) , 409 . s. heinz , a. merloni and j. schwab , ap . j. 658 ( 2007 ) , l9 . a. m. hillas , ann . a&a 22 ( 1984 ) 425 . h. kang , d. ryu and t. w. jones , ap . j. 456 ( 1996 ) 422 . y. y. kovalev , m. l. lister , d. c. homan and k. i. kellermann , ap . j. 668 ( 2007 ) , 27 . m. j. ledlow and f. n. owen , aj 112 ( 1996 ) 9 . m. lemoine , sf2a-2008 : proceedings of the french society of astronomy and astrophysics eds . : c. charbonnel , f. combes and r. samadi ( http://proc.sf2a.asso.fr , p.247 ) . m. lyutikov and r. ouyed , aph 27 ( 2007 ) , 473 . s. massaglia , ap&ss 287 ( 2003 ) 223 . s. massaglia , nuphs 165 ( 2007 ) , 130 . t. mauch and e. m. sadler , mnras 375 ( 2007 ) , 931 . m. milgrom and v. usov , ap . j. lett . 449 ( 1995 ) l37 . m. nagano and a.a . watson , rev . 72 no . 3 ( 2000 ) 689 . m. a. prieto , j. reunanen , th . beckert , k. tristram , n. neumayer , j. a. fernandez and j. acosta , aspc 373 ( 2007 ) , 600 . r. j. protheroe and a. p. szabo , phys . 69 ( 1992 ) 2885 . j. p. rachen and p. l. biermann , a&a 272 ( 1993 ) 161 . p. rossi , a. mignone , g. bodo , s. massaglia , a. ferrari , a&a 488 ( 2008 ) , 795 . m. sikora , m. c. begelman , g. m. madejski and j. -p . lasota , ap . j. 625 ( 2005 ) , 72 . c. m. urry and p. padovani , pasp 107 ( 1995 ) , 803 . m. vietri , ap . j. 453 ( 1995 ) 883 .

we present the general properties of the active galactic nuclei ( agns ) and discuss the origin and structure of jets that are associated to a fraction of these objects . we then we address the problems of particle acceleration at highly relativistic energies and set limits on the luminosity of agn jets for being origin of uhecrs .

almost all of the current semiconductor devices are based on manipulating electronic charges . the arising field of spintronics proposes to use the spin degree of freedom of electrons in place of / in addition to the charge degree of freedom for device applications in order to add new features and functionalities to semiconductors devices.@xcite the hope of the realization of the proposed spintronic devices is supported by the resent development of ultrafast nonlinear optical experiments where long spin dephasing time ( @xmath1 ns ) is reported . the functionalities of the semiconductor spintronic devices rely on the manipulation of the spin coherence . in order to realize these devices , one needs to thoroughly understand the spin dephasing mechanisms which tend to destroy the spin coherence . historically , three spin dephasing mechanisms have been proposed in semiconductors:@xcite the elliot - yafet ( ey ) mechanism,@xcite the dyakonov - perel ( dp ) mechanism,@xcite and the bir - aronov - pikus ( bap ) mechanism.@xcite all of the three mechanisms are either due to the spin flip ( sf ) scattering or are treated as effective sf scattering . the spin dephasing times of these mechanisms for low polarized system are calculated in the framework of single particle approximation.@xcite in additional to these single particle spin dephasing mechanisms , three years ago wu proposed a many - body spin dephasing mechanism which has long been overlooked in the literature . this mechanism is caused by irreversibly disrupting the phases between spin dipoles due to the inhomogeneous broadening together with the spin - conserving ( sc ) scattering , and is therefore a many - body effect . the inhomogeneous broadening can be introduced by the energy dependence of @xmath2-factor@xcite and/or the momentum @xmath3-dependence of the dp term.@xcite our recent works further show that this mechanism also plays an important role in the spin dephasing during the spin transport.@xcite very recently we performed a systematic investigation@xcite of the spin dephasing due to the dp effect in @xmath0-typed gaas ( 100 ) quantum wells for high temperatures ( @xmath4 k ) under magnetic fields in the voigt configuration by constructing and numerically solving the kinetic bloch equations . in these studies , we include all the sc scattering such as the electron - phonon , the electron - nonmagnetic impurity as well as the electron - electron coulomb scattering and investigate the spin dephasing under various conditions . the dephasing obtained from our theory contains both the single - particle dephasing caused by the effective sf scattering first proposed by dyakonov and perel,@xcite and the many - body one due to the inhomogeneous broadening provided by the dp term . we show that for the electron densities we studied , the spin dephasing rate is dominated by the many - body effect . moreover , as we include the electron - electron coulomb scattering , we are able to investigate the spin dephasing with extra large spin polarization ( up to 100 % ) which has not been discussed both theoretically and experimentally . we find that under moderate magnetic fields , the sdt increases dramatically with the initial spin polarization . for example , the sdt of a impurity free sample gets an increase of more than one order of magnitude when the initial spin polarization rises from about 0 to about 100 % at low temperature.@xcite the initial - spin - polarization dependence of the spin dephasing becomes more interesting when the magnetic field is increased to a few tens tesla where the sdt no longer increases monotonically with the initial spin polarization but shows an anomalous resonance peak versus the initial spin polarization.@xcite the dramatic increase and the anomalous resonance of sdt in the high spin polarization region is found to be due to the first order of the electron - electron interaction , _ i.e. _ , the hartree - fock ( hf ) contribution which provides an effective magnetic field that can reduce the spin dephasing and result in a fast increase of the sdt . moreover , under right condition , the hf term , the applied magnetic field as well as the dp term can reach to a resonance , and thus forms the anomalous peak . due to the small land @xmath2-factor in gaas , the resonance condition can only be achieved under very high magnetic fields . in this paper , we apply the kinetic theory to study the spin dephasing in the @xmath0-typed inas qw for high temperatures where the dp term is the leading dephasing mechanism . in qw system , the dp term is composed of the dresselhaus term@xcite and the rashba term.@xcite the dresselhaus term is due to the lack of inversion symmetry in the zinc - blende crystal brillouin zone and is sometimes referred to as bulk inversion asymmetry ( bia ) term . whereas the rashba term appears if the self - consistent potential within a qw is asymmetric along the growth direction and is therefore referred to as structure inversion asymmetry ( sia ) contribution . for qw s of wide band - gap semiconductors such as gaas , the dresselhaus term is the main spin dephasing mechanism . whereas for qw s composed of narrow band - gap semiconductors such as inas in the present case , the rashba term is dominant . as the rashba term is proportional to the interface electric field of the qw , therefore , the spin dephasing in the inas qw can be manipulated through applying an electric field perpendicular to the qw . moreover , as the land @xmath2-factor in inas is very large ( @xmath5 compared to 0.44 of gaas ) , one expects to achieve the resonance condition under a moderate magnetic field . we organize the paper as follows : we present our model and the kinetic equations in sec . then in in sec . iii(a ) we investigate how the sdt changes with the variation of the initial spin polarization . the temperature dependence of the sdt under different spin polarization is discussed in detail in sec . iii(b ) , where we also highlight the difference between the present many - body theory and the earlier simplified theory . in sec . iii(c ) we show the magnetic field dependence of the sdt . finally we discuss how the interface electric field affects the sdt . we present the conclusion and summary in sec . we start our investigation from an @xmath0-doped ( 100 ) inas qw with well width @xmath6 . the growth direction is assumed to be @xmath7-axis . a moderate magnetic field * b * is applied along the @xmath8 axis . due to the confinement of the qw , the momentum states along @xmath7 axis are quantized . therefore the electron states are characterized by a subband index @xmath0 and a two dimensional wave vector @xmath9 together with a spin index @xmath10 . in the present paper , we choose the electron density so that only the lowest subband is populated and the transition to the upper subbands is unimportant . therefore , one only needs to consider the lowest subband . with the dp term ( specifically the rashba term ) included , the hamiltonian of the electrons in the qw takes the form : @xmath11 \cdot{\vec{\bf \sigma}_{\sigma\sigma^{\prime}}\over 2}\biggr\ } c^{\dagger}_{{\bf k}\sigma}c_{{\bf k}\sigma^{\prime}}+h_i . \label{eq : hamiltonian}\ ] ] here @xmath12 is the energy of electron with wavevector @xmath3 and effective mass @xmath13 . @xmath14 are the pauli matrices . the rashba term @xmath15 can be written as @xmath16 in these equations , @xmath17 is proportional to the interface electric field @xmath18 along the growth direction : @xmath19 with the coefficient @xmath20 being inversely proportional to the energy gap and the effective mass.@xcite the interaction hamiltonian @xmath21 is composed of coulomb interaction @xmath22 , electron - phonon interaction @xmath23 , as well as electron - impurity scattering @xmath24 . their expressions can be found in textbooks.@xcite we construct the kinetic bloch equations by the nonequilibrium green function method@xcite as follows : @xmath25 here @xmath26 represents the single particle density matrix . the diagonal elements describe the electron distribution functions @xmath27 . the off - diagonal elements @xmath28 describe the inter - spin - band polarizations ( coherence ) of the spin coherence.@xcite note that @xmath29 . therefore , @xmath30 and @xmath26 are the quantities to be determined from bloch equations . the coherent part of the equation of motion for the electron distribution function and the spin coherence are given by @xmath31\mbox{im}\rho_{{\bf k}}+h_y({\bf k } ) \mbox{re}\rho_{{\bf k}}\bigr\ } + 4\sigma\mbox{im}\sum_{{\bf q}}v_{{\bf q}}\rho^{\ast}_{{\bf k}+{\bf q } } \rho_{{\bf k}},\ ] ] @xmath32 ( f_{{\bf k}{1\over 2}}-f_{{\bf k}-{1\over 2 } } ) + i\sum_{{\bf q}}v_{\bf q}\bigl[(f_{{\bf k}+{\bf q}{1\over 2 } } -f_{{\bf k}+{\bf q}-{1\over 2}})\rho_{{\bf k } } -\rho_{{\bf k}+{\bf q}}(f_{{\bf k}{1\over 2 } } -f_{{\bf k}-{1\over 2}})\bigr],\ ] ] respectively , where @xmath33 $ ] is the 2d coulomb matrix element under static screening . @xmath34 and @xmath35 is the static dielectric constant . the first term on the right hand side ( rhs ) of eq . ( [ eq : f_coh ] ) describes the spin precession of electrons under the magnetic field @xmath36 as well as the effective magnetic field @xmath15 due to the rashba effect . the scattering terms of the electron distribution function and the spin coherence are given by eqs . ( [ eq : f_scatt ] ) and ( [ eq : rho_scatt ] ) in the appendix . the initial conditions are taken at @xmath37 as : @xmath38 @xmath39 + 1\bigr\ } \label{eq : fk_init}\ ] ] where @xmath40 is the chemical potential for spin @xmath10 . the condition @xmath41 gives rise to the imbalance of the electron densities of the two spin bands the kinetic bloch equations form a set of nonlinear equations . all the unknowns to be solved appear in the scattering terms . specifically , the electron distribution function is no longer a fermi distribution because of the existence of the anisotropic rashba term @xmath42 . this term in the coherent part drives the electron distribution away from an isotropic fermi distribution . the scattering term attempts to randomize electrons in @xmath43-space . obviously , both the coherent part and the scattering terms have to be solved self - consistently to obtain the distribution function and the spin coherence . we numerically solve the kinetic bloch equations in such a self - consistent fashion to study the spin precession between the spin - up and -down bands . we include electron - phonon scattering and the electron - electron interaction throughout our computation . as we concentrate on the relatively high temperature regime in the present study , for electron - phonon scattering we only need to include electron - lo phonon scattering . electron - impurity scattering is sometimes excluded . as discussed in the previous paper,@xcite irreversible spin dephasing can be well defined by the slope of the envelope of the incoherently summed spin coherence @xmath44 . the material parameters of inas for our calculation are tabulated in table [ table1].@xcite the method of the numerical calculation has been laid out in detail in the previous paper on the dp mechanism in 3d systems.@xcite the difference is that here we are able to get the results quantitatively in stead of only qualitatively as in the previous 3d case , thanks to the smaller dimension in the momentum space . our main results are plotted in figs . [ fig3 ] to [ fig8 ] . in these calculations the total electron density @xmath45 and the applied magnetic field @xmath46 are chose to be @xmath47 @xmath48 and @xmath49 t respectively unless otherwise specified . .parameters used in the numerical calculations [ cols= " < , < , < , < , < , < , < " , ] we first study the spin polarization dependence of the sdt . as our theory is a many - body theory and we include all the scattering , especially the coulomb scattering , in our calculation , we are able to calculate the sdt with large spin polarization . in fig . [ fig3 ] , sdt @xmath50 is plotted against the initial spin polarization @xmath51 with @xmath52 [ fig . [ fig3](a ) ] and @xmath53 [ fig . [ fig3](b ) ] at different temperatures . the most striking feature of the impurity - free case is the huge anomalous peaks of the sdt in low temperatures . for @xmath54 k , the peak value of the sdt is about 6 times higher than that of low initial spin polarization . it is also seen from the figure that the anomalous peak is reduced with the increase of temperature and the peak shifts to higher polarization . for @xmath55 k there is no anomalous peak . the anomalous peak in the @xmath50-@xmath51 curve in low temperature region originates from the electron - electron interaction , specifically the hartree - fock ( hf ) self - energy [ _ i.e. _ , the last terms in the eq . ( [ eq : f_coh ] ) and ( [ eq : rho_coh ] ) ] . if one removes the hf term , the anomalous peak as well as the large increase of sdt disappears . it is pointed out in our previous paper that although the hf term itself does not contribute to the spin dephasing directly,@xcite it can alert the motion of the electrons as it behaves as an effective magnetic field @xmath56 . therefore , the hf term can affect the spin dephasing by combining with the dp term . for small spin polarization as commonly discussed in the literature , the contribution of the hf term is marginal . however , when the polarization gets higher , the hf contribution becomes larger . especially the effective magnetic field formed by the hf term contains a longitudinal component [ @xmath57 which can effectively reduce the `` detuning '' of the spin - up and -down electrons , and thus strongly reduces the spin dephasing , therefore the sdt increases with initial spin polarization.@xcite moreover , besides the initial polarization , @xmath58 and therefore @xmath59 are also affected by the applied magnetic field . with higher magnetic field , both gets larger . under the high magnetic field and when the initial spin polarization reaches to a right value , the effective magnetic field @xmath60 may reach the magnitude comparable to the contribution from the dp term as well as the applied magnetic field in the coherent parts of the bloch equations and reduces the anisotropic caused by the dp term . therefore , one gets much longer sdt . however , if one further increases the initial polarization , the hf term exceeds the resonance condition . as the result , the sdt decreases . therefore , one gets the anomalous peak which is similar to the resonance effect . it is noted at , as both the dp term and the hf term are @xmath61-dependent , the resonance is broadened . for high temperatures the hf term is smaller . in order to reach the resonance , one needs to go to higher polarization . therefore , as shown in the figure the anomalous peak shifts to the higher polarization . however , when the temperature is high enough , even largest polarization @xmath62 % can not make the hf term to reach the resonance condition . therefore , the peak disappears . the @xmath50-@xmath51 curve is much different when the impurities are introduced . it is seen from fig . [ fig3](b ) that , when the density of impurity is large , say @xmath53 , the fast rise in @xmath50-@xmath51 curve remains . nevertheless the increase is smaller than the corresponding one when the impurities are absent . in addition to the reduction of the rise in @xmath50-@xmath51 curves , the impurities destroy the anomaly too . one can easily see that , with the impurity level @xmath53 , when the temperature is 120 k , the anomalous peak is much flatter than the impurity free sample , while for all other temperatures we study , the sdt increases uniquely with the polarization . to further reveal the contribution of the impurity to the dephasing under different conditions , we plot the sdt as a function of the polarization for different impurity levels at @xmath54 k in fig . [ fig4 ] . the figure clearly shows that the impurity tends to remove the anomalous peak and to shift the peak to the larger initial spin polarization . this is because that the impurity reduces the hf term and therefore the resonance effect is also reduced . hence , in order to reach the maximum resonance , one has to increase the initial spin polarization . consequently , the peak shifts to larger @xmath51 . whereas when @xmath63 is raised to @xmath64 , the hf term is reduced too much to form a peak . above we discussed the dependence of spin dephasing on initial spin polarization for different temperatures . now we turn to the temperature dependence of the sdt under different initial spin polarizations . > from fig . [ fig3](a ) and ( b ) in sec . iii(b ) , one can see that for small polarization , the sdt increases with the temperature . whereas in high polarized region , the sdt decreases with the temperature . for moderate polarization , the temperature dependence is too complicated to be described by a monotonic function of temperature . to see more detail of how the spin dephasing depends on the temperature , we replot the sdt shown in fig . [ fig3 ] as a function of the temperature for different impurity levels and different spin polarizations in fig . [ fig5](a ) and ( b ) . it is seen from the figure that similar to the case of gaas where the dp term is composed of the dresselhaus term,@xcite for rashba term and for low spin polarization the sdt also increases with the temperature for all impurity levels . this property is again _ opposite _ to the results of earlier simplified treatments of the dp effect , where it was predicted that the spin lifetime decreases with the increase of temperature in the 2d system.@xcite the sdt based on the simplified model is given by@xcite @xmath65 in which @xmath66 and @xmath67d\theta\ .\ ] ] @xmath68 stands for scattering cross - section . for comparison , we plot the sdt predicated by the earlier model and by our present many - body theory in the inset of fig . [ fig5](a ) . > from the inset one can see that the sdt predicated by the earlier model is about one order of magnitude larger than the one predicated by our theory . in the mean time , the sdt of the earlier mode drops dramatically with the increase of the temperature . nevertheless , in our many - body treatment , it rises slightly with the temperature . the giant difference between two models lies on the fact that the earlier simplified model is based on the single particle picture which does not count for the dephasing due to the inhomogeneous broadening inherited in the rashba term , which is exactly the result of many body effect . by comparing the theoretical sdt predicated by the two models , we can see that the spin dephasing due to the inhomogeneous broadening is much more important . in the case we calculated , the spin dephasing is dominated by the inhomogeneous broadening . therefore , it is easy to understand why the earlier simplified treatment of the dp mechanism gives much slower spin dephasing . although there is no experiment results for inas , we have shown that for gaas the prediction of our many - body theory agrees both quantitatively and qualitatively with the experiment results.@xcite the temperature dependence of the sdt can be easily understood when the spin dephasing due to inhomogeneous broadening is taken into account : when the temperature increases , the inhomogeneous broadening is reduced as the electrons are distributed to the wider @xmath61-states . as a result , the number of electron occupation on each @xmath3 state is reduced . it is further noted that this reduction is mild as function of the temperature . therefore , the temperature dependence is quit mild unless it is within the regime of anomalous peak . in the region where hf term is important , in addition to the above mentioned two effects of the temperature acting on the spin dephasing , the temperature dependence of hf term should also be taken into account when we study the temperature dependence of the spin dephasing . in high spin polarization region , the sdt decreases with the temperature . however , in the moderately polarized region , the temperature dependence of sdt due to the combination of these three effects is too complicate to be described by a monotonic function . we replot the sdt as a function of the temperature in fig . [ fig5](b ) , for the high polarization @xmath69 % , which is near the anomalous peak shown in fig . [ fig3](a ) . we can see that due to the reduction of the hf term , the resonance is removed and the sdt drops dramatically with the increase of the temperature in the impurities free sample . while for the system with impurity concentration @xmath53 , the hf term is less important than that in the impurities free sample , and the sdt is less sensitive to the temperature . we now investigate the magnetic field dependence of the spin dephasing . in fig . [ fig6 ] , we plot the sdt versus the applied magnetic field for different impurity levels and different spin polarizations . it is seen that for all the cases we study , the sdt increases with the magnetic field . this is because in the presence of a magnetic field , the electron spins undergo a larmor precession around the magnetic field . this precession suppresses the precession about the effective magnetic field @xmath15.@xcite therefore the sdt increases with the magnetic field . it is pointed out that in 3d electron gas , the magnetic field also forces electrons to precess around it . this precession introduces additional symmetry in the momentum space that limits the @xmath3-space available to the dp effect which is anisotropic in it.@xcite this can further reduce the spin dephasing . however , it is expected that this effect in the 2d case is less effective than the 3d case as in @xmath7-direction the momentum is quantized and the momentum precession around the magnetic field should be suppressed . in additional to the above mentioned effect of the magnetic field on spin dephasing , one can further see from fig . [ fig6 ] , that for large polarization , the magnetic field also enhances the hf term . as we mentioned before , for large polarization , the contribution from the hf term is important . increase of the hf term serves as additional magnetic field which further suppresses the effect of the dp term @xmath15 , and therefore results in a faster rise in the @xmath50-@xmath46 curve in the region of @xmath70 t. when the applied magnetic field exceeds 4 t , the increase of hf term saturates , thus the slop of the @xmath50-@xmath46 curves in the region of @xmath71 t is reduced to that in the low polarization region . to reveal more concrete about the combining effect of the magnetic field and the hf term on spin dephasing , we plot the sdt as a function of polarization in fig . it is shown that the rise in the @xmath50-@xmath51 curve increases with the magnetic field . moreover , the position of the peak in @xmath50-@xmath51 shifts to a larger polarization . this is understood that , it needs a larger hf term , and hence a larger spin polarization , to achieve the resonance condition when the magnetic field increases . when the magnetic field is raised to 4 t , it is no longer possible to form the resonance for all of the polarization . as a result the sdt increases uniquely with the polarization and there is no peak in the @xmath50-@xmath51 curve . we now investigate how the interface electric field affects the spin dephasing . in fig . [ fig8 ] we plot the sdt as a function of the initial spin polarization for three different interface electric fields . it can be seen from the figure that when the interface electric field decreases from @xmath72 v / cm to @xmath73 v / cm , the sdt increases about 16 folds . it is understood that when the interface electric field increases , the rashba effect is enhanced , and consequently the spin dephasing is also enhanced . moreover , for fixed initial spin polarization , the hf term is reduced when @xmath18 is increased . as a result , the interface electric field @xmath18 also changes the anomalous peak in the @xmath50-@xmath51 curve as in order to achieve the resonance condition , one has to go to higher polarization in order to get large enough hf term . consequently the resonance peak is smoothed and the position moves to higher initial spin polarization region when @xmath18 increases , which is shown in the fig . in conclusion , we have performed a systematic investigation of the rashba effect on the spin dephasing of @xmath0-typed inas qw s under moderate magnetic fields in voigt configuration . based on the nonequilibrium green s function theory , we derived a set of kinetic bloch equations for a two - spin - band model . this model includes the electron - phonon , electron - impurity scattering as well as the electron - electron interaction . by numerically solving the kinetic bloch equations , we study the time evolution of electron densities in each spin band and the spin coherence the correlation between spin - up and -down bands . the spin dephasing time is calculated from the slope of the envelope of the time evolution of the incoherently summed spin coherence . we therefore are able to study in detail how this dephasing time is affected by spin polarization , temperature , impurity level , magnetic field and interface electric field . differing from the earlier studies on spin dephasing based on the single particle model which only considers the effective sf scattering , our theory also takes account of the contribution of many - body effect on the spin dephasing . in fact , for the @xmath0-typed semiconductors and the spin polarization studied in the experiments , this many - body dephasing effect is even more important than the effective sf scattering as it is one order of magnitude larger than the later . equally remarkable is that , as we include all the scattering , especially the coulomb scattering in our many - body theory , now we are able to calculate the spin dephasing with extra large ( up to 100 % ) initial spin polarization which is unable to be calculated by the former single particle theory . it is discovered that the sdt increases with the initial spin polarization . moreover , for low impurity level and low temperature , there is a giant anomalous resonant peak in the curve of the sdt versus the initial spin polarization . this resonant peak moves to high spin polarization and its magnitude is fast reduced ( enhanced ) until the whole resonance disappears if one increases the impurity density , the temperature and/or the interface electric field ( the magnetic field ) . it is discovered that this anomalous resonance peak originates from the hf contribution of the electron - electron coulomb interaction . under the right spin polarization , the contribution of hf term may reach the magnitude comparable to the contribution of the dp term as well as the magnetic field in the coherent part of the bloch equation and reduces the anisotropy caused by the rashba effect consequently reduces the spin dephasing . as the resonance is the combined effects of the hf term , the rashba term and the magnetic field , the magnitude and position of the resonance peak are affected by all the factors that can alert the magnitude of the hf term , such as temperature , impurity scattering , magnetic field as well as the interface electric field : for a given impurity concentration , when the temperature increases , the hf term reduces . consequently the @xmath50-@xmath51 curve is smoothed and the peak position is moved to higher spin polarization ; for impurity free sample , if the temperature is raised to 200 k , the hf term is reduced too much to form a resonance and the anomalous peak disappears ; the same situation happens when the impurity level increases at a given temperature as the scattering also lowers the hf term . when the impurity level is raised to @xmath64 there is no resonance in the temperature region we studied ; while the increase of the magnetic field enhances the hf term and results in a faster increase of the sdt as well as a higher resonant peak in @xmath50-@xmath51 curve . moreover , as the magnetic field becomes larger , it needs a larger hf term and hence a larger polarization in order to achieve the resonant condition . therefore the peak position is also moved to higher polarization ; when the interface electric field increases , the hf term is reduced . therefore the resonance peak in @xmath50-@xmath51 becomes flatter and its position moves to higher spin polarization . for low spin polarized regime , the sdt increases when the temperature rises . this is contrary to the result of earlier simplified single - particle calculation where the sdt always decreases with the increase of the temperature . moreover , the sdt predicted by our many - body calculation is one order of magnitude faster than the earlier result . the physics of this feature is due to the additional many - body spin dephasing channel due to the inhomogeneous broadening provided by the rashba term , which by combining with the sc scattering also causes spin dephasing . in the situation we studied , the spin dephasing is dominated by the many - body dephasing effect . with the increase of the temperature , the inhomogeneous broadening reduces and the sdt increases . in high spin polarization region , the contribution of the hf term should be taken into consideration . this brings more complication in the study of the spin dephasing . usually the sdt can not be described by a monotonic function of the temperature and the impurity concentration in high polarization regime . for polarization near the resonance peak in low temperature and impurities free samples , the sdt decreases dramatically with the temperature as the resonance is removed when the temperature increases . whereas when the impurity concentration is @xmath74 , the sdt is less sensitive to the temperature . as the magnetic field causes the electron spins to precess about it , this precession will suppress the precession about the effective magnetic field @xmath15 originated from the rashba effect . as a result the spin dephasing is reduced . the magnetic field also enhances the hf term , which servers as an additional magnetic field and further suppresses the rashba effect . therefore , the @xmath50-@xmath46 curve gets a faster increase in the high polarization region . our calculation also shows that when the interface electric field increases , the sdt decreases . this is because with the increasing of the interface electric field , the rashba term is strengthened . in summary we have performed a thorough investigation of the spin dephasing in @xmath0-typed inas qw s . many new features which have not been investigated both theoretically and experimentally are predicted in a wide range of parameters . mww is supported by the `` 100 person project '' of chinese academy of sciences and natural science foundation of china under grant no . he would also like to thank s. t. chui at bartol research institute , university of delaware for hospitality . the scattering terms of electron distribution functions in the markovian limit are given by @xmath75 \nonumber\\ & & -2\pi n_i\sum_{{\bf q}}u^2_{{\bf q } } \delta(\varepsilon_{{\bf k}}-\varepsilon_{{\bf k}-{\bf q } } ) \bigl[f_{{\bf k}\sigma}(1-f_{{\bf k}-{\bf q}\sigma})- \mbox{re}(\rho_{{\bf k}}\rho^{\ast}_{{\bf k}-{\bf q}})\bigr ] -2\pi\sum_{{\bf q}{\bf k}^{\prime}\sigma^{\prime}}v_{{\bf q}}^2 \delta(\varepsilon_{{\bf k}-{\bf q}}-\varepsilon_{{\bf k } } + \varepsilon_{{\bf k}^{\prime}}- \varepsilon_{{\bf k}^{\prime}-{\bf q } } ) \nonumber\\ & & \bigl [ ( 1-f_{{\bf k}-{\bf q}\sigma})f_{{\bf k}\sigma } ( 1-f_{{\bf k}^{\prime}\sigma^{\prime } } ) f_{{\bf k}^{\prime}-{\bf q}\sigma^{\prime } } + { 1\over 2}\rho_{{\bf k}}\rho^{\ast}_{{\bf k}-{\bf q } } ( f_{{\bf k}^{\prime}\sigma^{\prime}}- f_{{\bf k}^{\prime}-{\bf q}\sigma^{\prime } } ) + { 1\over 2}\rho_{{\bf k}^{\prime } } \rho^{\ast}_{{\bf k}^{\prime}-{\bf q } } ( f_{{\bf k}-{\bf q}\sigma}-f_{{\bf k}\sigma})\bigr ] \biggr\}\nonumber\\ & & -\{{\bf k}\leftrightarrow{\bf k}-{\bf q},{\bf k}^{\prime}\leftrightarrow{\bf k}^{\prime}-{\bf q}\},\end{aligned}\ ] ] in which @xmath76 stands for the same terms as in the previous @xmath77 but with the interchange @xmath78 and @xmath79 . the first term inside the braces on the rhs of eq . ( [ eq : f_scatt ] ) comes from the electron - phonon interaction . @xmath80 stands for the different phonon modes , _ i.e. _ , one longitude optical ( lo ) phonon mode , one longitudinal acoustic ( ac ) phonon mode due to the deformation potential , and two ac modes due to the transverse piezoelectric field . @xmath81 are the matrix elements of electron - phonon coupling for mode @xmath80 . for lo phonons , @xmath82\ } @xmath84 is the optical dielectric constant and @xmath85 is the lo phonon frequency . the form factor @xmath86 $ ] with @xmath87 . @xmath88 $ ] is the bose distribution of phonon mode @xmath80 at temperature @xmath89 . the second term inside the braces on the rhs of eq . ( [ eq : f_scatt ] ) results from the electron - impurity scattering under the random phase approximation with @xmath63 denoting the impurity concentration . @xmath90\bigr\}^2 |i(iq_z)|^2 $ ] is the electron - impurity interaction matrix element with @xmath91 stands for the charge number of the impurity . @xmath91 is assumed to be @xmath92 throughout our calculation . the third term is the contribution of the coulomb interaction . similarly , the scattering parts of the spin coherence are given by @xmath93 \nonumber \\ & & + \pi n_i\sum_{{\bf q}}u_{{\bf q}}^2 \delta(\varepsilon_{{\bf k}}-\varepsilon_{{\bf k}-{\bf q } } ) \bigl[(f_{{\bf k}{1\over 2}}+f_{{\bf k}-{1\over 2 } } ) \rho_{{\bf k}-{\bf q } } -(2-f_{{\bf k}-{\bf q}{1\over 2}}-f_{{\bf k}-{\bf q}-{1\over 2 } } ) \rho_{{\bf k}}\bigr]\nonumber\\ & & -\sum_{{\bf q}{\bf k}^{\prime}}\pi v_{{\bf q}}^2 \delta(\varepsilon_{{\bf k}-{\bf q}}-\varepsilon_{{\bf k}}+\varepsilon_{{\bf k}^{\prime}}-\varepsilon_{{\bf k}^{\prime}-{\bf q } } ) \biggl ( \bigl(f_{{\bf k}-{\bf q}{1\over 2}}\rho_{{\bf k } } + \rho_{{\bf k}-{\bf q}}f_{{\bf k}-{1\over 2 } } \bigr ) ( f_{{\bf k}^{\prime}{1\over 2}}- f_{{\bf k}^{\prime}-{\bf q}{1\over 2 } } + f_{{\bf k}^{\prime}-{1\over 2}}- f_{{\bf k}^{\prime}-{\bf q}-{1\over 2}})\nonumber\\ & & + \rho_{{\bf k}}\bigl [ ( 1-f_{{\bf k}^{\prime}{1\over 2}})f_{{\bf k}-{\bf q}{1\over 2 } } + ( 1-f_{{\bf k}^{\prime}-{1\over 2}})f_{{\bf k}-{\bf q}-{1\over 2 } } -2\mbox{re}(\rho^{\ast}_{{\bf k}^{\prime}}\rho_{{\bf k}-{\bf q } } ) \bigr ] - \rho_{{\bf k}-{\bf q}}\bigl [ f_{{\bf k}^{\prime}{1\over 2}}(1-f_{{\bf k}^{\prime}-{\bf q}{1\over 2 } } ) \nonumber\\ & & + ( 1-f_{{\bf k}^{\prime}-{1\over 2}})f_{{\bf k}^{\prime}-{\bf q}-{1\over 2 } } -2\mbox{re}(\rho^{\ast}_{{\bf k}^{\prime}}\rho_{{\bf k}^{\prime}-{\bf q } } ) \bigr]\biggl)\biggl\ } -\bigl\{{\bf k}\leftrightarrow { \bf k}-{\bf q},{\bf k}^{\prime}\leftrightarrow { \bf k}^{\prime}-{\bf q}\bigr\}\ .\end{aligned}\ ] ]

we perform a many - body investigation of the spin dephasing in @xmath0-typed inas quantum wells under moderate magnetic fields in the voigt configuration by constructing and numerically solving the kinetic bloch equations . we obtain the spin dephasing time due to the rashba effect together with the spin conserving scattering such as the electron - phonon , the electron - nonmagnetic impurity as well as the electron - electron coulomb scattering . by varying the initial spin polarization , temperature , impurity density , applied magnetic field and the interface electric field , we are able to study the spin dephasing time under various conditions . for the electron density and quantum well width we study , the many - body effect dominates the spin dephasing . moreover , we find an anomalous resonance peak in the spin dephasing time for high initial spin polarization under moderate magnetic fields .

as the instrumentation for millimeter - wave telescopes improves , it becomes progressively more straightforward to study individual molecular clouds in other galaxies . recent studies of local group galaxies have surveyed large numbers of molecular clouds in the large magellanic cloud @xcite , the small magellanic cloud @xcite , m33 @xcite , and a bevy of local group dwarf galaxies ( e.g. * ? ? ? * ; * ? ? ? these recent studies explore the nature of star formation on galactic scales by studying the properties of giant molecular clouds ( gmcs , @xmath1 ) throughout their host galaxies . such gmcs contain the majority of the molecular mass in the milky way s ism and are responsible for most of the star formation in the galaxy @xcite . the andromeda galaxy ( m31 ) is the second largest disk galaxy in the local group , after the milky way , and it subtends over 2 deg@xmath2 on the sky . its proximity ( 770 kpc , * ? ? ? * ) makes it an excellent target for studying extragalactic molecular clouds . numerous surveys of co emission have been conducted over a portion of m31 and a comprehensive list of the 24 co studies published up to 1999 is given in @xcite . this extensive list of surveys can be supplemented with a few major studies that have occurred since then . @xcite used the bima millimeter interferometer to study a @xmath3 field in the outer region of the galaxy ( @xmath4 kpc ) and find 6 molecular complexes similar to those found in the milky way . an extensive survey covering the entirety of the star - forming disk of m31 has been completed using the iram 30-m by ( * ? ? ? * see also references therein ) . finally , @xcite used the plateau de burre interferometer to examine the properties of molecular clouds in 9 fields . using the gaussclumps @xcite algorithm , they decompose the emission into 30 individual molecular clouds . previous high - resolution observations of co in m31 indicate that a large fraction of the molecular gas is found in gmcs . identifying individual gmcs requires a telescope beam with a projected size @xmath5 pc , the typical size of a gmc in the milky way @xcite , which requires an angular resolution of @xmath6 at the distance of m31 . there have been seven observational campaigns that observed co @xmath7 emission from m31 at sufficient resolution to distinguish molecular clouds : @xcite . with the exception of @xcite , all of these studies have found gmcs with properties similar to those found in the inner milky way and @xcite have argued that the differences observed by @xcite can be attributed to observational errors . indeed , @xcite presented the first direct observations of gmcs in any external galaxy using interferometric observations . subsequent studies with interferometers and single - dish telescopes confirmed that most co emission in m31 comes from gmcs and that the gmcs properties were similar to those found in the milky way @xcite . although the molecular gas in m31 has been extensively studied , there remains a gap connecting the large - scale , single - dish observations and the small - scale , interferometer observations . to address this gap , we completed co(@xmath8 ) observations of a large ( 20 kpc@xmath9 region ) along a spiral arm of m31 with high resolution ( @xmath10 pc ) . we then followed up on these observations using a more extended configuration of the interferometer yielding data with a resolution of @xmath11 pc . this paper presents the observational data of the both the survey and the follow - up observations ( [ obs ] ) . using only the follow - up data , we present the first results , namely a confirmation of previous studies that find gmcs in m31 are similar to those in the milky way ( [ analysis],[larson - sec ] ) . notably , this paper utilizes the techniques described in @xcite to correct the observational biases that plague extragalactic co observations , thereby placing derived cloud properties on a common scale that can be rigorously compared with gmc data from other galaxies . the follow - up observations are also used to examine the velocity gradients and angular momentum of the gmcs , which are then compared to the remainder of gas in the galaxy for insight into the gmc formation problem ( [ spang ] ) . we conclude the paper by examining the larger galactic environment of m31 to explore connections between the gmcs and the larger ism ( [ environment ] ) . subsequent work will explore the star formation properties of these gmcs and the formation of such clouds along the spiral arm using the data from the spiral arm survey . we observed @xmath12co(@xmath8 ) emission from m31 during the spring and fall observing seasons of 2002 with the d and c configurations of the bima millimeter interferometer @xcite . the observations consisted of an 81-field mosaic using the most compact ( d ) configuration with follow - up observations on seven sub - regions , covering 30 fields at higher resolution ( c array ) . the d - array survey spans a projected length of 6.7 kpc along a spiral arm in the galaxy . three of the seven follow - up , c - array fields targeted regions with known co emission from the d - array survey , and the remaining four fields targeted regions with strong co emission in the single - dish observations of @xcite over a range of galactocentric distances . the locations of the fields are indicated in figure [ fields - m31 ] . the d - array observations were completed in september and october 2002 over the course of four nights . each night roughly 20 pointings of the mosaic were observed . during the observations , the fields were observed for 46 seconds each , making two passes through the mosaic before returning to the phase calibrator ( 0102 + 504 , 2.6 jy ) every 30 minutes . this cycle was continued through the night , accumulating @xmath13 hours of integration time on m31 per night ( 18 minutes per field ) . the correlator was configured to span 200 mhz at 500 khz ( 1.3 km s@xmath14 ) resolution , easily encompassing the whole range of velocities expected from the m31 s rotation curve for this region @xcite . at the latitude of the hat creek radio observatory , m31 transits near zenith and can not be observed for 40 minutes a night . this time was used to perform flux calibrations using saturn and uranus . over the course of the observations , the derived flux of the phase calibrator was stable to 10% . the relative flux calibration of the data may be better than this level owing to the intrinsic variability of the phase calibrator . the data were reduced and inverted using the miriad software package @xcite , following the calibration procedures of the bima survey of nearby galaxies @xcite . the four nights of @xmath15 data were combined using a linear mosaicing technique . the inversion used natural weighting , and gridding in the @xmath15 plane was chosen to produce maps with @xmath16 pixels and @xmath17 km s@xmath14 channel width . the dirty maps were deconvolved with a steer - dewey - ito deconvolution algorithm optimized for mosaics ( mossdi2 in the miriad software package ) . each plane of the final image was cleaned to the 1.5@xmath18 significance level . the final resolution of the survey is @xmath6 , with negligible variation over the map . the mosaic is corrected for the gain of the primary beam . three regions within the d - array survey containing emission were observed with the more extended c - array configuration . in total , 26 fields were observed on five nights from october to december 2002 . the c - array observations used the same observing strategy as the low - resolution survey , observing each field for 46 seconds and making two or three passes ( depending on the size of the mosaic ) through the mosaic before observing the phase calibrator every 24 minutes . the correlator configuration and choice of calibrators were the same as for the survey . the follow - up observations used a subset of the d - array pointing centers to facilitate merging the two data sets . these data were combined with the d - array observations in the @xmath15 plane and inverted using uniform weighting . the data were cleaned using the same steer - dewey - ito deconvolution algorithm as for the d - array map . the final resolution of the combined data is @xmath19 with a 2.03 km s@xmath14 channel width . the images are corrected for the gain of the primary beam . prior to the d - array survey , four fields of m31 were observed using the c array ( spring 2002 ) . these observations targeted single fields known to contain bright co emission from @xcite over a range of galactocentric distance . the fields were chosen to span a variety of regions in the galaxy ( different spiral arms and galactocentric radii ) in an attempt to reveal any biases seen by focusing on one region of the galaxy . the observations used 0136 + 478 as a phase calibrator and mars to establish the flux scale of the phase calibrator . since the observations consisted exclusively of c - array data , they were inverted using natural weighting . the final synthesized beam size for these observations is @xmath20 . the observations produced one survey data cube and 7 high - resolution data cubes for analysis . the properties of these data sets are summarized in table [ m31-obsprops ] . since space constraints prohibit a full presentation of all the observational data , an excerpt from the center of field a is shown in figure [ data - example ] to illustrate typical quality of the data . the remainder of the paper emphasizes the high - resolution data ( fields a - g in table [ m31-obsprops ] ) . we defer presentation and analysis of the survey data to a subsequent paper . ccccccc survey & d & 00 41 48.1 , + 41 19 02 & @xmath21 & @xmath22 & 0.27 & 0.35 + a & c , d & 00 42 26.5 , + 41 28 45 & @xmath23 & @xmath24 & 0.64 & 0.35 + b & c , d & 00 42 12.3 , + 41 25 33 & @xmath25 & @xmath26 & 0.54 & 0.12 + c & c , d & 00 42 02.4 , + 41 22 13 & @xmath27 & @xmath28 & 0.76 & 0.17 + d & c & 00 43 01.8 , + 41 37 00 & @xmath29 & @xmath30 & 0.48 & 0.42 + e & c & 00 43 56.8 , + 41 26 12 & @xmath31 & @xmath30 & 0.94 & 0.23 + f & c & 00 44 44.3 , + 41 28 07 & @xmath32 & @xmath30 & 0.92 & 0.15 + g & c & 00 43 33.9 , + 41 09 58 & @xmath33 & @xmath30 & 1.38 & 0.19 + for data cubes a g in table [ m31-obsprops ] , we identified the emission using a dynamic thresholding technique . for every pixel in the cube , we first estimated the rms value for the noise using the method of @xcite ; this method accounts for gain variations in both position and frequency . we first estimate the noise at every position by measuring the standard deviation of the data values using iterative rejection of high signal pixels ( @xmath34 ) . we generate a noise map by smoothing the resulting pixel estimates with a boxcar kernel of the same size as the synthesized beam . then , we estimate the relative changes of the noise across the bandpass by estimating @xmath18 in each channel map and normalizing to the mean value of @xmath35 across the bandpass . we generate a cube of noise estimates by scaling the error map by the fractional change from the mean appropriate for that velocity . performing a pixel - wise division of the data cube by the cube of noise estimates yields a data set in significance units . we identify co emission in the data cubes by searching for contiguous regions of high significance in position - position - velocity space using a two - tiered thresholding scheme ( e.g. * ? ? ? the core pixels in the mask are identified as pairs of pixels in adjacent velocity channels with @xmath36 . the mask is then expanded in position and velocity space to include all pixels with @xmath37 that are connected by high significance pixels to the @xmath38 core of the mask . this masking process has been used in previous extragalactic , interferometric studies and has been shown to successfully identify emission with minimal inclusion of noise @xcite . eliminating noise from the emission under consideration is key to the methods used to determine cloud properties ( see [ cloudprops ] ) . the identification and analysis of gmcs in m31 may be complicated by the relatively high inclination of the galaxy : @xmath39 @xcite . the situation is similar , in part , to observing gmcs in the outer regions of the milky way ( @xmath40 ! ) , and the spatial resolution ( @xmath11 pc ) is comparable to what is found using small telescopes ( e.g. the cfa 1.2-m ) to observe distant ( @xmath41 kpc ) gmcs in the milky way @xcite . while study of gmcs has been conducted in the milky way with such data , the inclination being @xmath42 and not @xmath43 helps the situation significantly . the foreground and background spiral arms in m31 are separated from the arm under consideration by @xmath44 in projection . thus , spatial information is useful for separating the emission into gmcs . the clouds in the interarm region are likely too faint to be recovered by the deconvolution algorithm ( [ fluxrec ] ) . we conclude that the primary concern in decomposition is the blending of gmcs within the same spiral arm . @xcite estimate that the width of the spiral arms in the plane of the galaxy for this region is @xmath45 pc with a vertical thickness of 150 pc . there could be multiple , distinct clouds in the spiral arm along the same line of sight , so we make a simple estimate of how many gmcs are likely to be along a line of sight through the spiral arm . we model the arm as a cylinder with elliptical cross - section having major and minor axes of 500 pc and 150 pc respectively . for a viewing angle 77@xmath46 away from the minor axis of the cylinder , the path length through the cylinder is @xmath47 pc . we assume that the gmcs are distributed uniformly in the cylindrical volume with volume density @xmath48 and geometric cross - section @xmath49 . the mean number of clouds intersected by a line - of - sight through the arm is then @xmath50 . we estimate @xmath48 by counting the number of clouds ( @xmath51 ) in some trial volume along the arm @xmath52 where @xmath53,@xmath54 and @xmath55 represent the length , width ( major axis ) and height ( minor axis ) of the volume of the arm respectively . by assuming a surface density through a typical gmc of @xmath56 and comparing this to the observed surface density in the single - dish map @xmath57 , we can estimate the area filling fraction of gmcs in the arm when viewed from above the galaxy:@xmath58 . since @xmath59 is also given by @xmath60 , we can express @xmath61 significant confusion will occur for @xmath62 . for this condition to be met with @xmath63 pc and @xmath64 pc , @xmath65 . the data of @xcite show @xmath66 for this region requiring gmcs in m31 to have @xmath67 for significant blending to occur . we regard this possibility as unlikely since gmcs in other , less confused regions of m31 are found to have comparable properties to local group clouds @xcite where @xmath68 @xcite . unfortunately , one of the aims of the paper is to assess how similar clouds in m31 are to those in other galaxies ; so we can not reject the possibility that the clouds could be significantly blended along the line of sight although this would imply significantly larger sizes and lower column densities than is typical for gmcs in other galaxies . we conclude this discussion of inclination effects by noting that we will assume that gmc properties are independent of viewing angle , as is common in studies of gmcs in the milky way and beyond ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? as a result , the reported properties of gmcs do not need any correction for the high inclination of the galaxy . we identify emission in the data cubes from the high - resolution data ( fields a - g in table [ m31-obsprops ] ) using the signal - identification method described in [ m31-masking ] . maps of the high - resolution data fields appear in figures [ m31-cfield1 ] and [ m31-cfield2 ] . this emission was partitioned into individual molecular clouds using the segmentation algorithm described in ( * ? ? ? * rl06 ) to separate blended emission . the segmentation algorithm is a modified watershed algorithm , related to the original clumpfind @xcite . however , the adopted algorithm has been optimized to identify gmcs in a wide variety of observational data : the algorithm is significantly more robust to the presence of noise and does not over - segment gmcs as the original clumpfind is known to do @xcite . the rl06 algorithm establishes its parameters from physical scales in the data ( e.g. parsecs , km s@xmath14 ) as opposed to observationally determined scales ( beam and channel widths ) to minimize biases in comparing catalogs with different observational properties . for the high - resolution data , the algorithm identifies 67 molecular clouds in the seven cubes . cccccccc 1 & @xmath69 & 55 & 34 & 4.07 & 17.0 & 0.07 & @xmath70 + 2 & @xmath71 & 42 & 24 & 4.43 & 15.1 & 0.04 & @xmath72 + 3 & @xmath73 & 72 & 52 & 6.29 & 14.9 & 0.12 & @xmath74 + 4 & @xmath75 & 78 & 50 & 4.86 & 14.8 & 0.02 & @xmath76 + 5 & @xmath77 & 63 & 47 & 3.38 & 14.6 & 0.03 & @xmath78 + 6 & @xmath79 & 33 & 43 & 4.44 & 13.3 & 0.10 & @xmath80 + 7 & @xmath81 & 33 & 28 & 2.35 & 13.2 & 0.05 & @xmath82 + 8 & @xmath83 & 25 & 32 & 5.45 & 12.6 & 0.11 & @xmath84 + 9 & @xmath85 & 31 & 19 & 2.86 & 12.3 & 0.03 & @xmath86 + 10 & @xmath87 & 34 & 33 & 4.38 & 11.9 & 0.06 & @xmath88 + 11 & @xmath89 & 15 & 24 & 2.38 & 11.5 & 0.06 & @xmath90 + 12 & @xmath91 & 61 & 92 & 4.24 & 11.3 & 0.03 & @xmath92 + 13 & @xmath93 & 26 & 27 & 4.07 & 11.3 & 0.08 & @xmath94 + 14 & @xmath95 & 26 & 30 & 4.88 & 11.2 & 0.18 & @xmath96 + 15 & @xmath97 & 26 & 47 & 2.95 & 10.7 & 0.03 & @xmath98 + 16 & @xmath99 & 26 & 19 & 3.09 & 10.6 & 0.02 & @xmath100 + 17 & @xmath101 & 42 & 42 & 4.41 & 10.6 & 0.06 & @xmath102 + 18 & @xmath103 & 40 & 30 & 5.54 & 10.3 & 0.06 & @xmath104 + 19 & @xmath105 & 38 & 40 & 3.99 & 10.3 & 0.08 & @xmath106 + 20 & @xmath107 & 21 & 21 & 4.40 & 9.9 & 0.11 & @xmath108 + 21 & @xmath109 & 31 & 19 & 2.75 & 9.8 & 0.03 & @xmath110 + 22 & @xmath111 & 27 & 25 & 5.39 & 9.7 & 0.08 & @xmath112 + 23 & @xmath113 & 44 & 34 & 6.19 & 9.7 & 0.17 & @xmath114 + 24 & @xmath115 & 24 & 24 & 5.89 & 9.3 & 0.11 & @xmath116 + 25 & @xmath117 & 45 & 35 & 4.67 & 9.2 & 0.11 & @xmath118 + 26 & @xmath119 & 8 & & 4.64 & 8.8 & 0.16 & @xmath120 + 27 & @xmath121 & 22 & 28 & 4.13 & 8.7 & 0.09 & @xmath122 + 28 & @xmath123 & 10 & & 2.62 & 8.4 & 0.01 & @xmath102 + 29 & @xmath124 & 29 & & 6.53 & 8.2 & 0.21 & @xmath125 + 30 & @xmath126 & 3 & & 1.56 & 8.1 & 0.05 & @xmath127 + 31 & @xmath128 & 17 & 37 & 4.70 & 8.0 & 0.13 & @xmath129 + 32 & @xmath130 & 23 & 42 & 4.51 & 7.8 & 0.07 & @xmath131 + 33 & @xmath132 & 38 & 27 & 7.03 & 7.7 & 0.14 & @xmath133 + 34 & @xmath134 & 15 & & 2.38 & 7.6 & 0.04 & @xmath135 + 35 & @xmath136 & 12 & 29 & 3.15 & 7.5 & 0.05 & @xmath127 + 36 & @xmath137 & 14 & & 3.93 & 7.4 & 0.09 & @xmath138 + 37 & @xmath139 & 17 & 19 & 5.26 & 7.2 & 0.18 & @xmath140 + 38 & @xmath141 & 12 & 31 & 2.67 & 7.1 & 0.04 & @xmath142 + 39 & @xmath143 & 37 & 55 & 2.64 & 6.9 & 0.02 & @xmath144 + 40 & @xmath145 & 3 & & 1.50 & 6.7 & 0.07 & @xmath146 + 41 & @xmath147 & 2 & & 2.42 & 6.7 & 0.06 & @xmath148 + 42 & @xmath149 & 33 & 60 & 2.97 & 6.6 & 0.04 & @xmath84 + 43 & @xmath150 & 6 & & 3.62 & 6.6 & 0.05 & @xmath151 + 44 & @xmath152 & 12 & & 3.65 & 6.5 & 0.06 & @xmath153 + 45 & @xmath154 & 9 & 25 & 3.42 & 6.4 & 0.04 & @xmath155 + 46 & @xmath156 & 5 & & 1.58 & 6.3 & 0.02 & @xmath157 + 47 & @xmath158 & 16 & 35 & 2.08 & 6.3 & 0.04 & @xmath159 + 48 & @xmath160 & 8 & & 3.51 & 6.1 & 0.12 & @xmath161 + 49 & @xmath162 & 10 & & 2.97 & 6.1 & 0.11 & @xmath110 + 50 & @xmath163 & 3 & & 1.74 & 6.1 & 0.06 & @xmath164 + 51 & @xmath165 & 5 & 12 & 2.16 & 6.1 & 0.06 & @xmath166 + 52 & @xmath167 & 6 & & 2.24 & 6.1 & 0.05 & @xmath168 + 53 & @xmath169 & 2 & & 1.70 & 6.0 & 0.06 & @xmath170 + 54 & @xmath171 & 10 & & 6.27 & 5.9 & 0.15 & @xmath172 + 55 & @xmath173 & 8 & & 5.00 & 5.9 & 0.14 & @xmath142 + 56 & @xmath174 & 6 & 16 & 2.14 & 5.8 & 0.05 & @xmath175 + 57 & @xmath176 & 3 & & 1.49 & 5.5 & 0.00 & @xmath177 + 58 & @xmath178 & 3 & & 2.60 & 5.2 & 0.08 & @xmath179 + 59 & @xmath180 & 4 & & 2.38 & 5.1 & 0.04 & @xmath181 + 60 & @xmath182 & 2 & & 1.79 & 5.1 & 0.04 & @xmath183 + 61 & @xmath184 & 1 & & 0.99 & 5.1 & 0.01 & @xmath185 + 62 & @xmath186 & 4 & & 2.42 & 4.9 & 0.11 & @xmath187 + 63 & @xmath188 & 1 & & 0.98 & 4.8 & 0.04 & @xmath189 + 64 & @xmath190 & 3 & & 1.28 & 4.6 & 0.02 & @xmath191 + 65 & @xmath192 & 14 & 22 & 3.11 & 4.6 & 0.04 & @xmath193 + 66 & @xmath194 & 3 & & 0.65 & 4.5 & 0.01 & @xmath151 + 67 & @xmath195 & 4 & & 2.58 & 4.2 & 0.12 & @xmath196 + for each of these clouds , we measure three macroscopic properties : mass , radius ( @xmath197 ) , line width ( @xmath198 ) . these macroscopic properties are defined using moments of the intensity distribution ( see rl06 for details ) . the mass of the gmcs is determined by scaling the zeroth moment ( sum ) of the intensity distribution by a constant co - to - h@xmath199 conversion factor of @xmath200 @xcite . assuming a mean particle mass of 1.36@xmath201 @xcite gives a mass from the integrated flux over the cloud : @xmath202 the calculation uses a flux value @xmath203 that is extrapolated from the brightness threshold ( @xmath204 ) to 0 @xmath205 using variation in @xmath206 as a function of threshold @xcite . rl06 demonstrate that this extrapolation practically eliminates bias in the data due to low sensitivity , though incomplete flux recovery interferometer may still affect the resulting mass estimates ( [ fluxrec ] ) . an demonstration of the extrapolation method used in this paper is shown in figure [ extrapfig ] . in a similar fashion to the molecular mass , we use the moments of the emission distribution to measure the sizes and line widths of the resulting molecular clouds . following rl06 , the size is defined as the second moment of the emission distribution along the major and minor axes of the cloud : @xmath207 and @xmath208 respectively . like the total flux , the values of these moments are extrapolated to the 0 k km s@xmath14 pc@xmath2 intensity level to minimize the effects of the brightness threshold giving @xmath209 and @xmath210 ( see rl06 for details ) . we estimate the deconvolved size of the major and minor axes by subtracting the width of the beam in quadrature . the size of the cloud is the geometric mean of the `` deconvolved '' major- and minor - axis sizes : @xmath211^{1/2 } \label{rdefn}\ ] ] the radius of the cloud is estimated by scaling @xmath212 up by a factor of 1.91 ( * ? ? ? @xcite note that this radius estimate is insensitive to inclination and projection effects . the rms velocity dispersion is calculated as the second moment of the emission distribution in velocity space . again , the velocity dispersion is extrapolated to to the the 0 k km s@xmath14 pc@xmath2 intensity level to give the corrected velocity dispersion @xmath213 . the properties of the molecular clouds in the high - resolution data are summarized in table [ m31-cloudprops ] . the clouds are divided into two groups : ` high signal - to - noise clouds ' with @xmath214 where the macroscopic properties can be accurately recovered ( see [ fluxrec ] ) and ` low signal - to - noise clouds ' ( @xmath215 ) where the derived properties are suspect . since interferometers do not measure the total power of observed emission , they must rely upon single dish observations for accurate measurements of the co flux . in the absence of such single - dish data , deconvolution algorithms can be used to extrapolate into the unsampled short - spacing area of the @xmath216 plane @xcite . the accuracy of such extrapolations depends primarily on the signal - to - noise ratio of the data and to a lesser degree on the observational strategy and the structure of the object being observed . since this work uses flux information to calculate cloud masses as well as in the moments that determine the cloud properties , flux loss affects all stages of this analysis . we use the data from iram 30-m survey of m31 @xcite to quantify the amount of flux lost in each of the the observations . we report the fraction of flux in the bima maps relative to that found in the iram map in table [ m31-obsprops ] as the column labeled @xmath217 . the fractions recovered range from 12% to 42% . the relatively low fraction of flux recovered is a product of two factors that affect interferometric observations : spatial filtering of gmcs ( studied in * ? ? ? * ; * ? ? ? * ) and the non - linear recovery of signal in the low signal - to - noise regime ( explored in * ? ? ? the work of @xcite found that , because of their clumpy structure , spatial filtering did not hinder the accurate measurement of gmc shapes , fluxes and properties . however , their simulations were conducted in the high signal - to - noise regime . in the low signal - to - noise regime , @xcite found that even relatively compact simulated sources suffered from significant flux loss , owing to the inability for the deconvolution algorithms to isolate low - amplitude signal . these two effects ( spatial filtering of gmc emission and low signal - to - noise ) were studied together in rl06 who simulated bima c , d , and c+d observations of milky way gmcs ( orion , rosette , w3/4/5 ) as if those gmcs were located in m31 . their work showed that accurate recovery of cloud properties required signal - to - noise ratios of @xmath218 for less than a 20% loss in all the cloud properties . for clouds with lower significance , the flux loss is more severe . however , the dominant source of flux loss comes from clouds not detected in the observations at all , i.e. those with @xmath219 , the minimum amplitude to which the deconvolution algorithm operates . for these clouds , the deconvolution algorithm does not identify a significant local maximum for cleaning and the cloud contributes zero flux to the deconvolved map . such low mass clouds likely comprise most of the missing flux not seen in the interferometer data . when clouds are detected , cloud properties estimated from the observations should be relatively accurate , particularly if @xmath220 . in general , we only consider the properties of clouds in the study if this sensitivity condition is met ; exceptions to this criterion will be noted . for the high - resolution data , 19 gmcs meet the signal - to - noise criterion ( the ` high signal - to - noise clouds ' in table [ m31-cloudprops ] ) . we emphasize that for these m31 data interferometer observations act as a filter that preferentially detects gmcs , but the properties of those clouds are well - measured . upon measuring a set of gmc properties for clouds in another galaxy , the immediate question is whether these clouds are similar to those found in the milky way . ideally , the well - resolved clouds in the milky way would be analyzed using the methods of rl06 so that the results are directly comparable . however , since measuring cloud properties in the inner galaxy requires establishing a single distance by resolving the kinematic distance ambiguity , a detailed re - analysis has not been completed to date . the next best solution is to compare the results to an existing catalog . for comparison , we use the catalog of ( * ? ? ? * s87 ) to define the population of gmcs in the inner milky way . the work of s87 used similar methods to calculate cloud properties though their work adopts a fundamentally different procedure for segmenting gmcs . in addition , their cloud properties are not corrected for the effects of sensitivity as are the properties derived in this study . consequently , there may be systematic differences between their catalog and the present study . in general , the fluxes , line widths and radii of the s87 clouds will be smaller than would be derived using the methods of rl06 ; however , using interferometer data to observe the extragalactic clouds will reduce the fluxes , line widths and radii by a similar amount ( rl06 ) . we have rescaled the catalog of s87 to a different galactocentric solar radius : @xmath221 kpc to @xmath222 kpc and the co - to - h@xmath199 conversion factor adopted in this study . we also recalculated cloud sizes using the definition of rl06 . finally , it is noteworthy that the method of rl06 reports , in addition to unbiased values for the cloud properties , appropriate uncertainties for those properties facilitating a statistically based comparison of cloud properties in m31 to those in the milky way . @xcite first demonstrated that there were power - law scalings among the properties of molecular gas clouds . that gmcs followed these scalings became well established in a series of surveys and catalogs of the co emitting gas in the inner milky way ( * ? ? ? * ; * ? ? ? * ; * ? ? ? although there are several algebraically equivalent forms for expressing these relations , the most physically relevant relationships are generally agreed to be : @xmath223 and @xmath224 for the inner milky way , @xmath225 km s@xmath14 and @xmath226 ( s87 ) . the first relationship is the size - line width relationship and results from the supersonic turbulent motions in the gmcs , as first suggested by @xcite . the constant @xmath227 in equation [ virial ] is the virial parameter @xcite and since @xmath228 in the inner milky way the clouds are frequently interpreted as being self - gravitating . in figure [ larson ] , we plot these two relationships for the 19 clouds in m31 with sufficient signal to noise that their properties can be well - recovered in interferometric observations . we compare the relationships to those found in the catalog of s87 . in both the size - line width relationship and the value of the virial parameter , there is reasonably good agreement between the gmcs in m31 and the inner milky way gmcs of s87 . fitting the relationship for the 19 clouds in m31 that meet the sensitivity criterion ( [ fluxrec ] ) using the method of @xcite gives a size - line width relationship of @xmath229 for reference , a fit to the catalog of s87 using the same fitting method , albeit without uncertainties in the cloud properties , gives @xmath230 thus , the size - line width relationship for the molecular clouds is indistinguishable from the clouds found in the inner milky way . the similarity between size - line width relationship for clouds in m31 and the inner milky way has been reported on several occasions by other authors @xcite , though no previous study has reported an independent fit for the clouds in m31 with well characterized uncertainties in the cloud properties . the virial parameter for clouds in m31 is constant as a function of luminous mass . the uncertainty - weighted , mean value of the virial parameter for clouds in m31 is @xmath231 . for the sample of s87 , @xmath232 . the difference between the virial parameters is not statistically significant . moreover , any flux loss will affect the luminous mass estimates more than the virial estimate ( rl06 ) , so our measurement of the virial parameter is at worst an upper limit . it is unlikely that the true virial parameters of m31 cloud population differs significantly from the milky way , provided the co - to - h@xmath199 conversion factor is the same for both systems . alternatively , by assuming that the molecular clouds cataloged in this study are in the same dynamical state as are clouds in the inner milky way , we can justify our choice of co - to - h@xmath199 conversion factor . again , the similarity of the virial parameter has been noted by several authors in previous studies @xcite . in particular , the latter two studies find no systematic variation in the virial parameter for clouds over a large range of galactocentric radius in m31 . in addition to macroscopic properties like line width , radius and mass , interferometric observations of extragalactic gmcs can be used to measure the velocity gradients and angular momenta of the clouds . * rpeb ) analyzed the velocity gradients of gmcs in m33 to critically evaluate cloud formation theories . in this section , we analyze the velocity gradients for gmcs in m31 and compare our results to those of rpeb and the analysis of the velocity gradients in gmcs in the inner milky way by @xcite . we determine the velocity gradient by a least - squares fit of a plane to the velocity centroid surface @xmath233 the velocity centroid measured as a function of position across the cloud . the velocity centroid and its uncertainty are determined from the first moment of velocity weighted by brightness . the coefficients of the fit determine the magnitude ( @xmath234 ) and position angle ( @xmath235 ) of the velocity gradient for the molecular cloud ( see * ? ? ? * for details ) . the results of the fits are listed in table [ m31-cloudprops ] . for purposes of this analysis , we have included all clouds , since the velocity gradient is less affected by the interferometric effects which bias the macroscopic properties ( [ fluxrec ] ) . the velocity gradient is derived from line centroids and the first moments ( centroids ) are much less sensitive to the low surface brightness wings of the line than are the second moments . we do exclude all measurements of the velocity gradient where the uncertainty is larger than 0.15 km s@xmath14 pc@xmath14 . this upper limit was chosen from the maximum uncertainty for clouds with high signal - to - noise ( @xmath236 ) , including a total of 44 clouds in this portion of the analysis . uncertainties in the derived parameters are determined from the errors in the least - squares parameters given the uncertainty in the centroid . the magnitude of the velocity gradients range between 0 and 0.2 km s@xmath14 pc@xmath14 , in good agreement with both the values and distribution found for clouds in m33 ( rpeb ) and the milky way @xcite . we compare the magnitude of the velocity gradient to the velocity gradient that would be expected from the orbital motion of the gas in the galactic potential , ignoring streaming motions in the spiral arms . the local galactic velocity is estimated from the tangent plane to a surface defined by the line of sight projection of the galactic rotation velocity . we use the rotation curve of @xcite which is a compilation of previous co and studies using updated orientation information . on average , the rotation gradient of gmcs exceeds the circular galactic gradient by a factor of 3.0 . we measured the difference between the position angle of the gradient and that of the local galactic gradient expected from circular rotation for each of the clouds . while suggesting a slight alignment with the galaxy , the distribution of this difference is not significantly different from a random distribution according to a two - sided ks test ( * ? ? ? * @xmath237 ) . this is similar to the work of @xcite who found a random distribution of position angles in the milky way gmcs but different from rpeb who found significant alignment of the gradients with the galactic rotation . this analysis has assumed the decomposed data represent discrete clouds with independent velocity fields , and the factor of three difference between the magnitudes of the cloud and local galactic gradients supports that model . however , it is also possible that we are inappropriately decomposing a continuous gas flow into clouds and the large velocity gradients result from the change in the gas flowing into the density wave . in the presence of the radial motions that such a flow implies , the local galactic velocity gradient will change in magnitude by a factor of , at most , @xmath238 and in position angle by @xmath239 where @xmath240 and @xmath241 are the radial and azimuthal components of the gas flow . for gas in m31 , @xmath242 @xcite so these changes will not be significant and the factor of three excess of cloud gradient magnitude over the local rotational gradient implies discontinuous velocity fields . moreover , any changes should be systematic in nature , and the large range of position angles actually observed implies we are measuring the velocity gradients of discrete objects . while the magnitudes of the velocity gradients of the clouds are larger than the local rotation in the galaxy , the implied angular momentum of the gmcs is significantly smaller than that of an equivalent mass of atomic gas orbiting in the galactic gravitational potential at the gmc s position . the specific angular momentum for a cloud of gas with velocity gradient magnitude @xmath234 is @xmath243 where @xmath244 for a wide range of density distributions undergoing uniform rotation @xcite . if the velocity gradient arises solely from the turbulent motions in the cloud , @xmath245 is likely 23 times smaller @xcite . it is unlikely that the high galactic inclination affects the results . the near - random distribution of the position angles implies that the cloud angular momentum vectors are similarly close to random which would require scaling the angular momentum up by a factor of @xmath246 on average ( rpeb ) . a turbulent origin for the velocity gradients would not necessitate any correction . since any correction would be small , we make no correction for the orientation of the clouds . for the clouds in the velocity gradient analysis , the mean angular momentum is @xmath247 pc km s@xmath14 for a solid body rotator and 23 times smaller for the angular momentum of turbulent flows , comparable to the results for other systems ( rpeb , * ? ? ? we compare this to the angular momentum of progenitor material for the gmcs . the angular momentum of the gas in the galactic potential is given by rpeb ( equivalent to * ? ? ? * ) @xmath248 where @xmath249 is a form factor equal to 1/4 for a contraction of a uniform circular disk of material of radius @xmath250 . the length , @xmath250 , is the range of galactocentric radii over which material is accumulated to form a gmc . the minimum value of @xmath250 is set by the mass of the gmc under consideration and the surface density of gas that could comprise the progenitor material : @xmath251 . we take @xmath252 @xcite , including their estimates of both atomic and molecular material since small molecular clouds seen in co may be progenitor material for the larger gmcs . from the measured masses of gmcs and locations of gmcs , we calculate @xmath250 and the galactic shear from the rotation curve of @xcite . the mean value of @xmath253 for the clouds in the sample . on a cloud by cloud basis , the angular momentum of the galactic material is only a factor of 1.7 larger than the that of the gmcs , significantly smaller than the difference found in m33 where @xmath254 . the smaller margin is primarily due to the lower shear in m31 , reducing @xmath255 by a factor of 4 compared to that found for the clouds in m33 whereas the velocity gradients are comparable in the two systems . similarly , the cloud properties and gradients for clouds in the milky way are comparable to the clouds in m31 @xcite , but the shear is much larger in the inner milky way compared to the 11-kpc arm of m31 . consequently , the discrepancy between the angular momentum of progenitor material and the resulting gmcs is also larger in the inner milky way than it is in m31 ( see * ? ? ? the angular momentum defect for the m31 clouds is likely significant : the factor of 1.7 difference should be regarded as a minimum value since attributing the velocity gradients to turbulence or accumulating the progenitor material from a larger range of galactocentric radius will appreciably increase the margin . these results suggest a broad similarity between gmcs in m31 and the other disk galaxies in the local group . the primary goal of studying the velocity structure within the gmcs is to relate their angular momentum to that of the ism as a whole to evaluate formation mechanisms . the angular momentum defect between gmcs and the ism is common among the studied systems ( * ? ? ? * rpeb ) and suggests the importance of braking in molecular cloud formation . @xcite study the role of magnetic effects at removing angular momentum from forming clouds and conclude that magnetic effects can reduce the angular momentum to observed values . their unmagnetized simulations show no significant reduction in cloud angular momentum , although the density enhancements that represent proto - gmcs show @xmath256 , similar to the observed _ minimum _ value found in m31 . the relatively low - shear environment does not necessarily require magnetic braking to produce the observed cloud angular momenta , though our measurement of @xmath257 likely significantly underestimates the true value . with respect to their macroscopic properties and velocity gradients , gmcs in the 11-kpc , star - forming arm in m31 appears indistinguishable from the molecular clouds in the inner milky way . as our knowledge of gmcs in other galaxies grows , it becomes more apparent that such detailed agreement in the properties of gmcs is not universal and may actually represent exceptional cases @xcite . given the differences measured among extragalactic cloud populations , the most pressing question is : what physical mechanisms regulate the observed differences in gmcs properties ? these mechanisms must couple the gmcs to the ism on much larger scales , and future studies will examine the nature of this coupling in more detail . the similarity between gmcs in m31 and the milky way offers a useful point of reference for studying the relationship between the galactic ism and gmcs . finding environmental properties that vary significantly between the two systems while gmc properties remain the same would imply those properties are irrelevant to the regulation of molecular cloud properties . several features of the galactic environment are thought to play a role in regulating the structure of the cold ism and the properties of gmcs . here , we review several of these features and note the degree of similarity to the local milky way : * _ interstellar radiation field _ the interstellar radiation field in m31 has been previously noted to be poor in the ultraviolet relative to the milky way @xcite , though the m31 field may be significantly enhanced in the star - forming spiral arm imaged in the co observations . indeed , work by @xcite study [ ] emission in the star forming arms at @xmath258 kpc and note that the far ultraviolet field is consistent with being 100 times the local interstellar value ( i.e. @xmath259 ) around molecular clouds . thus , the observed fuv deficiency likely does not characterize the molecule rich region of the galaxy and the radiation environment is similar to clouds in the milky way . * _ metallicity _ the gas phase metallicity from regions in m31 at @xmath260 kpc is approximately [ o / h]-[o / h]@xmath261=0.0 dex with significant scatter , up to @xmath262 0.2 dex @xcite . the region shows no signs of being significantly enriched or depleted relative to the solar neighborhood . * _ dynamics _ the toomre @xmath0 parameter is frequently used to evaluate the stability of galactic disks with respect to the formation of self - gravitating structures . the parameter is defined as @xmath263 where @xmath264 is the epicyclic frequency , @xmath265 is the one - dimensional gas velocity dispersion and @xmath266 the surface density of the gas in the disk . most star - forming regions of galactic disks show @xmath267 @xcite and this region of m31 is no exception . for the rotation curve of @xcite , @xmath268 . we take @xmath269 @xcite . finally , we take @xmath270 from @xcite . combining these data give @xmath271 , well within the range of values observed in most galaxies and approximately the same value found in the solar neighborhood . * _ midplane density _ we can also compare the midplane particle density of this region in m31 to the solar neighborhood . the mean particle density is important for regulating thermal structure of the ism and the formation of the cold medium @xcite . adopting a gas scale height of 340 pc for this region of m31 @xcite gives @xmath272 in the spiral arm region , but falling to @xmath273 outside the arm . these densities are smaller than that found in the solar neighborhood ( @xmath274 , * ? ? ? the properties of the ism in this region of m31 appear , at least superficially , similar to those found in the solar neighborhood . hence , the derived similarity of molecular clouds in the two galaxies is not surprising . the only property that appears like it may be significantly different between the two environments is the midplane particle density in the neutral medium . if the difference is significant , the constancy of the gmc properties suggests that the particle density may be important in regulating the fraction of material found in the cold , neutral ism but perhaps not the properties of individual gmcs . while this environmental similarity does not isolate any irrelevant physical processes , noting the similarity in both clouds and environments illustrates that similar galactic environments do indeed produce similar gmcs this paper has presented bima millimeter interferometer observations of @xmath12co(@xmath8 ) emission from giant molecular clouds ( gmcs ) along a spiral arm at @xmath275 kpc in m31 . the observations were conducted in two parts : a low - resolution survey using the compact ( d ) configuration of the array and a high - resolution , follow - up study using a more extended ( c ) configuration of the array . all data from the study were segmented and analyzed with the analysis algorithms of @xcite , which correct the observational biases imprinted on the data by having relatively low signal - to - noise and marginally resolving the gmcs . the c - array follow - up spans 7.4 kpc@xmath276 finding 67 clouds . we report the following conclusions : \1 . the gmc population is very similar to that found in the inner milky way . both the size - line width relationship and distribution of virial parameters are statistically indistinguishable from those derived from the population of clouds in the inner milky way . this confirms the work of several authors @xcite while using techniques that correct for the biases imprinted by using interferometric observations . we measured the velocity gradients and angular momenta across the gmcs . like the milky way and m33 , there is significantly more angular momentum in possible progenitor material than in the resulting gmcs , though the discrepancy is smallest in m31 owing to the relatively low galactic shear at large galactocentric radii . all three systems suggest braking mechanisms act during the process of cloud formation to dissipate angular momentum . \3 . the galactic environment in m31 where the gmcs are found is similar in most respects that of the solar neighborhood and inner milky way . that gmcs in both environments are similar is consistent with the idea that the galactic environment regulates gmc properties . in systems where the galactic environment differs significantly from those studied here , the gmc properties are found to be different ( e.g. , * ? ? ? this work was supported by a nasa gsrp fellowship and an nsf postdoctoral fellowship ( ast-0502605 ) . i am extremely grateful to rainer beck , leo blitz , michel gulin , adam leroy , chris mckee , eve ostriker and an anonymous referee for carefully reading the manuscript and offering in - depth commentary . kartik sheth kindly discussed his current and future work on m31 , allowing me to read a preprint of a forthcoming paper . this work made extensive use of the nasa extragalactic database , the nasa abstract data service , and the mast archive . the molecular line observations used the bima millimeter array which was supported , in part , by nsf grant ast-0228963 to the radio astronomy laboratory at u.c . berkeley . , k. , vogel , s. n. , wilson , c. d. , & dame , t. m. 2000 , in the interstellar medium in m31 and m33 . proceedings 232 . we - heraeus seminar , 22 - 25 may 2000 , bad honnef , germany . edited by elly m. berkhuijsen , rainer beck , and rene a. m. walterbos . shaker , aachen , 2000 , p. 37 - 40 , 3740 , w. j. , thornton , d. d. , plambeck , r. l. , wright , m. c. h. , lugten , j. , urry , l. , fleming , m. , hoffman , w. , hudson , j. , lum , w. t. , forster , j. . r. , thatte , n. , zhang , x. , zivanovic , s. , snyder , l. , crutcher , r. , lo , k. y. , wakker , b. , stupar , m. , sault , r. , miao , y. , rao , r. , wan , k. , dickel , h. r. , blitz , l. , vogel , s. n. , mundy , l. , erickson , w. , teuben , p. j. , morgan , j. , helfer , t. , looney , l. , de gues , e. , grossman , a. , howe , j. e. , pound , m. , & regan , m. 1996 , , 108 , 93

we present berkeley illinois maryland association ( bima ) millimeter interferometer observations of giant molecular clouds ( gmcs ) along a spiral arm in m31 . the observations consist of a survey using the compact configuration of the interferometer and follow - up , higher - resolution observations on a subset of the detections in the survey . the data are processed using an analysis algorithm designed to extract gmcs and correct their derived properties for observational biases thereby facilitating comparison with milky way data . the algorithm identifies 67 gmcs of which 19 have sufficient signal - to - noise to accurately measure their properties . the gmcs in this portion of m31 are indistinguishable from those found in the milky way , having a similar size - line width relationship and distribution of virial parameters , confirming the results of previous , smaller studies . the velocity gradients and angular momenta of the gmcs are comparable to the values measured in m33 and the milky way ; and , in all cases , are below expected values based on the local galactic shear . the studied region of m31 has a similar interstellar radiation field , metallicity , toomre @xmath0 parameter , and midplane volume density as the inner milky way , so the similarity of gmc populations between the two systems is not surprising .

recently , there is increasing evidence of the excess of the 130gev gamma - ray line from the galactic center ( gc ) in the four - year fermi data @xcite . this may be interpreted as a signal of the dark matter ( dm ) , which annihilates or decays around the gc . an obstacle to construct a model of the annihilating / decaying dm which explains the observed gamma - ray line excess is that the branching ratio of the monochromatic photon production must be fairly large . it should be larger than around 0.01 @xcite . otherwise , continuum gamma - rays would hide the line gamma , and anti - protons may be overproduced . for instance , if the dm annihilation into photons takes place through loops of the standard model ( sm ) particles , it is difficult to realize such a large branching ratio @xcite . in this letter , we propose a model of the decaying dm which naturally explains the gamma - ray line excess without producing too much continuum gammas and anti - protons . a supersymmetric ( susy ) axion model @xcite is considered to solve the strong cp problem in the framework of the minimal susy sm ( mssm ) . the axino , which is a fermionic superpartner of the axion , is a suitable candidate of the dm , if it is the lightest susy particle ( lsp ) . by introducing small r - parity violations , the axino decays into a photon plus a neutrino , and the fermi gamma - ray line excess can be explained . it is stressed that the branching fraction of the axino decay into monochromatic photons typically becomes @xmath7 , and the constraints from the overproductions of the continuum gamma - ray and the antiproton are satisfied . this is in contrast to the decaying gravitino dm scenario , where the branching fraction of the monochromatic photon production is suppressed @xcite . moreover , the present scenario is cosmologically favored , because the lightest susy particle of the mssm ( mssm - lsp ) , e.g. , the lightest neutralino , decays by the r - parity violating effects before the big - bang nucleosynthesis ( bbn ) begins . this avoids the cosmological problem associated with a late decay of the mssm - lsp when the gravitino is lighter than the mssm - lsp . on the other hand , the morphology of the gamma - ray line signature from the gc seems to favor the annihilating dm scenario rather than that of the decaying dm @xcite . although relatively large gamma - ray signals are expected from the galactic halo in the decaying dm scenario , no such excesses have been observed . however , since there are potentially large uncertainties in the gamma - ray data and the dm density profile around the gc , it is premature to specify the dm model by the morphology @xcite . in the next section , the axino dm model will be introduced , and properties of the model will be explained , particularly paying attention to the r - parity violating effects . we consider the ksvz axion models @xcite . it will be shown that the model can explain the gamma - ray line excess . in addition , several cosmological aspects will be discussed . the last section will be devoted to the conclusion and discussion . let us first introduce r - parity violations . in this letter , we consider a bilinear type of the r - parity violation @xcite , which is characterized by the superpotential , @xmath8 where @xmath9 and @xmath10 are chiral superfields of the lepton doublet and the up - type higgs doublet , respectively . the index @xmath11 denotes the generation , and @xmath12 is a parameter with a mass dimension . here and hereafter , summation over @xmath13 is implicitly promised . by redefining @xmath9 and the down - type higgs superfield @xmath14 as @xmath15 and @xmath16 with @xmath17 , where @xmath18 is the higgsino mass parameter appearing in the superpotential as @xmath19 , the r - parity violating superpotential ( [ w_bi ] ) is eliminated . hereafter , for notational simplicity , the primes on the redefined fields are omitted . after the redefinition , the susy breaking potential becomes @xmath20 where @xmath21 is a scalar component of the superfield @xmath9 . the coefficients are @xmath22 and @xmath23 , where @xmath24 , @xmath25 and @xmath26 represent soft susy breaking parameters in the mssm , @xmath27 . due to the r - parity violating scalar potential ( [ lrpv ] ) , sneutrinos obtain non - zero vacuum expectation values ( vevs ) as @xmath28 where @xmath29 is a ratio of the vevs of the up- and down - type higgs fields , @xmath30gev , and @xmath31 is a sneutrino mass . before proceeding to discuss phenomenological aspects , several comments are in order . it is possible to introduce the bilinear r - parity violating soft terms , @xmath32 and @xmath33 in addition to , before the field redefinition . the coefficients in then have additional contributions , but the following analysis will not be affected as far as the r - parity violation is parametrized by the the sneutrino vev . next , trilinear r - parity violating terms , @xmath34 and @xmath35 , are also generated by the field redefinition . they are subdominant and will be ignored in the following study , because the terms are multiplied by the yukawa couplings . the sneutrino vevs ( [ nuvev ] ) induce mixings between the sm leptons and the gauginos . the sm neutrinos mix with the bino and the neutral wino , and the sm charged leptons mix with the charged winos . hence , the r - parity violating parameters are constrained . the neutrinos obtain masses of @xmath36 , where @xmath37 is a bino ( wino ) mass @xcite . for gaugino masses of @xmath38gev , @xmath39 is imposed to satisfy the experimental bound on the neutrino masses . also , the cosmological @xmath40 asymmetry is preserved for @xmath41 @xcite . other constraints are known to be weaker ( see e.g. , ref . as we will see , the size of the r - parity violation favored by the fermi gamma - ray line excess is much smaller as @xmath42 . the r - parity violation destabilizes the lsp . in this letter , we consider the axino lsp scenario in the ksvz axion models @xcite . the relevant interaction terms of the axino are @xmath43\tilde{b}b_{\mu \nu } + i \frac{\alpha_w c_w}{16 \pi f_a } \bar{\tilde{a}}\gamma_5[\gamma^{\mu},\gamma^{\nu}]\tilde{w}^aw^a_{\mu \nu } \label{ayy}\ ] ] where @xmath44 and @xmath45 are model - dependent coupling constants of order unity , @xmath46 is the fine structure constant of u(1)@xmath47 , @xmath48 is that of su(2 ) , @xmath49 is the pq scale , @xmath50 denotes the axino , @xmath51 is the bino ( wino ) , and @xmath52 is the field strength of the u(1)@xmath47 ( su(2 ) ) gauge boson . the axino lsp is stable as long as the r - parity is conserved , whereas it decays via the operators with the gaugino mixings with the sm leptons , once the r - parity violation is turned on . first , let us consider the case of @xmath53 ( see sec . [ sec : model ] for an explicit realization ) . the first term in ( [ ayy ] ) provides interactions of @xmath54 and @xmath55 . the r - parity violation opens a decay of the axino through the @xmath56 mixing as @xmath57 and @xmath58 . in the limit of @xmath59 , the axino decay rate becomes @xmath60 where @xmath61 is the axino mass , and @xmath62 is the weak mixing angle . here and hereafter , @xmath63 denotes a sum of the partial decay rates into @xmath64 and @xmath65 . the factor @xmath66 in the parenthesis comes from the bino - neutrino mixing , and @xmath67 from the u(1)@xmath47 gauge boson - photon mixing . similarly , we obtain @xmath68 where @xmath69 is the mass of the @xmath70 boson . for @xmath71 , the branching fractions are given by @xmath72 . from the above results , the axino lifetime is estimated as @xmath73 where the r - parity violating parameter is defined as @xmath74 . the two - body decay of the axino into a photon contributes to the monochromatic gamma signal of the fermi observation . if the axino mass is around 260gev , the photon produced by @xmath75 has an energy of about 130gev . according to ref . @xcite , the observed excess of the gamma - ray line is accounted for by a decaying dm , when its lifetime @xmath76 and the branching ratio @xmath77 are in the range of @xmath78sec and @xmath79 . the astrophysical constraints from the diffuse gamma - rays @xcite and neutrinos @xcite are also satisfied for such a parameter region . in the present model , the branching ratio is around 0.8 for @xmath53 , while the hadronic branch from @xmath80 is sufficiently small . thus , the lifetime and the branching fraction which are required to explain the gamma - ray line excess from the gc are realized by the axino dm . next , let us focus on the case of @xmath81 and neglect the contribution from the first term in ( [ ayy ] ) , i.e. , @xmath82 . the second term in ( [ ayy ] ) provides interactions of @xmath83 , @xmath84 and @xmath85 . the decays , @xmath86 and @xmath87 , proceed by these interactions with the mixings of @xmath88 and @xmath89 . in the limit of @xmath90 , the decay rate of the axino into a photon and a neutrino is given by @xmath91 where the factor @xmath92 in the parenthesis is the mixing between the wino and the neutrino , and @xmath93 the mixing between @xmath94 and the photon . similarly , the decay rate of @xmath95 becomes @xmath96 while that of @xmath97 , which is a sum of the rates of @xmath98 and @xmath99 , is @xmath100 where @xmath101 is the mass of the @xmath102 boson , and the factor @xmath103 represents the mixing between the charged wino and the lepton . thus , we obtain @xmath104 for @xmath105 . this results in the branching fraction of @xmath106 of around @xmath107 . in the case of @xmath108 , both the first and second terms in eq . contribute to the axino decay . the decay rates in such a generic case are summarized in app . [ sec : decay_rate ] . as shown there , the branching fractions are determined by a combination of @xmath109 for @xmath110 . [ fig : br_mass ] shows the branching ratios of the axino decay into @xmath111 , @xmath112 and @xmath113 as a function of @xmath114 for @xmath115 ( left ) and as a function of @xmath116 for @xmath117 ( right ) . one can confirm that the branching ratio of @xmath106 becomes @xmath118 for @xmath119 , while it becomes @xmath107 for large @xmath109 . in the intermediate regime , the branching ratio of @xmath106 decreases due to an interference effect and eventually vanishes at @xmath120 [ cf . ( [ eq : decay_limit ] ) ] . in most of the parameter space , however , @xmath121 and hence the model can explain the gamma - line without overproducing continuum gamma - ray and antiprotons . -1.cm into @xmath111 ( red ) , @xmath112 ( green ) and @xmath113 ( blue ) as a function of @xmath114 for @xmath122 ( left ) and as a function of @xmath123 for @xmath124 ( right ) . we assume @xmath125gev and @xmath126.,title="fig : " ] into @xmath111 ( red ) , @xmath112 ( green ) and @xmath113 ( blue ) as a function of @xmath114 for @xmath122 ( left ) and as a function of @xmath123 for @xmath124 ( right ) . we assume @xmath125gev and @xmath126.,title="fig : " ] -0.1 cm here , we briefly describe an explicit model of the susy axion . let us introduce pq superfields , @xmath127 and @xmath128 , with pq charges of @xmath129 and @xmath130 , respectively . also , pq quarks , @xmath131 and @xmath132 , are added , which have fundamental and anti - fundamental representations of the sm su(3 ) , respectively , and both of which have a pq charge of @xmath133 . the superpotential is given by @xmath134 where @xmath135 is a singlet superfield , @xmath136 and @xmath137 are coupling constants , and @xmath138 is a constant term with the gravitino mass @xmath139 and the reduced planck scale @xmath140 . the coupling constants are taken to be real and positive . including the susy breaking terms , @xmath141 and @xmath142 , the relevant terms of the scalar potential are @xmath143 where we have assumed the minimal khler potential , for simplicity . the vevs of the pq scalars are given by @xmath144 , which is related to the pq scale @xmath49 as @xmath145 , and the pq quarks obtain a mass of @xmath146 . the axion , which is a goldstone boson associated with the vevs of the pq scalars , has an anomaly - induced coupling to the gluon via pq quark loops , because the pq symmetry is anomalous under the qcd . thus it solves the strong cp problem . the coupling constants , @xmath44 and @xmath45 , depend on assignments of the u(1)@xmath47 and su(2 ) charge on the pq quarks . if they are a singlet under su(2 ) but have ( opposite ) u(1)@xmath47 charges , we obtain @xmath53 and @xmath147 . if they are embedded in the su(5 ) representation , both @xmath44 and @xmath45 are nonzero and satisfy @xmath115 . for instance , if the pq quarks are embedded in @xmath148 of su(5 ) , the coefficients become @xmath149 . in this model , the axino , that is a fermionic component of a linear combination of @xmath127 and @xmath128 , obtains a mass of @xmath150 , where the vev of @xmath135 is derived from . several additional effects can make the axino heavier or lighter . there can be other susy breaking contributions to the tadpole term of @xmath135 in ( [ vpq ] ) , which change the vev of @xmath135 and hence the axino mass . radiative correction from the @xmath151 loops can also modify the axino mass @xcite . in this letter , we assume that these effects slightly reduce the axino mass , and the axino becomes the lsp . in this section , we discuss several cosmological constraints on the decaying axino dm scenario . let us assume that the lightest neutralino is mostly composed of the bino and that it is the mssm - lsp . in the presence of the r - parity violation , the bino decays into @xmath152 and @xmath153 due to the sneutrino vev . the decay rate of the bino is given by @xcite @xmath154 in order for the bino decay not to disturb the bbn , i.e. , for the bino lifetime shorter than 0.1sec , we need @xmath155 for @xmath156 . the bino also decays into the axino through the r - parity conserving operators ( [ ayy ] ) as @xmath157 and @xmath158 . if this dominates the bino decays , the produced axinos may exceed the observed dm abundance . in order to avoid the axino overproduction , the production rate should be much less than @xmath159 . the decay rate of the bino into axinos with photons or @xmath70 bosons is totally given by @xcite @xmath160 the axino abundance produced by the bino decay in terms of the density parameter @xmath161 , where @xmath162 is the present energy density of the axino and @xmath163 is the present critical energy density , becomes @xmath164 where @xmath165 is the present hubble parameter in units of 100km / s / mpc , and @xmath166 is the bino abundance after the thermal decoupling evaluated as if the bino were stable . since the bino abundance is large in general , the axion abundance becomes too large , unless the branching ratio of the bino decay into the axino is suppressed , i.e. , @xmath167 . axinos are produced by scatterings of the gluons and the gluinos from the thermal bath at the reheating . the thermally produced axino abundance is given by @xcite @xmath168 where @xmath169 is the running qcd coupling constant at the @xmath170 scale , and @xmath170 is the reheating temperature after the inflation . this is valid as long as @xmath170 is larger than the axino mass . thus , the axino can be a dominant component of the dm for @xmath171 . the abundance of the axion coherent oscillation is estimated as @xcite @xmath172 where @xmath173 denotes the axion initial misalignment angle . for @xmath174gev , we need @xmath175 in order for the axion abundance to be lower than the dm abundance . the saxion , @xmath176 , belongs to a flat direction in the scalar potential ( [ vpq ] ) , which satisfies @xmath177 . it obtains a mass , @xmath178 , from the susy breaking effect . let us estimate the saxion abundance . the saxion sits around the minimum during the inflation , which is slightly displaced from the low - energy true minimum , and begins to oscillate around the true minimum when the hubble parameter decreases to @xmath178 with an initial amplitude of @xmath179 . the abundance of saxion coherent oscillation is @xmath180 where @xmath181 is the saxion energy density , and @xmath182 is the entropy density . here , we have assumed that the saxion oscillation starts before the reheating process of the inflation is finished , which is the case for @xmath183 . the saxion dominantly decays into the axion pair . the lifetime becomes @xcite @xmath184 with @xmath185 . note that the saxion also decays into a pair of the gluons with a branching fraction of @xmath186 @xcite , but it does not affect the bbn as long as @xmath187gev is satisfied for @xmath188gev .. this indicates @xmath189gev . ] the axions produced by the saxion decay contribute to the extra effective number of the neutrino species , @xmath190 @xcite , which is given by @xmath191^{1/3 } \left(\frac{\rho_\sigma}{\rho_{\rm rad}}\right)_{t = t_\sigma},\ ] ] where @xmath192 with @xmath193 denoting the temperature at which the saxion decays and @xmath194 being the radiation energy density . it is estimated as @xmath195 the contribution should satisfy @xmath196 . in other words , the recent claims of the existence of the extra light species , @xmath197 @xcite , can be explained by the non - thermal axions from the saxion decay . we assume @xmath198 for simplicity . and @xmath49 for the bino mssm - lsp with @xmath199 , @xmath200 , @xmath201gev , @xmath202tev , @xmath203gev and @xmath204 . in the light blue band , the fermi 130gev gamma - ray line excess is explained . on the right side of the black dashed line , the bino lifetime is shorter than @xmath205sec . above the orange dot - dashed line , the axino abundance produced by the bino decay satisfies @xmath206 , where @xmath207 is taken as a reference . for each @xmath49 , @xmath170 is set so that the axino becomes the dominant component of the dm . below the red horizontal line , @xmath208 is obtained from the axions produced by the saxon decay . above the blue dotted horizontal line , @xmath209 is satisfied to account for the axino dm . ( bottom ) same as the top panel , but for the wino mssm - lsp with @xmath210 , @xmath211 and @xmath212tev.,title="fig : " ] and @xmath49 for the bino mssm - lsp with @xmath199 , @xmath200 , @xmath201gev , @xmath202tev , @xmath203gev and @xmath204 . in the light blue band , the fermi 130gev gamma - ray line excess is explained . on the right side of the black dashed line , the bino lifetime is shorter than @xmath205sec . above the orange dot - dashed line , the axino abundance produced by the bino decay satisfies @xmath206 , where @xmath207 is taken as a reference . for each @xmath49 , @xmath170 is set so that the axino becomes the dominant component of the dm . below the red horizontal line , @xmath208 is obtained from the axions produced by the saxon decay . above the blue dotted horizontal line , @xmath209 is satisfied to account for the axino dm . ( bottom ) same as the top panel , but for the wino mssm - lsp with @xmath210 , @xmath211 and @xmath212tev.,title="fig : " ] the combined constraints on a plane of @xmath213 are shown in fig . [ fig : cont ] . in the top panel , we have taken @xmath199 , @xmath200 , @xmath201gev , @xmath202tev , @xmath203gev and @xmath204 . in the light blue band , the fermi 130gev gamma - ray line excess is explained . on the right side of the black dashed line , the bino lifetime is shorter than @xmath205sec , and the decay has no effects on the bbn . above the orange dot - dashed line , we have @xmath206 , and the axino abundance produced by the bino decay is sufficiently small . here , @xmath207 is taken as a reference value of the bino abundance as inferred from ref . @xcite . below the red horizontal line , @xmath208 is obtained from the axions produced by the saxion decay . in the figure , @xmath170 is set so that @xmath214 is realized for each @xmath49 . above the blue dotted horizontal line , @xmath209 and the axino dm is thermally produced . in the bottom panel , we have taken @xmath210 and @xmath211 and assumed that the wino is the mssm - lsp with its mass of @xmath212tev . note that since the thermal relic wino abundance is small , there is no constraint from the axino overproduction by the wino decay . in both cases , it is found that the fermi 130gev gamma - ray line excess is accounted for without suffering from the cosmological constraints for @xmath215 and @xmath216gev . here we briefly mention parameter dependences on these constraints . for larger neutralino mass , the axino lifetime becomes longer and the light - blue band moves to the bottom - right . also , the axino abundance from the neutralino decay ( [ omega_a_b ] ) becomes larger and the orange dot - dashed line moves to the top - right . for smaller saxion mass or larger initial amplitude of the saxion , the bound from @xmath217 ( [ neff ] ) becomes stronger and the red line moves to the bottom . for general values of @xmath45 and @xmath44 , the branching ratio of the axino decay into photon may become smaller , as shown in fig . [ fig : br_mass ] , and hence the light - blue band moves to the bottom - right to make the axino lifetime smaller . before closing this section , let us comment on the gravitino . the gravitinos are also produced by scatterings of the gluons and the gluinos from the thermal bath at the reheating . if the gravitino is lighter than the mssm - lsp , then it dominantly decays into the axino and the axion , and hence there is no bbn constraint @xcite . the nonthermal production of axinos by the gravitino decay is negligible , since the abundance of the thermally produced gravitino is less than that of the axino . on the other hand , if the gravitino is heavier than the mssm - lsp , its decay affects the bbn . the parameter range corresponding to @xmath218gev ( just below the red horizontal line in fig . [ fig : cont ] ) is constrained depending on the gravitino mass and the mssm mass spectrum @xcite . we have proposed the decaying axino dm scenario as a model to explain the fermi 130gev gamma - ray line excess from the gc . it is based on the susy ksvz axion model with the bilinear r - parity violation . the model realizes a fairly large branching fraction of the axino decay into a photon plus a neutrino . it was found that the fermi excess is accounted for while satisfying the other cosmological constraints for @xmath216gev and @xmath215 . compared to another well motivated decaying dm , i.e. , the decaying gravitino dm , the decaying axino dm typically has a larger branching fraction into the monochromatic gamma . the gravitino universally couples to the lepton and the higgs superfields , and hence the gravitino s decay into @xmath219 , @xmath220 and @xmath113 can not be suppressed in the presence of the bilinear r - parity violation . thus , the decaying gravitino dm is severely constrained by the observation of the antiproton flux @xcite . let us touch on the collider phenomenology . the mssm - lsp is stable in the detectors for the r - parity violation of @xmath221 ( see e.g. , ref . thus , when the mssm - lsp is neutral , the susy events would be detected by searching for signals with a large missing transverse momentum . if the mssm - lsp is a charged particle , it leaves a charged track , which is a characteristic signal in the detector . the sensitivities of the lhc searches are the same as the standard susy searches . as mentioned in the introduction , the morphology of the observed gamma - ray signature is still premature to refute the decaying dm scenario . if the uncertainties will be understood in future , such an analysis can be used for distinguishing the dm models particularly between the decaying and annihilating dm models . if the former scenario will become favored , the decaying axino dm model can be an attractive candidate . this work was supported by jsps kakenhi grant no . 23740172 ( m.e . ) , no . 21740164 ( k.h . ) , no . 22244021 ( k.h . ) , no . 22244030 ( k.n . ) and by the mext grant - in - aid no . 21111006 ( k.n . ) . the work of k.m . is supported in part by jsps research fellowships for young scientists . this work was supported by world premier international research center initiative ( wpi initiative ) , mext , japan . in this appendix , we provide the general formulae for the axino decay rate in the presence of the r - parity violation . the interaction between the axion and the su(2)@xmath222u(1 ) gauge supermultiplets is given by @xmath223 where @xmath224 stands for the fine structure constant of u(1)@xmath225 ( su(2 ) ) , @xmath49 is the pq scale , @xmath226 is the axion superfield , @xmath227 , and @xmath228 ( @xmath229 ) is the supersymmetric field strength of u(1 ) ( su(2 ) ) . the coefficients @xmath44 and @xmath45 are model - dependent constants of order unity . in terms of the component fields , the axino interactions become @xmath230\tilde{b}b_{\mu \nu } + i \frac{\alpha_w c_w}{16 \pi f_a } \bar{\tilde{a}}\gamma_5[\gamma^{\mu},\gamma^{\nu}]\tilde{w}^aw^a_{\mu \nu } , \label{eq : int}\end{aligned}\ ] ] where @xmath50 is the axino , @xmath231 ( @xmath232 ) is the gauge boson of u(1)@xmath47 ( su(2 ) ) , and @xmath233 and @xmath234 denote the bino and the wino , respectively . due to the sneutrino vev induced by the bilinear r - parity violation , the sm leptons and the gauginos mix with each other , and the axino lsp , @xmath50 , can decay into @xmath111 , @xmath112 and @xmath113 . the mass matrix of the neutralino and the neutrino becomes @xmath235 for @xmath236 with @xmath237 . here , the subscript @xmath238 stands for the generation , @xmath239 , @xmath240 , @xmath241 and @xmath242 with the weak mixing angle @xmath62 . the upper - left @xmath243 matrix is identical to the neutralino mass matrix of the mssm , and the neutrino masses are approximated to be zero . on the other hand , the mass matrix of the chargino and the lepton is given by @xmath244 for @xmath245 with @xmath246 and @xmath247 . here , @xmath248 is the lepton yukawa coupling constant , and the upper - left @xmath249 matrix is identical to the chargino mass matrix of the mssm . neglecting the small sm lepton masses in the phase space , one finds the decay rates as @xmath250 with @xmath251 \left| - \alpha_y c_y \sin \theta_w u _ { \nu_i \tilde b } + \alpha_w c_w \cos \theta_w u _ { \nu_i \tilde w } \right|^2&\mbox{for}~(g , l)=(z , \nu_i ) , \\[5pt ] \alpha_w^2 c_w^2 \ ( \left| u_{l_i \tilde w}^l \right|^2 + \left| u_{l_i \tilde w}^r \right|^2 \ ) . & \mbox{for}~(g , l ) = ( w , l_i ) . \end{cases } \label{eq : decay_each}\end{aligned}\ ] ] where @xmath252 is a mass of the gauge boson , and @xmath253 , @xmath254 and @xmath255 are the mixings that diagonalize the mass matrices of eqs . ( [ eq : mass_n ] ) and ( [ eq : mass_c ] ) . in the limit of @xmath126 , the mixings between the gauginos and the higgsinos become irrelevant . assuming that the r - parity violation is small , @xmath253 , @xmath254 and @xmath255 in eq . are approximated as @xmath256 \left| \alpha_y c_y \sin \theta_w \cfrac{g_y \ < \tilde \nu_i \>}{\sqrt{2 } m_{\tilde b } } + \alpha_w c_w \cos \theta_w \cfrac{g_2 \ < \tilde \nu_i \>}{\sqrt{2 } m_{\tilde w } } \right|^2 & \mbox{for}~(g , l ) = ( z,\nu_i ) , \\[20pt ] \alpha_w^2 c_w^2 \left| \cfrac{g_2 \ \nu_i \>}{m_{\tilde w } } \right|^2 & \mbox{for}~(g , l ) = ( w , l_i ) , \end{cases } \label{eq : decay_limit}\end{aligned}\ ] ] where @xmath257 is the sneutrino vev induced by the r - parity violation . as can be seen from , the branching fractions are determined by @xmath109 . when either @xmath45 or @xmath44 is sufficiently small , the decay rates are reduced to the expressions in sec . [ sec : r ] . if the pq quarks are embedded in a complete vector - like multiplet of the gut , they become @xmath115 . moreover , if the gut relation is assumed for the gaugino masses , they satisfy @xmath258 at the one - loop level of the renormalization group evolution . then , the decay rate of the axino to @xmath111 is found to vanish . this cancellation is seen in fig . [ fig : br_mass ] . 99 t. bringmann , x. huang , a. ibarra , s. vogl and c. weniger , jcap * 1207 * ( 2012 ) 054 [ arxiv:1203.1312 [ hep - ph ] ] . c. weniger , jcap * 1208 * , 007 ( 2012 ) [ arxiv:1204.2797 [ hep - ph ] ] . e. tempel , a. hektor and m. raidal , jcap * 1209 * , 032 ( 2012 ) [ addendum - ibid . * 1211 * , a01 ( 2012 ) ] [ arxiv:1205.1045 [ hep - ph ] ] . a. boyarsky , d. malyshev and o. ruchayskiy , arxiv:1205.4700 [ astro-ph.he ] . m. su and d. p. finkbeiner , arxiv:1206.1616 [ astro-ph.he ] . f. aharonian , d. khangulyan and d. malyshev , arxiv:1207.0458 [ astro-ph.he ] . a. hektor , m. raidal and e. tempel , arxiv:1207.4466 [ astro-ph.he ] . a. hektor , m. raidal and e. tempel , arxiv:1209.4548 [ astro-ph.he ] . d. p. finkbeiner , m. su and c. weniger , arxiv:1209.4562 [ astro-ph.he ] . w. buchmuller and m. garny , jcap * 1208 * , 035 ( 2012 ) [ arxiv:1206.7056 [ hep - ph ] ] . t. cohen , m. lisanti , t. r. slatyer and j. g. wacker , jhep * 1210 * , 134 ( 2012 ) [ arxiv:1207.0800 [ hep - ph ] ] . i. cholis , m. tavakoli and p. ullio , phys . d * 86 * , 083525 ( 2012 ) [ arxiv:1207.1468 [ hep - ph ] ] . m. asano , t. bringmann , g. sigl and m. vollmann , arxiv:1211.6739 [ hep - ph ] . for a review , see m. kawasaki and k. nakayama , arxiv:1301.1123 [ hep - ph ] . h. -b . kim and j. e. kim , phys . b * 527 * , 18 ( 2002 ) [ hep - ph/0108101 ] . d. hooper and l. -t . wang , phys . d * 70 * , 063506 ( 2004 ) [ hep - ph/0402220 ] . e. j. chun and h. b. kim , jhep * 0610 * , 082 ( 2006 ) [ hep - ph/0607076 ] . j. hasenkamp and j. kersten , phys . lett . b * 701 * , 660 ( 2011 ) [ arxiv:1103.6193 [ hep - ph ] ] . j. -c . park and s. c. park , arxiv:1207.4981 [ hep - ph ] . j. e. kim , phys . * 43 * , 103 ( 1979 ) ; m. a. shifman , a. i. vainshtein , v. i. zakharov , nucl . * b166 * , 493 ( 1980 ) . j. c. romao , m. a. diaz , m. hirsch , w. porod and j. w. f. valle , phys . d * 61 * ( 2000 ) 071703 [ arxiv : hep - ph/9907499 ] . f. takayama and m. yamaguchi , phys . b * 476 * ( 2000 ) 116 , [ arxiv : hep - ph/9910320 ] . m. hirsch and j. w. f. valle , new j. phys . * 6 * ( 2004 ) 76 . [ arxiv : hep - ph/0405015 ] b. a. campbell , s. davidson , j. r. ellis and k. a. olive , phys . b * 256 * ( 1991 ) 457 . w. fischler , g. f. giudice , r. g. leigh and s. paban , phys . b * 258 * ( 1991 ) 45 . h. k. dreiner and g. g. ross , nucl . b * 410 * ( 1993 ) 188 [ hep - ph/9207221 ] . m. endo , k. hamaguchi and s. iwamoto , jcap * 1002 * , 032 ( 2010 ) [ arxiv:0912.0585 [ hep - ph ] ] . w. buchmuller , l. covi , k. hamaguchi , a. ibarra and t. yanagida , jhep * 0703 * , 037 ( 2007 ) [ hep - ph/0702184 [ hep - ph ] ] . m. endo and t. shindou , jhep * 0909 * , 037 ( 2009 ) [ arxiv:0903.1813 [ hep - ph ] ] . s. bobrovskyi , w. buchmuller , j. hajer and j. schmidt , jhep * 1010 * , 061 ( 2010 ) [ arxiv:1007.5007 [ hep - ph ] ] . t. f. - : m. ackermann _ et al . _ [ lat collaboration ] , astrophys . j. * 761 * , 91 ( 2012 ) [ arxiv:1205.6474 [ astro-ph.co ] ] . l. covi , m. grefe , a. ibarra and d. tran , jcap * 1004 * , 017 ( 2010 ) [ arxiv:0912.3521 [ hep - ph ] ] . t. goto and m. yamaguchi , phys . b * 276 * , 103 ( 1992 ) . k. ishiwata , s. matsumoto and t. moroi , phys . d * 78 * , 063505 ( 2008 ) [ arxiv:0805.1133 [ hep - ph ] ] . l. covi , h. -b . kim , j. e. kim and l. roszkowski , jhep * 0105 * ( 2001 ) 033 [ hep - ph/0101009 ] . a. brandenburg and f. d. steffen , jcap * 0408 * , 008 ( 2004 ) [ hep - ph/0405158 ] . a. strumia , jhep * 1006 * , 036 ( 2010 ) [ arxiv:1003.5847 [ hep - ph ] ] . t. asaka and t. yanagida , phys . b * 494 * , 297 ( 2000 ) [ hep - ph/0006211 ] . e. j. chun and a. lukas , phys . b * 357 * , 43 ( 1995 ) [ hep - ph/9503233 ] . k. ichikawa , m. kawasaki , k. nakayama , m. senami and f. takahashi , jcap * 0705 * , 008 ( 2007 ) [ hep - ph/0703034 [ hep - ph ] ] . k. choi , e. j. chun and j. e. kim , phys . b * 403 * , 209 ( 1997 ) [ hep - ph/9608222 ] . e. j. chun , d. comelli and d. h. lyth , phys . d * 62 * , 095013 ( 2000 ) [ hep - ph/0008133 ] . m. kawasaki , n. kitajima and k. nakayama , phys . d * 83 * , 123521 ( 2011 ) [ arxiv:1104.1262 [ hep - ph ] ] . m. kawasaki , k. miyamoto , k. nakayama and t. sekiguchi , jcap * 1202 * , 022 ( 2012 ) [ arxiv:1107.4962 [ astro-ph.co ] ] . k. s. jeong and f. takahashi , jhep * 1208 * , 017 ( 2012 ) [ arxiv:1201.4816 [ hep - ph ] ] . t. moroi and m. takimoto , phys . b * 718 * , 105 ( 2012 ) [ arxiv:1207.4858 [ hep - ph ] ] . k. choi , k. -y . choi and c. s. shin , phys . d * 86 * , 083529 ( 2012 ) [ arxiv:1208.2496 [ hep - ph ] ] . z. hou , c. l. reichardt , k. t. story , b. follin , r. keisler , k. a. aird , b. a. benson and l. e. bleem _ et al . _ , arxiv:1212.6267 [ astro-ph.co ] . m. kawasaki , k. kohri , t. moroi and a. yotsuyanagi , phys . d * 78 * , 065011 ( 2008 ) [ arxiv:0804.3745 [ hep - ph ] ] . s. asai , y. azuma , m. endo , k. hamaguchi and s. iwamoto , jhep * 1112 * , 041 ( 2011 ) [ arxiv:1103.1881 [ hep - ph ] ] .

we show that decaying axino dark matter with r - parity violation can explain the observed excess of the 130gev gamma - ray line from the galactic center in the fermi data . the branching fraction of the axino decay into monochromatic photons can be @xmath0 , and constraints from continuum gamma - rays and the anti - proton flux are ameliorated . the peccei - quinn scale of @xmath1gev and the r - parity violation parameter of @xmath2 are cosmologically favored . .5 in ut-13 - 03 + ipmu 13 - 0024 * axino dark matter with r - parity violation and + 130 gev gamma - ray line * .5 in motoi endo@xmath3 , koichi hamaguchi@xmath3 , seng pei liew@xmath4 , + kyohei mukaida@xmath4 and kazunori nakayama@xmath3 .3 in _ @xmath5department of physics , university of tokyo , bunkyo - ku , tokyo 113 - 0033 , japan + @xmath6kavli institute for the physics and mathematics of the universe , university of tokyo , kashiwa 277 - 8583 , japan + _ .5 in

hidden in batse s superb gamma - ray burst lightcurves in different energy bands are temporal and spectral signatures of the fundamental physical processes which produced the observed emission . various techniques have been applied to the batse data to extract these signatures , such as : auto- and crosscorrelations of lightcurves in different energies@xcite ; fourier transforms@xcite ; lightcurve averaging@xcite ; cross - fourier transforms@xcite and pulse fitting@xcite . here we propose to use linear state space models ( lssm ) to study the gamma - ray burst lightcurves . lssm estimates a time series underlying autoregressive ( ar ) process in the presence of observational noise . an ar process assumes that the real time series is a linear function of its past values ( `` autoregression '' ) in addition to `` noise , '' a stochastic component of the process . since the noise adds information to the system , it is sometimes called the `` innovation''@xcite . a moving average of the previous noise terms is equivalent to autoregression , and therefore these models are often called arma ( autoregressive , moving average ) processes@xcite . while arma processes are simply mathematical models of a time series , the resulting model can be interpreted physically , which is the purpose of their application to astrophysical systems . for example , the noise may be the injection of energy into an emission region , while the autoregression may be the response of the emission region to this energy injection , such as exponential cooling . the application of lssm to burst lightcurves can be viewed as an exploration of burst phenomenology devoid of physical content : how complicated an ar process is necessary to model burst lightcurves ? can all bursts be modeled with the same ar process ? however , because different types of ar processes can be interpreted as the response of a system to a stochastic excitation , characterizing bursts in terms of ar processes has physical implications . since we have lightcurves in different energy bands , we can compare the response at different energies . for example , the single coefficient in the ar[1 ] process ( the nomenclature is described below ) is a function of an exponential decay constant . if the lightcurves in all energy bands can be modeled by ar[1 ] then we have decay constants for every energy band . since most bursts undergo hard - to - soft spectral evolution@xcite and temporal structure is narrower at high energy than at low energy@xcite , we expect the decay constants to be shorter for the high energy bands . the purpose of the lssm methodology is to recover the hidden ar process . if the time series @xmath0 is an ar[p ] process then @xmath1 where time is assumed to advance in integral units . the `` noise '' ( or `` innovation '' ) @xmath2 is uncorrelated and possesses a well - defined variance @xmath3 ; the noise is usually assumed to be gaussian . since the burst count rate can not be negative , we expect the noise also can not be negative . a kolmogorov - smirnov test is used to determine when p is large enough to model the system adequately@xcite . if p=1 , the system responds exponentially to the noise with a decay constant @xmath4 , and @xmath5 the p=2 system is a damped oscillator with period @xmath6 and relaxation time @xmath4 , @xmath7 thus , the lowest order ar processes lend themselves to obvious physical interpretations . unfortunately , we do not detect @xmath0 directly , but a quantity @xmath8 which is a linear function of @xmath0 and observational noise : @xmath9 where in our case @xmath10 is an irrelevant multiplicative factor and @xmath11 is a zero - mean noise term with variance @xmath12 ; @xmath11 is also often assumed to be gaussian . the lssm code uses the expectation - maximization algorithm@xcite . we have thus far applied our lssm code@xcite to 17 gamma - ray bursts . we used the 4-channel batse lad discriminator lightcurves extracted from the discsc , preb , and discla datatypes , which have 64 ms resolution ; the energy ranges are 2550 , 50100 , 100300 and 3002000 kev . each channel was treated separately , resulting in 68 lightcurves . of these lightcurves , 52 could be modeled by ar[1 ] , 13 by ar[2 ] and 3 by ar[4 ] . thus there is a preference for the simplest model , ar[1 ] . note that chernenko @xcite found an exponential response to a source function in their soft component . figure 1 presents the normalized relaxation time constants for the bursts in our sample , as well as their average . even for models more complicated that ar[1 ] a relaxation time constant can be identified . as expected , the averages of these time constants become shorter as the energy increases from channel 1 to channel 4 , consistent with the trend found in quantitative studies of spectral evolution@xcite and the qualitative inspection of burst lightcurves . in figure 2 we present the analysis of grb 940217 , the burst with an 18 gev photon 90 minutes after the lower energy gamma - ray emission ended@xcite . as can be seen , the residuals are much smaller than the model and are consistent with fluctuations around 0 ; plots for the data and the model are indistinguishable , and only one is presented . the amplitude of the residuals increases as the count rate increases ( attributable in part to counting statistics ) , but there is no net deviation from 0 . we plan to apply the lssm code to a large number of bursts . we will compare the order of the underlying ar process and the resulting coefficients obtained for the different energy lightcurves of the same burst and for different bursts . in this way we can search for hidden classes of bursts and explore the universality of the physical processes . the `` noise '' @xmath13 might be a measure of the energy supplied to the emission region ( although which physical processes are the noise and which the response is model dependent ) . therefore characterizing @xmath13 may probe a deeper level of the burst phenomenon . the @xmath13 lightcurves for the different energy bands should be related ; we expect major events to occur at the same time in all the energy bands , although the relative intensities may differ . many of the bursts consist of well - separated spikes or complexes of spikes . we will apply the lssm code to each part of the burst to determine whether the same order ar process characterizes the entire burst , and if so , whether the ar process has the same coefficients . this will test whether the physical processes remain the same during the burst . d. band s gamma - ray burst research is supported by the _ cgro _ guest investigator program and nasa contract nas8 - 36081 . band , d. l. , _ ap . j. _ * 486 * , 928 ( 1997 ) . shaviv , n. , ph.d . thesis , technion ( 1996 ) . mitrofanov , i. , , _ ap . j. _ * 459 * , 570 ( 1996 ) . kouveliotou , c. , , in _ gamma - ray bursts , aip conf . 265 _ , eds . w. paciesas and g. fishman ( new york : aip ) , 299 ( 1992 ) . norris , j. p. , , _ ap . j. _ * 459 * , 393 ( 1996 ) . scargle , j. d. , _ ap . j. suppl . _ * 45 * , 1 ( 1981 ) . ford , l. a. , , _ ap . j. _ * 439 * , 307 ( 1995 ) . fenimore , e. e. , , _ ap . j. lett . _ * 448 * , l101 ( 1995 ) . honerkamp , j. , _ stochastic dynamical systems _ , ( new york : vch publ . ) ( 1993 ) . knig , m. , and timmer , j. , _ astron . astrophys . _ * 124 * , 589 ( 1997 ) . chernenko , a. , , these proceedings ( 1998 ) . hurley , k. , , _ nature _ * 372 * , 652 ( 1994 ) .

linear state space modeling determines the hidden autoregressive ( ar ) process in a noisy time series ; for an ar process the time series current value is the sum of current stochastic `` noise '' and a linear combination of previous values . we present preliminary results from modeling a sample of 4 channel batse lad lightcurves . we determine the order of the ar process necessary to model the bursts . the comparison of decay constants for different energy bands shows that structure decays more rapidly at high energy . the resulting models can be interpreted physically ; for example , they may reveal the response of the burst emission region to the injection of energy .

chiral lagrangian for low lying pseudoscalar mesons@xcite@xcite as the most successful effective field theory is now widely used in various strong , weak and electromagnetic processes . to match the increasing demand for higher precision in low energy description of qcd , the applications of the low energy expansion of the chiral lagrangian is extended from early time discussions on the leading @xmath3 and next to leading @xmath1 orders to present @xmath0 order . for the latest review , see ref.@xcite . in the chiral lagrangian , there are many unknown phenomenological low energy constants ( lecs ) which appear in front of each goldstone field dependent operators and the number of the lecs increases rapidly when we go to the higher orders of the low energy expansion . for example for the three flavor case , the @xmath3 and @xmath1 order chiral lagrangian have 2 and 10 lecs respectively , while the normal part of @xmath0 order chiral lagrangian have 90 lecs . such a large number of lecs is very difficult to fix from the experiment data . this badly reduces the predictive power of the chiral lagrangian and blur the check of its convergence . the area of estimating @xmath0 order lecs is where most improvement is needed in the future of higher order chiral lagrangian calculations . a way to increase the precision of the low energy expansion and improve the present embarrassed situation is studying the relation between the chiral lagrangian and the fundamental principles of qcd . we expect that this relation will be helpful for understanding the origin of these lecs and further offer us their values . in previous paper @xcite , based on a more earlier study of deriving the chiral lagrangian from the first principles of qcd @xcite in which lecs are defined in terms of certain green s functions in qcd , we have developed techniques and calculated the @xmath3 and @xmath1 order lecs approximately from qcd . our simple approach involves the approximations of taking the large-@xmath4 limit , the leading order in dynamical perturbation theory , and the improved ladder approximation , thereby the relevant green s functions relate to lecs are expressed in terms of the quark self energy @xmath2 . the result chiral lagrangian in terms of the quark self energy is proved equivalent to a gauge invariant , nonlocal , dynamical ( gnd ) quark model@xcite . by solving the schwinger - dyson equation ( sde ) for @xmath2 , we obtain the approximate qcd predicted lecs which are consistent with the experimental values . with these results , generalization of the calculations to @xmath0 order lecs becomes the next natural step . considering that the algebraic derivations for those formulae to express lecs in terms of the quark self energy at @xmath1 order are lengthy ( they need at least several months of handwork ) , it is almost impossible to achieve the similar works for the @xmath0 order calculations just by hand . therefore , to realize the calculations for the @xmath0 order lecs , we need to computerize the original calculations and this is a very hard task . the key difficulty comes from that the formulation developed in ref.@xcite and exploited in ref.@xcite not automatically keeps the local chiral covariance of the theory and one has to adjust the calculation procedure by hand to realize the covariance of the results . to match with the computer program , we need to change the original formulation to a chiral covariant one . in ref.@xcite , we have built and developed such a formulation , followed by next several year s efforts , we now successfully encode the formulation into computer programs . with the help of these computer codes we can reproduce analytical results on the computer originally derived by hand in ref.@xcite within 15 minutes now . this not only confirms the reliability of the program itself , but also checks the correctness of our original formulae . based on these progresses , in this paper , we generalize our previous works on calculating the @xmath1 order lecs to computing the @xmath0 order lecs of chiral lagrangian both for two and three flavor pseudo - scalar mesons . this generalization not only produces new numerical predictions for the @xmath0 order lecs , but also forces us to reexamine our original formulation from a new angle in dealing with @xmath3 and @xmath1 order lecs . this paper is organized as follows : in sec.ii , we review our previous calculations on the @xmath3 and @xmath1 order lecs . then , in sec.iii , based on the technique developed in ref.@xcite , we reformulate the original low energy expansion used in ref.@xcite into a chiral covariant one suitable for computer derivation . in sec.iv , from present @xmath0 order viewpoint , we reexamine the formulation we taken before and show that if we sum all higher order anomaly part contributions terms together , their total contributions to the normal part of the chiral lagrangian vanish . this leads a change the role of finite @xmath1 order anomaly part contributions which originally are subtracted in the chiral lagrangian in ref.@xcite and now must be used to cancel divergent higher order anomaly part contributions . we reexhibit the numerical result of the @xmath1 order lecs without subtraction of @xmath1 order anomaly part contributions . in sec.v , we present general @xmath0 order chiral lagrangian in terms of rotated sources and express the @xmath0 order lecs in terms of the quark self energy . sec.vi is a part where we give numerical results for @xmath0 order lecs in the normal part of chiral lagrangian both for two and three flavor pseudo scalar mesons . in sec . vii , we apply and compare with our results to some individuals and combinations of lecs proposed and estimated in the literature , checking the correctness of our numerical predictions . sec.viii is a summary . in appendices , we list some necessary formulae and relations . with the analytical formulae for lecs of @xmath53 and @xmath54 for @xmath55 as functions of @xmath25 , we can suitably choose running coupling constant @xmath30 , solve sde ( [ eq0 ] ) numerically obtaining quark self energy @xmath25 , then calculate the numerical values of all @xmath3 and @xmath1 order lecs . to obtain the final numerical result in ref.@xcite , we have assumed @xmath69mev as inputmev for two - flavour case . for detail , see the discussion of eq.(58 ) ] to fix the dimensional parameter @xmath70 appear in running coupling constant @xmath30 and taken cutoff parameter @xmath51 appear in ( [ trln ] ) equal to infinity and 1gev respectively . the final obtained values are consistent with those fixed phenomenologically . eq.([trln ] ) is the starting point of our reformulation in this section . in ref.@xcite , we expand ( [ trln ] ) up to the @xmath1 order and obtain analytical result . this expansion is not explicitly chiral covariant , since the operator @xmath71 appears in the formula is not always covariant under the local chiral symmetry transformations . for example , when @xmath71 acts on a constant number 1 , it gives @xmath72 which is not covariant since @xmath73 itself behaves as the gauge field in the local chiral symmetry transformations . only when they combined into commutators , such as @xmath74 $ ] or @xmath75 $ ] , the covariance recovers back . therefore in the detail calculation , we need to confirm that all @xmath71s appear in the result do can be arranged into some commutators . this is a conjecture . in the original work of ref.@xcite , we have found that this conjecture is valid up to some terms with coefficients being expressed as integration over some total derivatives , i.e. form of @xmath76 . if we ignore these total derivative terms , up to order of @xmath1 , we can explicitly prove the conjecture . at the stage of our earlier works , we do not question the reason that why we can drop out those total derivative terms ( in fact , in eq.(74 ) of ref.@xcite , we have shown that in order to obtain the well - known pagels - stokar formula , a total derivative term must be dropped out ) . this leads the further discussions on the role of total derivative terms in the quantum field theory @xcite . later in this section , we will give the correct reason of dropping out those total derivative terms . arranging various @xmath71 into commutators is a very tricky and complex task which is very hard to be achieved by computer . in order to computerize the calculation , we need to find a way which can automatically arrange all @xmath71s into some commutators . this leads the developments given in ref.@xcite , where we have introduced @xmath77 in which @xmath78 ( i\bar{\nabla}_x^\mu)\bigg)\nonumber\\ & = & \frac{1}{2}(\nu\mu)\frac{\partial}{\partial k^{\nu } } -\frac{i}{3}(\lambda\nu\mu)\frac{\partial^2}{\partial k^{\lambda}\partial k^{\nu } } -\frac{1}{8 } ( \rho\lambda\nu\mu ) \frac{\partial^3}{\partial k^{\rho}\partial k^{\lambda}\partial k^{\nu } } + \frac{i}{30}(\sigma\rho\lambda\nu\mu ) \frac{\partial^4}{\partial k^\sigma\partial k^{\rho}\partial k^{\lambda}\partial k^{\nu}}\notag\\ & & + \frac{1}{144}(\delta\sigma\rho\lambda\nu\mu ) \frac{\partial^5}{\partial k^\delta\partial k^\sigma\partial k^{\rho}\partial k^{\lambda}\partial k^{\nu}}+o(p^7)\;,\\ & & \hspace{-2.3cm}f(z)=\sum^\infty_{n=2}\frac{z^{n-1}}{n!}\hspace{3 cm } [ ad(b)]^n(c)\equiv[\underbrace{b,[b,\cdots,[b}_{\mbox{n times}},c]\cdots]]\;,\notag\\ & & \hspace{-2.3cm}(\mu_n\mu_{n-1}\cdots\mu_2\mu_1 ) \equiv[\nx^{\mu_n},[\nx^{\mu_{n-1}},\cdots,[\nx^{\mu_2},\nx^{\mu_1}]\cdots]]\;,\notag \end{aligned}\ ] ] where the default set of lorentz indices for @xmath79 is the supperscripts , in some cases , we need subscript , we will use @xmath80 to denote the corresponding subscript for @xmath81 . note that in present notation for @xmath79 , we do nt explicitly write @xmath82s , but only their greek superscripts for short . if we use other symbols , such as @xmath83 appeared in @xmath84 and @xmath85 in @xmath86 , then we take definition that @xmath87 $ ] and @xmath88 $ ] . with the help of ( [ debt ] ) , ( [ d2ebt ] ) and ( [ d3ebt])-([d6ebt ] ) , as long as the @xmath118 are known , ( [ ebexp0 ] ) is known and we can substitute it back into ( [ db1 ] ) to calculate the real part of @xmath63 $ ] order by orders up to the @xmath0 order in the low energy expansion . to obtain @xmath119 , ( [ bdef ] ) tells us that the difficulty is the low energy expansion for @xmath120 . to achieve it , we expand the argument of @xmath120 as @xmath121 in which @xmath122 since we are only interested in the terms not higher than @xmath0 , we find that those traceless terms of @xmath123 and @xmath0 orders will not make contributions to the final result . so to save space and simplify the computations , we do not explicitly write down the detail structure of them , just represent these terms with symbol @xmath124 and remove traceless term in @xmath125 and @xmath126 . further introduce @xmath127 as , @xmath128 then @xmath129\bigg]_{t=0 } + \frac{1}{2!}\bigg[\frac{d^2}{d t^2}\sigma[a(t)]\bigg]_{t=0 } + \frac{1}{3!}\bigg[\frac{d^3}{d t^3}\sigma[a(t)]\bigg]_{t=0 } + \frac{1}{4!}\bigg[\frac{d^4}{d t^4}\sigma[a(t)]\bigg]_{t=0}+\ldots\;.\label{sigmaexp}\end{aligned}\ ] ] now , we need to know @xmath130\bigg]_{t=0}$ ] , using the following formula @xmath131&=&\sigma[s+a(t)]\bigg|_{s=0 } = e^{a(t)\frac{\partial}{\partial s}}\sigma(s ) e^{-a(t)\frac{\partial}{\partial s}}\bigg|_{s=0 } = e^{a(t)\frac{\partial}{\partial s}}\sigma(s)\bigg|_{s=0}\ ; , \end{aligned}\ ] ] then @xmath132\bigg]_{t=0 } = \bigg[\frac{d^m}{d t^m}e^{a(t)\frac{\partial}{\partial s } } \bigg]_{t=0}\sigma(s)\bigg|_{s=0}\;. \end{aligned}\ ] ] therefore to compute @xmath130\bigg]_{t=0}$ ] , we only need to calculate @xmath133_{t=0}\sigma(s)\bigg|_{s=0}$ ] which is just equivalent to replace @xmath134 in ( [ debt]),([d2ebt ] ) and ( [ d3ebt])-([d6ebt ] ) , followed by multiplying an extra factor @xmath135 at the r.h.s . and vanishing parameter @xmath136 after finishing all differential operations . follow this calculation road map , the detail calculation gives @xmath137\bigg]_{t=0 } & = & e^{ad(a_0\frac{\partial}{\partial s})}(f[ad(-a_0\frac{\partial}{\partial s } ) ] a_1)\sigma'(s+a_0)\bigg|_{s=0}=0\;,\\ \frac{1}{2}\bigg[\frac{d^2}{d t^2}\sigma[a(t)]\bigg]_{t=0 } & = & \frac{1}{2}e^{ad(a_0\frac{\partial}{\partial s})}\bigg[\bigg(e^{-a_0\frac{\partial}{\partial s } } \frac{d}{d t}e^{a(t)\frac{\partial}{\partial s}}\bigg|_{t=0 } f[ad(-a_0\frac{\partial}{\partial s})](a_1\frac{\partial}{\partial s } ) + \frac{d f}{d t}[ad(-a(t)\frac{\partial}{\partial s})]\bigg|_{t=0}(a_1\frac{\partial}{\partial s})\nonumber\\ & & + f[ad(-a_0\frac{\partial}{\partial s } ) ] ( a_2\frac{\partial}{\partial s})\bigg)\bigg]\sigma(s+a(t))\bigg|_{s=0}=- ( \mu,\nu)k_{\mu}\sigma'_k \frac{\partial}{\partial k^{\nu}}\ ; , \end{aligned}\ ] ] where @xmath138 . for more higher orders , we list the results of @xmath139\bigg]_{t=0}$ ] , @xmath140\bigg]_{t=0}$ ] , @xmath141\bigg]_{t=0}$ ] and @xmath142\bigg]_{t=0}$ ] in appendix [ ebexp ] . with these results , we finally obtain the low energy expansion of @xmath100 , @xmath143\tau -i ( d^{\mu}a_\omega^\nu - d^{\nu}a_\omega^{\mu})\gamma_{\mu}\gamma_{\nu}\gamf \tau + 4 s_\omega \tau \sk+2 ( \mu \nu ) \tau k_{\mu}\frac{\partial}{\partial k^{\nu } } + 4 ( \mu a_\omega^{\nu } ) \gamma_{\nu}\gamf \tau k_{\mu}\skp\notag\\ & & + 4 ( \mu a_\omega^{\nu } ) \gamma_{\nu}\gamf \tau \sk \frac{\partial}{\partial k^{\mu } } + 2i ( \mu a_{\omega\mu } ) \gamf \tau + 4i ( \mu a_\omega^{\nu } ) \gamf \tau k_{\nu}\frac{\partial}{\partial k^{\mu}}+4i ( \mu \nu ) \gamma_{\mu}\tau k_{\nu}\skp + 4 ( \mu \nu ) \tau k_{\mu}\sk \skp \frac{\partial}{\partial k^{\nu}}\;.\end{aligned}\ ] ] we list @xmath144 in appendix.[ebexp ] . with these explicit expressions for @xmath118 , using ( [ debt ] ) , ( [ d2ebt ] ) and ( [ d3ebt])-([d6ebt ] ) , we get ( [ ebexp0 ] ) and further substitute ( [ ebexp0 ] ) back into ( [ db1 ] ) , we can obtain the real part of @xmath63 $ ] order by orders up to the @xmath0 order in the low energy expansion . the analytical results of @xmath3 and @xmath1 orders are the same as those given by ( 34),(35 ) and ( 36 ) in ref.@xcite , except some total derivative terms which , as we mentioned before , can be ignored as long as we take finite cutoff @xmath51 . in the last section , we have introduced a chiral covariant method to calculate @xmath34 $ ] which is already computerized now . with the help of computer , for the @xmath3 and @xmath1 order analytical formulae in the low energy expansion , we can get results within 15 minutes , while for the @xmath0 order terms , we need roughly 13 hours to output all expansion results . from our general result ( [ seff1 ] ) , the term @xmath63 $ ] is the normal part . to get the full result of the chiral lagrangian , we need to calculate the remaining anomaly part contributions @xmath145-in_c\mathrm{tr}\ln[i\slashed{\partial}+j]$ ] . as the discussion of ref.@xcite , in 1980s there is a class of works ( see references given in @xcite ) identifying this part as the full chiral lagrangian , and in ref.@xcite we refer them as the anomaly approach of calculating lecs . in our previous work @xcite , we pointed out that this anomaly part contributions are completely canceled by the normal part contribution , left nontrivial pure @xmath25 dependent terms contribute to the chiral lagrangian . for the anomaly part contributions , the key is to calculate the @xmath18 field dependent term @xmath35 $ ] which , as we mentioned before , can be obtained by vanishing @xmath25 in @xmath34 $ ] . in practice , the limit was taken by first assuming @xmath25 being a constant mass @xmath146 and then letting @xmath147 . for @xmath3 order , this operation gives null result , while for @xmath1 order , it gives the result originally presented in anomaly approach . now in this work , naively what we need to do is to generalize the calculation to @xmath0 order . but to our surprise , we get many terms with divergent coefficients . checking the calculation carefully , we find that the reason of appearance of infinities is due to the fact that most of the coefficients in front of the @xmath0 order operators have dimension of @xmath148 which goes to infinity when we take limit @xmath149 . note that the @xmath0 terms may also have coefficients of @xmath150 which are finite in the limit of @xmath147 , although they vanish when we take @xmath151 . these terms are irrelevant to our discussion on the divergence of @xmath0 order terms and therefore we do not need to care about them . applying the argument on @xmath148 dependence of the @xmath0 order coefficients back to the @xmath3 and @xmath1 order results we discussed before , coefficients in front of @xmath3 order operators have dimension of @xmath152 which goes to zero , this explains the phenomena that anomaly approach can not produce @xmath3 order terms . for @xmath1 order , the coefficients in front of operators are dimensionless and therefore the @xmath146 dependence is at most logarithmic of form @xmath153 which implies existence of a logarithmic ultraviolet divergence . since we know that in the large @xmath4 limit , the @xmath1 order lecs ( non - contact coefficients ) are not divergent , the @xmath153 term then can not appear in the final expression of these lecs , therefore in @xmath1 order , anomaly approach leads finite result lecs . in general for a @xmath154 order operator , the corresponding coefficient should has dimension @xmath155 . this implies that the infinity in the anomaly part contributions will be a general phenomena , when we go to the higher orders of the low energy expansion , since the more higher the order is , the more negative powers of @xmath146 dependence the coefficient will have and these negative powers of @xmath146 will result in infinities as we take limit @xmath149 . the appearance of these high order infinities provides another evidence that the anomaly approach is not a correct formulation in calculating lecs , at least not for the @xmath0 and more higher order lecs . since high order divergence term is as an addition part of our general result ( [ seff1 ] ) , we can not avoid them in our computations . how to deal with these high order infinities from negative powers of @xmath146 ? there exists an alternative way , not relying on the low energy expansion , to examine this anomaly part contributions in which we must exploit the first equation of ( [ jomega ] ) and we find @xmath156-\mathrm{tr}\ln[i\slashed{\partial}+j ] & = & \ln\mathrm{det}[i\slashed{\partial}+j_{\omega}]-\ln\mathrm{det}[i\slashed{\partial}+j]\nonumber\\ & = & \ln\mathrm{det}\bigg[[\omega p_r+\omega^{\dag}p_l][j+i\slashed{\partial}][\omega p_r+\omega^\dag p_l]\bigg]-\ln\mathrm{det}[i\slashed{\partial}+j]\nonumber\\ & = & \ln\mathrm{det}\bigg[[\omega p_r+\omega^{\dag}p_l][\omega p_r+\omega^\dag p_l]\bigg]=\mathrm{tr}\ln\bigg[[\omega p_r+\omega^{\dag}p_l][\omega p_r+\omega^\dag p_l]\bigg]\;.\end{aligned}\ ] ] for our interests , we are only interested in the real part of it , then @xmath157-\mathrm{retr}\ln[i\slashed{\partial}+j ] & = & \frac{1}{2}\mathrm{tr}\ln\bigg[[\omega p_r+\omega^{\dag}p_l][\omega p_r+\omega^\dag p_l]\bigg]+\frac{1}{2}\mathrm{tr}\ln\bigg[[\omega p_r+\omega^{\dag}p_l]^\dag[\omega p_r+\omega^\dag p_l]^\dag\bigg]\nonumber\\ & = & \frac{1}{2}\mathrm{tr}\ln\bigg[[\omega p_r+\omega^{\dag}p_l][\omega p_r+\omega^\dag p_l]\bigg]+\frac{1}{2}\mathrm{tr}\ln\bigg[[\omega^\dag p_r+\omega p_l][\omega^\dag p_r+\omega p_l]\bigg]\nonumber\\ & = & \frac{1}{2}\mathrm{tr}\ln\bigg[[p_r+p_l][p_r+p_l]\bigg]=\frac{1}{2}\mathrm{tr}\ln 1=0\;.\end{aligned}\ ] ] which shows that the compact form of anomaly part contributions to normal part of the chiral lagrangian is zero ! the third method is first taking taylor expansion in terms the power of @xmath164 and @xmath165 which corresponds performing the low energy expansion and then finishing integration , finally vanishing @xmath146 , @xmath170 we see that there are negative power of @xmath146 terms which will cause divergence when we take limit @xmath171 . this is just what has happened for the high order terms in the anomaly part contributions . so if we calculate term by terms in above expansion , we will meet infinities which seems contradict with results obtained in first two methods . the only way left to escape this contradiction is to sum all these divergences together , to see that what will happen after the summation , we introduce a series @xmath172 in which @xmath173 and @xmath174 which will go to negative infinity when @xmath171 . with the help of relation @xmath175 and boundary condition @xmath176 , we find @xmath177\;,\notag\\ & & g'(x , c)=(x-1)e^c[-\mathbf{ei}(cx - c)+\mathbf{ei}(-c)]+\frac{1}{c}(e^{cx}-1)\;,\notag\\ & & g(x , c)=\frac{1}{2}(x-1)^2e^c[-\mathbf{ei}(cx - c)+\mathbf{ei}(-c)]+\frac{x-1}{2c}e^{cx}+\frac{1}{2c } + \frac{1}{2c^2}(e^{cx}-1)-\frac{x}{c}\;.\end{aligned}\ ] ] then ( [ i1 ] ) becomes @xmath178\nonumber\\ & = & \lim_{m\rightarrow0}\bigg[\frac{\lambda^4}{2}-c\lambda^2-\frac{1}{2}c^2\mathbf{ei}(-\frac{m^2}{\lambda^2 } ) + \frac{1}{2}(bm+c+m^2)^2[-\mathbf{ei}(\frac{-bm - c - m^2}{\lambda^2})+\mathbf{ei}(-\frac{m^2}{\lambda^2 } ) ] \nonumber\\ & & + \frac{1}{2}\lambda^2(-bm - c - m^2)e^{\frac{-bm - c - m^2}{\lambda^2 } } + \frac{1}{2}\lambda^2m^2e^{-\frac{m^2}{\lambda^2}}+\frac{1}{2}\lambda^4 ( e^{\frac{-bm - c - m^2}{\lambda^2}}-e^{-\frac{m^2}{\lambda^2}})+\lambda^2(bm+c)\bigg]\nonumber\\ & = & -\frac{1}{2}c^2\mathbf{ei}(-\frac{c}{\lambda^2 } ) -\frac{1}{2}\lambda^2ce^{-\frac{c}{\lambda^2}}+\frac{1}{2}\lambda^4e^{-\frac{c}{\lambda^2}}\;. \end{aligned}\ ] ] it is the same as the results obtained from first two methods , i.e. summing all those infinities together , we obtain correct finite result . with this new viewpoint on all anomaly part contributions , we need to modify our original numerical results on @xmath1 order lecs , since it takes into account of the finite values of anomaly part contributions and now we know that these nontrivial values must be used to cancel the infinities come from all higher order terms . in table i , we list our modified @xmath1 order lecs for cutoff @xmath1811000@xmath182mev . the @xmath183 variation of the cutoff is considered in our calculation to examine the effects of cutoff dependence and the result change can be treated as the error of our calculations . the result lecs are taken the values at @xmath184gev with superscript the difference caused at @xmath185gev and subscript the difference caused at @xmath186gev , i.e. , @xmath187 . the obtained values of the @xmath1 coefficients @xmath188 for three flavor quarks and @xmath189 for two flavor quarks + where @xmath190 for @xmath191 , @xmath192770mev and @xmath193 are given in ref.@xcite . since @xmath194 , we calculate @xmath195 instead of @xmath196 . + together with the experimental values given in ref.@xcite and our old result given in ref.@xcite for comparisons . + @xmath70 , @xmath51 and @xmath197 are in units of mev , and @xmath198 are in units of @xmath199 . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] in obtaining the result , we have taken the running coupling constant as model a given by ( 40 ) of ref.@xcite and the low energy value of this @xmath200 is already chosen well above the critical value to trigger the s@xmath201sb of the theory . it should be noted that @xmath200 depends on the number of quark flavors , so does for @xmath25 from sde . in fixing the @xmath70 we have taken input @xmath202mev . the reason of taking this value is that if the final @xmath203 is around value of 93mev , then our formula shows @xmath204 must be located around 87mev . in sec.vi , we will exhibit this phenomena explicitly . the general form of @xmath0 order chiral lagrangian was first introduced in ref.@xcite and then discussed in ref.@xcite . now we can express the normal part of it in terms of our rotated sources as what we have done in ( [ p4 ] ) for the @xmath1 order terms . considering that our computation is done under large @xmath4 limit , within this approximation , terms in the chiral lagrangian with two and more traces vanish when we not apply the equation of motion . to avoid unnecessary complicities , in this paper we only write down those terms with one trace @xmath205~+~o(\frac{1}{n_c})\bigg ] \label{sp6ours}\end{aligned}\ ] ] with @xmath206 being @xmath0 order operator we could get in our calculation and @xmath207 being corresponding coefficient , @xmath208 are consist of the most multi - traces terms . our computations then give the explicit expressions of @xmath207 in terms of quark self energy . the detail expressions are given in ( [ zsigma ] ) . and the definitions of operators @xmath206 for @xmath209 are given in table.ii , @xmath210 where some operators have @xmath211 in front of them to insure their coefficients being real . in ref.@xcite , @xmath0 order operator was denoted by @xmath212 for the case of n flavor with coefficient @xmath213 @xcite , @xmath214 for the case of three flavor with coefficient @xmath215 and @xmath216 for the case of two flavors with coefficient @xmath217 , @xmath218 consider that our parametrization of the @xmath0 order chiral lagrangian ( [ sp6ours ] ) is general to the case of n flavor quarks , there exist some relations among our coefficients and n flavor coefficients given in ( [ sp6ref ] ) . with the help of computer derivations , we have worked out these relations and list them in apendix[kz ] . as a check , we vanish the quark self energy @xmath25 in the codes which corresponds to take @xmath219 before other further calculations and find null @xmath0 result . this verify the analytical result discussed in sec.iv that the anomaly part contributions do not the contribute to normal part of chiral lagrangian . another consistency check is done for those operators which have two terms combined together by @xmath165 and @xmath220 symmetry requirements . for @xmath221 flavor case , such operators are @xmath222 , @xmath223 , @xmath224 , @xmath225 , @xmath226 , @xmath227 , @xmath228 , @xmath229 , @xmath230 , @xmath231 , @xmath232 , @xmath233 , @xmath234 , @xmath235 , @xmath236 , @xmath237 , @xmath238 , @xmath239 , @xmath240 , @xmath241 , @xmath242 , @xmath243 , @xmath244 , @xmath245 , @xmath246 , @xmath247 , @xmath248 , @xmath249 , @xmath250 , @xmath251 , @xmath252 . since in each of these operators , there are two terms , we can compute the coefficients in front of each terms and check if they are same . we have done the checks for all these operators and all obtain the same analytical expressions for the two terms in the same operator . this partly verifies the correctness of our result given in ( [ zsigma ] ) . from n flavors to three flavors , there are some extra constraints ( see ( b1 ) in ref.@xcite)which make some operators depending on others . further from three flavors to two flavors , there are also some extra constraints ( see ( b3 ) in ref.@xcite)which make some more operators depending on others . the sequence number for n flavors , three flavors and two flavors are different , their comparisons are list in table 2 in ref.@xcite . with all above preparations in previous sections , we now come to the stage of giving numerical values to the @xmath0 order lecs in the normal part of the chiral lagrangian . note the necessary input and process of the present computations for the @xmath0 order lecs are the same as those for the @xmath3 and @xmath1 order lecs given in the end of sec.iv , we list the numerical result in table.iii , as done in table.i , the result lecs are taken for the values at @xmath184gev with superscript the difference caused at @xmath185gev and subscript the difference caused at @xmath186gev , @xmath253 we further list result of the @xmath0 order lecs at @xmath254 in table.iv . consider that in the limit of @xmath254 , dropping out momentum total derivative terms in eq.(14 ) is problematic , we only take result lecs at @xmath254 as a reference . since the terms of three flavors and two flavors may have different sequence numbers , as done in ref.@xcite , we put them in the same line in our table . since the number of independent operators in the two flavors is smaller than that in the three flavors , there are some operators in three flavors being independent operators , but being dependent in two flavors , then these operators will not have their two flavor counter parts in our table , these leave the r.h.s . some empty blanks in the corresponding two flavor columns . for two flavor case , ref.@xcite further propose a new relation among operators , @xmath255 which implies that one of the operators appears in above formula should be further dependent operator . due to ignorance of the values of the coefficients in front of these operators , ref.@xcite arbitrarily chooses this operator being @xmath256 . now our computations show that @xmath257 , so original choice is not suitable . considering that @xmath258 in our computation , we instead now take @xmath259 as that dependent operator . @xmath259 now is a dependent operator , its name then is deleted in our table.iii . @xmath273 as we have mentioned in the introduction of this paper , present experiment data is far enough to fix the @xmath0 order lecs . but there do exist some combinations of the lecs which already have their experiment or model calculation values . usually , these lecs are labeled by dimensionless parameters with convention and @xmath274 are used to denote the renormalized lecs in some literatures.]of @xmath275 or @xmath276 . in this section , we collect those combinations of lecs in the literature which have their experiment or model calculation values and compare them with our numerical results obtained in the last section with finite cutoff , we have checked that qualitative feature of the comparisons results of this section will not change . ] as the check of our computations . from the investigation of @xmath277 scattering amplitudes , one can work out the values of some combinations of @xmath0 order lecs . ref.@xcite introduces following combinations , @xmath279 and gives the values of them by two theoretical methods of the resonance - saturation ( rs)@xcite and pure dimensional analysis ( nd ) which only accounts for the order of magnitude and in table.v . @xmath280 we see that all coefficients obtained from our calculations are consistent with those more precise rs results given in ref.@xcite . with our predictions for @xmath1 and @xmath0 order lecs , we can directly calculate the scattering lengths @xmath281 and slope parameters @xmath282 which relate to @xmath1 and @xmath0 order lecs through formulae given in appendix c and d. of ref.@xcite . we list experimental and our results in table.vi . in our results , as done in table.iii , we take @xmath283mev , but to match the result given in ref.@xcite where @xmath81 is taken at @xmath284gev , we also take @xmath285mev for comparison . we take two options , one only includes @xmath1 order contributions and the other combines in @xmath0 order contributions . for @xmath0 order contributions , for comparisons , we consider the cases of without and with @xmath286 coefficients . @xmath287 we see that the contributions from @xmath0 order lecs are rather small and only change the third digit of the result . further , ref.@xcite introduce coefficients in @xmath278 scattering @xmath288 in the table 1 . of ref.@xcite , in terms of @xmath289 , three constraints of @xmath0 order lecs are fixed from @xmath278 subthreshold parameters , @xmath277 amplitude and a resonance model . and in the table 2 . of ref.@xcite , in terms of @xmath290 , another three constraints of @xmath0 order lecs are fixed from the dispersive calculations and a resonance model , @xmath291 in which for l.h.s . of the table vii . , except @xmath292 , all other lecs or combinations of lecs obtained by us have the same signs and orders of magnitudes as those from ref.@xcite . while for r.h.s . of the table , our results are not consistent with those obtained from the dispersive calculations . in ref.@xcite , in dealing with the vector form factor of the pion , @xmath293 and @xmath294 are introduced into theory which relate to @xmath0 order lecs through @xmath295 while for the scalar form factor , people introduce @xmath296 and @xmath297 relate to @xmath0 order lecs by @xmath298 in ref.@xcite , discussion of the decay of @xmath299 further introduces @xmath300 and @xmath301 relate to @xmath0 order lecs by @xmath302 in ref.@xcite , a naive estimation of @xmath303 is made from scalar meson dominance ( smd ) of the pion scalar form - factor and @xmath304 is estimated through @xmath305 in @xmath306 measurements ( see eq.(8.11 ) in ref.@xcite ) . while in ref.@xcite , @xmath307 and @xmath308 are also estimated from the @xmath278 form factors . in table.viii , we list the numerical results for above combinations of lecs given by our calculations based on table.iii in last section and by ref.@xcite,@xcite,@xcite . @xmath309 from which we see that among ten parameters between our predictions and values given in the literature , four of them have the same orders of magnitudes and signs ( @xmath310 , @xmath311 , @xmath312 and @xmath313 ) , another one of them has different orders of magnitudes but the same signs ( @xmath303 in ref.@xcite ) , the left five of them have opposite signs ( @xmath314 , @xmath315 , @xmath316 , @xmath317 and @xmath303 in ref.@xcite ) . + further in fig.[vectorformfactor ] , we compare the experimental data for vector form factors collected in figure 4 . and figure 5 . of ref.@xcite with our results . in obtaining our numerical predictions , we have exploited the formula given by eq.(3.16 ) in ref.@xcite which especially depends on @xmath0 order lecs through @xmath318 defined in ( [ rvdef ] ) and we input the formula @xmath1 and @xmath0 lecs obtained in table.iii of the last section . order with lecs obtained in table.iii of this paper . the red dashed line is the result by vanishing @xmath0 order lecs in corresponding red solid curve . the blue dot - dashed curve corresponds to predictions from chiral perturbation up to @xmath1 order with lecs obtained in table.i of this paper . the blue dotted line is the result by vanishing @xmath1 order lecs in corresponding blue dot - dashed curve.the black x - axis of with @xmath319 corresponds to predictions from @xmath3 order chiral perturbation.,title="fig : " ] order with lecs obtained in table.iii of this paper . the red dashed line is the result by vanishing @xmath0 order lecs in corresponding red solid curve . the blue dot - dashed curve corresponds to predictions from chiral perturbation up to @xmath1 order with lecs obtained in table.i of this paper . the blue dotted line is the result by vanishing @xmath1 order lecs in corresponding blue dot - dashed curve.the black x - axis of with @xmath319 corresponds to predictions from @xmath3 order chiral perturbation.,title="fig : " ] from fig.[vectorformfactor ] , we see that @xmath0 order lecs explicitly improve the @xmath1 and @xmath3 order chiral perturbation predictions and making them being more consistent with experimental data . in ref.@xcite , discussion of the photon - photon collision @xmath320 introduces @xmath321 and @xmath322 relate to @xmath0 order lecs by @xmath323 while in ref.@xcite , calculation of the photon - photon collision @xmath324 introduces another type of @xmath321 and @xmath322 relate to @xmath0 order lecs by @xmath325 in table.ix , we list the numerical results for above combinations of lecs given by our calculations based on table.iii in last section and by ref.@xcite and @xcite . @xmath326 for which we see that among six parameters between our predictions and values given in the literature , except two have opposite signs , other four all have the same orders of magnitudes and signs . in ref.@xcite , through reanalysis of the radiative pion decay , a group of @xmath0 order lecs are fixed . @xmath327 we find that all lecs and combination of lecs from our predictions have the same signs and orders of magnitudes as those from experiment values . except above phenomenological estimations on the values of some lecs , there are model calculations for some others of them and most of these analysis use a ( single ) resonance approximation . in contrast , our calculations do not rely on the assumption of existence of resonances . in this subsection , we list down these calculation values we can collect from the literature and compare with our results . ref.@xcite estimates values of some lecs . @xmath328 for @xmath329 and @xmath330 , ref.@xcite gives the value for their combination @xmath331 which , compares to our result of @xmath332 , is at the same order of magnitude and has the same sign . for @xmath333 , there are several works to estimate its values , we list them in table.xii . @xmath334 where @xmath335 given in ref.@xcite and @xcite are in form of @xmath336 in unit of gev@xmath337 , we have transformed them into our expression of @xmath333 with @xmath338 . further , ref.@xcite exploits resonance lagrangian estimates values of lecs @xmath339 , @xmath340 , @xmath336 , @xmath341 , @xmath342 , @xmath343 . @xmath344 we find that our results are consistent with those obtained from resonance lagrangian . ref.@xcite estimates the value of @xmath345 and gives @xmath346 which is also consistent with our result of @xmath347 . in terms of resonance exchange , ref.@xcite proposes some relations among different @xmath0 order lecs , @xmath348 to check the validity of these relations for our results , in table . xiv , we write corresponding values obtained in our calculations @xmath349 we see that except @xmath350 , all the other lecs satisfy the relations . in this paper , we revise our original formulation of calculating lecs from the first principle of qcd to a chiral covariant one suitable to computerize . with the help of computer , we successfully obtain the analytical expressions for all the @xmath0 order lecs in the normal part of chiral lagrangian for pseudo scalar mesons on the quark self energy @xmath351 . the ambiguities for anomaly part contributions to the normal part of the chiral lagrangian are clarified and we prove that this part totally should vanish and therefore need not to be considered in our computations . since our calculation is done under large @xmath4 limit , only operators of @xmath0 order with one trace and some multi - traces from the equation of motion survive in our formulation . we set up relations among the coefficients in front of these operators and lecs defined in ref.@xcite . then with input of @xmath35287mev to fix the @xmath70 in the running coupling constant of @xmath353 appear in the kernel of sde and choose cutoff of the theory being @xmath265mev and @xmath254 , we calculate all @xmath0 order lecs numerically both for two flavor and three flavor cases . compare our result lecs with those combinations which we can find experimental or model calculation values in the literature , we find that except few of them have wrong signs , most of our predicted combinations of @xmath0 order lecs have the same signs and orders of magnitudes with experiment or model calculation values . this sets the solid basis for our @xmath0 order computations . for those combinations with wrong signs or wrong order of magnitudes with experiment values , we need further investigations . based on these obtained @xmath0 order lecs , we expect a very large number of predictions for various pseudo scalar meson physics in the near future . this work was supported by national science foundation of china ( nsfc ) under grant no.10875065 . we thank prof . y.p.kuang for the helpful discussions . s.weinberg , physica , * 96a * , 327 ( 1979 ) . j.gasser and h.leutwyler , ann . phys . * 158 * , 142 ( 1984 ) ; nucl . * b 250 * , 465 ( 1985 ) . j.bijnens , prog . part . . phys . * 58 * , 521(2007 ) . h. yang , q. wang , y-p.kuang and q. lu , phys . rev . * d66 * , 014019(2002 ) . q. wang , y - p . kuang , x - l . wang and m. xiao , phys . rev . * d61 * , 054011(2000 ) . h. yang , q. wang and q. lu , phys.lett . * b532 * , 240(2002 ) . q. lu , h. yang and q. wang , commun . . phys . * 38 * , 185(2002 ) . an , h. yang and q , wang , eur . j. * c29 * , 65(2003 ) . y. zhang and q. wang , chin . . lett . * 25 * , 1227(2008 ) . q. lu , y. zhang and q. wang , commun . theor . phys . * 49 * , 431(2008 ) . ma and q. wang , phys . b560 * , 188(2003 ) . q. wang , int . * a20 * , 1627(2005 ) . h. w. fearing and s. scherer , phys . rev . * d53 * , 315(1996 ) . j. bijnens , g. colangelo and g. ecker , jhep*02 * , 020(1999 ) . j. bijnens , g. colangelo and g. ecker , ann . phys . * 280 * , 100(2000 ) . c.haefeli , m. a. ivanov , m. schmid , and g. ecker , arxiv:0705.0576 [ hep - ph ] . j. bijnens , g. colangelo and p. talavera , jhep*05 * , 014(1998 ) . j.bijnens , g.colangelo , g.ecker , j.gasser and m.e.sainio , nucl . b508,263(1997),erratum , _ ibid . _ * b517 * , 639(1998 ) . o. dumbrajs et al . . phys . * b216 * , 277(1983 ) . j.bijnens , p.dhonte , p.talaveta , jhep*0405 * , 036(2004 ) . k. kampf and b. moussallam , eur . j. * c47 * , 723(2006 ) . j. bijnens and p. talavera , nucl . b669 * , 341(2003 ) . m. jamin , j. a. oller and a. pich , jhep*02 * , 047(2004 ) . j. gasser , m.a . ivanov and m.e . sainio , nucl . phys . * b728 * , 31(2005 ) . j. gasser , m.a . ivanov and m.e . sainio , nucl . phys . * b745 * , 84(2006 ) . r. unterdorfer and hannes pichl , eur . j. * c55 * , 273(2008 ) . g. amor@xmath354s , j. bijnens and p. talavera , nucl . phys . * b568 * , 319(2000 ) . j. prades , poskaon:022,(2008 ) , arxiv:0707.1789[hep - ph ] . a. pich , i. rosell and j. j. sanz - cillero , jhep*07 * , 014(2008 ) . p. masjuan and s. peris , phys . lett . * b663 * , 61(2008 ) . m. gonz@xmath355lez - alonso , a. pich and j. prades , phys . rev . * d78 * , 116012(2008 ) . c. q. geng , i - l . ho and t. h. wu , nucl b684 * , 281(2004 ) . i. rosell , a. pich and j. j. sanz - cillero , jhep*01 * , 039(2007 ) . v. cirigliano , g. ecker , m. eidem@xmath356ller , r. kaiser , a. pich and j. portol@xmath357s , nucl . b753 * , 139(2006 ) . in this appendix , we first list down the @xmath359 , @xmath1 , @xmath123 and @xmath0 order low energy expansion result for @xmath358 used in ( [ ebexp0 ] ) . @xmath360(b_1 ) + \frac{d}{d t}e^{b(t)}\bigg|_{t=0}\bigg\ { 2\frac{d f}{d t}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 2f[ad(-b_0)](b_2)\bigg\}\notag\\ & & + e^{b_0}\bigg\{\frac{d^2 f}{d t^2}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 2\frac{d f}{d t}[ad(-b(t))]\bigg|_{t=0}(b_2)+f[ad(-b_0)](b_3)\bigg\}\;,\label{d3ebt } \end{aligned}\ ] ] @xmath361(b_1 ) + \frac{d^2}{d t^2}e^{b(t)}\bigg|_{t=0 } \bigg\{3\frac{d f}{dt}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 3f[ad(-a_0)](b_2)\bigg\}\nonumber\\ & & + \frac{d}{d t}e^{b(t)}\bigg|_{t=0}\bigg\ { 3\frac{d^2 f}{d t^2}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 6\frac{d f}{d t}[ad(-b(t))]\bigg|_{t=0}(b_2 ) + 3f[ad(-b_0)](b_3)\bigg\}\nonumber\\ & & + e^{b_0}\bigg\{\frac{d^3 f}{d t^3}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 3\frac{d^2 f}{d t^2}[ad(-b(t))]\bigg|_{t=0}(b_2)\nonumber\\ & & + 3\frac{d f}{d t}[ad(-b(t))]\bigg|_{t=0}(b_3 ) + f[ad(-b_0)](b_4)\bigg\}\;,\label{d4ebt } \end{aligned}\ ] ] @xmath362(b_1 ) + \frac{d^3}{d t^3}e^{b(t)}\bigg|_{t=0 } \bigg\{4\frac{d f}{dt}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 4f[ad(-a_0)](b_2)\bigg\}\notag\\ & & + \frac{d^2}{d t^2}e^{b(t)}\bigg|_{t=0 } \bigg\{6\frac{d^2 f}{d t^2}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 12\frac{d f}{d t}[ad(-b(t))]\bigg|_{t=0}(b_2 ) + 6f[ad(-b_0)](b_3)\bigg\}\nonumber\\ & & + \frac{d}{d t}e^{b(t)}\bigg|_{t=0 } \bigg\{4\frac{d^3 f}{d t^3}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 12\frac{d^2 f}{d t^2}[ad(-b(t))]\bigg|_{t=0}(b_2)+12\frac{d f}{d t}[ad(-b(t))]\bigg|_{t=0}(b_3)\nonumber\\ & & + 4f[ad(-b_0)](b_4)\bigg\}+e^{b_0}\bigg\{\frac{d^4 f}{d t^4}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 4\frac{d^3 f}{d t^3}[ad(-b(t))]\bigg|_{t=0}(b_2)\notag\\ & & + 6\frac{d^2 f}{d t^2}[ad(-b(t))]\bigg|_{t=0}(b_3 ) + 4\frac{d f}{d t}[ad(-b(t))]\bigg|_{t=0}(b_4 ) + f[ad(-b_0)](b_5)\bigg\}\;,\label{d5ebt}\end{aligned}\ ] ] @xmath363(b_1 ) + \frac{d^4}{d t^4}e^{b(t)}\bigg|_{t=0 } \bigg\{5\frac{d f}{dt}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 5f[ad(-b_0)](b_2)\bigg\}\nonumber\\ & & + \frac{d^3}{d t^3}e^{b(t)}\bigg|_{t=0 } \bigg\{10\frac{d^2 f}{d t^2}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 20\frac{d f}{d t}[ad(-b(t))]\bigg|_{t=0}(b_2 ) + 10f[ad(-b_0)](b_3)\bigg\}\nonumber\\ & & + \frac{d^2}{d t^2}e^{b(t)}\bigg|_{t=0 } \bigg\{10\frac{d^3 f}{d t^3}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 30\frac{d^2 f}{d t^2}[ad(-b(t))]\bigg|_{t=0}(b_2)\nonumber\\ & & + 30\frac{d f}{d t}[ad(-b(t))]\bigg|_{t=0}(b_3 ) + 10f[ad(-b_0)](b_4)\bigg\}+\frac{d}{d t}e^{b(t)}\bigg|_{t=0 } \bigg\{5\frac{d^4 f}{d t^4}[ad(-b(t))]\bigg|_{t=0}(b_1)\notag\\ & & + 20\frac{d^3 f}{d t^3}[ad(-b(t))]\bigg|_{t=0}(b_2 ) + 30\frac{d^2 f}{d t^2}[ad(-b(t))]\bigg|_{t=0}(b_3 ) + 20\frac{d f}{d t}[ad(-b(t))]\bigg|_{t=0}(b_4 ) \notag\\ & & + 5f[ad(-b_0)](b_5)\bigg\}+e^{b_0}\bigg\{\frac{d^5 f}{d t^5}[ad(-b(t))]\bigg|_{t=0}(b_1 ) + 5\frac{d^4 f}{d t^4}[ad(-b(t))]\bigg|_{t=0}(b_2)\notag\\ & & + 10\frac{d^3 f}{d t^3}[ad(-b(t))]\bigg|_{t=0}(b_3 ) + 10\frac{d^2 f}{d t^2}[ad(-b(t))]\bigg|_{t=0}(b_4)+5\frac{d f}{d t}[ad(-b(t))]\bigg|_{t=0}(b_5)\notag\\ & & + f[ad(-b_0)](b_6)\bigg\}\;.\label{d6ebt}\end{aligned}\ ] ] then , we list down the @xmath359 , @xmath1 , @xmath123 and @xmath0 order low energy expansion result for @xmath120 used in ( [ sigmaexp ] ) . note traceless terms in @xmath123 and @xmath0 orders are dropped out . @xmath364\bigg]_{t=0 } & = & \frac{1}{6}e^{ad(a_0\pps)}\bigg[e^{-a_0\pps}\frac{d^2}{d t^2}e^{a(t)\pps}\bigg|_{t=0}f[ad(-a_0\pps)](a_1\pps)\nonumber\\ & & + e^{-a_0\pps}\frac{d}{d t}e^{a(t)\pps}\bigg|_{t=0}\bigg\ { 2\frac{d f}{d t}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps ) + 2f[ad(-a_0\pps)](a_2\pps)\bigg\}\notag\\ & & + \bigg\{\frac{d^2 f}{d t^2}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps ) + 2\frac{d f}{d t}[ad(-a(t)\pps)]\bigg|_{t=0}(a_2\pps ) \nonumber\\ & & + f[ad(-a_0\pps)](a_3\pps)\bigg\}\bigg ] \sigma(s+a(t))\bigg|_{s=0}\nonumber\\ & = & -\frac{i}{3 } ( \mu \underline{\mu } \nu ) k_{\nu}\skpp + \frac{2i}{3 } ( \mu \nu \lambda ) k_{\mu}k_{\nu}\skpp \frac{\partial}{\partial k^{\lambda } } + \frac{2i}{3 } ( \mu \nu \lambda ) k_{\nu}\skp \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } + \frac{i}{3 } ( \mu \underline{\mu } \nu ) \skp \frac{\partial}{\partial k^{\nu}}\;,\end{aligned}\ ] ] @xmath365\bigg]_{t=0 } & = & e^{ad(a_0\pps)}\bigg[e^{-a_0\pps}\frac{d^3}{d t^3}e^{a(t)\pps}\bigg|_{t=0}~f[ad(-a_0\pps)](a_1\pps ) \nonumber\\ & & + e^{-a_0\pps}\frac{d^2}{d t^2}e^{a(t)\pps}\bigg|_{t=0 } \bigg\{3\frac{d f}{dt}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps ) + 3f[ad(-a_0\pps)](a_2\pps)\bigg\}\nonumber\\ & & + e^{-a_0\pps}\frac{d}{d t}e^{a(t)\pps}\bigg|_{t=0}\bigg\ { 3\frac{d^2 f}{d t^2}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps ) + 6\frac{d f}{d t}[ad(-a(t)\pps)]\bigg|_{t=0}(a_2\pps ) \nonumber\\ & & + 3f[ad(-a_0\pps)](a_3\pps)\bigg\ } + \bigg\{\frac{d^3 f}{d t^3}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps ) + 3\frac{d^2 f}{d t^2}[ad(-a(t)\pps)]\bigg|_{t=0}(a_2\pps)\nonumber\\ & & + 3\frac{d f}{d t}[ad(-a(t)\pps)]\bigg|_{t=0}(a_3\pps ) + f[ad(-a_0\pps)](a_4\pps)\bigg\}\bigg]_{t=0}\sigma(s+a(t))\bigg|_{s=0}\nonumber\\ & = & -\frac{1}{4 } ( \mu \nu ) ( \underline{\mu } \lambda ) k_{\nu}\skpp \frac{\partial}{\partial k^{\lambda } } + \frac{1}{2 } ( \mu \nu ) ( \lambda \rho ) k_{\mu}k_{\lambda}\skpp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } -\frac{1}{8 } ( \mu \nu \underline{\nu } \lambda ) k_{\lambda}\skpp \frac{\partial}{\partial k^{\mu } } \notag\\ & & -\frac{1}{8 } ( \mu \nu \underline{\mu } \lambda ) k_{\lambda}\skpp \frac{\partial}{\partial k^{\nu } } + \frac{1}{4 } ( \mu \underline{\mu } \nu \lambda ) k_{\nu}\skpp \frac{\partial}{\partial k^{\lambda } } + \frac{1}{4 } ( \mu \nu \lambda \rho ) k_{\nu}k_{\lambda}\skpp \frac{\partial^2}{\partial k^{\mu}\partial k^{\rho } } \notag\\ & & + \frac{1}{4 } ( \mu \nu \lambda \rho ) k_{\mu}k_{\lambda}\skpp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } + \frac{1}{4 } ( \mu \nu \lambda \rho ) k_{\lambda}\skp \frac{\partial^3}{\partial k^{\mu}\partial k^{\nu}\partial k^{\rho } } + \frac{1}{4 } ( \mu \nu ) ( \underline{\mu } \underline{\nu } ) \skpp \notag\\ & & + \frac{1}{4 } ( \mu \nu ) ( \underline{\mu } \lambda ) k_{\lambda}\skpp \frac{\partial}{\partial k^{\nu } } + \frac{1}{4 } ( \mu \nu ) ( \underline{\mu } \lambda ) \skp \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } + \frac{1}{8 } ( \mu \nu \underline{\nu } \lambda ) k_{\mu}\skpp \frac{\partial}{\partial k^{\lambda } } \notag\\ & & + \frac{1}{8 } ( \mu \nu \underline{\nu } \lambda ) \skp \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } + \frac{1}{8 } ( \mu \nu \underline{\mu } \lambda ) k_{\nu}\skpp \frac{\partial}{\partial k^{\lambda } } + \frac{1}{8 } ( \mu \nu \underline{\mu } \lambda ) \skp \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } \notag\\ & & -\frac{1}{6 } ( \mu \nu \underline{\nu } \lambda ) k_{\mu}k_{\lambda}\skppp -\frac{1}{6 } ( \mu \nu \underline{\mu } \lambda ) k_{\nu}k_{\lambda}\skppp + \frac{1}{3 } ( \mu \nu \lambda \rho ) k_{\mu}k_{\nu}k_{\lambda}\skppp \frac{\partial}{\partial k^{\rho } } + \frac{1}{3 } ( \mu \nu ) ( \underline{\mu } \lambda ) k_{\nu}k_{\lambda}\skppp\ ; , \end{aligned}\ ] ] @xmath366\bigg]_{t=0}\nonumber\\ & & \hspace{-0.4cm}=\frac{1}{5!}e^{ad(a_0\pps)}\bigg [ e^{-a_0\pps}\frac{d^4}{d t^4}e^{a(t)\pps}\bigg|_{t=0}~f[ad(-a_0\pps)](a_1\pps ) + e^{-a_0\pps}\frac{d^3}{d t^3}e^{a(t)\pps}\bigg|_{t=0 } \bigg\{4\frac{d f}{dt}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps)\nonumber\\ & & + 4f[ad(-a_0\pps)](a_2\pps)\bigg\}+e^{-a_0\pps}\frac{d^2}{d t^2}e^{a(t)\pps}\bigg|_{t=0 } \bigg\{6\frac{d^2 f}{d t^2}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps ) \nonumber\\ & & + 12\frac{d f}{d t}[ad(-a(t)\pps)]\bigg|_{t=0}(a_2\pps ) + 6f[ad(-a_0\pps)](a_3\pps)\bigg\}+e^{-a_0\pps}\frac{d}{d t}e^{a(t)\pps}\bigg|_{t=0 } \bigg\{4\frac{d^3 f}{d t^3}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps)\nonumber\\ & & + 12\frac{d^2 f}{d t^2}[ad(-a(t)\pps)]\bigg|_{t=0}(a_2\pps ) + 12\frac{d f}{d t}[ad(-a(t)\pps)]\bigg|_{t=0}(a_3\pps ) + 4f[ad(-a_0\pps)](a_4\pps)\bigg\}\notag\\ & & + \bigg\{\frac{d^4 f}{d t^4}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps ) + 4\frac{d^3 f}{d t^3}[ad(-a(t)\pps)]\bigg|_{t=0}(a_2\pps ) + 6\frac{d^2 f}{d t^2}[ad(-a(t)\pps)]\bigg|_{t=0}(a_3\pps)\notag\\ & & + 4\frac{d f}{d t}[ad(-a(t)\pps)]\bigg|_{t=0}(a_4\pps ) + f[ad(-a_0\pps)](a_5\pps)\bigg\}\bigg]\sigma(s+a(t))\bigg|_{s=0}\nonumber\\ & & \hspace{-0.4cm}=\mbox{traceless terms}\ ; , \end{aligned}\ ] ] @xmath367\bigg]_{t=0}\nonumber\\ & & \hspace{-0.4cm}=e^{ad(a_0\pps)}\bigg [ e^{-a_0\pps}\frac{d^5}{d t^5}e^{a(t)\pps}\bigg|_{t=0}[ad(-a_0\pps)](a_1\pps ) + e^{-a_0\pps}\frac{d^4}{d t^4}e^{a(t)\pps}\bigg|_{t=0 } \bigg\{5\frac{d f}{dt}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps)\notag\\ & & + 5f[ad(-a_0\pps)]\bigg|_{t=0}(a_2\pps)\bigg\ } + e^{-a_0\pps}\frac{d^3}{d t^3}e^{a(t)\pps}\bigg|_{t=0 } \bigg\{10\frac{d^2 f}{d t^2}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps ) \nonumber\\ & & + 20\frac{d f}{d t}[ad(-a(t)\pps)]\bigg|_{t=0}(a_2\pps ) + 10f[ad(-a_0\pps)](a_3\pps)\bigg\}\nonumber\\ & & + e^{-a_0\pps}\frac{d^2}{d t^2}e^{a(t)\pps}\bigg|_{t=0 } \bigg\{10\frac{d^3 f}{d t^3}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps ) + 30\frac{d^2 f}{d t^2}[ad(-a(t)\pps)]\bigg|_{t=0}(a_2\pps)\nonumber\\ & & + 30\frac{d f}{d t}[ad(-a(t)\pps)]\bigg|_{t=0}(a_3\pps ) + 10f[ad(-a_0\pps)](a_4\pps)\bigg\}\notag\\ & & + e^{-a_0\pps}\frac{d}{d t}e^{a(t)\pps}\bigg|_{t=0 } \bigg\{5\frac{d^4 f}{d t^4}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps ) + 20\frac{d^3 f}{d t^3}[ad(-a(t)\pps)]\bigg|_{t=0}(a_2\pps)\notag\\ & & + 30\frac{d^2 f}{d t^2}[ad(-a(t)\pps)]\bigg|_{t=0}(a_3\pps ) + 20\frac{d f}{d t}[ad(-a(t)\pps)]\bigg|_{t=0}(a_4\pps)+5f[ad(-a_0\pps)](a_5\pps)\bigg\}\notag\\ & & + \bigg\{\frac{d^5 f}{d t^5}[ad(-a(t)\pps)]\bigg|_{t=0}(a_1\pps ) + 5\frac{d^4 f}{d t^4}[ad(-a(t)\pps)]\bigg|_{t=0}(a_2\pps ) + 10\frac{d^3 f}{d t^3}[ad(-a(t)\pps)]\bigg|_{t=0}(a_3\pps)\notag\\ & & + 10\frac{d^2 f}{d t^2}[ad(-a(t)\pps)]\bigg|_{t=0}(a_4\pps ) + 5\frac{d f}{d t}[ad(-a(t)\pps)]\bigg|_{t=0}(a_5\pps ) + f[ad(-a_0\pps)](a_6\pps)\bigg\}\bigg]\sigma(s+a_0)\bigg|_{s=0}\nonumber\\ & & \hspace{-0.4cm}=\frac{1}{9 } ( \mu \underline{\mu } \nu ) ( \lambda \underline{\nu } \underline{\lambda } ) \skppp -\frac{1}{6 } ( \mu \underline{\mu } \nu \lambda ) ( \underline{\nu } \underline{\lambda } ) \skppp -\frac{1}{6 } ( \mu \nu \underline{\mu } \lambda ) ( \underline{\nu } \underline{\lambda } ) \skppp -\frac{1}{6 } ( \mu \nu \underline{\nu } \lambda ) ( \underline{\mu } \underline{\lambda } ) \skppp -\frac{4}{27 } ( \mu \nu \lambda ) ( \underline{\mu } \underline{\nu } \underline{\lambda } ) \skppp -\frac{4}{27 } ( \mu \nu \lambda ) ( \underline{\nu } \underline{\mu } \underline{\lambda } ) \skppp \notag\\ & & -\frac{1}{24 } ( \mu \nu \underline{\nu } \lambda ) ( \underline{\mu } \rho ) k_{\lambda}k_{\rho}\skpppp -\frac{1}{24 } ( \mu \nu \underline{\mu } \lambda ) ( \underline{\nu } \rho ) k_{\lambda}k_{\rho}\skpppp -\frac{1}{18 } ( \mu \underline{\mu } \nu ) ( \lambda \underline{\lambda } \rho ) k_{\nu}k_{\rho}\skpppp -\frac{1}{4 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \underline{\rho } ) k_{\mu}k_{\nu}\skpppp \notag\\ & & + \frac{1}{4 } ( \mu \nu \lambda \rho ) ( \underline{\nu } \underline{\lambda } ) k_{\mu}k_{\rho}\skpppp + \frac{1}{4 } ( \mu \nu \lambda \rho ) ( \underline{\mu } \underline{\lambda } ) k_{\nu}k_{\rho}\skpppp -\frac{1}{4 } ( \mu \underline{\mu } \nu \lambda ) ( \underline{\nu } \rho ) k_{\lambda}k_{\rho}\skpppp -\frac{7}{54 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \underline{\rho } ) k_{\mu}k_{\lambda}\skpppp \notag\\ & & -\frac{2}{9 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \underline{\lambda } ) k_{\mu}k_{\rho}\skpppp + \frac{2}{9 } ( \mu \nu \lambda ) ( \rho \underline{\mu } \underline{\nu } ) k_{\lambda}k_{\rho}\skpppp + \frac{2}{9 } ( \mu \nu \lambda ) ( \underline{\nu } \underline{\lambda } \rho ) k_{\mu}k_{\rho}\skpppp -\frac{2}{9 } ( \mu \nu \lambda ) ( \underline{\mu } \underline{\nu } \rho ) k_{\lambda}k_{\rho}\skpppp \notag\\ & & + \frac{11}{54 } ( \mu \underline{\mu } \nu ) ( \lambda \underline{\nu } \rho ) k_{\lambda}k_{\rho}\skpppp + \frac{5}{24 } ( \mu \nu \underline{\nu } \lambda ) ( \underline{\lambda } \rho ) k_{\mu}k_{\rho}\skpppp + \frac{5}{24 } ( \mu \nu \underline{\mu } \lambda ) ( \underline{\lambda } \rho ) k_{\nu}k_{\rho}\skpppp -\frac{1}{12 } ( \mu \nu ) ( \underline{\mu } \lambda ) ( \underline{\lambda } \rho ) k_{\nu}k_{\rho}\skpppp \notag\\ & & -\frac{1}{12 } ( \mu \nu ) ( \lambda \rho ) ( \underline{\mu } \underline{\lambda } ) k_{\nu}k_{\rho}\skpppp + \frac{1}{8 } ( \mu \nu \underline{\nu } \lambda ) ( \underline{\lambda } \rho ) k_{\rho}\skppp \frac{\partial}{\partial k^{\mu } } + \frac{1}{12 } ( \mu \nu \underline{\nu } \lambda ) ( \underline{\lambda } \rho ) k_{\mu}\skppp \frac{\partial}{\partial k^{\rho } } + \frac{1}{24 } ( \mu \nu \underline{\nu } \lambda ) ( \underline{\mu } \rho ) k_{\rho}\skppp \frac{\partial}{\partial k^{\lambda } } \notag\\ & & + \frac{1}{12 } ( \mu \nu \underline{\nu } \lambda ) ( \underline{\mu } \rho ) k_{\lambda}\skppp \frac{\partial}{\partial k^{\rho } } + \frac{1}{8 } ( \mu \nu \underline{\mu } \lambda ) ( \underline{\lambda } \rho ) k_{\rho}\skppp \frac{\partial}{\partial k^{\nu } } + \frac{1}{12 } ( \mu \nu \underline{\mu } \lambda ) ( \underline{\lambda } \rho ) k_{\nu}\skppp \frac{\partial}{\partial k^{\rho } } + \frac{1}{24 } ( \mu \nu \underline{\mu } \lambda ) ( \underline{\nu } \rho ) k_{\rho}\skppp \frac{\partial}{\partial k^{\lambda } } \notag\\ & & + \frac{1}{12 } ( \mu \nu \underline{\mu } \lambda ) ( \underline{\nu } \rho ) k_{\lambda}\skppp \frac{\partial}{\partial k^{\rho } } + \frac{1}{18 } ( \mu \underline{\mu } \nu ) ( \lambda \underline{\lambda } \rho ) k_{\rho}\skppp \frac{\partial}{\partial k^{\nu } } + \frac{1}{18 } ( \mu \underline{\mu } \nu ) ( \lambda \underline{\lambda } \rho ) k_{\nu}\skppp \frac{\partial}{\partial k^{\rho } } -\frac{1}{6 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \underline{\rho } ) k_{\nu}\skppp \frac{\partial}{\partial k^{\mu } } \notag\\ & & -\frac{1}{6 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \underline{\rho } ) k_{\mu}\skppp \frac{\partial}{\partial k^{\nu } } + \frac{1}{6 } ( \mu \nu \lambda \rho ) ( \underline{\nu } \underline{\lambda } ) k_{\rho}\skppp \frac{\partial}{\partial k^{\mu } } + \frac{1}{6 } ( \mu \nu \lambda \rho ) ( \underline{\nu } \underline{\lambda } ) k_{\mu}\skppp \frac{\partial}{\partial k^{\rho } } + \frac{1}{6 } ( \mu \nu \lambda \rho ) ( \underline{\mu } \underline{\lambda } ) k_{\rho}\skppp \frac{\partial}{\partial k^{\nu } } \notag\\ & & + \frac{1}{6 } ( \mu \nu \lambda \rho ) ( \underline{\mu } \underline{\lambda } ) k_{\nu}\skppp \frac{\partial}{\partial k^{\rho } } -\frac{1}{12 } ( \mu \underline{\mu } \nu \lambda ) ( \underline{\nu } \rho ) k_{\rho}\skppp \frac{\partial}{\partial k^{\lambda } } -\frac{2}{27 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \underline{\rho } ) k_{\lambda}\skppp \frac{\partial}{\partial k^{\mu } } -\frac{4}{27 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \underline{\rho } ) k_{\mu}\skppp \frac{\partial}{\partial k^{\lambda } } \notag\\ & & -\frac{4}{27 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \underline{\lambda } ) k_{\rho}\skppp \frac{\partial}{\partial k^{\mu } } -\frac{4}{27 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \underline{\lambda } ) k_{\mu}\skppp \frac{\partial}{\partial k^{\rho } } + \frac{4}{27 } ( \mu \nu \lambda ) ( \rho \underline{\mu } \underline{\nu } ) k_{\rho}\skppp \frac{\partial}{\partial k^{\lambda } } + \frac{4}{27 } ( \mu \nu \lambda ) ( \rho \underline{\mu } \underline{\nu } ) k_{\lambda}\skppp \frac{\partial}{\partial k^{\rho } } \notag\\ & & + \frac{4}{27 } ( \mu \nu \lambda ) ( \underline{\nu } \underline{\lambda } \rho ) k_{\rho}\skppp \frac{\partial}{\partial k^{\mu } } + \frac{4}{27 } ( \mu \nu \lambda ) ( \underline{\nu } \underline{\lambda } \rho ) k_{\mu}\skppp \frac{\partial}{\partial k^{\rho } } -\frac{4}{27 } ( \mu \nu \lambda ) ( \underline{\mu } \underline{\nu } \rho ) k_{\rho}\skppp \frac{\partial}{\partial k^{\lambda } } + \frac{4}{27 } ( \mu \underline{\mu } \nu ) ( \lambda \underline{\nu } \rho ) k_{\rho}\skppp \frac{\partial}{\partial k^{\lambda } } \notag\\ & & + \frac{2}{27 } ( \mu \underline{\mu } \nu ) ( \lambda \underline{\nu } \rho ) k_{\lambda}\skppp \frac{\partial}{\partial k^{\rho } } -\frac{2}{27 } ( \mu \nu \lambda ) ( \rho \underline{\mu } \underline{\rho } ) k_{\nu}\skppp \frac{\partial}{\partial k^{\lambda } } + \frac{4}{27 } ( \mu \nu \lambda ) ( \underline{\nu } \underline{\mu } \rho ) k_{\lambda}\skppp \frac{\partial}{\partial k^{\rho } } -\frac{2}{27 } ( \mu \underline{\mu } \nu ) ( \underline{\nu } \lambda \rho ) k_{\lambda}\skppp \frac{\partial}{\partial k^{\rho } } \notag\\ & & -\frac{1}{6 } ( \mu \nu ) ( \underline{\mu } \underline{\nu } ) ( \lambda \rho ) k_{\lambda}\skppp \frac{\partial}{\partial k^{\rho } } -\frac{1}{12 } ( \mu \nu ) ( \lambda \rho ) ( \underline{\mu } \underline{\lambda } ) k_{\nu}\skppp \frac{\partial}{\partial k^{\rho } } -\frac{1}{12 } ( \mu \nu ) ( \lambda \rho ) ( \underline{\lambda } \underline{\rho } ) k_{\mu}\skppp \frac{\partial}{\partial k^{\nu } } -\frac{1}{12 } ( \mu \nu ) ( \underline{\mu } \lambda ) ( \underline{\lambda } \rho ) k_{\nu}\skppp \frac{\partial}{\partial k^{\rho } } \notag\\ & & + \frac{1}{16 } ( \mu \nu \underline{\nu } \lambda ) ( \underline{\lambda } \rho ) \skpp \frac{\partial^2}{\partial k^{\mu}\partial k^{\rho } } -\frac{1}{16 } ( \mu \nu \underline{\nu } \lambda ) ( \underline{\mu } \rho ) \skpp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\rho } } + \frac{1}{16 } ( \mu \nu \underline{\mu } \lambda ) ( \underline{\lambda } \rho ) \skpp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } -\frac{1}{16 } ( \mu \nu \underline{\mu } \lambda ) ( \underline{\nu } \rho ) \skpp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\rho } } \notag\\ & & -\frac{1}{18 } ( \mu \underline{\mu } \nu ) ( \lambda \underline{\lambda } \rho ) \skpp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } -\frac{1}{8 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \underline{\rho } ) \skpp \frac{\partial^2}{\partial k^{\mu}\partial k^{\nu } } + \frac{1}{8 } ( \mu \nu \lambda \rho ) ( \underline{\nu } \underline{\lambda } ) \skpp \frac{\partial^2}{\partial k^{\mu}\partial k^{\rho } } + \frac{1}{8 } ( \mu \nu \lambda \rho ) ( \underline{\mu } \underline{\lambda } ) \skpp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } \notag\\ & & -\frac{1}{8 } ( \mu \underline{\mu } \nu \lambda ) ( \underline{\nu } \rho ) \skpp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\rho } } -\frac{1}{9 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \underline{\rho } ) \skpp \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } -\frac{1}{9 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \underline{\lambda } ) \skpp \frac{\partial^2}{\partial k^{\mu}\partial k^{\rho } } + \frac{1}{9 } ( \mu \nu \lambda ) ( \rho \underline{\mu } \underline{\nu } ) \skpp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\rho } } \notag\\ & & + \frac{1}{9 } ( \mu \nu \lambda ) ( \underline{\nu } \underline{\lambda } \rho ) \skpp \frac{\partial^2}{\partial k^{\mu}\partial k^{\rho } } -\frac{1}{9 } ( \mu \nu \lambda ) ( \underline{\mu } \underline{\nu } \rho ) \skpp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\rho } } -\frac{1}{6 } ( \mu \nu ) ( \underline{\mu } \lambda ) ( \underline{\nu } \rho ) k_{\lambda}\skppp \frac{\partial}{\partial k^{\rho } } -\frac{1}{8 } ( \mu \nu ) ( \underline{\mu } \lambda ) ( \underline{\lambda } \rho ) \skpp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } \notag\\ & & -\frac{1}{8 } ( \mu \nu ) ( \underline{\mu } \lambda ) ( \underline{\nu } \rho ) \skpp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\rho } } -\frac{16}{45 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\mu}k_{\lambda}k_{\rho}k_{\sigma}\skppppp -\frac{2}{5 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\mu}k_{\nu}k_{\rho}k_{\sigma}\skppppp\notag\\ & & -\frac{1}{4 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\nu}k_{\rho}k_{\sigma}\skpppp \frac{\partial}{\partial k^{\mu } } -\frac{1}{4 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\mu}k_{\rho}k_{\sigma}\skpppp \frac{\partial}{\partial k^{\nu } } -\frac{2}{9 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\lambda}k_{\rho}k_{\sigma}\skpppp \frac{\partial}{\partial k^{\mu } } \notag\\ & & -\frac{2}{9 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\mu}k_{\lambda}k_{\sigma}\skpppp \frac{\partial}{\partial k^{\rho } } -\frac{1}{27 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\mu}k_{\lambda}k_{\rho}\skpppp \frac{\partial}{\partial k^{\sigma } } + \frac{1}{12 } ( \mu \nu \lambda \rho ) ( \underline{\nu } \sigma ) k_{\mu}k_{\lambda}k_{\sigma}\skpppp \frac{\partial}{\partial k^{\rho } } \notag\\ & & + \frac{1}{12 } ( \mu \nu \lambda \rho ) ( \underline{\mu } \sigma ) k_{\nu}k_{\lambda}k_{\sigma}\skpppp \frac{\partial}{\partial k^{\rho } } + \frac{1}{6 } ( \mu \nu \underline{\nu } \lambda ) ( \rho \sigma ) k_{\mu}k_{\lambda}k_{\rho}\skpppp \frac{\partial}{\partial k^{\sigma } } + \frac{1}{6 } ( \mu \nu \underline{\mu } \lambda ) ( \rho \sigma ) k_{\nu}k_{\lambda}k_{\rho}\skpppp \frac{\partial}{\partial k^{\sigma } } \notag\\ & & + \frac{1}{27 } ( \mu \nu \lambda ) ( \rho \underline{\mu } \sigma ) k_{\nu}k_{\rho}k_{\sigma}\skpppp \frac{\partial}{\partial k^{\lambda } } + \frac{1}{9 } ( \mu \nu \lambda ) ( \rho \underline{\rho } \sigma ) k_{\mu}k_{\nu}k_{\sigma}\skpppp \frac{\partial}{\partial k^{\lambda } } + \frac{1}{9 } ( \mu \underline{\mu } \nu ) ( \lambda \rho \sigma ) k_{\nu}k_{\lambda}k_{\rho}\skpppp \frac{\partial}{\partial k^{\sigma } } \notag\\ & & -\frac{1}{4 } ( \mu \nu ) ( \underline{\mu } \lambda ) ( \rho \sigma ) k_{\nu}k_{\lambda}k_{\rho}\skpppp \frac{\partial}{\partial k^{\sigma } } -\frac{1}{12 } ( \mu \nu ) ( \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\mu}k_{\rho}k_{\sigma}\skpppp \frac{\partial}{\partial k^{\nu } } -\frac{5}{27 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\mu}k_{\rho}k_{\sigma}\skpppp \frac{\partial}{\partial k^{\lambda } } \notag\\ & & + \frac{5}{27 } ( \mu \nu \lambda ) ( \underline{\nu } \rho \sigma ) k_{\mu}k_{\lambda}k_{\rho}\skpppp \frac{\partial}{\partial k^{\sigma } } -\frac{1}{6 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\mu}k_{\nu}k_{\sigma}\skpppp \frac{\partial}{\partial k^{\rho } } -\frac{1}{12 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\nu}k_{\sigma}\skppp \frac{\partial^2}{\partial k^{\mu}\partial k^{\rho } } \notag\\ & & -\frac{1}{12 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\mu}k_{\sigma}\skppp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } -\frac{4}{27 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\lambda}k_{\sigma}\skppp \frac{\partial^2}{\partial k^{\mu}\partial k^{\rho } } + \frac{2}{27 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\mu}k_{\lambda}\skppp \frac{\partial^2}{\partial k^{\rho}\partial k^{\sigma } } \notag\\ & & + \frac{1}{12 } ( \mu \nu \lambda \rho ) ( \underline{\nu } \sigma ) k_{\lambda}k_{\sigma}\skppp \frac{\partial^2}{\partial k^{\mu}\partial k^{\rho } } -\frac{1}{6 } ( \mu \nu \lambda \rho ) ( \underline{\nu } \sigma ) k_{\mu}k_{\lambda}\skppp \frac{\partial^2}{\partial k^{\rho}\partial k^{\sigma } } + \frac{1}{12 } ( \mu \nu \lambda \rho ) ( \underline{\mu } \sigma ) k_{\lambda}k_{\sigma}\skppp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } \notag\\ & & -\frac{1}{6 } ( \mu \nu \lambda \rho ) ( \underline{\mu } \sigma ) k_{\nu}k_{\lambda}\skppp \frac{\partial^2}{\partial k^{\rho}\partial k^{\sigma } } + \frac{1}{8 } ( \mu \nu \underline{\nu } \lambda ) ( \rho \sigma ) k_{\lambda}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\mu}\partial k^{\sigma } } -\frac{1}{8 } ( \mu \nu \underline{\nu } \lambda ) ( \rho \sigma ) k_{\mu}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\sigma } } \notag\\ & & + \frac{1}{8 } ( \mu \nu \underline{\mu } \lambda ) ( \rho \sigma ) k_{\lambda}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\nu}\partial k^{\sigma } } -\frac{1}{8 } ( \mu \nu \underline{\mu } \lambda ) ( \rho \sigma ) k_{\nu}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\sigma } } -\frac{4}{27 } ( \mu \nu \lambda ) ( \rho \underline{\mu } \sigma ) k_{\nu}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\sigma } } \notag\\ & & + \frac{1}{9 } ( \mu \nu \lambda ) ( \rho \underline{\rho } \sigma ) k_{\nu}k_{\sigma}\skppp \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } -\frac{1}{9 } ( \mu \nu \lambda ) ( \rho \underline{\rho } \sigma ) k_{\mu}k_{\nu}\skppp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\sigma } } -\frac{1}{9 } ( \mu \underline{\mu } \nu ) ( \lambda \rho \sigma ) k_{\lambda}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\nu}\partial k^{\sigma } } \notag\\ & & + \frac{1}{9 } ( \mu \underline{\mu } \nu ) ( \lambda \rho \sigma ) k_{\nu}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\sigma } } -\frac{1}{6 } ( \mu \nu ) ( \underline{\mu } \lambda ) ( \rho \sigma ) k_{\lambda}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\nu}\partial k^{\sigma } } -\frac{1}{12 } ( \mu \nu ) ( \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\mu}k_{\sigma}\skppp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } \notag\\ & & + \frac{1}{12 } ( \mu \nu ) ( \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\mu}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\nu}\partial k^{\sigma } } -\frac{1}{4 } ( \mu \underline{\mu } \nu \lambda ) ( \rho \sigma ) k_{\nu}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\sigma } } + \frac{4}{27 } ( \mu \nu \lambda ) ( \underline{\nu } \rho \sigma ) k_{\lambda}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\mu}\partial k^{\sigma } } \notag\\ & & -\frac{4}{27 } ( \mu \nu \lambda ) ( \underline{\mu } \rho \sigma ) k_{\nu}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\sigma } } -\frac{2}{27 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\rho}k_{\sigma}\skppp \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } -\frac{1}{6 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\mu}k_{\nu}\skppp \frac{\partial^2}{\partial k^{\rho}\partial k^{\sigma } } \notag\\ & & -\frac{4}{27 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\mu}k_{\sigma}\skppp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\rho } } -\frac{4}{27 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\mu}k_{\rho}\skppp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\sigma } } -\frac{1}{6 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\rho}k_{\sigma}\skppp \frac{\partial^2}{\partial k^{\mu}\partial k^{\nu } } \notag\\ & & -\frac{1}{8 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\nu}\skpp \frac{\partial^3}{\partial k^{\mu}\partial k^{\rho}\partial k^{\sigma } } -\frac{1}{8 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\mu}\skpp \frac{\partial^3}{\partial k^{\nu}\partial k^{\rho}\partial k^{\sigma } } -\frac{1}{9 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\sigma}\skpp \frac{\partial^3}{\partial k^{\mu}\partial k^{\lambda}\partial k^{\rho } } \notag\\ & & -\frac{1}{9 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\rho}\skpp \frac{\partial^3}{\partial k^{\mu}\partial k^{\lambda}\partial k^{\sigma } } + \frac{1}{9 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\lambda}\skpp \frac{\partial^3}{\partial k^{\mu}\partial k^{\rho}\partial k^{\sigma } } -\frac{1}{9 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) k_{\mu}\skpp \frac{\partial^3}{\partial k^{\lambda}\partial k^{\rho}\partial k^{\sigma } } \notag\\ & & -\frac{1}{8 } ( \mu \nu \lambda \rho ) ( \underline{\nu } \sigma ) k_{\lambda}\skpp \frac{\partial^3}{\partial k^{\mu}\partial k^{\rho}\partial k^{\sigma } } -\frac{1}{8 } ( \mu \nu \lambda \rho ) ( \underline{\mu } \sigma ) k_{\lambda}\skpp \frac{\partial^3}{\partial k^{\nu}\partial k^{\rho}\partial k^{\sigma } } -\frac{1}{8 } ( \mu \nu \underline{\nu } \lambda ) ( \rho \sigma ) k_{\rho}\skpp \frac{\partial^3}{\partial k^{\mu}\partial k^{\lambda}\partial k^{\sigma } } \notag\\ & & -\frac{1}{8 } ( \mu \nu \underline{\mu } \lambda ) ( \rho \sigma ) k_{\rho}\skpp \frac{\partial^3}{\partial k^{\nu}\partial k^{\lambda}\partial k^{\sigma } } -\frac{2}{9 } ( \mu \nu \lambda ) ( \rho \underline{\mu } \sigma ) k_{\nu}\skpp \frac{\partial^3}{\partial k^{\lambda}\partial k^{\rho}\partial k^{\sigma } } -\frac{1}{9 } ( \mu \nu \lambda ) ( \rho \underline{\rho } \sigma ) k_{\nu}\skpp \frac{\partial^3}{\partial k^{\mu}\partial k^{\lambda}\partial k^{\sigma } } \notag\\ & & -\frac{1}{9 } ( \mu \underline{\mu } \nu ) ( \lambda \rho \sigma ) k_{\rho}\skpp \frac{\partial^3}{\partial k^{\nu}\partial k^{\lambda}\partial k^{\sigma } } -\frac{1}{8 } ( \mu \nu ) ( \underline{\mu } \lambda ) ( \rho \sigma ) k_{\rho}\skpp \frac{\partial^3}{\partial k^{\nu}\partial k^{\lambda}\partial k^{\sigma } } -\frac{1}{8 } ( \mu \nu ) ( \lambda \rho ) ( \underline{\lambda } \sigma ) k_{\mu}\skpp \frac{\partial^3}{\partial k^{\nu}\partial k^{\rho}\partial k^{\sigma } } \notag\\ & & -\frac{1}{8 } ( \mu \nu \lambda \rho ) ( \underline{\lambda } \sigma ) \skp \frac{\partial^4}{\partial k^{\mu}\partial k^{\nu}\partial k^{\rho}\partial k^{\sigma } } -\frac{1}{9 } ( \mu \nu \lambda ) ( \rho \underline{\nu } \sigma ) \skp \frac{\partial^4}{\partial k^{\mu}\partial k^{\lambda}\partial k^{\rho}\partial k^{\sigma } } + \frac{1}{6 } ( \mu \nu ) ( \lambda \rho ) ( \underline{\mu } \sigma ) k_{\nu}k_{\lambda}\skppp \frac{\partial^2}{\partial k^{\rho}\partial k^{\sigma } } \notag\\ & & -\frac{1}{3 } ( \mu \nu \lambda \rho ) ( \sigma \delta ) k_{\mu}k_{\nu}k_{\lambda}k_{\sigma}\skpppp \frac{\partial^2}{\partial k^{\rho}\partial k^{\delta } } -\frac{2}{9 } ( \mu \nu \lambda ) ( \rho \sigma \delta ) k_{\mu}k_{\nu}k_{\rho}k_{\sigma}\skpppp \frac{\partial^2}{\partial k^{\lambda}\partial k^{\delta } } -\frac{1}{4 } ( \mu \nu \lambda \rho ) ( \sigma \delta ) k_{\nu}k_{\lambda}k_{\sigma}\skppp \frac{\partial^3}{\partial k^{\mu}\partial k^{\rho}\partial k^{\delta } } \notag\\ & & -\frac{1}{4 } ( \mu \nu \lambda \rho ) ( \sigma \delta ) k_{\mu}k_{\lambda}k_{\sigma}\skppp \frac{\partial^3}{\partial k^{\nu}\partial k^{\rho}\partial k^{\delta } } -\frac{2}{9 } ( \mu \nu \lambda ) ( \rho \sigma \delta ) k_{\nu}k_{\rho}k_{\sigma}\skppp \frac{\partial^3}{\partial k^{\mu}\partial k^{\lambda}\partial k^{\delta } } -\frac{2}{9 } ( \mu \nu \lambda ) ( \rho \sigma \delta ) k_{\mu}k_{\nu}k_{\sigma}\skppp \frac{\partial^3}{\partial k^{\lambda}\partial k^{\rho}\partial k^{\delta } } \notag\\ & & -\frac{1}{4 } ( \mu \nu \lambda \rho ) ( \sigma \delta ) k_{\lambda}k_{\sigma}\skpp \frac{\partial^4}{\partial k^{\mu}\partial k^{\nu}\partial k^{\rho}\partial k^{\delta } } -\frac{2}{9 } ( \mu \nu \lambda ) ( \rho \sigma \delta ) k_{\nu}k_{\sigma}\skpp \frac{\partial^4}{\partial k^{\mu}\partial k^{\lambda}\partial k^{\rho}\partial k^{\delta } } -\frac{1}{6 } ( \mu \nu ) ( \lambda \rho ) ( \sigma \delta ) k_{\mu}k_{\lambda}k_{\sigma}\skppp \frac{\partial^3}{\partial k^{\nu}\partial k^{\rho}\partial k^{\delta}}\notag\\ & & + \mbox{traceless terms}\;. \end{aligned}\ ] ] finally , we list down the @xmath359 , @xmath1 , @xmath123 , @xmath0 order low energy expansion result for @xmath100 used in ( [ bexp ] ) , @xmath368 @xmath369 \gamf \tau + 24i ( \mu d^{\nu}s_\omega)\gamma_{\nu}\tau \frac{\partial}{\partial k^{\mu } } + 24 ( \mu d^{\nu}p_\omega ) \gamma_{\nu}\gamf \tau \frac{\partial}{\partial k^{\mu } } -24 a_\omega^{\mu}(\nu s_\omega)\gamma_{\mu}\gamf \tau \frac{\partial}{\partial k^{\nu}}\notag\\ & & -24 ( \mu a_\omega^{\nu } ) s_\omega \gamma_{\nu}\gamf \tau \frac{\partial}{\partial k^{\mu } } + 24i a_\omega^{\mu}(\nu p_\omega)\gamma_{\mu}\tau \frac{\partial}{\partial k^{\nu } } + 24i ( \mu a_\omega^{\nu } ) p_\omega \gamma_{\nu}\tau \frac{\partial}{\partial k^{\mu } } -24 s_\omega ( \mu a_\omega^{\nu } ) \gamma_{\nu}\gamf \tau \frac{\partial}{\partial k^{\mu } } \notag\\ & & -24 ( \mu s_\omega)a_\omega^{\nu}\gamma_{\nu}\gamf \tau \frac{\partial}{\partial k^{\mu } } + 24i p_\omega ( \mu a_\omega^{\nu } ) \gamma_{\nu}\tau \frac{\partial}{\partial k^{\mu } } + 24i ( \mu p_\omega)a_\omega^{\nu}\gamma_{\nu}\tau \frac{\partial}{\partial k^{\mu } } -6 ( \mu \nu \lambda \rho ) \gamma_{\lambda}\gamma_{\rho}\tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\nu } } \notag\\ & & + 6 a_\omega^{\mu}(\nu \lambda a_\omega^{\rho } ) \gamma_{\mu}\gamma_{\rho}\tau \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } + 12 ( \mu a_\omega^{\nu } ) ( \lambda a_\omega^{\rho } ) \gamma_{\nu}\gamma_{\rho}\tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } + 6 ( \mu \nu a_\omega^{\lambda } ) a_\omega^{\rho}\gamma_{\lambda}\gamma_{\rho}\tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\nu } } \notag\\ & & -6 a_\omega^{\mu}(\nu \lambda a_\omega^{\rho } ) \gamma_{\rho}\gamma_{\mu}\tau \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } -12 ( \mu a_\omega^{\nu } ) ( \lambda a_\omega^{\rho } ) \gamma_{\rho}\gamma_{\nu}\tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } -6 ( \mu \nu a_\omega^{\lambda } ) a_\omega^{\rho}\gamma_{\rho}\gamma_{\lambda}\tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\nu } } \notag\\ & & + 6i ( \mu\nu(d^{\lambda}a_\omega^{\rho}-d^{\rho}a_\omega^{\lambda}))\gamma_{\lambda}\gamma_{\rho}\gamf \tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\nu } } -3 ( \mu \nu \underline{\nu } \lambda ) \tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } -3 ( \mu \nu \underline{\mu } \lambda ) \tau \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } + 12 a_\omega^{\mu}(\nu \lambda a_{\omega\mu } ) \tau \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda}}\notag\\ & & + 24 ( \mu a_\omega^{\nu } ) ( \lambda a_{\omega\nu } ) \tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } + 12 ( \mu \nu a_\omega^{\lambda } ) a_{\omega\lambda}\tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\nu } } -12 ( \mu \nu ) ( \underline{\mu } \underline{\nu } ) \tau \sk \skpp -12 ( \mu \nu ) ( \underline{\mu } \underline{\nu } ) \tau \skp^2 \notag\\ & & -6 ( \mu \nu ) ( \underline{\mu } \lambda ) \tau \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } + 8 ( \mu \underline{\mu } \nu ) a_\omega^{\lambda}\gamma_{\lambda}\gamf \tau k_{\nu}\skpp -8 ( \mu \underline{\mu } \nu ) a_\omega^{\lambda}\gamma_{\lambda}\gamf \tau \skp \frac{\partial}{\partial k^{\nu } } -24 ( \mu \underline{\mu } s_\omega)\tau \skp-24i ( \mu \underline{\mu } p)\gamf \tau \skp \notag\\ & & -48 ( \mu \nu s_\omega)\tau k_{\mu}k_{\nu}\skpp -48i ( \mu \nu p_\omega)\gamf \tau k_{\mu}k_{\nu}\skpp -24 ( \mu \nu s_\omega)\tau k_{\mu}\skp \frac{\partial}{\partial k^{\nu } } -24i ( \mu \nu p_\omega)\gamf \tau k_{\mu}\skp \frac{\partial}{\partial k^{\nu } } \notag\\ & & -24 ( \mu \nu s_\omega)\tau k_{\nu}\skp \frac{\partial}{\partial k^{\mu } } -24i ( \mu \nu p_\omega)\gamf \tau k_{\nu}\skp \frac{\partial}{\partial k^{\mu } } -24 ( \mu \nu s_\omega)\tau \sk \frac{\partial^2}{\partial k^{\mu}\partial k^{\nu } } -8i ( \mu \nu \underline{\nu } \lambda ) \gamma_{\mu}\tau k_{\lambda}\skpp \notag\\ & & + 8i ( \mu \nu \underline{\nu } \lambda ) \gamma_{\mu}\tau \skp \frac{\partial}{\partial k^{\lambda } } + 8 a_\omega^{\mu}(\nu \underline{\nu } \lambda ) \gamma_{\mu}\gamf \tau k_{\lambda}\skpp -8 a_\omega^{\mu}(\nu \underline{\nu } \lambda ) \gamma_{\mu}\gamf \tau \skp \frac{\partial}{\partial k^{\lambda } } + 8 ( \mu \nu \underline{\nu } \lambda ) \tau k_{\mu}k_{\lambda}\sk \skppp\notag\\ & & + 8 ( \mu \nu \underline{\mu } \lambda ) \tau k_{\nu}k_{\lambda}\sk \skppp + 24 ( \mu \nu \underline{\nu } \lambda ) \tau k_{\mu}k_{\lambda}\skp \skpp + 24 ( \mu \nu \underline{\mu } \lambda ) \tau k_{\nu}k_{\lambda}\skp \skpp + 6 ( \mu \nu \underline{\nu } \lambda ) \tau k_{\lambda}\sk \skpp \frac{\partial}{\partial k^{\mu } } \notag\\ & & + 6 ( \mu \nu \underline{\mu } \lambda ) \tau k_{\lambda}\sk \skpp \frac{\partial}{\partial k^{\nu } } -12 ( \mu \underline{\mu } \nu \lambda ) \tau k_{\nu}\sk \skpp \frac{\partial}{\partial k^{\lambda } } -6 ( \mu \nu \underline{\nu } \lambda ) \tau k_{\mu}\sk \skpp \frac{\partial}{\partial k^{\lambda } } -6 ( \mu \nu \underline{\mu } \lambda ) \tau k_{\nu}\sk \skpp \frac{\partial}{\partial k^{\lambda } } \notag\\ & & + 6 ( \mu \nu \underline{\nu } \lambda ) \tau k_{\lambda}\skp^2 \frac{\partial}{\partial k^{\mu } } + 6 ( \mu \nu \underline{\mu } \lambda ) \tau k_{\lambda}\skp^2 \frac{\partial}{\partial k^{\nu } } -12 ( \mu \underline{\mu } \nu \lambda ) \tau k_{\nu}\skp^2 \frac{\partial}{\partial k^{\lambda } } -6 ( \mu \nu \underline{\nu } \lambda ) \tau k_{\mu}\skp^2 \frac{\partial}{\partial k^{\lambda}}\notag\\ & & -6 ( \mu \nu \underline{\mu } \lambda ) \tau k_{\nu}\skp^2 \frac{\partial}{\partial k^{\lambda } } -24 ( \mu \nu ) s_\omega \tau k_{\mu}\skp \frac{\partial}{\partial k^{\nu } } + 24i ( \mu \nu ) p_\omega \gamf \tau k_{\mu}\skp \frac{\partial}{\partial k^{\nu } } -6 ( \mu \nu \lambda \rho ) \tau k_{\lambda}\frac{\partial^3}{\partial k^{\mu}\partial k^{\nu}\partial k^{\rho } } \notag\\ & & -6 ( \mu \nu \underline{\nu } \lambda ) \tau \sk \skp \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } -6 ( \mu \nu \underline{\mu } \lambda ) \tau \sk \skp \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } -12i ( \mu \nu ) ( \lambda a_{\omega\mu } ) \gamf \tau \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } -16 ( \mu \nu \underline{\nu } a_\omega^{\lambda } ) \gamma_{\lambda}\gamf \tau k_{\mu}\skpp \notag\\ & & -8 ( \mu \nu \underline{\nu } a_\omega^{\lambda } ) \gamma_{\lambda}\gamf \tau \skp \frac{\partial}{\partial k^{\mu } } -16 ( \mu \nu \underline{\mu } a_\omega^{\lambda } ) \gamma_{\lambda}\gamf \tau k_{\nu}\skpp -16 ( \mu \underline{\mu } \nu a_\omega^{\lambda } ) \gamma_{\lambda}\gamf \tau k_{\nu}\skpp -32 ( \mu \nu \lambda a^{\rho } ) \gamma_{\rho}\gamf \tau k_{\mu}k_{\nu}k_{\lambda}\skppp\notag\\ & & -16 ( \mu \nu \lambda a_\omega^{\rho } ) \gamma_{\rho}\gamf \tau k_{\nu}k_{\lambda}\skpp \frac{\partial}{\partial k^{\mu } } -8 ( \mu \nu \underline{\mu } a_\omega^{\lambda } ) \gamma_{\lambda}\gamf \tau \skp \frac{\partial}{\partial k^{\nu } } -16 ( \mu \nu \lambda a_\omega^{\rho } ) \gamma_{\rho}\gamf \tau k_{\mu}k_{\lambda}\skpp \frac{\partial}{\partial k^{\nu } } \notag\\ & & -8 ( \mu \nu \lambda a_\omega^{\rho } ) \gamma_{\rho}\gamf \tau k_{\lambda}\skp \frac{\partial^2}{\partial k^{\mu}\partial k^{\nu } } -8 ( \mu \underline{\mu } \nu a_\omega^{\lambda } ) \gamma_{\lambda}\gamf \tau \skp \frac{\partial}{\partial k^{\nu } } -16 ( \mu \nu \lambda a_\omega^{\rho } ) \gamma_{\rho}\gamf \tau k_{\mu}k_{\nu}\skpp \frac{\partial}{\partial k^{\lambda}}\notag\\ & & -8 ( \mu \nu \lambda a^{\rho } ) \gamma_{\rho}\gamf \tau k_{\nu}\skp \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } -8 ( \mu \nu \lambda a_\omega^{\rho } ) \gamma_{\rho}\gamf \tau k_{\mu}\skp \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } -8 ( \mu \nu \lambda a_\omega^{\rho } ) \gamma_{\rho}\gamf \tau \sk \frac{\partial^3}{\partial k^{\mu}\partial k^{\nu}\partial k^{\lambda } } \notag\\ & & -8i ( \mu \nu \lambda ) a_{\omega\nu}\gamf \tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } -12i ( \mu a_\omega^{\nu } ) ( \underline{\nu } \lambda ) \gamf \tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } -8i a_\omega^{\mu}(\nu \underline{\mu } \lambda ) \gamf \tau \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } -24 s_\omega ( \mu \nu ) \tau k_{\mu}\skp \frac{\partial}{\partial k^{\nu } } \notag\\ & & -24i p_\omega ( \mu \nu ) \gamf \tau k_{\mu}\skp \frac{\partial}{\partial k^{\nu } } -16 ( \mu \nu ) ( \underline{\mu } \lambda ) \tau k_{\nu}k_{\lambda}\sk \skppp -48 ( \mu \nu ) ( \underline{\mu } \lambda ) \tau k_{\nu}k_{\lambda}\skp \skpp + 12 ( \mu \nu ) ( \underline{\mu } \lambda ) \tau k_{\nu}\sk \skpp \frac{\partial}{\partial k^{\lambda } } \notag\\ & & -12 ( \mu \nu ) ( \underline{\mu } \lambda ) \tau k_{\lambda}\sk \skpp \frac{\partial}{\partial k^{\nu } } + 12 ( \mu \nu ) ( \underline{\mu } \lambda ) \tau k_{\nu}\skp^2 \frac{\partial}{\partial k^{\lambda } } -12 ( \mu \nu ) ( \underline{\mu } \lambda ) \tau k_{\lambda}\skp^2 \frac{\partial}{\partial k^{\nu } } \notag\\ & & -12 ( \mu \nu ) ( \underline{\mu } \lambda ) \tau \sk \skp \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } -16 ( \mu \nu \lambda ) a_\omega^{\rho}\gamma_{\rho}\gamf \tau k_{\mu}k_{\nu}\skpp \frac{\partial}{\partial k^{\lambda } } -24 ( \mu \nu ) ( \lambda a_\omega^{\rho } ) \gamma_{\rho}\gamf \tau k_{\mu}\skp \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } \notag\\ & & -16 ( \mu \nu \lambda ) a_\omega^{\rho}\gamma_{\rho}\gamf \tau k_{\nu}\skp \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } -24 ( \mu a_\omega^{\nu } ) ( \underline{\mu } \lambda ) \gamma_{\nu}\gamf \tau \skp \frac{\partial}{\partial k^{\lambda } } -48 ( \mu a_\omega^{\nu } ) ( \lambda \rho ) \gamma_{\nu}\gamf \tau k_{\mu}k_{\lambda}\skpp \frac{\partial}{\partial k^{\rho } } \notag\\ & & -24 ( \mu a_\omega^{\nu } ) ( \lambda \rho ) \gamma_{\nu}\gamf \tau k_{\lambda}\skp \frac{\partial^2}{\partial k^{\mu}\partial k^{\rho } } -4i ( \mu \nu \lambda a_{\omega\mu } ) \gamf \tau \frac{\partial^2}{\partial k^{\nu}\partial k^{\lambda } } -4i ( \mu \nu \lambda a_{\omega\nu } ) \gamf \tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\lambda } } \notag\\ & & -4i ( \mu \nu \lambda a_{\omega\lambda } ) \gamf \tau \frac{\partial^2}{\partial k^{\mu}\partial k^{\nu } } -8i ( \mu \nu \lambda a_\omega^{\rho } ) \gamf \tau k_{\rho}\frac{\partial^3}{\partial k^{\mu}\partial k^{\nu}\partial k^{\lambda } } + 16i ( \mu \nu \lambda \rho ) \gamma_{\mu}\tau k_{\nu}k_{\lambda}\skpp \frac{\partial}{\partial k^{\rho } } \notag\\ & & + 16i ( \mu \nu \lambda \rho ) \gamma_{\mu}\tau k_{\lambda}\skp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } -24i ( \mu \nu ) ( \lambda \rho ) \gamma_{\mu}\tau k_{\lambda}\skp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } + 24i ( \mu \nu ) ( \lambda \rho ) \gamma_{\lambda}\tau k_{\mu}\skp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } \notag\\ & & -16i ( \mu \underline{\mu } \nu \lambda ) \gamma_{\nu}\tau k_{\lambda}\skpp -8i ( \mu \underline{\mu } \nu \lambda ) \gamma_{\nu}\tau \skp \frac{\partial}{\partial k^{\lambda } } + 16i ( \mu \nu \underline{\mu } \lambda ) \gamma_{\lambda}\tau k_{\nu}\skpp + 16i ( \mu \nu \underline{\nu } \lambda ) \gamma_{\lambda}\tau k_{\mu}\skpp\notag\\ & & -32i ( \mu \nu \lambda \rho ) \gamma_{\lambda}\tau k_{\mu}k_{\nu}k_{\rho}\skppp -16i ( \mu \nu \lambda \rho ) \gamma_{\lambda}\tau k_{\mu}k_{\nu}\skpp \frac{\partial}{\partial k^{\rho } } + 8i ( \mu \nu \underline{\mu } \lambda ) \gamma_{\lambda}\tau \skp \frac{\partial}{\partial k^{\nu } } -16i ( \mu \nu \lambda \rho ) \gamma_{\lambda}\tau k_{\mu}k_{\rho}\skpp \frac{\partial}{\partial k^{\nu } } \notag\\ & & -8i ( \mu \nu \lambda \rho ) \gamma_{\lambda}\tau k_{\mu}\skp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } + 8i ( \mu \nu \underline{\nu } \lambda ) \gamma_{\lambda}\tau \skp \frac{\partial}{\partial k^{\mu } } -16i ( \mu \nu \lambda \rho ) \gamma_{\lambda}\tau k_{\nu}k_{\rho}\skpp \frac{\partial}{\partial k^{\mu } } -8i ( \mu \nu \lambda \rho ) \gamma_{\lambda}\tau k_{\nu}\skp \frac{\partial^2}{\partial k^{\mu}\partial k^{\rho } } \notag\\ & & -8i ( \mu \nu \lambda \rho ) \gamma_{\lambda}\tau k_{\rho}\skp \frac{\partial^2}{\partial k^{\mu}\partial k^{\nu } } -16 a^{\mu}(\nu \lambda \rho ) \gamma_{\mu}\gamf \tau k_{\nu}k_{\lambda}\skpp \frac{\partial}{\partial k^{\rho } } -16 a^{\mu}(\nu \lambda \rho ) \gamma_{\mu}\gamf \tau k_{\lambda}\skp \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } \notag\\ & & -16 ( \mu \nu \lambda \rho ) \tau k_{\mu}k_{\nu}k_{\lambda}\sk \skppp \frac{\partial}{\partial k^{\rho } } -48 ( \mu \nu \lambda \rho ) \tau k_{\mu}k_{\nu}k_{\lambda}\skp \skpp \frac{\partial}{\partial k^{\rho } } -12 ( \mu \nu \lambda \rho ) \tau k_{\nu}k_{\lambda}\sk \skpp \frac{\partial^2}{\partial k^{\mu}\partial k^{\rho } } \notag\\ & & -12 ( \mu \nu \lambda \rho ) \tau \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } -12 ( \mu \nu \lambda \rho ) \tau k_{\nu}k_{\lambda}\skp^2 \frac{\partial^2}{\partial k^{\mu}\partial k^{\rho } } -12 ( \mu \nu \lambda \rho ) \tau k_{\mu}k_{\lambda}\skp^2 \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } \notag\\ & & -12 ( \mu \nu \lambda \rho ) \tau \frac{\partial^3}{\partial k^{\mu}\partial k^{\nu}\partial k^{\rho } } + 24i ( \mu \nu ) ( \underline{\mu } \lambda ) \gamma_{\nu}\tau \skp \frac{\partial}{\partial k^{\lambda } } -48i ( \mu \nu ) ( \lambda \rho ) \gamma_{\mu}\tau k_{\nu}k_{\lambda}\skpp \frac{\partial}{\partial k^{\rho } } \notag\\ & & -24 ( \mu \nu ) ( \lambda \rho ) \tau \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho } } -24 ( \mu \nu ) ( \lambda \rho ) \tau k_{\mu}k_{\lambda}\skp^2 \frac{\partial^2}{\partial k^{\nu}\partial k^{\rho}}\ ; , \end{aligned}\ ] ] @xmath370 @xmath371 @xmath373\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{2}=\int dk\bigg[2 \tau^3 -\frac{1}{3 } \tau^4 k^2 -12 \tau^4 \sk^2 -\frac{1}{18 } \tau^5 k^4 + \frac{7}{3 } \tau^5 k^2 \sk^2 + 8 \tau^5 \sk^4 + \frac{1}{180 } \tau^6 k^6 -\frac{2}{45 } \tau^6 k^4 \sk^2 -\frac{4}{5 } \tau^6 k^2 \sk^4 -\frac{16}{15 } \tau^6 \sk^6\bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{3}=\int dk\bigg[- \tau^3 + \frac{1}{2 } \tau^4 k^2 + 6 \tau^4 \sk^2 -\frac{1}{18 } \tau^5 k^4 -\frac{4}{3 } \tau^5 k^2 \sk^2 -4 \tau^5 \sk^4 + \frac{1}{360 } \tau^6 k^6 + \frac{1}{15 } \tau^6 k^4 \sk^2 + \frac{2}{5 } \tau^6 k^2 \sk^4 + \frac{8}{15 } \tau^6 \sk^6\bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{4}= \int dk\bigg[- \tau^3 + \frac{1}{6 } \tau^4 k^2 + 6 \tau^4 \sk^2 - \tau^5 k^2 \sk^2 -4 \tau^5 \sk^4+\frac{1}{360 } \tau^6 k^6 -\frac{1}{45 } \tau^6 k^4 \sk^2 + \frac{2}{5 } \tau^6 k^2 \sk^4 + \frac{8}{15 } \tau^6 \sk^6 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{5}= \int dk\bigg[\frac{1}{3 } \tau^3 -\frac{1}{6 } \tau^4 k^2 -2 \tau^4 \sk^2 + \frac{1}{3 } \tau^5 k^2 \sk^2 + \frac{4}{3 } \tau^5 \sk^4 + \frac{1}{1080 } \tau^6 k^6 + \frac{1}{45 } \tau^6 k^4 \sk^2 -\frac{2}{15 } \tau^6 k^2 \sk^4 -\frac{8}{45 } \tau^6 \sk^6 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{6}= \int dk\bigg[-\frac{4}{15 } \tau^3 + \frac{1}{3 } \tau^4 k^2 + \frac{2}{5 } \tau^4 \sk^2 -\frac{1}{10 } \tau^5 k^4 + \frac{2}{3 } \tau^5 k^4 \sk^2 \skp^2 -\frac{2}{5 } \tau^5 k^2 \sk^2 + \frac{1}{120 } \tau^6 k^6 + \frac{1}{15 } \tau^6 k^4 \sk^2 -\frac{1}{30 } \tau^6 k^6 \sk^2 \skp^2\notag\\ & & \hspace{0.4 cm } -\frac{4}{15 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{7}= \int dk\bigg[-\frac{11}{5 } \tau^3 + 3 \tau^3 k^2 \skp^2 + \frac{13}{10 } \tau^4 k^2 + \frac{92}{15 } \tau^4 \sk^2 -18 \tau^4 k^2 \sk^2 \skp^2 -\frac{2}{9 } \tau^4 k^4 \skp^2 -\frac{17}{90 } \tau^5 k^4 -\frac{11}{5 } \tau^5 k^2 \sk^2 -\frac{8}{5 } \tau^5 \sk^4 \notag\\ & & + \frac{14}{9 } \tau^5 k^4 \sk^2 \skp^2 + 12 \tau^5 k^2 \sk^4 \skp^2 + \frac{1}{120 } \tau^6 k^6 + \frac{2}{15 } \tau^6 k^4 \sk^2 + \frac{2}{5 } \tau^6 k^2 \sk^4 -\frac{1}{30 } \tau^6 k^6 \sk^2 \skp^2 -\frac{8}{15 } \tau^6 k^4 \sk^4 \skp^2 -\frac{8}{5 } \tau^6 k^2 \sk^6 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{8}= \int dk\bigg[\frac{4}{15 } \tau^3 + \frac{1}{5 } \tau^4 k^2 -\frac{2}{5 } \tau^4 \sk^2 -\frac{1}{10 } \tau^5 k^4 + \frac{2}{3 } \tau^5 k^4 \sk^2 \skp^2 -\frac{2}{5 } \tau^5 k^2 \sk^2 + \frac{1}{120 } \tau^6 k^6 + \frac{1}{15 } \tau^6 k^4 \sk^2 -\frac{1}{30 } \tau^6 k^6 \sk^2 \skp^2\notag\\ & & \hspace{0.4 cm } -\frac{4}{15 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{9}= \int dk\bigg[-\frac{23}{15 } \tau^3 + \frac{59}{60 } \tau^4 k^2 + \frac{17}{15 } \tau^4 \sk^2 -\frac{2}{9 } \tau^4 k^4 \skp^2 -\frac{7}{45 } \tau^5 k^4 + \frac{8}{9 } \tau^5 k^4 \sk^2 \skp^2 -\frac{2}{5 } \tau^5 k^2 \sk^2 + \frac{1}{120 } \tau^6 k^6 + \frac{1}{15 } \tau^6 k^4 \sk^2 \notag\\ & & \hspace{0.4cm}-\frac{1}{30 } \tau^6 k^6 \sk^2 \skp^2-\frac{4}{15 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{10}= \int dk\bigg[-\frac{4}{5 } \tau^3 + \frac{19}{30 } \tau^4 k^2 + \frac{8}{15 } \tau^4 \sk^2 -\frac{2}{9 } \tau^4 k^4 \skp^2 -\frac{7}{45 } \tau^5 k^4 + \frac{8}{9 } \tau^5 k^4 \sk^2 \skp^2 -\frac{2}{5 } \tau^5 k^2 \sk^2 + \frac{1}{120 } \tau^6 k^6 + \frac{1}{15 } \tau^6 k^4 \sk^2 \notag\\ & & \hspace{0.4cm}-\frac{1}{30 } \tau^6 k^6 \sk^2 \skp^2-\frac{4}{15 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{11}= \int dk\bigg[-\frac{2}{5 } \tau^3 + \frac{1}{30 } \tau^4 k^2 + \frac{4}{15 } \tau^4 \sk^2 + \frac{2}{9 } \tau^4 k^4 \skp^2 -\frac{1}{30 } \tau^5 k^4 -\frac{2}{3 } \tau^5 k^4 \sk^2 \skp^2 + \frac{2}{5 } \tau^5 k^2 \sk^2 + \frac{1}{120 } \tau^6 k^6 -\frac{1}{15 } \tau^6 k^4 \sk^2\notag\\ & & \hspace{0.4cm}-\frac{1}{30 } \tau^6 k^6 \sk^2 \skp^2 + \frac{4}{15 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{12}=\int dk\bigg[\frac{16}{15 } \tau^3 -\frac{2}{5 } \tau^4 k^2 -\frac{14}{15 } \tau^4 \sk^2 + \frac{2}{9 } \tau^4 k^4 \skp^2 -\frac{1}{30 } \tau^5 k^4 -\frac{2}{3 } \tau^5 k^4 \sk^2 \skp^2 + \frac{2}{5 } \tau^5 k^2 \sk^2 + \frac{1}{120 } \tau^6 k^6 -\frac{1}{15 } \tau^6 k^4 \sk^2\notag\\ & & \hspace{0.4cm}-\frac{1}{30 } \tau^6 k^6 \sk^2 \skp^2 + \frac{4}{15 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{13}= \int dk\bigg[-\frac{28}{15 } \tau^3 + \frac{9}{10 } \tau^4 k^2 -\frac{2}{9 } \tau^4 k^4 \skp^2 + \frac{4}{5 } \tau^4 \sk^2 -\frac{7}{45 } \tau^5 k^4 + \frac{8}{9 } \tau^5 k^4 \sk^2 \skp^2 -\frac{2}{5 } \tau^5 k^2 \sk^2 + \frac{1}{120 } \tau^6 k^6 + \frac{1}{15 } \tau^6 k^4 \sk^2\notag\\ & & \hspace{0.4cm}-\frac{1}{30 } \tau^6 k^6 \sk^2 \skp^2 -\frac{4}{15 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{14}= \int dk\bigg[\frac{13}{15 } \tau^3 + \frac{23}{60 } \tau^4 k^2 -\frac{7}{15 } \tau^4 \sk^2 -\frac{2}{9 } \tau^4 k^4 \skp^2 -\frac{7}{45 } \tau^5 k^4 + \frac{8}{9 } \tau^5 k^4 \sk^2 \skp^2 -\frac{2}{5 } \tau^5 k^2 \sk^2 + \frac{1}{120 } \tau^6 k^6 + \frac{1}{15 } \tau^6 k^4 \sk^2\notag\\ & & \hspace{0.4cm}-\frac{1}{30 } \tau^6 k^6 \sk^2 \skp^2 -\frac{4}{15 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{15}= \int dk\bigg[\frac{11}{5 } \tau^3 -3 \tau^3 k^2 \skp^2 -\frac{19}{30 } \tau^4 k^2 -\frac{92}{15 } \tau^4 \sk^2 + 18 \tau^4 k^2 \sk^2 \skp^2 + \frac{4}{9 } \tau^4 k^4 \skp^2 + \frac{11}{5 } \tau^5 k^2 \sk^2 + \frac{8}{5 } \tau^5 \sk^4-\frac{4}{3 } \tau^5 k^4 \sk^2 \skp^2 \notag\\ & & \hspace{0.4 cm } -12 \tau^5 k^2 \sk^4 \skp^2+\frac{1}{120 } \tau^6 k^6 -\frac{2}{15 } \tau^6 k^4 \sk^2 -\frac{2}{5 } \tau^6 k^2 \sk^4 -\frac{1}{30 } \tau^6 k^6 \sk^2 \skp^2 + \frac{8}{15 } \tau^6 k^4 \sk^4 \skp^2 + \frac{8}{5 } \tau^6 k^2 \sk^6 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{16}= \int dk\bigg[-\frac{11}{5 } \tau^3 + 3 \tau^3 k^2 \skp^2 + \frac{13}{10 } \tau^4 k^2 + \frac{92}{15 } \tau^4 \sk^2 -18 \tau^4 k^2 \sk^2 \skp^2 -\frac{4}{9 } \tau^4 k^4 \skp^2 -\frac{11}{45 } \tau^5 k^4 -\frac{11}{5 } \tau^5 k^2 \sk^2 -\frac{8}{5 } \tau^5 \sk^4 \notag\\ & & \hspace{0.4cm}+\frac{16}{9 } \tau^5 k^4 \sk^2 \skp^2 + 12 \tau^5 k^2 \sk^4 \skp^2 + \frac{1}{120 } \tau^6 k^6 + \frac{2}{15 } \tau^6 k^4 \sk^2 + \frac{2}{5 } \tau^6 k^2 \sk^4 -\frac{1}{30 } \tau^6 k^6 \sk^2 \skp^2 -\frac{8}{15 } \tau^6 k^4 \sk^4 \skp^2 -\frac{8}{5 } \tau^6 k^2 \sk^6 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{17}= \int dk\bigg[-\frac{11}{15 } \tau^3 + \frac{49}{60 } \tau^4 k^2 + \frac{3}{5 } \tau^4 \sk^2 -\frac{2}{9 } \tau^4 k^4 \skp^2 -\frac{17}{90 } \tau^5 k^4 + \frac{10}{9 } \tau^5 k^4 \sk^2 \skp^2 -\frac{8}{15 } \tau^5 k^2 \sk^2 + \frac{1}{90 } \tau^6 k^6 + \frac{4}{45 } \tau^6 k^4 \sk^2\notag\\ & & \hspace{0.4cm}-\frac{2}{45 } \tau^6 k^6 \sk^2 \skp^2 -\frac{16}{45 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{18}= \int dk\bigg[-\frac{29}{15 } \tau^3 + \frac{67}{60 } \tau^4 k^2 + \frac{11}{15 } \tau^4 \sk^2 -\frac{2}{9 } \tau^4 k^4 \skp^2 -\frac{17}{90 } \tau^5 k^4 + \frac{10}{9 } \tau^5 k^4 \sk^2 \skp^2 -\frac{8}{15 } \tau^5 k^2 \sk^2 + \frac{1}{90 } \tau^6 k^6 + \frac{4}{45 } \tau^6 k^4 \sk^2\notag\\ & & \hspace{0.4cm}-\frac{2}{45 } \tau^6 k^6 \sk^2 \skp^2 -\frac{16}{45 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{19}= \int dk\bigg[\frac{4}{3 } \tau^3 -\frac{2}{5 } \tau^4 k^2 -\frac{2}{3 } \tau^4 \sk^2 + \frac{1}{9 } \tau^4 k^4 \skp^2 -\frac{1}{45 } \tau^5 k^4 -\frac{4}{9 } \tau^5 k^4 \sk^2 \skp^2 + \frac{4}{15 } \tau^5 k^2 \sk^2 + \frac{1}{180 } \tau^6 k^6 -\frac{2}{45 } \tau^6 k^4 \sk^2\notag\\ & & \hspace{0.4cm}-\frac{1}{45 } \tau^6 k^6 \sk^2 \skp^2 + \frac{8}{45 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{20}= \int dk\bigg[-\frac{4}{5 } \tau^3 + 2 \tau^3 k^2 \skp^2 + \frac{7}{10 } \tau^4 k^2 + \frac{58}{15 } \tau^4 \sk^2 -12 \tau^4 k^2 \sk^2 \skp^2 -\frac{2}{9 } \tau^4 k^4 \skp^2 -\frac{13}{90 } \tau^5 k^4 -\frac{22}{15 } \tau^5 k^2 \sk^2 -\frac{16}{15 } \tau^5 \sk^4 \notag\\ & & \hspace{0.4cm}+\frac{10}{9 } \tau^5 k^4 \sk^2 \skp^2 + 8 \tau^5 k^2 \sk^4 \skp^2 + \frac{1}{180 } \tau^6 k^6 + \frac{4}{45 } \tau^6 k^4 \sk^2 + \frac{4}{15 } \tau^6 k^2 \sk^4 -\frac{1}{45 } \tau^6 k^6 \sk^2 \skp^2 -\frac{16}{45 } \tau^6 k^4 \sk^4 \skp^2 -\frac{16}{15 } \tau^6 k^2 \sk^6 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{21}= \int dk\bigg[\frac{1}{15 } \tau^3 -\frac{1}{30 } \tau^4 k^2 + \frac{2}{9 } \tau^4 k^4 \skp^2 -\frac{4}{15 } \tau^4 \sk^2 -\frac{1}{45 } \tau^5 k^4 -\frac{4}{9 } \tau^5 k^4 \sk^2 \skp^2 + \frac{4}{15 } \tau^5 k^2 \sk^2 + \frac{1}{180 } \tau^6 k^6 -\frac{2}{45 } \tau^6 k^4 \sk^2 \notag\\ & & \hspace{0.4cm}-\frac{1}{45 } \tau^6 k^6 \sk^2 \skp^2+\frac{8}{45 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{22}= \int dk\bigg[-2 \tau^3 k^2 \skp^2 + \frac{4}{5 } \tau^3 -\frac{58}{15 } \tau^4 \sk^2 -\frac{1}{5 } \tau^4 k^2 + 12 \tau^4 k^2 \sk^2 \skp^2 + \frac{1}{3 } \tau^4 k^4 \skp^2 + \frac{22}{15 } \tau^5 k^2 \sk^2 + \frac{16}{15 } \tau^5 \sk^4 -\frac{8}{9 } \tau^5 k^4 \sk^2 \skp^2 \notag\\ & & \hspace{0.4cm}-8 \tau^5 k^2 \sk^4 \skp^2+\frac{1}{180 } \tau^6 k^6 -\frac{4}{45 } \tau^6 k^4 \sk^2 -\frac{4}{15 } \tau^6 k^2 \sk^4 -\frac{1}{45 } \tau^6 k^6 \sk^2 \skp^2 + \frac{16}{45 } \tau^6 k^4 \sk^4 \skp^2 + \frac{16}{15 } \tau^6 k^2 \sk^6 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{23}= \int dk\bigg[-\frac{1}{15 } \tau^3 -\frac{1}{10 } \tau^4 k^2 + \frac{2}{9 } \tau^4 k^4 \skp^2 + \frac{4}{15 } \tau^4 \sk^2 -\frac{1}{45 } \tau^5 k^4 -\frac{4}{9 } \tau^5 k^4 \sk^2 \skp^2 + \frac{4}{15 } \tau^5 k^2 \sk^2 + \frac{1}{180 } \tau^6 k^6 -\frac{2}{45 } \tau^6 k^4 \sk^2 \notag\\ & & \hspace{0.4cm}-\frac{1}{45 } \tau^6 k^6 \sk^2 \skp^2+\frac{8}{45 } \tau^6 k^4 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{24}= \int dk\bigg[4 \tau^2 - \tau^3 k^2 -16 \tau^3 \sk^2 + \frac{4}{3 } \tau^4 k^2 \sk^2 + \frac{16}{3 } \tau^4 \sk^4\bigg]\;,\hspace{3 cm } \mathcal{z}_{25}= \int dk\bigg[4 \tau^2 - \tau^3 k^2 -4 \tau^3 \sk^2\bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{26}= \int dk\bigg[2 \tau^2 -8 \tau^3 \sk^2 + \frac{2}{3 } \tau^4 k^2 \sk^2 + \frac{8}{3 } \tau^4 \sk^4 \bigg]\;,\hspace{1cm}\mathcal{z}_{27}= \int dk\bigg[2 \tau^2 \bigg]\;,\hspace{1cm}\mathcal{z}_{28}= \int dk\bigg[-4 \tau^2 + \tau^3 k^2 + 4 \tau^3 \sk^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{29}= \int dk\bigg[2 \tau^2 \bigg]\;,\hspace{2.5cm}\mathcal{z}_{30}= \int dk\bigg[-\frac{1}{3 } \tau^3 k^2 \skp \bigg]\;,\hspace{2.5cm}\mathcal{z}_{31}= \int dk\bigg[-\frac{1}{3 } \tau^3 k^2 \skp \bigg];,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{32}= \int dk\bigg[-6 \tau^3 \sk + \frac{2}{3 } \tau^3 k^2 \skp + \frac{4}{3 } \tau^4 k^2 \sk + \frac{8}{3 } \tau^4 \sk^3 -\frac{2}{9 } \tau^4 k^4 \skp\bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{33}= \int dk\bigg[4 \tau^3 \sk + \frac{4}{3 } \tau^3 k^2 \skp -\frac{4}{3 } \tau^4 k^2 \sk -\frac{8}{3 } \tau^4 \sk^3 -\frac{2}{9 } \tau^4 k^4 \skp \bigg]\;,\hspace{2cm}\mathcal{z}_{34}= \int dk\bigg[\frac{4}{3 } \tau^3 \sk + \frac{1}{3 } \tau^3 k^2 \skp \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{35}= \int dk\bigg[-2 \tau^3 \sk + \frac{1}{3 } \tau^3 k^2 \skp \bigg]\;,\hspace{2cm}\mathcal{z}_{36}= \int dk\bigg[-\frac{2}{3 } \tau^3 k^2 \skp \bigg]\;,\hspace{2cm}\mathcal{z}_{37}= \int dk\bigg[\frac{2}{3 } \tau^3 \sk \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{38}=\int dk\bigg[-\frac{1}{3 } \tau^3 k^2 \skp + \frac{1}{9 } \tau^4 k^4 \skp + \frac{2}{3 } \tau^4 k^4 \sk \skp^2 -\frac{2}{3 } \tau^4 k^2 \sk + \frac{1}{9 } \tau^5 k^4 \sk -\frac{4}{9 } \tau^5 k^4 \sk^3 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{39}= \int dk\bigg[6 \tau^3 \sk -10 \tau^3 k^2 \sk \skp^2 -2 \tau^4 k^2 \sk -\frac{8}{3 } \tau^4 \sk^3 + \frac{1}{9 } \tau^4 k^4 \skp + \frac{2}{3 } \tau^4 k^4 \sk \skp^2 + \frac{40}{3 } \tau^4 k^2 \sk^3 \skp^2 + \frac{1}{9 } \tau^5 k^4 \sk \notag\\ & & \hspace{0.4cm}+\frac{2}{3 } \tau^5 k^2 \sk^3 -\frac{4}{9 } \tau^5 k^4 \sk^3 \skp^2 -\frac{8}{3 } \tau^5 k^2 \sk^5 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{40}= \int dk\bigg[- \tau^3 k^2 \skp + \frac{4}{3 } \tau^3 \sk + \frac{1}{9 } \tau^4 k^4 \skp + \frac{2}{3 } \tau^4 k^4 \sk \skp^2 -\frac{2}{3 } \tau^4 k^2 \sk + \frac{1}{9 } \tau^5 k^4 \sk -\frac{4}{9 } \tau^5 k^4 \sk^3 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{41}= \int dk\bigg[10 \tau^3 \sk -\frac{8}{3 } \tau^4 k^2 \sk -\frac{40}{3 } \tau^4 \sk^3 + \frac{1}{9 } \tau^5 k^4 \sk + \frac{4}{3 } \tau^5 k^2 \sk^3 + \frac{8}{3 } \tau^5 \sk^5 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{42}= \int dk\bigg[-10 \tau^3 \sk + \frac{4}{3 } \tau^4 k^2 \sk + \frac{40}{3 } \tau^4 \sk^3 + \frac{1}{9 } \tau^5 k^4 \sk -\frac{4}{3 } \tau^5 k^2 \sk^3 -\frac{8}{3 } \tau^5 \sk^5 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{43}= \int dk\bigg[10 \tau^3 \sk -2 \tau^4 k^2 \sk -\frac{40}{3 } \tau^4 \sk^3 + \frac{1}{9 } \tau^5 k^4 \sk + \frac{4}{3 } \tau^5 k^2 \sk^3 + \frac{8}{3 } \tau^5 \sk^5 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{44}= \int dk\bigg[-\frac{1}{12 } \tau^3 -\frac{1}{6 } \tau^3 \sk \skp -\frac{4}{3 } \tau^3 k^2 \skp^2 + \frac{29}{120 } \tau^4 k^2 + \frac{1}{3 } \tau^4 k^2 \sk^2 \skp^2 -\frac{29}{540 } \tau^5 k^4 + \frac{19}{90 } \tau^5 k^4 \sk^2 \skp^2 -\frac{1}{135 } \tau^5 k^4 \sk^3 \skp^3 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{45}= \int dk\bigg[-\frac{1}{3 } \tau^2 \skp^2 + \frac{2}{3 } \tau^3 \sk \skp + \frac{5}{6 } \tau^3 k^2 \skp^2 -\frac{1}{3 } \tau^3 k^2 \sk \skp^3 -\frac{2}{5 } \tau^3 + \frac{4}{3 } \tau^3 \sk^2 \skp^2 + \frac{7}{45 } \tau^4 k^2 -\frac{8}{45 } \tau^4 \sk^2+\frac{4}{9 } \tau^4 k^2 \sk^3 \skp^3 \notag\\ & & \hspace{0.4 cm } -\frac{4}{9 } \tau^4 \sk^3 \skp-\frac{1}{3 } \tau^4 k^2 \sk^2 \skp^2 -\frac{4}{9 } \tau^4 \sk^4 \skp^2 + \frac{1}{270 } \tau^5 k^4 + \frac{1}{45 } \tau^5 k^2 \sk^2 -\frac{1}{45 } \tau^5 k^4 \sk^2 \skp^2-\frac{2}{135 } \tau^5 k^4 \sk^3 \skp^3 -\frac{2}{15 } \tau^5 k^2 \sk^4 \skp^2\notag\\ & & \hspace{0.4 cm } -\frac{4}{45 } \tau^5 k^2 \sk^5 \skp^3 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{46}= \int dk\bigg[\frac{7}{36 } \tau^3 + \frac{5}{6 } \tau^3 \sk \skp -\frac{5}{18 } \tau^4 k^2 -\frac{2}{9 } \tau^4 k^2 \sk^2 \skp^2 + \frac{1}{9 } \tau^4 k^4 \skp^2 + \frac{2}{9 } \tau^4 k^4 \sk \skp^3 + \frac{1}{18 } \tau^5 k^4 -\frac{2}{9 } \tau^5 k^4 \sk^2 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{47}= \int dk\bigg[\frac{11}{12 } \tau^3 -\frac{1}{18 } \tau^3 \sk \skp + \frac{2}{9 } \tau^3 k^2 \skp^2 -\frac{113}{360 } \tau^4 k^2 -\frac{1}{9 } \tau^4 k^2 \sk^2 \skp^2 + \frac{31}{540 } \tau^5 k^4 -\frac{7}{30 } \tau^5 k^4 \sk^2 \skp^2 -\frac{1}{135 } \tau^5 k^4 \sk^3 \skp^3 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{48}= \int dk\bigg[\frac{35}{18 } \tau^3 -\frac{1}{9 } \tau^3 \sk \skp -\frac{14}{9 } \tau^3 k^2 \skp^2 -\frac{211}{180 } \tau^4 k^2 -\frac{4}{3 } \tau^4 \sk^2 + \frac{85}{18 } \tau^4 k^2 \sk^2 \skp^2 + \frac{1}{9 } \tau^4 k^4 \skp^2 + \frac{2}{9 } \tau^4 k^4 \sk \skp^3 \notag\\ & & \hspace{0.4cm}+\frac{59}{540 } \tau^5 k^4 + \frac{1}{3 } \tau^5 k^2 \sk^2-\frac{13}{30 } \tau^5 k^4 \sk^2 \skp^2 + \frac{1}{135 } \tau^5 k^4 \sk^3 \skp^3 -\frac{4}{3 } \tau^5 k^2 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{49}= \int dk\bigg[\frac{1}{36 } \tau^3 -\frac{1}{18 } \tau^3 \sk \skp + \frac{5}{9 } \tau^3 k^2 \skp^2 -\frac{3}{20 } \tau^4 k^2 -\frac{2}{3 } \tau^4 k^2 \sk^2 \skp^2 + \frac{1}{9 } \tau^4 k^4 \skp^2 + \frac{2}{9 } \tau^4 k^4 \sk \skp^3 + \frac{7}{135 } \tau^5 k^4\notag\\ & & \hspace{0.4cm}-\frac{1}{5 } \tau^5 k^4 \sk^2 \skp^2 + \frac{2}{135 } \tau^5 k^4 \sk^3 \skp^3 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{50}= \int dk\bigg[-\frac{2}{45 } \tau \sk \skppp -\frac{2}{15 } \tau \skp \skpp + \frac{2}{15 } \tau^2 \sk \skpp + \frac{7}{15 } \tau^2 \skp^2 + \frac{4}{15 } \tau^2 \sk^2 \skp \skpp + \frac{4}{15 } \tau^2 \sk \skp^3 + \frac{1}{18 } \tau^2 k^4 \skpp^2 -\frac{1}{45 } \tau^3 \notag\\ & & \hspace{0.4 cm } -\frac{2}{15 } \tau^3 \sk \skp + \frac{1}{3 } \tau^3 k^2 \skp^2-\frac{4}{15 } \tau^3 \sk^2 \skp^2 + \frac{4}{9 } \tau^3 k^2 \sk \skp^3 -\frac{8}{45 } \tau^3 \sk^3 \skp^3 + \frac{4}{27 } \tau^3 k^4 \skp^4 -\frac{1}{18 } \tau^4 k^4 \skp^2 -\frac{4}{27 } \tau^4 k^4 \sk \skp^3\notag\\ & & \hspace{0.4cm}-\frac{2}{27 } \tau^4 k^4 \sk^2 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{51}= \int dk\bigg[\frac{1}{135 } \tau \sk \skppp + \frac{1}{45 } \tau \skp \skpp -\frac{1}{45 } \tau^2 \sk \skpp -\frac{1}{45 } \tau^2 \skp^2 -\frac{2}{45 } \tau^2 \sk^2 \skp \skpp -\frac{2}{45 } \tau^2 \sk \skp^3 -\frac{1}{18 } \tau^2 k^4 \skpp^2 -\frac{22}{135 } \tau^3 \notag\\ & & \hspace{0.4 cm } + \frac{1}{45 } \tau^3 \sk \skp -\frac{1}{9 } \tau^3 k^2 \skp^2+\frac{2}{45 } \tau^3 \sk^2 \skp^2 + \frac{4}{135 } \tau^3 \sk^3 \skp^3 -\frac{4}{27 } \tau^3 k^4 \skp^4 + \frac{1}{24 } \tau^4 k^2 + \frac{1}{18 } \tau^4 k^4 \skp^2 -\frac{1}{6 } \tau^4 k^2 \sk^2 \skp^2\notag\\ & & \hspace{0.4cm}+\frac{4}{27 } \tau^4 k^4 \sk \skp^3 + \frac{2}{27 } \tau^4 k^4 \sk^2 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{52}= \int dk\bigg[\frac{1}{135 } \tau \sk \skppp + \frac{1}{45 } \tau \skp \skpp -\frac{1}{45 } \tau^2 \sk \skpp -\frac{1}{45 } \tau^2 \skp^2 -\frac{2}{45 } \tau^2 \sk^2 \skp \skpp -\frac{2}{45 } \tau^2 \sk \skp^3 -\frac{1}{18 } \tau^2 k^4 \skpp^2 + \frac{1}{270 } \tau^3 \notag\\ & & \hspace{0.4cm}+\frac{1}{45 } \tau^3 \sk \skp -\frac{1}{9 } \tau^3 k^2 \skp^2+\frac{2}{45 } \tau^3 \sk^2 \skp^2 + \frac{4}{135 } \tau^3 \sk^3 \skp^3 -\frac{4}{27 } \tau^3 k^4 \skp^4 + \frac{1}{18 } \tau^4 k^4 \skp^2 + \frac{4}{27 } \tau^4 k^4 \sk \skp^3 + \frac{2}{27 } \tau^4 k^4 \sk^2 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{53}= \int dk\bigg[-\frac{1}{18 } \tau^3 + \frac{17}{180 } \tau^4 k^2 -\frac{1}{18 } \tau^4 k^2 \sk^2 \skp^2 -\frac{4}{81 } \tau^4 k^4 \sk \skp^3 + \frac{1}{162 } \tau^4 k^6 \skp^4 + \frac{1}{54 } \tau^4 k^6 \sk^2 \skpp^2 -\frac{1}{36 } \tau^5 k^4 \notag\\ & & \hspace{0.4cm}+\frac{29}{270 } \tau^5 k^4 \sk^2 \skp^2 + \frac{29}{405 } \tau^5 k^4 \sk^3 \skp^3 + \frac{1}{81 } \tau^5 k^6 \sk \skp^3+\frac{2}{81 } \tau^5 k^6 \sk^2 \skp^4 + \frac{1}{540 } \tau^6 k^6 -\frac{2}{135 } \tau^6 k^6 \sk^2 \skp^2-\frac{8}{405 } \tau^6 k^6 \sk^3 \skp^3\notag\\ & & \hspace{0.4 cm } -\frac{4}{405 } \tau^6 k^6 \sk^4 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{54}= \int dk\bigg[-\frac{1}{6 } \tau^3 + \frac{5}{36 } \tau^4 k^2 -\frac{1}{6 } \tau^4 k^2 \sk^2 \skp^2 -\frac{4}{81 } \tau^4 k^4 \sk \skp^3 + \frac{1}{162 } \tau^4 k^6 \skp^4 + \frac{1}{54 } \tau^4 k^6 \sk^2 \skpp^2 -\frac{17}{540 } \tau^5 k^4 \notag\\ & & \hspace{0.4 cm } + \frac{7}{54 } \tau^5 k^4 \sk^2 \skp^2 + \frac{7}{81 } \tau^5 k^4 \sk^3 \skp^3 + \frac{1}{81 } \tau^5 k^6 \sk \skp^3+\frac{2}{81 } \tau^5 k^6 \sk^2 \skp^4 + \frac{1}{540 } \tau^6 k^6 -\frac{2}{135 } \tau^6 k^6 \sk^2 \skp^2 -\frac{8}{405 } \tau^6 k^6 \sk^3 \skp^3\notag\\ & & \hspace{0.4cm}-\frac{4}{405 } \tau^6 k^6 \sk^4 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{55}= \int dk\bigg[\frac{1}{18 } \tau^2 k^4 \skpp^2 -\frac{1}{5 } \tau^3 + \frac{1}{2 } \tau^3 k^2 \skp^2 + \frac{11}{9 } \tau^3 k^2 \sk \skp^3 -\frac{2}{9 } \tau^3 k^4 \sk^2 \skpp^2 + \frac{2}{27 } \tau^3 k^4 \skp^4 + \frac{37}{180 } \tau^4 k^2 + \frac{11}{45 } \tau^4 \sk^2 \notag\\ & & \hspace{0.4 cm } -\frac{1}{18 } \tau^4 k^4 \skp^2 -\frac{14}{9 } \tau^4 k^2 \sk^2 \skp^2-\frac{26}{81 } \tau^4 k^4 \sk \skp^3 -\frac{44}{27 } \tau^4 k^2 \sk^3 \skp^3 + \frac{1}{108 } \tau^4 k^6 \sk^2 \skpp^2 + \frac{1}{324 } \tau^4 k^6 \skp^4 + \frac{2}{27 } \tau^4 k^4 \sk^4 \skpp^2 \notag\\ & & \hspace{0.4cm}-\frac{4}{9 } \tau^4 k^4 \sk^2 \skp^4 -\frac{1}{30 } \tau^5 k^4 -\frac{1}{9 } \tau^5 k^2 \sk^2 + \frac{41}{135 } \tau^5 k^4 \sk^2 \skp^2 + \frac{22}{45 } \tau^5 k^2 \sk^4 \skp^2+\frac{1}{162 } \tau^5 k^6 \sk \skp^3 + \frac{182}{405 } \tau^5 k^4 \sk^3 \skp^3 \notag\\ & & \hspace{0.4 cm } + \frac{44}{135 } \tau^5 k^2 \sk^5 \skp^3 + \frac{1}{81 } \tau^5 k^6 \sk^2 \skp^4 + \frac{8}{27 } \tau^5 k^4 \sk^4 \skp^4 + \frac{1}{1080 } \tau^6 k^6 + \frac{1}{135 } \tau^6 k^4 \sk^2 -\frac{1}{135 } \tau^6 k^6 \sk^2 \skp^2 -\frac{8}{135 } \tau^6 k^4 \sk^4 \skp^2 \skp^3\notag\\ & & \hspace{0.4cm}-\frac{4}{405 } \tau^6 k^6 \sk^3 \skp^3-\frac{32}{405 } \tau^6 k^4 \sk^5-\frac{2}{405 } \tau^6 k^6 \sk^4 \skp^4 -\frac{16}{405 } \tau^6 k^4 \sk^6 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{56}= \int dk\bigg[-\frac{1}{18 } \tau^3 + \frac{5}{72 } \tau^4 k^2 -\frac{1}{18 } \tau^4 k^2 \sk^2 \skp^2 -\frac{4}{81 } \tau^4 k^4 \sk \skp^3 + \frac{1}{162 } \tau^4 k^6 \skp^4 + \frac{1}{54 } \tau^4 k^6 \sk^2 \skpp^2 -\frac{1}{45 } \tau^5 k^4 \notag\\ & & \hspace{0.4cm}+\frac{2}{27 } \tau^5 k^4 \sk^2 \skp^2 + \frac{4}{81 } \tau^5 k^4 \sk^3 \skp^3 + \frac{1}{81 } \tau^5 k^6 \sk \skp^3+\frac{2}{81 } \tau^5 k^6 \sk^2 \skp^4 + \frac{1}{540 } \tau^6 k^6 -\frac{2}{135 } \tau^6 k^6 \sk^2 \skp^2 -\frac{8}{405 } \tau^6 k^6 \sk^3 \skp^3\notag\\ & & \hspace{0.4cm}-\frac{4}{405 } \tau^6 k^6 \sk^4 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{57}= \int dk\bigg[-\frac{1}{36 } \tau^3 + \frac{5}{144 } \tau^4 k^2 -\frac{1}{36 } \tau^4 k^2 \sk^2 \skp^2 -\frac{2}{81 } \tau^4 k^4 \sk \skp^3 + \frac{1}{324 } \tau^4 k^6 \skp^4 + \frac{1}{108 } \tau^4 k^6 \sk^2 \skpp^2 -\frac{1}{90 } \tau^5 k^4 \notag\\ & & \hspace{0.4cm}+\frac{1}{27 } \tau^5 k^4 \sk^2 \skp^2 + \frac{2}{81 } \tau^5 k^4 \sk^3 \skp^3 + \frac{1}{162 } \tau^5 k^6 \sk \skp^3+\frac{1}{81 } \tau^5 k^6 \sk^2 \skp^4 + \frac{1}{1080 } \tau^6 k^6 -\frac{1}{135 } \tau^6 k^6 \sk^2 \skp^2 -\frac{4}{405 } \tau^6 k^6 \sk^3 \skp^3\notag\\ & & \hspace{0.4cm}-\frac{2}{405 } \tau^6 k^6 \sk^4 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{58}= \int dk\bigg[-\frac{1}{36 } \tau^3 + \frac{5}{144 } \tau^4 k^2 -\frac{1}{36 } \tau^4 k^2 \sk^2 \skp^2 -\frac{2}{81 } \tau^4 k^4 \sk \skp^3 + \frac{1}{324 } \tau^4 k^6 \skp^4 + \frac{1}{108 } \tau^4 k^6 \sk^2 \skpp^2 -\frac{1}{90 } \tau^5 k^4 \notag\\ & & \hspace{0.4 cm } + \frac{1}{27 } \tau^5 k^4 \sk^2 \skp^2 + \frac{2}{81 } \tau^5 k^4 \sk^3 \skp^3 + \frac{1}{162 } \tau^5 k^6 \sk \skp^3+\frac{1}{81 } \tau^5 k^6 \sk^2 \skp^4 + \frac{1}{1080 } \tau^6 k^6 -\frac{1}{135 } \tau^6 k^6 \sk^2 \skp^2 -\frac{4}{405 } \tau^6 k^6 \sk^3 \skp^3\notag\\ & & \hspace{0.4cm}-\frac{2}{405 } \tau^6 k^6 \sk^4 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{59}= \int dk\bigg[\frac{1}{18 } \tau^2 k^4 \skpp^2 -\frac{2}{5 } \tau^3 + \frac{2}{9 } \tau^3 k^2 \sk \skp^3 + \frac{2}{27 } \tau^3 k^4 \skp^4 -\frac{2}{9 } \tau^3 k^4 \sk^2 \skpp^2 + \frac{31}{180 } \tau^4 k^2 -\frac{5}{9 } \tau^4 k^2 \sk^2 \skp^2 -\frac{8}{27 } \tau^4 k^2 \sk^3 \skp^3\notag\\ & & \hspace{0.4 cm } -\frac{1}{18 } \tau^4 k^4 \skp^2 -\frac{26}{81 } \tau^4 k^4 \sk \skp^3-\frac{4}{9 } \tau^4 k^4 \sk^2 \skp^4 + \frac{2}{45 } \tau^4 \sk^2 + \frac{1}{324 } \tau^4 k^6 \skp^4 + \frac{1}{108 } \tau^4 k^6 \sk^2 \skpp^2 + \frac{2}{27 } \tau^4 k^4 \sk^4 \skpp^2\notag\\ & & \hspace{0.4 cm } -\frac{1}{45 } \tau^5 k^4 + \frac{32}{135 } \tau^5 k^4 \sk^2 \skp^2 + \frac{164}{405 } \tau^5 k^4 \sk^3 \skp^3 + \frac{8}{27 } \tau^5 k^4 \sk^4 \skp^4 -\frac{2}{45 } \tau^5 k^2 \sk^2+\frac{4}{45 } \tau^5 k^2 \sk^4 \skp^2 + \frac{8}{135 } \tau^5 k^2 \sk^5 \skp^3\notag\\ & & \hspace{0.4 cm } + \frac{1}{162 } \tau^5 k^6 \sk \skp^3 + \frac{1}{81 } \tau^5 k^6 \sk^2 \skp^4 + \frac{1}{1080 } \tau^6 k^6 -\frac{1}{135 } \tau^6 k^6 \sk^2 \skp^2 -\frac{4}{405 } \tau^6 k^6 \sk^3 \skp^3-\frac{2}{405 } \tau^6 k^6 \sk^4 \skp^4 + \frac{1}{135 } \tau^6 k^4 \sk^2\notag\\ & & \hspace{0.4 cm } -\frac{8}{135 } \tau^6 k^4 \sk^4 \skp^2-\frac{32}{405 } \tau^6 k^4 \sk^5 \skp^3 -\frac{16}{405 } \tau^6 k^4 \sk^6 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{60}= \int dk\bigg[\frac{5}{18 } \tau^3 -\frac{1}{20 } \tau^4 k^2 + \frac{5}{18 } \tau^4 k^2 \sk^2 \skp^2 -\frac{2}{81 } \tau^4 k^4 \sk \skp^3 + \frac{1}{324 } \tau^4 k^6 \skp^4 + \frac{1}{108 } \tau^4 k^6 \sk^2 \skpp^2 -\frac{1}{108 } \tau^5 k^4 \notag\\ & & \hspace{0.4 cm } + \frac{7}{270 } \tau^5 k^4 \sk^2 \skp^2 + \frac{7}{405 } \tau^5 k^4 \sk^3 \skp^3 + \frac{1}{162 } \tau^5 k^6 \sk \skp^3+\frac{1}{81 } \tau^5 k^6 \sk^2 \skp^4 + \frac{1}{1080 } \tau^6 k^6 -\frac{1}{135 } \tau^6 k^6 \sk^2 \skp^2\notag\\ & & \hspace{0.4cm}-\frac{4}{405 } \tau^6 k^6 \sk^3 \skp^3 -\frac{2}{405 } \tau^6 k^6 \sk^4 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{61}= \int dk\bigg[\frac{1}{18 } \tau^2 k^4 \skpp^2 + \frac{2}{9 } \tau^3 k^2 \sk \skp^3 -\frac{1}{15 } \tau^3 + \frac{2}{27 } \tau^3 k^4 \skp^4 -\frac{2}{9 } \tau^3 k^4 \sk^2 \skpp^2 + \frac{4}{45 } \tau^4 k^2 -\frac{2}{9 } \tau^4 k^2 \sk^2 \skp^2 \notag\\ & & \hspace{0.4 cm } -\frac{8}{27 } \tau^4 k^2 \sk^3 \skp^3 -\frac{1}{18 } \tau^4 k^4 \skp^2 -\frac{26}{81 } \tau^4 k^4 \sk \skp^3-\frac{4}{9 } \tau^4 k^4 \sk^2 \skp^4 + \frac{2}{45 } \tau^4 \sk^2 + \frac{1}{324 } \tau^4 k^6 \skp^4 + \frac{1}{108 } \tau^4 k^6 \sk^2 \skpp^2 \notag\\ & & \hspace{0.4 cm } + \frac{2}{27 } \tau^4 k^4 \sk^4 \skpp^2 -\frac{1}{45 } \tau^5 k^4 + \frac{32}{135 } \tau^5 k^4 \sk^2 \skp^2 + \frac{164}{405 } \tau^5 k^4 \sk^3 \skp^3 + \frac{8}{27 } \tau^5 k^4 \sk^4 \skp^4 -\frac{2}{45 } \tau^5 k^2 \sk^2+\frac{4}{45 } \tau^5 k^2 \sk^4 \skp^2 \notag\\ & & \hspace{0.4 cm } + \frac{8}{135 } \tau^5 k^2 \sk^5 \skp^3 + \frac{1}{162 } \tau^5 k^6 \sk \skp^3 + \frac{1}{81 } \tau^5 k^6 \sk^2 \skp^4 + \frac{1}{1080 } \tau^6 k^6 -\frac{1}{135 } \tau^6 k^6 \sk^2 \skp^2 -\frac{4}{405 } \tau^6 k^6 \sk^3 \skp^3\notag\\ & & \hspace{0.4cm}-\frac{2}{405 } \tau^6 k^6 \sk^4 \skp^4 + \frac{1}{135 } \tau^6 k^4 \sk^2 -\frac{8}{135 } \tau^6 k^4 \sk^4 \skp^2-\frac{32}{405 } \tau^6 k^4 \sk^5 \skp^3 -\frac{16}{405 } \tau^6 k^4 \sk^6 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{62}= \int dk\bigg[\frac{1}{45 } \tau^4 k^2 -\frac{4}{81 } \tau^4 k^4 \sk \skp^3 + \frac{1}{162 } \tau^4 k^6 \skp^4 + \frac{1}{54 } \tau^4 k^6 \sk^2 \skpp^2 -\frac{2}{135 } \tau^5 k^4 + \frac{4}{135 } \tau^5 k^4 \sk^2 \skp^2 \notag\\ & & \hspace{0.4cm}+\frac{8}{405 } \tau^5 k^4 \sk^3 \skp^3 + \frac{1}{81 } \tau^5 k^6 \sk \skp^3 + \frac{2}{81 } \tau^5 k^6 \sk^2 \skp^4 + \frac{1}{540 } \tau^6 k^6-\frac{2}{135 } \tau^6 k^6 \sk^2 \skp^2 -\frac{8}{405 } \tau^6 k^6 \sk^3 \skp^3 -\frac{4}{405 } \tau^6 k^6 \sk^4 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{63}= \int dk\bigg[\frac{1}{18 } \tau^3 -\frac{1}{90 } \tau^4 k^2 + \frac{1}{18 } \tau^4 k^2 \sk^2 \skp^2 -\frac{2}{81 } \tau^4 k^4 \sk \skp^3 + \frac{1}{324 } \tau^4 k^6 \skp^4 + \frac{1}{108 } \tau^4 k^6 \sk^2 \skpp^2 -\frac{1}{180 } \tau^5 k^4 \notag\\ & & \hspace{0.4 cm } + \frac{1}{270 } \tau^5 k^4 \sk^2 \skp^2 + \frac{1}{405 } \tau^5 k^4 \sk^3 \skp^3 + \frac{1}{162 } \tau^5 k^6 \sk \skp^3+\frac{1}{81 } \tau^5 k^6 \sk^2 \skp^4 + \frac{1}{1080 } \tau^6 k^6 -\frac{1}{135 } \tau^6 k^6 \sk^2 \skp^2\notag\\ & & \hspace{0.4cm}-\frac{4}{405 } \tau^6 k^6 \sk^3 \skp^3 -\frac{2}{405 } \tau^6 k^6 \sk^4 \skp^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{64}= \int dk\bigg[-\frac{1}{9 } \tau^3 k^2 \skp + \frac{4}{9 } \tau^3 k^2 \sk \skp^2 -\frac{2}{9 } \tau^3 k^4 \skp^3 + \frac{1}{9 } \tau^4 k^4 \skp + \frac{2}{9 } \tau^4 k^4 \sk \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{65}= \int dk\bigg[\frac{4}{3 } \tau^3 \sk -\frac{7}{9 } \tau^3 k^2 \skp -\frac{20}{9 } \tau^3 k^2 \sk \skp^2 -\frac{2}{9 } \tau^3 k^4 \skp^3 -\frac{1}{3 } \tau^4 k^2 \sk + \frac{1}{9 } \tau^4 k^4 \skp + \frac{2}{9 } \tau^4 k^4 \sk \skp^2 + \frac{4}{3 } \tau^4 k^2 \sk^3 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{66}= \int dk\bigg[-\frac{4}{9 } \tau^3 k^2 \skp -\frac{2}{9 } \tau^3 k^2 \sk \skp^2 -\frac{2}{9 } \tau^3 k^4 \skp^3 + \frac{1}{9 } \tau^4 k^4 \skp + \frac{2}{9 } \tau^4 k^4 \sk \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{67}= \int dk\bigg[\frac{5}{3 } \tau^2 - \tau^2 k^2 \skp^2 -\frac{4}{3 } \tau^3 \sk^2 -\frac{1}{3 } \tau^3 k^2 + 4 \tau^3 k^2 \sk^2 \skp^2 + \frac{1}{3 } \tau^4 k^2 \sk^2 -\frac{4}{3 } \tau^4 k^2 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{68}= \int dk\bigg [ \tau^2 + \tau^2 k^2 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{69}= \int dk\bigg[\frac{1}{3 } \tau^3 + \frac{2}{3 } \tau^3 k^2 \skp^2 -\frac{2}{3 } \tau^3 k^2 \sk \skp^3 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{70}= \int dk\bigg[- \tau^2 \sk \skpp - \tau^2 \skp^2 + \frac{5}{3 } \tau^3 \sk \skp -\frac{1}{2 } \tau^3 + 4 \tau^3 \sk^2 \skp^2 - \tau^3 k^2 \skp^2 + \frac{2}{3 } \tau^3 \sk^3 \skpp + \frac{1}{6 } \tau^4 k^2 + \frac{1}{3 } \tau^4 \sk^2 \notag\\ & & \hspace{0.4 cm } + \frac{4}{3 } \tau^4 k^2 \sk^2 \skp^2-\frac{4}{3 } \tau^4 \sk^3 \skp -\frac{4}{3 } \tau^4 \sk^4 \skp^2 + \frac{2}{9 } \tau^4 k^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{71}= \int dk\bigg[\frac{13}{6 } \tau^3 + \frac{1}{3 } \tau^3 \sk \skp + \frac{1}{3 } \tau^3 k^2 \skp^2 -\frac{7}{12 } \tau^4 k^2 -\frac{4}{3 } \tau^4 \sk^2 -\frac{5}{3 } \tau^4 k^2 \sk^2 \skp^2 + \frac{2}{9 } \tau^4 k^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{72}= \int dk\bigg[-2 \tau^3 + \tau^3 \sk \skp -3 \tau^3 k^2 \skp^2 + \frac{13}{24 } \tau^4 k^2 + \frac{4}{3 } \tau^4 \sk^2 + \frac{11}{6 } \tau^4 k^2 \sk^2 \skp^2 + \frac{2}{9 } \tau^4 k^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{73}= \int dk\bigg[-\frac{11}{12 } \tau^3 + \frac{2}{3 } \tau^3 k^2 \skp^2 + \frac{23}{48 } \tau^4 k^2 + \frac{2}{3 } \tau^4 \sk^2 + \frac{1}{12 } \tau^4 k^2 \sk^2 \skp^2 + \frac{1}{9 } \tau^4 k^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{74}= \int dk\bigg[\frac{1}{12 } \tau^4 k^2 + \frac{1}{3 } \tau^4 \sk^2 -\frac{1}{9 } \tau^4 k^4 \skp^2 + \frac{2}{3 } \tau^4 k^2 \sk^2 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{75}= \int dk\bigg[-\frac{1}{3 } \tau \skpp + \frac{2}{3 } \tau^2 \skp + 2 \tau^2 \sk \skp^2 + \frac{2}{3 } \tau^2 \sk^2 \skpp + 2 \tau^3 k^2 \sk \skp^2 -\frac{4}{3 } \tau^3 \sk^2 \skp + \frac{2}{3 } \tau^3 \sk -\frac{4}{3 } \tau^3 \sk^3 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{76}= \int dk\bigg[-\frac{2}{3 } \tau^2 \skp + 2 \tau^3 \sk -\frac{8}{9 } \tau^3 k^2 \skp + \frac{20}{9 } \tau^3 k^2 \sk \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{77}= \int dk\bigg[-\frac{2}{3 } \tau^2 \skp + \frac{1}{9 } \tau^3 k^2 \skp + \frac{2}{9 } \tau^3 k^2 \sk \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{78}= \int dk\bigg[-\frac{2}{3 } \tau^3 -\frac{1}{6 } \tau^3 k^2 \skp^2 + \frac{43}{120 } \tau^4 k^2 -\frac{1}{18 } \tau^4 k^4 \skp^2 -\frac{1}{9 } \tau^4 k^4 \sk \skp^3 -\frac{11}{180 } \tau^5 k^4 + \frac{23}{90 } \tau^5 k^4 \sk^2 \skp^2 + \frac{1}{45 } \tau^5 k^4 \sk^3 \skp^3 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{79}= \int dk\bigg[\frac{31}{15 } \tau^3 - \tau^3 k^2 \skp^2 + 2 \tau^3 k^2 \sk \skp^3 -\frac{13}{30 } \tau^4 k^2 + \frac{2}{5 } \tau^4 \sk^2 -\frac{8}{3 } \tau^4 k^2 \sk^3 \skp^3 -\frac{1}{45 } \tau^5 k^4 -\frac{2}{15 } \tau^5 k^2 \sk^2 \notag\\ & & \hspace{0.4 cm } + \frac{2}{15 } \tau^5 k^4 \sk^2 \skp^2 + \frac{4}{5 } \tau^5 k^2 \sk^4 \skp^2+\frac{4}{45 } \tau^5 k^4 \sk^3 \skp^3 + \frac{8}{15 } \tau^5 k^2 \sk^5 \skp^3 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{80}= \int dk\bigg[-\frac{4}{3 } \tau^3 + \frac{5}{6 } \tau^3 k^2 \skp^2 + \frac{3}{40 } \tau^4 k^2 -\frac{4}{3 } \tau^4 k^2 \sk^2 \skp^2 + \frac{1}{18 } \tau^4 k^4 \skp^2 + \frac{1}{9 } \tau^4 k^4 \sk \skp^3 + \frac{7}{180 } \tau^5 k^4 -\frac{11}{90 } \tau^5 k^4 \sk^2 \skp^2\notag\\ & & \hspace{0.4cm}+\frac{1}{15 } \tau^5 k^4 \sk^3 \skp^3 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{81}= \int dk\bigg[\frac{8}{3 } \tau^3 -2 \tau^3 k^2 \skp^2 -\frac{6}{5 } \tau^4 k^2 -\frac{4}{3 } \tau^4 \sk^2 + 5 \tau^4 k^2 \sk^2 \skp^2 + \frac{1}{9 } \tau^4 k^4 \skp^2 + \frac{2}{9 } \tau^4 k^4 \sk \skp^3 + \frac{1}{10 } \tau^5 k^4+\frac{1}{3 } \tau^5 k^2 \sk^2 \notag\\ & & \hspace{0.4 cm } -\frac{17}{45 } \tau^5 k^4 \sk^2 \skp^2+\frac{2i}{45 } \tau^5 k^4 \sk^3 \skp^3 -\frac{4}{3 } \tau^5 k^2 \sk^4 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{82}= \int dk\bigg[-6 \tau^3 \sk -\frac{2}{3 } \tau^3 k^2 \skp + \frac{4}{3 } \tau^4 k^2 \sk + \frac{8}{3 } \tau^4 \sk^3 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{83}= \int dk\bigg[-\frac{20}{3 } \tau^3 \sk + \frac{4}{3 } \tau^3 k^2 \skp + \frac{4}{3 } \tau^4 k^2 \sk + \frac{8}{3 } \tau^4 \sk^3 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{84}= \int dk\bigg[\frac{17}{15 } \tau^3 -\frac{47}{60 } \tau^4 k^2 -\frac{73}{15 } \tau^4 \sk^2 + \frac{1}{9 } \tau^5 k^4 + \tau^5 k^2 \sk^2 + \frac{4}{3 } \tau^5 \sk^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{85}= \int dk\bigg[-\frac{8}{5 } \tau^3 + \frac{11}{30 } \tau^4 k^2 + \frac{76}{15 } \tau^4 \sk^2 - \tau^5 k^2 \sk^2 -\frac{4}{3 } \tau^5 \sk^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{86}= \int dk\bigg[\frac{46}{15 } \tau^3 -\frac{43}{30 } \tau^4 k^2 -\frac{28}{5 } \tau^4 \sk^2 + \frac{1}{9 } \tau^5 k^4 + \tau^5 k^2 \sk^2 + \frac{4}{3 } \tau^5 \sk^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{87}= \int dk\bigg[\frac{4}{5 } \tau^3 + \frac{2}{15 } \tau^4 k^2 -\frac{26}{5 } \tau^4 \sk^2 -\frac{1}{9 } \tau^5 k^4 + \tau^5 k^2 \sk^2 + \frac{4}{3 } \tau^5 \sk^4 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{88}= \int dk\bigg[-4 \tau^2 \sk + \frac{8}{3 } \tau^3 \sk^3 \bigg]\;,\hspace{1.9cm}\mathcal{z}_{89}= \int dk\bigg[-4 \tau^2 \sk \bigg]\;,\hspace{1.9cm}\mathcal{z}_{90}=0\hspace{1.9cm}\mathcal{z}_{91}=0\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{92}= \int dk\bigg[6 \tau^3 \sk -10 \tau^3 k^2 \sk \skp^2 -2 \tau^4 k^2 \sk -\frac{8}{3 } \tau^4 \sk^3 + \frac{4}{9 } \tau^4 k^4 \sk \skp^2 + \frac{40}{3 } \tau^4 k^2 \sk^3 \skp^2 + \frac{1}{9 } \tau^5 k^4 \sk+\frac{2}{3 } \tau^5 k^2 \sk^3\notag\\ & & \hspace{0.4 cm } -\frac{4}{9 } \tau^5 k^4 \sk^3 \skp^2 -\frac{8}{3 } \tau^5 k^2 \sk^5 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{93}= \int dk\bigg[-\frac{2}{3 } \tau^3 \sk + \frac{4}{9 } \tau^4 k^4 \sk \skp^2 -\frac{2}{3 } \tau^4 k^2 \sk + \frac{1}{9 } \tau^5 k^4 \sk -\frac{4}{9 } \tau^5 k^4 \sk^3 \skp^2 \bigg]\;,\notag\\ & & \hspace{-0.5cm}\mathcal{z}_{94}= \int dk\bigg[\frac{2}{3 } \tau^3 \sk + \frac{4}{9 } \tau^4 k^4 \sk \skp^2 -\frac{2}{3 } \tau^4 k^2 \sk + \frac{1}{9 } \tau^5 k^4 \sk -\frac{4}{9 } \tau^5 k^4 \sk^3 \skp^2 \bigg]\;,\label{zsigma } \end{aligned}\ ] ] where @xmath374 as we mentioned before , since the momentum integrations in ( [ zsigma ] ) are convergent due to suppression factor @xmath375 , the integrand is only accurate up to some total derivatives , but if we insist on taking the limit of @xmath151 , the dropping out momentum space total derivatives becomes problematic . we choose these momentum space total derivatives to reduce the high order self energy derivatives as much as possible . for example , following relation with @xmath376 being zero or positive integers can be used to reduce the term with a @xmath377 to the terms with lower order self energy derivatives , @xmath378 similarly , we have a series relations to reduce the term with a @xmath379 , or a @xmath380 , a @xmath381 , a @xmath382 to the terms with lower order self energy derivatives , @xmath383 with the help of above relations , the highest self energy derivatives appear in the final result ( [ zsigma ] ) is @xmath380 . @xmath384 to help understanding mutual relation between the definition of symbols in our formulation and those in ref.@xcite , in table xiii , we give a comparison . @xmath385 in obtaining ( [ kzrelation ] ) , some dependent terms must be reduced into independent terms , in table xv , the first column is the dependent operators and the second column is the sum of its corresponding independent operators . @xmath386

we present results of computing the @xmath0 order low energy constants in the normal part of chiral lagrangian both for two and three flavor pseudo - scalar mesons . this is a generalization of our previous work on calculating the @xmath1 order coefficients of the chiral lagrangian in terms of the quark self energy @xmath2 approximately from qcd . we show that most of our results are consistent with those we can find in the literature .

consider that alice holds the qubit system and bob holds the qutrit system . like in the @xmath26-qubit scenario here also the following two cases are possible : ( a - i ) all the povm elements are rank one operators ; ( a - ii ) some of the povm elements may have more than one rank . * case(a - i ) * : alice and bob perform three - outcome rank one povms on their respective parts of the shared qubit - qutrit state . like the @xmath105 scenario , consider @xmath70 as basis for the alice s qubit system . for bob s qutrit system consider @xmath106 as basis . then the other vectors on bob s side can be expressed as : @xmath107 forms a basis for the @xmath54 tensor product hilbert space . according to the conditions ( [ h31])-([h33 ] ) the qubit - qutrit state exhibiting hardy s test ( [ h3 ] ) must be orthogonal to the following nine vectors : among the above nine vectors the set @xmath109 are linearly independent.and the rest four vectors ( i.e. @xmath110 ) can be expressed in terms of these vectors provided the following conditions are satisfied : @xmath111 the unique qubit - qutrit state orthogonal to the subspace spanned by the set @xmath112 reads as : @xmath113 being a product state the above qubit - qutrit state can not manifest the hardy s argument ( [ hardy3 ] ) . * case(a - ii ) * : in this case some povm elements are greater than rank one operators . the analysis goes similar as case(ii ) . the @xmath54 state exhibiting hardy s argument ( [ hardy3 ] ) need to be orthogonal to the support of the each product operator in table-([table1 ] ) . for some cases the ranges of theses operators together span the six dimension of the @xmath54 hilbert space and hence in such cases , there is no possibility for hardy s state . for rest of the cases the ranges together span five dimension and hence the cases boil down to the case(a - i ) . a , k. ekert , https://journals.aps.org/prl/abstract/10.1103/physrevlett.67.661[phys . lett . * 67 * , 661 ( 1991 ) ] ; c. h. bennett , g. brassard , and n. d. mermin , https://journals.aps.org/prl/abstract/10.1103/physrevlett.68.557[phys . rev . lett . * 68 * , 557 ( 1992 ) ] ; n. d. mermin , in _ fundamental problems in quantum theory _ , edited by d. m. greenberger and a. zeilinger , special issue of http://onlinelibrary.wiley.com/doi/10.1111/j.1749-6632.1995.tb39001.x/abstract;jsessionid=d3564388b18ff3c9248317b4109a9686.f03t04[ann . n. y. acad . 755 , 616 ( 1995 ) ] . at this point it is important to note that , in the case of bb84 cryptography protocol if the required correlation results from measurements in the @xmath114 and @xmath115 bases on qubit pairs then security follows only if it is obtained from @xmath26-qubit maximally entangled state . however , the same correlation can also be obtained from higher dimensional separable state and the security analysis breaks down ( see @xcite ) . j. barrett , l. hardy , and a. kent , https://journals.aps.org/prl/abstract/10.1103/physrevlett.95.010503[phys . lett . * 95 * , 010503 ( 2005 ) ] . a. acin , n. gisin , and l. masanes , https://journals.aps.org/prl/abstract/10.1103/physrevlett.97.120405[phys . * 97 * , 120405 ( 2006 ) ] . a. acin , n. brunner , n. gisin , s. massar , s. pironio , and v. scarani , https://journals.aps.org/prl/abstract/10.1103/physrevlett.98.230501[phys . lett . * 98 * , 230501 ( 2007 ) ] . r. rahaman , m. g. parker , p. mironowicz , m. pawowski , http://arxiv.org/abs/1308.6447[arxiv:1308.6447 ] . s. pironio _ , http://www.nature.com/nature/journal/v464/n7291/full/nature09008.html[nature ( london ) * 464 * , 1021 ( 2010 ) ] . a. acn , s. massar , and s. pironio , http://journals.aps.org/prl/abstract/10.1103/physrevlett.108.100402[phys . lett . * 108 * , 100402 ( 2012 ) ] . s. fehr , r. gelles , and c. schaffner , http://journals.aps.org/pra/abstract/10.1103/physreva.87.012335[phys . a * 87 * , 012335 ( 2013 ) ] . s. pironio , and s. massar , http://journals.aps.org/pra/abstract/10.1103/physreva.87.012336[phys . a * 87 * , 012336 ( 2013 ) ] . m. banik , http://arxiv.org/abs/1401.1338[arxiv:1401.1338 ] .

schmidt rank of bipartite pure state serves as a testimony of entanglement . it is a monotone under local operation @xmath0 classical communications ( locc ) and puts restrictions in locc convertibility of quantum states . identifying the schmidt rank of an unknown quantum state therefore seek importance from information theoretic perspective . in this work it is shown that a modified version of hardy s argument , which reveals the contradiction of quantum theory with _ local realism _ , turns out to be useful for inspecting the minimal schmidt rank of the unknown state and hence also the minimal dimension of the system . use of hardy s test in such task provides a practical advantage : the schmidt rank can be determined without knowing the detailed functioning of the experimental devices i.e. , hardy s test suffices to be a device independent schmidt rank witness . among various counterintuitive features of quantum mechanics , certainly , one of the most bizarre property is quantum entanglement @xcite . this holistic property of compound quantum systems , which involves non - classical correlations among subsystems , has potential for many quantum processes , including canonical ones : quantum cryptography @xcite , quantum teleportation @xcite , and dense coding @xcite . according to the quantum formalism , the total hilbert space @xmath1 of the @xmath2 separate systems is a tensor product of the subsystem spaces , i.e. , @xmath3 . when the number of the involved subsystems are two , the pure state @xmath4 of the bipartite system can always be described by its schmidt decomposition , i.e. , the representation of @xmath4 in an orthogonal product basis with minimal number of terms @xcite . a bipartite pure state @xmath5 , with @xmath6 and @xmath7 , has schmidt rank @xmath8 if its schmidt decomposition reads : @xmath9 , where @xmath10 , @xmath11 , @xmath12 and @xmath13 is an orthonormal set of vectors in the hilbert space @xmath14 and @xmath15 is an orthonormal set of vectors in the hilbert space @xmath16 . the schmidt number for a bipartite mixed state @xmath17 is the number @xmath18 such that : ( a ) for any decomposition @xmath19 of @xmath20 with @xmath21 at least one of the vectors @xmath22 has at least schmidt rank @xmath18 , and ( b ) and there exists a decomposition of @xmath20 with all vectors @xmath22 of schmidt rank at most @xmath18 @xcite . here , @xmath23 denotes the collection of positive , trace-@xmath24 operators acting on @xmath25 . the schmidt rank is the number of non vanishing terms in schmidt decomposition . this decomposition gives a clear insight into the number of degrees of freedom that are entangled between both parties if the schmidt rank is greater than unity then the pure bipartite state must be entangled . furthermore it has been proved that the schmidt number is non - increasing under local operations and classical communication ( locc ) , i.e. , it is a monotone under locc . hence , it puts restriction in locc convertibility of states @xcite . a necessary condition for a pure state to be convertible by locc to another pure state is that the schmidt rank of the later can not be larger than that of the previous one @xcite . from an information theoretic point of view , the schmidt rank of a state , therefore , can be considered as a resource . identifying the schmidt rank of a state is also important for quantifying the power of quantum correlations , a central issue in quantum information theory . in this work the problem of determining the schmidt rank of an unknown bipartite state has been addressed . interestingly , it has been shown that considering a modified version of the hardy s paradox , recently introduced by chen _ et.al _ @xcite , one can know the minimal schmidt rank of the given unknown state . it also provides information about the minimal hilbert space dimension of the concerned system . the use of nonlocality argument in this task comes up with a novel advantage . the schmidt rank can be determined from measurement data alone , in a scenario in which all devices used in the experiment , including the measurement device , are uncharacterized or in other words no assumption about the internal working of the devices is needed . the original hardy s argument was defined for _ dichotomic _ observables , i.e. , observables with two outcomes @xcite . the authors in ref.@xcite have generalized it for observables with arbitrary many outcomes . moreover , they have shown that , unlike the original hardy s argument , the success probability of the many - outcome argument increases with the increase of the system s dimension . in this work the three - outcome hardy s argument has been considered . firstly , it has been shown that neither a @xmath26-qubit state nor a qubit - qutrit state exhibits this argument for three - outcome generalized measurement , i.e. , positive - operator - valued - measurement ( povm ) @xcite . using this result it has been further shown that this argument can be designed as a device independent schmidt rank as well as hilbert space dimension witness . before going to the main result a quick overview on the hardy s argument has been presented . l. hardy provided an elegant argument which , like bell s inequality @xcite , reveals nonlocality within quantum mechanics @xcite and it is commonly called ` hardy paradox ' . hardy s proof is usually considered `` the simplest form of bell s theorem '' @xcite . the argument requires two spatially separated observers , say alice and bob , each with two measurements ( the measurements for alice and bob are denoted by @xmath27 and @xmath28 respectively , with @xmath29 ) , each with two possible outcomes denoted by ` @xmath30 ' and ` @xmath24 ' . it puts restrictions on a certain choice of @xmath31 out of @xmath32 joint probabilities in the correlation matrix . one such choice is : @xmath33 [ hardy ] here @xmath34 denotes the conditional joint probability of obtaining outcome ` @xmath35 ' by alice and outcome ` @xmath2 ' by bob when they perform measurement @xmath27 and @xmath28 , respectively ; @xmath36 . the non zero probability in eq.([hardy ] ) ( i.e. left hand side of eq.([h4 ] ) ) is called hardy s success probability , @xmath37 . for @xmath26-qubit system the maximum achievable value of hardy s success is @xmath38 @xcite . it is important to note that for @xmath26-qudit system this maximum success probability remains same @xcite , i.e. , for showing the contradiction of quantum mechanics with _ local realism _ , higher dimensional systems give no advantage in experimental implementation of such a test . recently , the authors in @xcite have introduced a hardy like argument for @xmath39-outcome measurements , i.e. , @xmath40 . denoting the joint conditional probability as @xmath41 the argument reads as @xmath42 , @xmath43 , @xmath44 , @xmath45 . for two outcomes the argument boils down to the original hardy s argument , i.e. eq.([hardy ] ) . for three outcomes their argument explicitly looks : @xmath46 [ hardy3 ] likewise ( [ h4 ] ) , the left hand side in ( [ h34 ] ) measures the success probability of three - outcome hardy s test and similarly for the @xmath39-outcome cases . higher value of this quantity implies that experimentally it is easier to demonstrate the contradiction of quantum mechanics with _ local realism_. in quantum theory the joint conditional probabilities are calculated as : @xmath47 where , @xmath48 and @xmath49 are povms acting on alice s and bob s side respectively and @xmath50 is the shared state between alice and bob . considering @xmath26-qudit pure states and projective measurements the authors in @xcite find the optimal success probability for @xmath39-outcome hardy s test . as for example , for @xmath51-outcome case the optimal achievable success probability is @xmath52 , which is strictly greater than the optimal two - outcome hardy s success probability . this value can be achieved by performing three - outcome projective measurements on @xmath26-qutrit system . it seems to imply that the success probability increases with increasing system s dimension . however this implication is not conclusive . cause it has not yet been proved that by sharing a @xmath53 state ( or @xmath54 state ) and performing three - outcome generalized measurement one can not exhibit the hardy s paradox ( [ hardy3 ] ) with success probability greater than @xmath55 . in the following a more powerful result has been proved that neither a @xmath26-qubit state nor a qubit - qutrit state exhibits the three - outcome hardy s paradox ( [ hardy3 ] ) . first the 2-qubit case has been considered . let alice and bob share a 2-qubit state . both of them perform two three - outcome povms , with outcomes denoted by @xmath56 respectively . let @xmath57 denote the povm element corresponding to alice s outcome @xmath35 and similarly @xmath58 for bob s outcome @xmath2 . thus we have : @xmath59 there are following possible cases : ( i ) all the povm elements are rank one operators ; ( ii ) some of the povm elements may have more than one rank . in the course of analysis , intuitively , it will become clear that if the measurements with rank one povm elements do not pass the three - outcome hardy s test then it is even more difficult for the measurements with higher rank povm elements to pass it . * case ( i ) * : in this case all the povm elements can be considered as proportional to projection operators on some ray vectors , i.e. , @xmath60,~~\mbox{and}~~\mathcal{f}^n_j\propto\pi[\phi_j^n],\ ] ] where @xmath61\equiv|\psi_i^m\rangle\langle\psi_i^m|$ ] and @xmath62\equiv|\phi_j^n\rangle\langle\phi_j^n|$ ] . the 2-qubit state that exhibits the hardy s argument ( [ hardy3 ] ) must satisfy the conditions ( [ h31])-([h33 ] ) , which imply that each term on the left hand side of these equations must be zero . the condition @xmath63 implies that the concerned 2-qubit state is orthogonal to the product vector @xmath64 and similar is true for other cases . thus the conditions ( [ h31])-([h33 ] ) altogether imply that the concerned state must be orthogonal to the following nine vectors : @xmath65 @xmath66 @xmath67 @xmath68 @xmath69 without loss of generality , consider @xmath70 as the basis for alice s qubit system . other states on alice s side when written in linear combination of this basis , read as : @xmath71 similarly choosing @xmath72 as the basis for bob s qubit : @xmath73 consider the set latexmath:[$\{|\psi_1 ^ 0\rangle\otimes|\phi_1 ^ 0\rangle,|\psi_1 ^ 0\rangle\otimes|\phi_1 ^ 1\rangle,|\psi_1 ^ 1\rangle\otimes|\phi_1 ^ 0\rangle , product space @xmath53 . the nine vectors in the set @xmath75 written in the above basis read as : @xmath76 among the above nine vectors if it turns out that four are linearly independent then those four vectors span the whole 2-qubit tensor product hilbert space and hence there will be no vector to exhibit the hardy s argument ( [ hardy3 ] ) . however , from the above expressions it is clear that the set of vectors @xmath77 is linearly independent . if we consider the set @xmath78 then it will be linearly dependent provided @xmath79=-\alpha_1 ^ 2\beta_1 ^ 2\gamma_2 ^ 1=0 $ ] . eq.([a1 ] ) tells that neither @xmath80 nor @xmath81 can be zero . but , no such restriction applies for @xmath82 to be nonzero . in the similar way , analyzing the criteria for linear dependence of the different sets @xmath83 with @xmath84 , we obtain the following conditions : @xmath85 for exhibiting the hardy s argument ( [ hardy3 ] ) the concerned 2-qubit state must be orthogonal to the subspace spanned by the set @xmath77 and this unique state reads as : @xmath86 as the state turns out to be a product state it can not manifest the hardy s argument ( [ hardy3 ] ) . * case(ii ) * : here all the povm elements are in general not rank one operator . to satisfy the conditions ( [ h31])-([h33 ] ) the concerned 2-qubit state must be orthogonal to the subspace spanned by the following nine product operators : .any @xmath26-qubit state that manifest the hardy s argument ( [ hardy3 ] ) must be orthogonal to the support of each of these nine product operators . [ cols="^,^,^,^,^",options="header " , ] if @xmath87 is rank two and all others are rank one then it is straightforward to argue that ranges of these operators together span four dimension of the tensor product hilbert space @xmath53 and thus there is no space left for exhibiting hardy s argument ( [ hardy3 ] ) . similar is true for other cases . in some cases ( e.g. @xmath88 is rank two and rest are rank one ) ranges of these operators together span three dimension and in such cases the analysis boils down to the case(i ) . form the analysis so far presented , it is clear that no 2-qubit state exhibits the three - outcome hardy s argument . this fact provides information about the hilbert space dimension of the composite system s , i.e. , any system that manifests the hardy s argument ( [ hardy3 ] ) can not be a @xmath53 system . at this point one can not make any comment about the schmidt rank of the system , since a @xmath54 state may also have schmidt rank @xmath26 and can exhibit the hardy paradox ( [ hardy3 ] ) . similar analysis shows that no @xmath53 exhibits the hardy paradox ( [ hardy3 ] ) ( see the appendix ) . moving further it will be now shown that the hardy s argument ( [ hardy3 ] ) is useful for witnessing the minimal schmidt rank in device independent manner . in device independent scenario one does not have detailed knowledge about the experimental apparatus . so a black - box description of the experiment has been shown in fig([fig1 ] ) . on each side , the experimental device is depicted like a box with some knobs . a knob with different positions on each device , denoted respectively by @xmath27 and @xmath28 , allows alice and bob to change the parameters of each measuring apparatus . each measurement performed by alice and bob has @xmath39 possible outcomes . finally , the frequencies @xmath34 of occurrence of a given pair of outcomes for each pair of measurements have been collected . after some calculations with the observed frequency the aim is to make some conclusion about the schmidt rank of the state shared between the two devices . for a given unknown bipartite state alice and bob are asked to perform two different measurements with three outcomes . the resulting statistics @xmath89 have been collected . it will be checked whether the collected statistics satisfy the conditions described in eq.([hardy3 ] ) . according to these conditions the left hand side of ( [ h34 ] ) must be strictly greater than zero . among the three terms of ( [ h34 ] ) if one is nonzero ( say , @xmath90 ) and the rest two are zero then also the required condition is satisfied . it is important to note that sharing a @xmath26-qubit pure entangled state @xcite ( i.e. a state with schmidt rank two ) and performing suitable two - outcome measurements on each side the required conditions can be satisfied . here each of alice and bob will assign zero probability for the third outcome . in this way the maximum value of left hand side of ( [ h34 ] ) can reach up to @xmath55 @xcite . one can impose a more stringent restriction that all the three terms in the left hand side of ( [ h34 ] ) should be nonzero . the previous analysis tells that in quantum theory this stringent conditions can not be satisfied by performing three - outcome generalized measurements on @xmath26-qubit ( qubit - qutrit ) state . however , alice and bob can have the following strategy . suppose they share the following higher dimensional state : @xmath91 where the particles @xmath92 and @xmath93 are in alice s lab and the particles @xmath94 and @xmath95 are in bob s lab . each @xmath96 is a @xmath26-qubit pure entangled states and hence each of them are of schmidt rank two . here the primed particles behave as _ flag _ variable . whenever the primed particles are in the state @xmath97 ( which alice and bob can know by performing a von neumann measurement in @xmath98 basis ) then alice and bob certainly know that the unprimed particles are in the state @xmath99 . note that in the bipartition @xmath100 vs @xmath101 the schmidt number is two . if the state is @xmath102 , then on their respective particle ( unprimed ) they perform suitable two - outcome measurements , which exhibits the two - outcome hardy s argument ( [ hardy ] ) and rename the outcomes accordingly . thus they are able to make the first term in the left hand side of ( [ h34 ] ) nonzero . similarly , when the unprimed particles are in the state @xmath103 ( @xmath104 ) , alice and bob can make the second ( third ) term in ( [ h34 ] ) nonzero by performing suitable measurements and renaming the outcomes accordingly @xcite . sharing this type of states the stringent conditions , that all the terms in ( [ h34 ] ) are nonzero , can be satisfied . but , due to convexity the success probability can not be greater than @xmath55 . thus , whenever the success probability is strictly greater than @xmath55 , the shared state must have schmidt rank greater than @xmath26 . this provides the information about the minimal schmidt rank of the shared bipartite system and importantly it has been done in device independent manner . it also gives information about the minimal dimension of the shared quantum system , that the resulting statistics can not be obtained from a @xmath26-qubit or a qubit - qutrit state . besides revealing the the contradiction of quantum mechanics with local - realism hardy s argument also finds applications in various information theoretic tasks . it has been proved to be useful in witnessing post quantum correlations @xcite . in the recent times various device - independent @xcite information theoretic protocols like cryptography @xcite , randomness certification @xcite , hilbert spaces dimension witness @xcite make use of nonlocality arguments @xcite . in this work it has been shown that such an argument turns out to be useful for inspecting the minimal schmidt rank as well as the minimal hilbert space dimension in device independent manner . * acknowledgments * : it is a pleasure to thank guruprasad kar for various simulating discussions and useful suggestions . we also thank sibasish ghosh , samir kunkri , and ramij rahaman for useful discussions . am acknowledge support from the csir project 09/093(0148)/2012-emr - i .

recently , the study of the dynamics of @xmath3 on the @xmath2-character variety of @xmath5 for a compact hyperbolizable @xmath6-manifold @xmath0 with nonempty boundary and no toroidal boundary components , has led to the somewhat surprising discovery that often the dynamical and geometric decompositions of the @xmath2-character variety do not coincide . the @xmath2-character variety of @xmath5 denoted @xmath7 is the geometric quotient of @xmath8 by inner automorphisms of @xmath9 sitting inside @xmath10 is @xmath11 the set of discrete and faithful representations , which can equivalently be thought of as the deformation space of hyperbolic @xmath6-manifolds homotopy equivalent to @xmath12 the action of @xmath3 on the interior of @xmath13 is properly discontinuous ( see canary @xcite ) , but if @xmath0 has a primitive essential annulus , the action can not be properly discontinuous on all of @xmath13 ( canary - storm @xcite ) . minsky in @xcite was the first to observe that there exists a domain of discontinuity , called the set of primitive - stable representations , for @xmath3 containing the interior of @xmath13 as well as points on @xmath14 when @xmath0 is a handlebody . in particular , the set of primitive - stable representations contains both discrete and faithful representations and dense representations . jeon - kim - ohshika - lecuire @xcite gave a complete criterion for a discrete and faithful representation to be primitive - stable in terms of disc - busting properties of the ending lamination and parabolic loci of the associated hyperbolic @xmath6-manifold . in her thesis @xcite , the second author gave a generalization of the primitive - stable condition to representations of the fundamental group of a compression body into @xmath2 , called the separable - stable condition , and she showed that the set of separable - stable representations is a domain of discontinuity for the action of @xmath3 that contains the interior of @xmath13 as well as point on @xmath15 in this paper , we give a characterization of which discrete and faithful representations of @xmath1 into @xmath2 are separable - stable analogous to the jeon - kim - ohshika - lecuire criterion . a compression body @xmath0 is the boundary connect sum of a @xmath6-ball , a collection of trivial @xmath16-bundles over surfaces , and a handlebody , where the other components are connected to the @xmath6-ball along disjoint discs . its fundamental group is a free product of closed surface groups and a free group . if @xmath0 is not the boundary connect sum of two trivial @xmath16-bundles over closed surfaces , then an element in @xmath1 is separable if it lies in a proper factor of a free decomposition of @xmath4 if @xmath0 is the boundary connect sum of @xmath17 and @xmath18 where @xmath19 and @xmath20 are closed surfaces , then an element is separable if it lies in a factor of a decomposition @xmath21 where @xmath22 is homotopic to a simple closed curve on @xmath19 or @xmath20 . a homomorphism @xmath23 is separable - stable if every geodesic defined by a separable element in the cayley graph of @xmath24 is mapped to a uniform quasi - geodesic in @xmath25 . see section [ sep - stab ] for a precise definition . for any discrete and faithful representation @xmath26 we let @xmath27 denote the corresponding hyperbolic @xmath6-manifold obtained from taking the quotient @xmath28 the main theorem of this article is : [ thm : main ] let @xmath0 be a nontrivial hyperbolizable compression body without toroidal boundary components that is not the boundary connect sum of @xmath17 and @xmath18 where @xmath19 and @xmath20 are closed surfaces with genus at least two . let @xmath29 be a discrete and faithful representation of @xmath1 into @xmath2 . then , @xmath29 is separable - stable if and only if in @xmath30 each component of the parabolic loci and each ending lamination is disc - busting . a measured lamination @xmath31 is _ disc - busting _ if there exists @xmath32 such that @xmath33 for all unweighted meridians @xmath34 . otherwise it is _ disc - dodging_. we say that an ending lamination is disc - busting if it is the support of a measured lamination that is disc - busting . such a lamination is used by jeon - kim - ohshika - lecuire in @xcite for the characterization of primitive stable representations of free groups , and has a root in doubly incompressible laminations ( see kim - lecuire - oshika @xcite or lecuire @xcite ) . in @xcite , the second author showed that representations obtained by pinching masur domain curves or masur domain laminations on @xmath35 are separable - stable . for the case where @xmath0 is the boundary connect sum of @xmath36 and @xmath37 where @xmath19 and @xmath20 are closed surfaces , we establish the following analogous characterization of the set of discrete and faithful representations that are separable - stable . [ thm : special case ] let @xmath0 be the boundary connect sum of @xmath17 and @xmath18 where @xmath19 and @xmath20 are closed surfaces with genus at least two . let @xmath26 be a discrete and faithful representation . then , @xmath29 is separable - stable if and only if in @xmath30 each component of the parabolic loci and each ending lamination is annulus - busting . a measured lamination @xmath31 is _ annulus busting _ if there exists @xmath38 such that @xmath39 for any essential annulus @xmath40 a _ compression body _ is a compact , orientable , irreducible @xmath6-manifold @xmath0 with a boundary component , @xmath41 called the _ exterior boundary _ , such that the inclusion @xmath42 induces a surjection on the level of fundamental groups @xmath43 . the other boundary components are called _ interior boundary components_. equivalently , a compression body is the boundary connect sum of a @xmath6-ball , a collection of trivial @xmath16-bundles over surfaces , and a handlebody , where the other components are connected to the @xmath6-ball along disjoint discs . the fundamental group of a compression body can be decomposed as a free product , @xmath44 where @xmath45 is isomorphic to a closed surface group for @xmath46 and @xmath47 is infinite cyclic @xmath48 . by grushko s theorem ( @xcite ) and kurosh s subgroup theorem ( @xcite ) , any other decomposition of the fundamental group into a free product , @xmath49 where each factor is freely indecomposable , satisfies @xmath50 and @xmath51 , up to re - ordering . a compression body is _ trivial _ if @xmath52 is an isomorphism , i.e. if @xmath0 is a trivial @xmath16-bundle . we say that @xmath0 is _ uniquely freely decomposable _ if @xmath0 is the boundary connect sum of two trivial @xmath16-bundles over closed surfaces since , in this case , the decomposition of @xmath1 is essentially unique . an _ essential disc _ in @xmath0 is a properly embedded disc whose boundary is nontrivial in @xmath53 a compression body @xmath0 has the property that any splitting of @xmath1 as a free product or as an hnn - extension over the trivial group can be realized by an essential disc in @xmath0 in the following sense . suppose that @xmath54 is a separating essential disc and @xmath55 if @xmath56 for @xmath57 then @xmath58 where @xmath59 is inclusion and the base point is chosen to lie in @xmath60 in this case , we say that @xmath54 realizes the splitting of the fundamental group . as we can move the basepoint around , this is only well - defined up to conjugation . if @xmath54 is a non - separating essential disc and @xmath61 then @xmath62 we say that @xmath54 realizes this hnn - extension . [ lem : freeprod ] let @xmath0 be a compression body and let @xmath63 be a nontrivial splitting of @xmath1 into a free product . then , the splitting is realizable by an essential disc . [ lem : hnn ] let @xmath0 be a compression body . any splitting of the form @xmath64 is realizable by an essential disc . recall that @xmath65 is isomorphic to @xmath66 by lemma [ lem : freeprod ] , we can realize the latter splitting by an essential disc @xmath67 . then , @xmath67 separates @xmath68 into two components @xmath69 and @xmath70 where @xmath71 is conjugate to @xmath72 and @xmath70 is a compact , oriented , irreducible @xmath6-manifold with infinite cyclic fundamental group , i.e. a solid torus . then , the essential disc @xmath54 in the solid torus , realizes the original splitting . in a free group , an element is called _ primitive _ if it can be completed to a free generating set and an element is _ separable _ if it lies in a free factor . the notion of separability is originally due to stallings ( @xcite ) in his study of separable sets in free groups that generalized whitehead s ( @xcite ) study of primitive elements of free groups . although the notion of a primitive element does not generalize to other freely decomposable groups , the notion of a separable element does . if @xmath0 is a compression body that is not uniquely freely decomposable , then an element in @xmath1 is _ separable _ if it lies in a proper factor of a decomposition of @xmath1 into a nontrivial free product . if @xmath0 is the boundary connect sum of two trivial @xmath16-bundles over closed surfaces @xmath19 and @xmath20 , then an element in @xmath1 is _ separable _ if it lies in a factor of a decomposition @xmath73 where @xmath22 is homotopic to a simple closed curve on @xmath19 or @xmath20 one can think of separability in the following geometric way . if @xmath0 is not uniquely freely decomposable , by lemma [ lem : freeprod ] , an element @xmath74 of @xmath1 lies in a proper factor of a free decomposition if and only if its associated curve in @xmath0 is homotopic to one that misses an essential disc . similarly , lying in a factor of a decomposition @xmath73 where @xmath22 is homotopic to a simple closed curve on @xmath19 or @xmath20 is equivalent to missing the corresponding essential annulus @xmath75 up to homotopy . essential annulus _ in @xmath0 is a properly embedded incompressible annulus that is not homotopic , relative to its boundary , into the boundary of @xmath12 let @xmath24 denote the fundamental group of @xmath0 and let @xmath76 be a finite symmetric generating set for @xmath77 the _ cayley graph of @xmath24 with respect to @xmath76 _ , denoted @xmath78 , is a graph where the vertices are in one - to - one correspondence with the elements of @xmath24 and there is an edge between @xmath74 and @xmath79 if there exists an element @xmath80 in the generating set @xmath76 such that @xmath81 . the group @xmath24 acts on @xmath78 by left multiplication on the vertices . since @xmath24 is torsion - free and gromov hyperbolic each element @xmath74 of @xmath24 has two fixed points on @xmath82 . for each @xmath74 in @xmath24 , let @xmath83 and @xmath84 denote the repelling and attracting fixed points of @xmath74 acting on @xmath82 . let @xmath85 denote the set of geodesics connecting @xmath83 and @xmath84 and @xmath86 denote the set of geodesics @xmath87 in @xmath78 such that @xmath87 is contained in @xmath85 for some separable element @xmath74 . given a representation @xmath88 and a basepoint @xmath89 in @xmath25 , there exists a unique @xmath29-equivariant map @xmath90 taking the identity to @xmath89 and edges to geodesic segments . a representation @xmath88 is called _ @xmath91-separable - stable _ if there exists a basepoint @xmath89 in @xmath25 such that @xmath92 is a @xmath91-quasi - geodesic for any @xmath87 in @xmath93 we will call @xmath29 , a representation , _ separable - stable _ if there exists @xmath91 such that @xmath29 is @xmath91-separable - stable . separable - stability is independent of the choice of basepoint in @xmath25 and the choice of generators @xmath76 for @xmath24 , and separable - stability is invariant under conjugation ( see lemma 16 in lee @xcite ) . the set of separable - stable representations forms a domain of discontinuity for the action of @xmath3 strictly larger than the interior of @xmath13 . namely the following is true . let @xmath0 be a nontrivial compression body without toroidal boundary components . the outer automorphism group @xmath3 acts properly discontinuously on the set of separable - stable representations , which contains the interior of @xmath13 as well as points on the boundary of @xmath94 in this section , we describe otal s ( @xcite ) generalization of the whitehead graph for a free group to the whitehead graph for a compression body , which we will use in proving the sufficient condition . let @xmath0 be a nontrivial compression body without toroidal boundary components . a _ system of meridians _ @xmath95 on @xmath0 is a collection of pairwise nonisotopic and pairwise disjoint meridians @xmath96 bounding discs @xmath97 such that @xmath98 is a disjoint union of trivial @xmath16-bundles over closed surfaces , where @xmath99 is a regular neighborhood of @xmath97 . by thurston s hyperbolization theorem ( see kapovich @xcite ) and marden ( @xcite ) , we can fix @xmath100 a convex cocompact representation of @xmath1 such that @xmath101 is homeomorphic to @xmath0 , where @xmath102 is the conformal boundary of @xmath103 let @xmath104 be the limit set of @xmath105 let @xmath106 be a closed subset of @xmath107 that is @xmath108-invariant and also invariant under switching the two factors . in this paper , @xmath106 will be one of the following two examples . * if @xmath109 is a closed geodesic in @xmath110 , then @xmath111 is the set of endpoints of all the lifts of @xmath112 * if @xmath31 is a lamination on @xmath35 that is in tight position with respect to @xmath95 , then @xmath113 is the set of endpoints of all lifts of all leaves of @xmath114 see the discussion before proposition [ discbusting ] for the definition of a lamination in tight position and lemma [ endpoints ] to see why @xmath113 is well - defined . identify @xmath35 with @xmath102 . let @xmath95 be a system of meridians bounding the discs @xmath115 and suppose @xmath116 is the disjoint union of @xmath117 where each @xmath118 is a closed surface of genus at least two . the _ whitehead graph for @xmath0 of @xmath106 with respect to @xmath95 _ , denoted @xmath119 is a collection of not necessarily connected graphs , @xmath120 where the elements in the collection are in one - to - one correspondence with the components of @xmath116 . in @xmath116 , there are two copies @xmath121 and @xmath122 of each @xmath97 in @xmath54 . given a component @xmath123 of @xmath116 , the vertices in the corresponding graph @xmath124 are in one - to - one correspondence with the components of @xmath125 in the frontier of @xmath123 . abusing notation , relabel the vertices @xmath126 fix a jordan curve @xmath127 in @xmath128 that is invariant under a conjugate of @xmath129 , which we will continue to denote @xmath129 . let @xmath130 denote the boundary component of @xmath123 coming from @xmath131 . fix a lift @xmath132 of each @xmath133 in @xmath134 such that @xmath135 lies in the component of the preimage of @xmath130 in @xmath134 containing @xmath127 on its boundary . let @xmath136 be the open set in @xmath134 , bounded by @xmath132 not containing @xmath127 . the edges from @xmath137 to @xmath138 will be in one - to - one correspondence with elements @xmath74 in @xmath129 such that @xmath139 is nonempty . we will denote such an edge @xmath140 . although these edges are directed , for each edge from @xmath136 to @xmath141 labeled @xmath74 , there is an edge from @xmath141 to @xmath136 labeled @xmath142 . [ otal ] a connected component of @xmath143 is _ strongly connected _ if there exists a cycle that represents a nontrivial element of @xmath129 . a connected component of @xmath143 has a _ strong cutpoint _ if we can express the graph as the union of two graphs @xmath144 and @xmath145 that intersect in a single vertex such that either @xmath144 or @xmath145 is not strongly connected . we take the convention that a cycle @xmath146 corresponds to the group element @xmath147 . although we made several choices when defining @xmath148 , the two properties defined above , strong connectedness and the presence of a strong cutpoint are independent of these choices ( see section 3.1.2 in lee @xcite ) . otal in @xcite , also , gave a dichotomy between the whitehead graph of a separable element and the whitehead graph of a disc - busting lamination in tight position with respect to a system of meridians . in this section , we define and collect the relevant properties of a lamination in tight position . the results in this section are essentially all due to otal in @xcite . since @xcite is difficult to obtain for each result we either give an alternate source containing a proof or we provide a proof . a lamination @xmath31 is in _ tight position with respect to a system of meridians @xmath95 _ if there are no waves disjoint from @xmath31 where a _ wave _ is an arc @xmath149 satisfying the following two conditions : 1 . @xmath149 has endpoints on @xmath95 , but its interior is disjoint from @xmath95 and , 2 . @xmath149 is homotopic , relative to its endpoints , in @xmath0 but not in @xmath35 into @xmath150 for any disc - busting lamination @xmath31 we can find a system of meridians @xmath95 such that @xmath31 is in tight position with respect to @xmath150 otal proved this for masur domain laminations but his proof applies in this more general case . [ discbusting ] if @xmath31 is a disc - busting lamination , there exists a system of meridians @xmath95 with respect to which @xmath31 is in tight position . suppose that @xmath31 is not in tight position with respect to a system of meridians @xmath95 . let @xmath149 be a wave and suppose its endpoints lie on @xmath151 then the endpoints of @xmath149 split @xmath96 into two arcs @xmath152 and @xmath153 such that @xmath154 and @xmath155 are the endpoints of @xmath156 since @xmath149 is homotopic into @xmath96 through @xmath0 but not in @xmath157 we have that @xmath158 and @xmath159 are both meridians . since @xmath31 is disc - busting , @xmath160 and @xmath161 by construction both @xmath162 and @xmath163 are disjoint from @xmath164 for any @xmath165 replacing @xmath96 with either @xmath162 or @xmath163 will produce a new collection of meridians @xmath166 and @xmath167 respectively , whose intersection number with @xmath31 is at least @xmath168 less than @xmath169 we claim that at least one of these new collections of meridians @xmath166 or @xmath167 is a system of meridians . by construction @xmath170 and @xmath163 bound a pair of pants on @xmath53 since they are all meridians and @xmath0 is irreducible , the discs they bound , @xmath171 and @xmath172 bound a ball in @xmath0 . so @xmath173 contains a component that is a ball @xmath174 . we will use a @xmath175 superscript to denote the copy of @xmath97 ( or @xmath176 ) that lies on the boundary of the ball , i.e. @xmath177 lie on the boundary of @xmath178 since @xmath179 is a collection of trivial @xmath16-bundles over closed surfaces , @xmath180 must consist of a collection of trivial @xmath16-bundles over closed surfaces and two balls , one of which is the ball @xmath174 described above . the other ball must contain either @xmath181 or @xmath182 on its boundary ( possibly both ) . if it contains @xmath183 then @xmath167 will be a system of meridians , and vice versa . so , replacing @xmath96 with one of @xmath162 and @xmath163 results in a system of meridians whose intersection number with @xmath31 is at least @xmath168 less than @xmath169 if @xmath31 is not in tight position with respect to this new system of meridians , repeat this process . since @xmath184 for any system of meridians @xmath185 , eventually , this process must terminate . an important property of disc - busting , minimal laminations in tight position with respect to a system of meridians @xmath95 is that any lift of a leaf has well - defined endpoints . [ endpoints ] let @xmath0 be a nontrivial compression body . let @xmath186 be a convex cocompact representation uniformizing @xmath0 . let @xmath31 be a lamination on @xmath102 in tight position with respect to @xmath150 let @xmath87 be a leaf @xmath31 such that any half leaf of @xmath87 intersects @xmath150 then , any lift @xmath187 of @xmath87 to @xmath188 has two well - defined endpoints in @xmath189 propositions [ discbusting ] and lemma [ endpoints ] imply that if @xmath31 is a disc - busting ending lamination , then @xmath113 , the set of endpoints of lifts of leaves of @xmath190 is well - defined . in particular , we will refer to the whitehead graph of @xmath31 as the whitehead graph of @xmath191 otal proved the following two results , which provide a dichotomy between separable elements and disc - busting laminations . [ separableelement ] let @xmath0 be a nontrivial compression body that is not the boundary connect sum of two trivial @xmath16-bundles over closed surfaces . if @xmath74 is separable , then for any system of meridians @xmath95 some connected component of @xmath192 is either not strongly connected , or has a strong cut - point . [ otal ] if @xmath31 is a disc - busting lamination in tight position with respect to a system of meridians @xmath193 then the whitehead graph of @xmath31 with respect to @xmath95 is strongly connected and without strong cutpoints . in @xcite , otal states proposition [ otal ] only for masur domain laminations , but in fact his proof applies in this more general case . in general , for @xmath194 a discrete and faithful representation , the associated hyperbolic @xmath6-manifold @xmath27 will be homotopy equivalent but not necessarily homeomorphic to @xmath12 since otal s results only apply in the case when @xmath0 is a compression body , in this section we associate a whitehead graph to any hyperbolizable @xmath6-manifold @xmath195 homotopy equivalent , but not necessarily homeomorphic to @xmath0 . we can do this by using the characteristic compression body introduced by bonahon in @xcite . we also describe how to compare whitehead graphs for @xmath0 and @xmath195 . let @xmath196 denote the set of compact , orientable , irreducible , atoroidal ( unmarked ) @xmath6-manifolds homotopy equivalent to @xmath0 , up to homeomorphism . let @xmath197 , \ldots , [ m_n]$ ] be the elements in @xmath198 where @xmath199 is homeomorphic to @xmath12 for each @xmath200 fix a convex cocompact representation @xmath201 such that @xmath202 is homeomorphic to the interior of @xmath203 let @xmath204 be an orbit map for @xmath201 as in the definition of separable - stable representations where @xmath205 . then , @xmath206 is a quasi - isometric embedding and there is a continuous extension , @xmath207 which is a homeomorphism . we suppressed the basepoint in @xmath206 since @xmath208 is independent of the choice of basepoint . then , let @xmath209 be the homeomorphism from @xmath210 to @xmath211 in order to compare the whitehead graph of @xmath199 and @xmath200 we need a way to compare systems of meridians on the two manifolds . define a map @xmath212 from the set @xmath213 of unweighted meridians on @xmath200 to the set @xmath214 of unweighted meridians on @xmath199 in the following way ( see figure [ pic ] ) . on the characteristic compression body @xmath215 of @xmath200 is mapped to the system of meridians @xmath216 on @xmath217,width=491 ] let @xmath218 denote the essential disc bounded by @xmath150 first suppose that @xmath95 is a meridian in @xmath200 such that @xmath218 separates @xmath203 then , @xmath218 realizes a splitting @xmath219 up to conjugation . in @xmath199 we can realize the splitting by an essential disc @xmath67 . let @xmath220 be the boundary of @xmath221 this is well - defined up to isotopy . indeed if we consider @xmath220 on the conformal boundary of @xmath222 then some lift @xmath223 separates @xmath224 into two components one of which contains all the attracting fixed points of any element whose reduced word representation starts with an element in @xmath72 and the other contains all the attracting fixed points of any element whose reduced word representation starts with an element in @xmath178 since such points are dense in @xmath225 this determines @xmath226 up to isotopy on @xmath227 now suppose that @xmath218 is non - separating . then , @xmath218 realizes an hnn extension , @xmath228 by lemma [ lem : hnn ] , we can realize such a splitting in @xmath199 by an essential disc @xmath67 . as before , let @xmath220 be the boundary of @xmath221 again , to see that @xmath220 is well - defined notice that some lift @xmath229 separates @xmath134 into two components one of which contains all the attracting fixed points of any element whose reduced word representation starts with an element of @xmath72 or @xmath230 and the other contains all the attracting fixed points of any element whose reduced word representation starts with @xmath231 where @xmath230 is the new generator associated to the hnn - extension . any manifold @xmath0 with compressible boundary has a _ characteristic compression body _ @xmath232 , unique up to isotopy that has the following properties ( see bonahon @xcite or mccullough - miller @xcite ) . the exterior boundary of @xmath232 coincides with @xmath35 . the closure of @xmath233 contains no essential compression discs and no component of @xmath234 is a 3-ball . let @xmath215 denote the characteristic compression body of @xmath235 with any trivial compression body components removed . [ lem : mermap ] let @xmath236 be a system of meridians on @xmath237 then , the collection of meridians @xmath238 is a system of meridians on @xmath217 first we will show that the meridians @xmath238 can be realized such that they are pairwise disjoint and pairwise nonisotopic . to see that we can realize @xmath239 and @xmath240 disjointly , observe that since @xmath164 and @xmath241 are disjoint , any two lifts @xmath242 and @xmath243 , have the property that there exist @xmath141 a component of @xmath244 and @xmath245 a component of @xmath246 such that @xmath247 and @xmath248 are each non - empty and disjoint from each other . there exists lifts @xmath249 and @xmath250 and components @xmath251 and @xmath252 of @xmath253 and @xmath254 respectively , such that @xmath255 and @xmath256 then , we can homotope , @xmath239 and @xmath240 to be disjoint . to see that @xmath239 and @xmath240 are not isotopic , notice that if they were isotopic , then there exists lifts @xmath249 and @xmath250 that bound a region in @xmath134 containing no limit points . then , the same would be true about the corresponding lifts @xmath242 and @xmath257 which contradicts that they are not isotopic . let @xmath258 be the disc bounded by @xmath239 in @xmath217 finally , we want to see that @xmath259 is a collection of trivial @xmath16-bundles over closed surfaces . we will first show that @xmath260 is a collection of trivial @xmath16-bundles over closed surfaces . recall that @xmath261 where @xmath215 is the characteristic compression body and @xmath262 has incompressible boundary . by definition , @xmath263 is a collection of trivial @xmath16-bundles over closed surfaces . we claim that the components of @xmath262 are all trivial @xmath16-bundles over closed surfaces . let @xmath127 be a component of @xmath264 and let @xmath265 be a component of @xmath266 since @xmath127 has incompressible boundary and can not be a @xmath6-ball , we have that @xmath267 lies in a surface group factor of some maximal free decomposition of @xmath1 i.e. @xmath267 is a subgroup of some @xmath268 . the image of @xmath269 must be finite index in @xmath268 since they are both closed surface groups . this implies that @xmath269 is finite index in @xmath270 which in turn implies that @xmath127 is an @xmath16-bundle ( theorem 10.5 in hempel @xcite ) . so @xmath127 is either a trivial @xmath16-bundle or twisted @xmath16-bundle . it remains to check that @xmath127 can not be a twisted @xmath16-bundle . if @xmath127 were a twisted @xmath16-bundle , then the fundamental group of @xmath127 would be the fundamental group of a non - orientable surface , but this is impossible as @xmath267 is finite index in @xmath271 so , we have shown that the components of @xmath262 are all trivial @xmath16-bundles over surfaces . now @xmath272 is formed by gluing @xmath262 and @xmath263 along their boundaries , so the result will be a collection of trivial @xmath16-bundles over closed surfaces . now , consider @xmath273 a component of @xmath274 . we want to see that @xmath273 is a trivial @xmath16-bundle over a closed surface . if not , then either @xmath273 is a @xmath6-ball , or @xmath273 has compressible boundary . in the former case , let @xmath238 be the components of @xmath220 that lie on the boundary of @xmath273 ( notice that we can have two copies of @xmath164 on the boundary of @xmath273 , relabel accordingly ) . fix a lift @xmath275 of @xmath276 then , the corresponding lifts @xmath277 bound a connected region in @xmath134 containing no limit points . the corresponding lifts of @xmath278 of @xmath95 would have the same property . this implies that there is a @xmath6-ball component of @xmath279 a contradiction . if @xmath273 has compressible boundary , then there exists a closed curve @xmath109 in @xmath273 such that @xmath109 does not lie in any surface group factor of a free decomposition of @xmath4 moreover , any lift @xmath280 of @xmath109 has the property that both endpoints lie in the same component of @xmath253 for any lift of any @xmath281 since @xmath109 misses any @xmath282 then , we would also have that @xmath280 has both endpoints in the same component of @xmath283 so , @xmath109 misses all the meridians @xmath281 a contradiction , since @xmath284 consists of trivial @xmath16-bundles over closed surfaces . using lemma [ lem : mermap ] we can compare the whitehead graphs of @xmath215 and @xmath217 recall from section [ white ] the definition of the whitehead graph of a compression body . for @xmath285 we will abuse notation , and let @xmath286 denote the image of @xmath106 under @xmath287 [ whotherhomeo ] let @xmath200 lie in @xmath196 and let @xmath215 be the characteristic compression body of @xmath203 if @xmath288 is strongly connected and without strong cutpoint , then @xmath289 is strongly connected and without strong cutpoint . recall that we have a @xmath24-equivariant homeomorphism @xmath290 although @xmath291 is only defined on the limit set of @xmath292 it induces a natural map on lifts of @xmath293 to lifts of @xmath294 for example , given a lift @xmath242 of @xmath164 , there exists a corresponding lift @xmath249 of @xmath295 that partitions @xmath296 in the same way as @xmath242 via @xmath297 we ll abuse notation to denote such a lift @xmath298 as @xmath299 more precisely , suppose that @xmath300 then , @xmath301 is the lift @xmath249 of @xmath239 such that if @xmath302 then @xmath303 where @xmath304 up to switching @xmath305 and @xmath306 suppose that @xmath123 is a component of @xmath307 recall that when defining the whitehead graph , we fixed @xmath127 a jordan curve in @xmath134 invariant under @xmath129 . we let @xmath130 denote the boundary component of @xmath123 intersecting @xmath95 and we fixed the lift @xmath308 of @xmath130 whose boundary is @xmath309 if @xmath242 corresponds to a vertex of @xmath310 then map it to @xmath299 we want to show that this is a well - defined map on the vertices . first notice that since @xmath311 is a system of meridians , @xmath312 consists of trivial @xmath16-bundles over surfaces . one of these components @xmath313 must have fundamental group conjugate to @xmath314 by the uniqueness of the free decomposition of @xmath1 ( see section [ sec : compbody ] ) . let @xmath315 be the jordan curve invariant under @xmath129 . note that @xmath316 let @xmath317 be the component of @xmath123 intersecting @xmath220 and let @xmath318 denote the component of the preimage of @xmath317 with @xmath315 on its boundary . then , we claim that @xmath301 lies on @xmath319 indeed , this is clear from the fact that @xmath301 partitions @xmath296 in the same way as @xmath242 via @xmath320 as described above . so we see that the map on the vertices is well - defined . now suppose that @xmath321 is an edge in @xmath322 this implies that the intersection of @xmath106 and @xmath323 is nonempty . by construction , this is true if and only if the same is true for @xmath286 and @xmath324 so , we have a bijection between the whitehead graph of @xmath288 and that of @xmath325 where an edge in @xmath288 labeled @xmath74 is mapped to an edge in @xmath325 with the same label . in particular , we see that @xmath288 is strongly connected and without cutpoint if and only if the same is true for @xmath326 recently , mj ( @xcite ) has proved the existence of cannon - thurston maps for general kleinian groups . in particular , given a kleinian group @xmath24 and a discrete and faithful representation @xmath327 , he showed that there exists a continuous extension @xmath328 of @xmath329 moreover , he gives a characterization , which we describe below , of which points are mapped non - injectively by @xmath330 . before stating mj s result we recall some facts from hyperbolic geometry that we will need . we will restrict to the case that @xmath24 has no non - cyclic abelian subgroups since this is the case in this paper . there exists a constant @xmath331 called the margulis constant , such that for any hyperbolic @xmath6-manfiold , and any @xmath332 each component of the @xmath333-thin part of @xmath334 is a metric neighborhood of a closed geodesic or a parabolic cusp homeomorphic to @xmath335 given @xmath336 the _ non - cuspidal part of @xmath27 _ , denoted @xmath337 is the submanifold of @xmath27 obtained by removing the noncompact components of the @xmath333-thin part of @xmath338 when @xmath339 has parabolics , @xmath340 has a _ relative compact core , _ @xmath341 that is a compact submanifold whose inclusion is a homotopy equivalence and that intersects each cusp neighborhood along an annulus whose core curve we call a _ parabolic curve _ ( mccullough @xcite ) . let @xmath342 denote the union of these annuli . moreover , by the tameness ( agol @xcite or calegari - gabai @xcite ) and the uniqueness of cores ( mccullough - miller - swarup @xcite ) @xmath27 is homeomorphic to the interior of @xmath127 . each frontier component @xmath343 of @xmath344 in @xmath345 faces an end @xmath346 with a well - defined ending lamination @xmath347 obtained from taking the support of a projective limit of any sequence of simple closed geodesics exiting the end @xmath348 the ending lamination @xmath349 is a filling minimal lamination on @xmath343 ( canary @xcite ) . let @xmath350 be the union of @xmath349 and the parabolic curves , and assume that each is disc - busting . if we fix a convex cocompact structure @xmath110 on @xmath127 , then we can identify @xmath351 with a geodesic lamination on @xmath102 . by lemma [ endpoints ] any lift @xmath187 of a leaf @xmath87 of @xmath349 for any @xmath59 has two well defined endpoints at infinity on @xmath128 . since @xmath100 is convex cocompact , we can identify @xmath82 with @xmath128 . we introduce a relation @xmath352 for points in @xmath128 ( and so also on @xmath82 ) such that @xmath353 if and only if either @xmath354 and @xmath355 are the endpoints of a leaf of @xmath356 or ideal vertices of a complementary region of @xmath356 . in contrast to the case of groups without parabolics , there are complementary regions of @xmath357 which are ideal polygons with infinitely many sides . these are exactly regions touching lifts of parabolic curves , which are isolated leaves . the relation @xmath358 may not be transitive . we define @xmath359 to be the transitive closure of @xmath352 , which is an equivalence relation . mj s theorem about the identified points of a cannon - thurston map can be adapted to our case as follows . [ mj2 ] let @xmath23 be a discrete and faithful representation . the identification of @xmath78 with its image under @xmath360 extends continuously to a map @xmath361 . moreover , for @xmath362 , we have @xmath363 if and only if @xmath364 . in this section , we use the dichotomy between the whitehead graph of a separable element and a disc - busting lamination to show the sufficient condition of theorem [ thm : main ] , namely that when @xmath1 is not uniquely freely decomposable , a representation is separable - stable if each end invariant is disc - busting . for the sufficient condition , we ll use mj s result described in the previous section . the following lemma , whose proof can be found in lee @xcite lemma 21 , sets the stage for using mj s result . [ notss ] let @xmath26 be discrete and faithful . if @xmath29 is not separable - stable , then there exists a sequence of separable elements @xmath365 such that if @xmath366 and @xmath367 are the endpoints of @xmath368 then @xmath369 and @xmath370 converge , up to subsequence , to the same point in @xmath371 but @xmath366 and @xmath367 converge to @xmath372 and @xmath373 in @xmath374 where @xmath375 [ sufficient ] let @xmath0 be a nontrivial compression body , without toroidal boundary components , that is not the boundary connect sum of two trivial @xmath16-bundles over closed surfaces . let @xmath26 be a discrete and faithful representation . if each parabolic locus and ending lamination of @xmath27 is disc - busting , then @xmath29 is separable - stable . notice that the statement of the above proposition does not make any assumption about the homeomorphism type of @xmath338 suppose that @xmath27 is homeomorphic to the interior of @xmath376 where @xmath377 \in a(m).$ ] let @xmath378 be the union of the parabolic loci and ending laminations on @xmath379 if @xmath29 is not separable - stable , then by lemma [ notss ] there exists a sequence of separable elements @xmath365 such that the fixed points of @xmath365 in @xmath82 converge to @xmath372 and @xmath380 but the endpoints of @xmath381 converge to the same point in @xmath382 by theorem [ mj2 ] we have that @xmath383 and @xmath384 are each an endpoint of a leaf of one of the end invariants , although not necessarily the same end - invariant . suppose that @xmath383 contains an endpoint of a leaf of @xmath385 let @xmath386 be the set of limit points of @xmath387 then @xmath388 is @xmath389-invariant and also invariant under switching the two factors . moreover , @xmath390 lies in @xmath391 this implies that @xmath392 contains @xmath393 for any system of meridians @xmath95 ( see the proof of proposition 22 in lee @xcite ) . fix a system of meridians @xmath95 such that @xmath31 is in tight position with respect to @xmath150 then , @xmath393 is strongly connected and without any strong cutpoints by proposition [ otal ] . by lemma [ whotherhomeo ] , the whitehead graph @xmath394 is strongly connected and without strong cutpoint . observe that @xmath393 is a finite graph since edges correspond to homotopy classes of arcs in @xmath190 so @xmath394 is also finite . since the graph @xmath392 contains @xmath393 , for @xmath59 large enough @xmath395 contains the whitehead graph @xmath396 this implies @xmath397 contains @xmath398 for @xmath59 large enough . but then the whitehead graph @xmath397 is strongly connected and without strong cutpoint . notice , that @xmath399 is the same as @xmath400 this contradicts proposition [ separableelement ] . in this section , we show that when @xmath1 is not uniquely freely decomposable , if @xmath26 is discrete and faithful , then the condition that each end invariant of @xmath27 is disc - busting is also necessary . notice that if each end invariant is disc - busting , then , in particular , @xmath27 contains only one compressible boundary component . we will make use of the following characterization of which discrete and faithful representations are separable - stable . [ compact ] let @xmath26 be discrete and faithful . then , @xmath29 is separable - stable if and only if every separable curve is homotopic to a closed geodesic and the set of separable geodesics is contained in a compact set of @xmath338 if @xmath31 is a minimal lamination that is not a simple closed curve , then @xmath401 the _ supporting surface of _ @xmath190 is the unique minimal compact subsurface with geodesic boundary containing @xmath114 let @xmath402 be the unique , up to isotopy , maximum , simple , multi - curve of @xmath403 disjoint from @xmath114 the multi - curve @xmath402 contains @xmath404 . [ necessary ] let @xmath0 be a nontrivial compression body , without toroidal boundary components , that is not the boundary connect sum of two trivial @xmath16-bundles over closed surfaces . suppose @xmath405 is a discrete and faithful representation . if @xmath29 is separable - stable , then each end invariant of @xmath27 is disc - busting . the proof of the necessary condition follows almost exactly the proof in the case when @xmath0 is a handlebody from jeon - kim - ohshika - lecuire @xcite ( see section 5 ) . we include it here for completeness . by lemma [ compact ] , if a representation maps a separable element of @xmath24 to a parabolic element in @xmath406 then the representation is not separable - stable . since any disc - dodging , closed curve is separable , if there is a disc - dodging parabolic curve in @xmath27 , then @xmath29 is not separable - stable . so we may assume that each parabolic curve is disc - busting . similarly , if there exists an ending lamination @xmath31 that misses an essential disc @xmath54 , then @xmath31 is the limit of separable elements . in particular , there exists a sequence of separable elements exiting every compact set of @xmath338 by lemma [ compact ] , @xmath29 is not separable - stable . so , we may also assume that @xmath407 for any essential disc @xmath54 and any ending lamination @xmath31 of @xmath338 then , proposition [ necessary ] follows from the following two lemmas whose proofs we include below . we first explain how the necessary condition follows from the following two lemmas , and we include the proofs of the lemmas , afterwards . [ esan ] let @xmath0 have compressible boundary . for any essential annulus @xmath72 in @xmath0 , there is a meridian which is disjoint from @xmath72 . [ limitannuli ] let @xmath26 be discrete and faithful . suppose that @xmath31 is a disc - dodging ending lamination of @xmath27 . if some component @xmath22 of @xmath402 is disc - busting , then @xmath31 is the unique minimal component in the hausdorff limit of a sequence @xmath408 where @xmath409 is an essential annulus . indeed , by lemma [ limitannuli ] , if @xmath27 has an ending lamination that is disc - dodging , then there exists a sequence of simple closed curves @xmath410 , that are homotopic to core curves of essential annuli , such that @xmath411 exits every compact set . by lemma [ esan ] these curves are separable . hence , there exists a sequence of separable curves exiting every compact set . by lemma [ compact ] , the representation is not separable - stable . let @xmath72 be an essential annulus in @xmath12 since @xmath0 has compressible boundary there exists an essential disc @xmath60 we can isotope @xmath54 so that there are no inessential intersections between @xmath54 and @xmath72 . if @xmath412 , then the boundary of @xmath54 is a meridian disjoint from @xmath40 if not , consider an outermost arc @xmath149 in @xmath413 then , @xmath149 together with an arc in @xmath414 bounds a disc @xmath415 . if both of the endpoints of @xmath149 lie on the same component of @xmath416 , then together with an arc in @xmath417 the arc @xmath149 cuts off a disc @xmath418 from @xmath419 and @xmath420 is an essential disc which can be isotoped off @xmath72 . if @xmath149 connects two components of @xmath416 , then we can boundary - compress @xmath72 along @xmath415 , and to get an essential disc disjoint from @xmath40 the proof of lemma [ limitannuli ] makes use of a result of lecuire in @xcite that states the following . if @xmath0 is a compact hyperbolizable @xmath6-manifold , and there exists a lamination @xmath421 such that * each closed leaf has weight at most @xmath422 and * for any compressing disc @xmath54 , we have @xmath423 then , the following two conditions are equivalent . 1 . there exists @xmath168 such that @xmath424 for any essential annulus @xmath40 2 . if @xmath425 and @xmath426 are disjoint geodesic rays on @xmath427 such that two lifts @xmath428 and @xmath429 share the same endpoint in @xmath371 then they are asymptotic . the first step is finding a lamination containing @xmath31 that satisfies the first two conditions ( a ) and ( b ) . let @xmath430 be the union of @xmath31 and @xmath431 where we place a weight of @xmath422 on each closed curve . if there exists an essential annulus @xmath72 that is disjoint from @xmath432 then cut @xmath0 along @xmath72 and let @xmath433 be the component containing @xmath114 we claim that @xmath31 is still disc - dodging on @xmath434 indeed , suppose that @xmath435 is a sequence of meridians on @xmath0 such that @xmath436 then , as in the proof of lemma [ esan ] we can find a sequence of meridians @xmath437 disjoint from @xmath419 such that @xmath438 hence , @xmath31 is still disc dodging on @xmath434 let @xmath439 be the union of @xmath430 and the core curve or curves of @xmath419 depending on whether @xmath72 separates @xmath0 . put a weight of @xmath422 on each core curve that is added . notice that this new sequence of meridians also has the property that @xmath440 if there exists an essential annulus on @xmath433 disjoint from @xmath441 we repeat the procedure above . since there exists at most finitely many disjoint non - isotopic essential annuli on @xmath68 this procedure must terminate . let @xmath442 be the manifold obtained at the end of this procedure , and let @xmath443 be the corresponding lamination . then , any closed leaf has weight at most @xmath444 by construction . to see that @xmath445 for any essential disc @xmath446 it suffices to show that @xmath414 intersects an element in @xmath402 twice . indeed , this implies that @xmath447 and since @xmath448 we have that @xmath449 since there is a disc - busting component @xmath22 of @xmath431 we know that @xmath22 intersects @xmath54 at least once . if @xmath22 intersects @xmath54 exactly once , then a regular neighborhood of @xmath450 is a solid torus , @xmath451 we can form an essential disc missing @xmath22 by taking the closure of @xmath452 , a contradiction since @xmath22 is disc - busting . so @xmath443 satisfies conditions @xmath453 and @xmath454 above . secondly , we want to find two geodesic rays satisfying the hypothesis of condition @xmath455 above . namely , we want to find two disjoint , geodesic rays on @xmath456 that have lifts with a common endpoint in @xmath382 recall that there exists a sequence of meridians @xmath457 such that @xmath458 and @xmath459 up to subsequence , @xmath457 converges to a geodesic lamination @xmath460 in the hausdorff topology . by casson s criterion ( see casson - long @xcite or theoreme b1 in lecuire @xcite ) , @xmath106 contains a homoclinic leaf @xmath461 we want to show that @xmath79 has two half - leaves @xmath462 and @xmath463 that are disjoint from @xmath464 then , if @xmath22 is the disc - busting component in @xmath431 we have that @xmath462 and @xmath463 will lie on @xmath465 since @xmath466 is an incompressible subsurface , any lift @xmath467 has a well - defined endpoint in @xmath468 where @xmath469 is the subgroup of @xmath470 corresponding to the subsurface @xmath471 since @xmath79 is homoclinic , @xmath462 and @xmath463 must have the same endpoint . to see why two such half - leaves exist , first note that if @xmath79 intersects @xmath31 transversely , then it contains an arc @xmath149 such that @xmath472 this implies that for @xmath59 large enough , @xmath473 which is a contradiction since @xmath474 so @xmath79 does not intersect @xmath31 transversely . it follows that @xmath79 has two half - leaves @xmath475 that are disjoint from @xmath476 similarly , if @xmath79 intersects a leaf @xmath109 of @xmath443 outside of @xmath401 then @xmath477 for @xmath59 large enough , a contradiction since @xmath459 so @xmath79 has two half - leaves @xmath475 that are disjoint from @xmath464 next , we want to see that @xmath443 does not satisfy condition ( c ) above . indeed we want to show that if we take lifts @xmath478 and @xmath479 with the same endpoint , then they are not asymptotic . suppose they are asymptotic and take a sequence of geodesic arcs @xmath480 joining @xmath478 and @xmath479 such that the length of @xmath480 goes to zero . let @xmath481 be the projection of @xmath480 to @xmath482 . then @xmath483 since @xmath31 is minimal and not a simple closed curve . since @xmath478 and @xmath479 are asymptotic and do not transversely intersect @xmath31 we can assume that @xmath481 are homotopic relative to their endpoints and that the homotopy is through arcs transverse to @xmath114 then , we must have @xmath484 for all @xmath485 . in particular , the arc @xmath486 is disjoint from @xmath114 then , adjusting @xmath487 so that it begins at @xmath488 , if we take @xmath489 this projects to a nontrivial curve @xmath34 on @xmath482 disjoint from @xmath31 that is trivial in @xmath442 . it follows from the proof of dehn s lemma that there is a compressing disc disjoint from @xmath190 a contradiction . since condition ( c ) is not true , we have that condition ( c ) is also not true . in particular , there exists a sequence @xmath409 of essential annuli such that @xmath490 since each component of @xmath491 is a closed curve with weight @xmath422 , we have that @xmath492 for @xmath59 large enough . up to subsequence , @xmath493 converges , in the hausdorff topology , to a geodesic lamination @xmath494 then , @xmath495 and @xmath31 can not intersect transversely . since @xmath496 we can not have that @xmath495 and @xmath31 are disjoint . since @xmath492 for @xmath59 large enough , we have that @xmath497 in particular , @xmath495 lies on @xmath498 since @xmath31 is filling on @xmath499 , we have that @xmath31 and @xmath495 coincide as geodesic laminations . since @xmath493 is disjoint from all the essential annuli used to construct @xmath442 , we can view @xmath495 as a lamination on @xmath35 and we are done . this section deals with the missing case when @xmath1 is uniquely freely decomposable , i.e. when @xmath0 is the boundary connect sum of @xmath17 and @xmath18 where @xmath343 is a closed surfaces of genus at least two for @xmath500 here an element of @xmath1 is separable if and only if it misses an essential annulus in one of the two surface group factors . by lemma [ esan ] for any essential annulus @xmath419 there exists an essential disc disjoint from @xmath40 since there is only one essential disc , @xmath446 up to isotopy , in fact , @xmath72 lies in one of the two surface group factors . if @xmath72 is contained in @xmath501 then @xmath72 is either an essential annulus in @xmath502 or homotopic ( relative to its boundary ) into the boundary of @xmath503 if the latter is true , since @xmath72 is essential in @xmath68 if @xmath504 and @xmath72 lies in @xmath505 then @xmath416 bounds an annulus in @xmath506 containing @xmath60 in either case , if a curve or lamination misses @xmath419 it misses an essential annulus in @xmath17 of @xmath507 we will consider annulus - busting laminations on @xmath35 . recall that a measured lamination @xmath31 is annulus - busting if there exists an @xmath168 such that @xmath508 for any essential annulus @xmath40 let @xmath0 be the boundary connect sum of @xmath17 and @xmath18 where @xmath19 and @xmath20 are closed surfaces of genus at least two . let @xmath26 be discrete and faithful . then , @xmath29 is separable - stable if and only if each end invariant of @xmath27 is annulus - busting . we will start with the sufficient condition , and we will follow the general outline of proposition [ sufficient ] . in this case , we have the following simplifications . firstly , @xmath196 has only one element , i.e. any manifold homotopy equivalent to @xmath0 is homeomorphic to @xmath0 . secondly , we have only one meridian @xmath95 . if @xmath31 is annulus - busting , then it must intersect @xmath193 and so it is automatically in tight position with respect to @xmath150 now suppose that @xmath26 is a discrete and faithful representation such that each end invariant and parabolic locus of @xmath27 is annulus - busting . if @xmath29 is not separable - stable , by lemma [ notss ] , there exists a sequence of separable elements @xmath365 with endpoints @xmath509 such that @xmath510 and @xmath511 then , by theorem [ mj2 ] , the image @xmath512 is an endpoint of one of the end invariants @xmath31 . here @xmath31 can either be an ending lamination or a simple closed curve that is mapped to a parabolic element under @xmath513 as before , this implies that for @xmath59 large enough @xmath514 contains @xmath515 if @xmath31 is annulus - busting , then @xmath516 intersects each essential annulus in the following sense . suppose that @xmath72 is an annulus in @xmath503 let @xmath517 . in defining the whitehead graph , we fixed lifts @xmath518 of @xmath343 . if we take a lift @xmath519 of @xmath520 and the lift @xmath521 of @xmath522 with the same endpoints as @xmath519 , then @xmath523 forms a loop in @xmath134 . we will say that an edge @xmath524 intersects @xmath72 if there exists lifts @xmath519 and @xmath521 in @xmath518 as above such that @xmath136 and @xmath525 lie in different components of @xmath526 . since @xmath514 contains @xmath516 for @xmath59 large enough , we have that @xmath514 `` intersects '' any essential annulus . this implies that the geodesic representative @xmath410 of @xmath365 intersects any essential annulus , a contradiction . for the necessary condition , suppose that @xmath527 is a discrete , faithful and separable - stable representation and some end invariant of @xmath27 is annulus - dodging . if a parabolic curve @xmath22 is annulus - dodging , then @xmath22 is separable , a contradiction to lemma [ compact ] . so suppose that all parabolic curves are annulus - busting , but there is an ending lamination @xmath31 that is annulus - dodging . if @xmath31 is disjoint from an essential annulus @xmath419 then we can find a sequence of separable elements approaching @xmath190 a contradiction to lemma [ compact ] . so we can assume that @xmath528 for any essential annulus . since @xmath31 is annulus - dodging , there exists a sequence @xmath409 of essential annuli such that @xmath529 since each essential annulus misses the essential disc , @xmath446 up to a subsequence , we can assume that @xmath409 is contained in one of the two surface group factors for all @xmath530 without loss of generality , assume that @xmath409 is contained in @xmath531 for all @xmath530 let @xmath351 be the limit , up to subsequence , of @xmath493 in @xmath532 the space of projective measured laminations . since @xmath493 is disjoint from the meridian for all @xmath59 , so is @xmath351 . in particular , @xmath351 lies on @xmath533 since @xmath534 either @xmath31 and @xmath351 are disjoint or @xmath535 since @xmath31 is minimal . if @xmath31 and @xmath351 are disjoint , then we can find a simple closed curve @xmath22 on @xmath19 that is disjoint from @xmath114 then , @xmath22 is the boundary of an essential annulus @xmath536 that misses @xmath190 a contradiction . if @xmath537 then @xmath31 lies on @xmath19 and misses any essential annulus in @xmath18 a contradiction we thank the organizers of the workshop on ergodic decompositions of representation varieties " held in princeton in october 2011 , where the layout of the paper was conceived . the second author thanks dick canary for several useful conversations . 99 i. agol , _ tameness of hyperbolic 3-manifolds _ , preprint available at arxiv : math\0405568v1 f. bonahon , _ cobordism of automorphisms of surfaces _ , ann . sci . cole norm . ( 4 ) * 16 * ( 1983 ) , no . 2 , 237270 . d. calegari and d. gabai , _ shrinkwrapping and the taming of hyperbolic 3-manifolds _ , j. amer . * 19 * ( 2006 ) , no . 2 , 385446 . r. d. canary , _ ends of hyperbolic @xmath6-manifolds _ , j. amer . math . * 6 * ( 1993 ) , no . 1 , 135 . r.d . canary , dynamics on character varieties : a survey , 2013 , preprint available at arxiv:1306.5832 [ math.gt ] . canary and p. storm , moduli spaces of hyperbolic 3-manifolds and dynamics on character varieties , to appear in commentarii mathematici helvetici . a. casson and d. long , _ algorithmic compression of surface automorphisms _ , invent . math . 81 ( 1985 ) , no . 2 , 295303 . t. gelander , _ on deformations of fn in compact lie groups _ , israel j. math . 167 ( 2008 ) , 1526 . w. goldman , _ an ergodic action of the outer automorphism group of a free group _ , gafa , 17 ( 2007 ) , 793805 . i. a. grushko , on the bases of a free product of groups . matematicheskii sbornik , * 8 * ( 1940 ) , 169182 . j. hempel , _ @xmath6-manifolds _ , princeton univ . press , princeton , nj , 1976 . w. jeon , i. kim and k. ohshika with c. lecuire , _ primitive stable representations of free kleinian groups _ , to appear in israel jour math . m. kapovich , _ hyperbolic manifolds and discrete groups _ , progress in mathematics , 183 , birkhuser boston , boston , ma , 2001 . i. kim , c. lecuire and k. ohshika , _ convergence of freely decomposable kleinian groups _ , submitted . g. kleineidam and j. souto , _ algebraic convergence of function groups _ , comment.math . 77 ( 2002 ) , 244269 . a. kurosch , die untergruppen der freien produkte von beliebigen gruppen , math . * 109 * ( 1934 ) , no . 1 , 647660 . c. lecuire , _ plissage des varit hyperboliques de dimension 3 _ , invent . 164 ( 2006 ) , no . 1 , 85141 . c. lecuire , _ an extension of the masur domain . spaces of kleinian groups _ , london math . lecture note ser . 329 , ( 2006 ) , 4973 . m. lee , dynamics on @xmath2-character varieties of certain hyperbolic @xmath6-manifolds , ph.d . thesis , university of michigan , 2012 , available at http://hdl.handle.net/2027.42/94050 m. lee , _ dynamics on the @xmath2-character variety of a compression body _ , to appear in algebraic and geometric topology . a. marden , the geometry of finitely generated kleinian groups , ann . of math . ( 2 ) * 99 * ( 1974 ) , 383462 . d. mccullough , _ compact submanifolds of @xmath6-manifolds with boundary _ , quart . oxford ser . ( 2 ) * 37 * ( 1986 ) , no . 147 , 299307 . d. mccullough and a. miller , _ homeomorphisms of @xmath6-manifolds with compressible boundary _ , mem . * 61 * ( 1986 ) , no . 344 d. mccullough , a. miller and g. a. swarup , _ uniqueness of cores of noncompact @xmath6-manifolds _ , j. london math . soc . ( 2 ) * 32 * ( 1985 ) , y. n. minsky , on dynamics of @xmath538 on @xmath2 characters , israel j. math . * 193 * ( 2013 ) , no . 1 , 4770 . m. mj , _ cannon - thurston maps for kleinian groups _ , preprint 2010 available at arxiv:1002.0996 [ math.gt ] . j. p. otal , _ courants godsiques et produits libres _ , thse detat , universit de paris - sud , orsay ( 1988 ) . f. palesi , ergodic actions of mapping class groups on moduli spaces of representations of non - orientable surfaces , geom . dedicata * 151 * ( 2011 ) , 107140 . j. r. stallings , whitehead graphs on handlebodies , in _ geometric group theory down under ( canberra , 1996 ) _ , 317330 , de gruyter , berlin . d. sullivan , _ on the ergodic theory at infinity of an arbitrary discrete group of hyperbolic motions _ , riemann surfaces and related topics : proceedings of the 1978 stony brook conference , ann . of math . 97 , princeton , 1981 . d. pickrell and e. z. xia , _ ergodicity of mapping class group actions on representation varieties . i. closed surfaces , _ comment . * 77 * ( 2002 ) , no . 2 , 339362 . j. h. c. whitehead , _ on certain sets of elements in a free group _ , london math . 41 ( 1936 ) , 4856 . 782 - 800 .

let @xmath0 be a hyperbolizable , nontrivial compression body without toroidal boundary components . in this paper , we characterize which discrete and faithful representations of @xmath1 into @xmath2 are separable - stable . the set of separable - stable representations forms a domain of discontinuity for the action of @xmath3 on the @xmath2-character variety of @xmath4

psr b1706 - 44 is a young ( @xmath6 yr ) , energetic 102 ms pulsar originally discovered by @xcite . it is one of several sources with spin down ages 10@xmath710@xmath8 yrs which are referred to as vela - like pulsars , due to their similar emission properties ( @xcite ) . the source is known to display glitches ( @xcite ) and is plausibly associated with the supernova remnant g343.1 - 2.3 ( @xcite ; @xcite ; @xcite ) . according to the new @xcite model , the pulsar is @xmath9 kpc away , which agrees with the kinematic distance in the range 2.43.2 kpc inferred from h i absorption ( @xcite ) . vla images ( @xcite , @xcite ) indicated that the pulsar is located inside a synchrotron plerionic nebula about @xmath10 in size . evidence for a more extended x - ray compact nebula ( with radius @xmath11 ) was also found in _ rosat_-hri and _ chandra _ images ( @xcite , @xcite ) . psr b1706 - 44 is one of only eight radio pulsars which are known to emit in the gev range ( @xcite ) , and one of only three detected in the tev range , although this detection is marginal ( @xcite , @xcite ) . @xcite found pulsations at the radio period in the _ egret_-cgro data . the light curve above 400 mev is complex , with evidence of two , possibly three peaks , none of which are in phase with the radio peak . we note that the light curve of vela in the gamma - ray range shows two peaks ( @xcite ) , and at least three in the x - rays ( @xcite ) . particular characteristics of the vela - like pulsars , with respect to the crab - like pulsars , are that the pulse profiles at different energies are shifted in phase with respect to each other . an unpulsed x - ray source at the radio - pulsar position had been detected with _ rosat_-pspc ( @xcite ) , asca ( @xcite ) and bepposax ( @xcite ) . more recently , deeper _ chandra _ observations have been presented by gotthelf , halpern & dodson ( 2002 ) . these authors discovered a broad sinusoidal x - ray pulsation at the radio - period , with a pulsed fraction of @xmath12 . the phasing of the radio pulse was consistent with that of the center of the broad x - ray peak . the high spectral and spatial resolution of _ chandra _ allowed a multi - component fit of the x - ray spectrum revealing the presence of a thermal component : the x - ray spectrum was found to be well - fit with a two component model , with a blackbody temperature of @xmath13 k and a power law index of @xmath14 . the blackbody radius @xmath15 km determined from the model parameters , suggests that the emission is from a hot spot , or that a fit with an atmospheric model is required ( @xcite ) . hydrogen atmospheres are in general harder than a blackbody since free - free absorption dominates in the x - ray band . therefore , the temperature fitted by a hydrogen atmosphere is lower than that resulting from a blackbody fit ( hence the former fit yields a larger radius ) . in this paper we report on _ xmm - newton _ observations of psr b1706 - 44 . results from the spectral and timing analyses are presented in the following sections . psr b1706 - 44 was observed with _ xmm - newton _ for @xmath16 ks on 2002 march 12 , and @xmath17 ks on 2002 march 13 , as a part of the guaranteed time program . the following analysis uses data from the three european photon imaging camera ( epic ) instruments : two epic mos detectors ( @xcite ) and the epic pn detector ( @xcite ) . the thin optical blocking filter was used on the pn . to minimize pile - up the pn was operated in _ small window _ mode , which gives a temporal resolution of 6 ms . the mos1 was operated in _ full window _ ( imaging ) mode with a time resolution of 1.4 s. in order to obtain better temporal resolution the mos2 was operated in _ timing mode _ ; in this mode data from the central ccd are collapsed into a one - dimensional row to achieve a 1.5 ms time resolution . the medium filter was used for both mos observations . observations of psr b1706 - 44 were taken with the resolution grating spectrometer ( rgs ) , however we did not detect enough photons for a meaningful analysis . due to an optically bright source in the field the blocking filter was used on the optical monitor ( om ) . we reduced the epic data with the _ xmm - newton _ science analysis system ( sas v5.4.1 ) . to maximize the signal - to - noise we filtered the data to include only single , double , triple and quadruple photon events for mos , and only single and double photon events for pn . we also filtered the events files to exclude times of high background . psr b1706 - 44 is known to be surrounded by a pulsar wind nebula ( @xcite ) . in an ideal situation analysis of the source should be carried out on data which are not contaminated by the diffuse emission . however , the nebula is very compact , with the diffuse emission visible between @xmath18 and @xmath19 ( see @xcite ) . therefore the spatial resolution of _ xmm - newton _ is not high enough to be able to separate the pulsar from its putative synchrotron nebula . in order to quantify this effect , we have compared psr b1706 - 44 s emission with that for a point source . we simulated the expected pn data for a point source with similar column depth and spectral characteristics as psr b1706 - 44 using _ scisim_. the radial profiles for psr b1706 - 44 from 2002 march 12 and 13 , and the simulated source , are shown in fig . [ fig : radial ] . the background has been subtracted from both the psr b1706 - 44 and point source data . we have normalized the total point source counts to the total counts from psr b1706 - 44 by performing a @xmath20 minimization to determine the appropriate scaling factor required to fit only the core of the pulsar ( @xmath21 ) . our results show marginal evidence for an excess of diffuse emission between @xmath22 and @xmath23 . an extraction radius of @xmath21 does not give a sufficient number of photons for analysis purposes , hence , we have chosen an extraction radius of @xmath23 . we note that this radius will therefore include emission from the pulsar wind nebula . as only @xmath24 % of the energy from the source is encircled in this radius the measured fluxes have been corrected in the following analyses . spectra for psr b1706 - 44 were extracted from both the pn and mos1 data , and regrouped by requiring at least 30 counts per spectral bin . we subtracted a background which was extracted from an annulus around the source fiducial region . the corresponding photon redistribution matrix ( rmf ) and ancillary region file ( arf ) were created . we simultaneously fit the four spectra , two from the pn and two from mos1 , with thermal and power - law models , modified by photoelectric absorption . results are shown in fig . [ fig : spec ] and the best - fit parameters are summarized in table [ spec ] . the column density was first fixed at the value given in @xcite , @xmath25 @xmath26 , for the single blackbody , single power - law and blackbody plus power - law fits . as this value was not very well constrained in @xcite , we also allowed @xmath27 to vary for the above model fits . we find that the data are poorly fit with a single blackbody for free and fixed @xmath27 , and with a single power - law with fixed @xmath27 . the single power - law model with free @xmath27 fits the data reasonably well , however the two - component blackbody plus power - law model with either free or fixed @xmath27 gives a better fit . we find the two - component fit where the column density is allowed to vary results in @xmath28 @xmath26 , with a temperature @xmath29 k , power - law index @xmath30 , and @xmath31 for 658 d.o.f . the x - ray luminosity is @xmath32 ergs s@xmath33 in the @xmath34 kev band , and the blackbody contribution is @xmath35 ergs s@xmath33 . the distance to psr b1706 - 44 is uncertain , with values in the range @xmath36 kpc ( @xcite ) . by adopting the value determined from the @xcite free - electron model of the galaxy , @xmath37 kpc , we obtain an emitting radius @xmath38 km , too small to be compatible with a neutron star equation of state . the results for the column density , temperature and power - law index are in good agreement with those obtained by @xcite based on _ chandra _ data . while the radius we determine is smaller than that from the fits to the _ chandra _ data , @xcite also found that the radius of the emitting region inferred from the blackbody fit of the thermal component indicates a hot region on a cooler surface . in order to try a more physical description of the thermal component , we fitted the data with a neutron star hydrogen atmosphere plus power - law model . the magnetic field of psr b1706 - 44 , as inferred from the spin - down rate , is @xmath39 g , high enough to have a substantial effect on the opacities ( @xcite ) . we used a grid of pure - h , atmospheric cooling models computed for @xmath40 g and different effective temperatures , provided by v. zavlin ( v.e . zavlin 2003 , private communication ) . we first fixed the neutron star mass and radius at @xmath41 , @xmath42 km ; all other parameters were allowed to vary . the resulting best - fit parameters are @xmath43 , @xmath44 @xmath26 and @xmath45 k , where @xmath46 is the effective temperature evaluated at the star surface ( see table [ spec ] ) . this gives a temperature at infinity @xmath47 k. the source distance resulting from this fit is @xmath48 kpc , lower than that computed by @xcite . the reliability of the pulsar distance derived using the dispersion measure method is uncertain , however , the distance issue can only be firmly addressed by parallax measurements , which are not available for psr b1706 - 44 . we therefore fit the data assuming a distance to the source of @xmath49 kpc . by fixing @xmath50 km and repeating the atmospheric plus power - law model fit we find @xmath51 , @xmath52 @xmath26 and @xmath53 k [ @xmath54 k ] . the distance to the pulsar from this fit is @xmath55 kpc , in agreement with the value obtained from the dispersion measure . due to the goodness of fit for both the blackbody plus power - law , and the magnetic atmosphere plus power - law models we are unable to distinguish between these two scenarios on only a statistical basis . taken at face value , these results indicate that the emission is either from the whole neutron star surface with radius @xmath56 km , or that the thermal x - rays originate from a smaller region , i.e. a hot spot . for the timing analysis the event files were filtered on energy , to include only photons in the range @xmath34 kev . the filtered event files were then barycentrically corrected . we extracted data for the pn from a circular region centered on the source of radius @xmath23 . selection of the photons from the mos2 event files was achieved by extracting 90% of the flux within a rectangular region centered on the source . the fluxes have been corrected for this in the following temporal analysis . to increase the signal - to - noise we have combined the pn data from 2002 march 12 and 13 . we also combined the mos2 datasets . we determine a predicted pulse period at the epoch of our _ xmm - newton_observations for psr b1706 - 44 , assuming a linear spin - down rate and using the radio measurements ( @xcite ) . we find @xmath57 ms ( @xmath58 hz ) at the start of our first observation ( mjd 52345.0 ) , and @xmath59 ms ( @xmath60 hz ) at the end of our second observation ( mjd 52347.0 ) . as the frequency of the predicted period varies over the duration of our observations , and glitches and/or deviations from a linear spin - down may alter the period , we searched for a pulsed signal in the pn and mos2 data over a wider frequency range centered on @xmath61 hz . we have employed two methods in our search for pulsed emission from psr b1706 - 44 . in both methods we have included the frequency derivative from @xcite in our calculations to determine the best - fit frequency . in the first method we implement the @xmath62 test ( @xcite ) , with the number of harmonics @xmath63 being varied from 1 to 5 . in the second method we calculate the rayleigh statistic ( see @xcite ) and then calculate the maximum likelihood periodogram ( mlp ) using the @xmath64-statistic ( @xcite ) to determine significant periodicities in the datasets ( see @xcite ) . the most significant @xmath62-statistic for the combined pn data occurs for @xmath65 . with the number of harmonics equal to one , the @xmath62-statistic corresponds to the well known rayleigh statistic . we find a peak at @xmath66 hz ( see fig . [ fig : psearch ] , _ top left panel _ ) . the quoted 90% uncertainty range is determined from the error in the position of the peak . within errors , the peak in the @xmath67 periodogram lies in the range of predicted frequencies . the @xmath67-statistic for this peak is 41.82 , which has a probability of chance occurrence of @xmath68 . the corresponding peak in the mlp periodogram occurs at @xmath69 hz ( fig . [ fig : psearch ] , _ bottom left panel _ ) . within the 90% confidence limit , this value is consistent with the predicted radio frequency , and the frequency found from the @xmath67 periodogram . we generated @xmath62 and mlp periodograms for the mos2 data as for the pn data . there is a peak at @xmath70 hz in the @xmath67 and mlp periodograms , however it is not the most significant peak . there are several peaks present in both periodograms , the significance of which are low compared to the noise level . the individual pn datasets from 2002 march 12 and 13 have @xmath71 and @xmath72 , respectively . the mos2 data from the same dates both have @xmath73 . the low signal - to - noise ratio of the mos2 data is the cause for the lack of a significant peak at the predicted frequency . as our _ xmm - newton _ observations were taken more than a year outside the valid range of the radio ephemeris given in @xcite we are unable to determine a phase relationship between the x - ray and radio pulse profiles . hence , we have used an arbitrary ephemeris @xmath74 mjd @xmath75 . we folded the pn and mos2 data on the frequency found from the @xmath67 periodogram , @xmath70 hz . the effects of the drifting period are taken into account by including the frequency derivative in the calculations . in fig . [ fig : prof ] we show the combined pn , mos2 and pn plus mos2 pulse profiles . we modeled the folded light curves with one and two sinusoids ; the resulting @xmath76 values are given in table [ sinfit ] . as it can be seen , the fits to the data are improved with the double sinusoidal model . to determine the statistical significance of the second sinusoid we employed an f - test . we find f - test probabilities of @xmath77 , @xmath78 and @xmath79 , for the pn , mos2 and pn plus mos2 fits respectively . these values indicate that the pn and pn plus mos2 data are best - fit with the two sinusoid model , while no firm conclusion can be derived from the low s / n mos2 data . the pulsed fractions of the two peaks are @xmath4 and @xmath5 . in an attempt to investigate if the two peaks in the folded light curve are thermal or non - thermal in origin , we determined from the two - component model fits to the spectra the energy at which the power - law starts to dominate over the thermal emission . this occurs at @xmath80 kev , and @xmath81 kev in the blackbody plus power - law and magnetic atmosphere plus power - law models , respectively . by filtering the pn and mos2 event files on energy we produced light curves in the following energy ranges @xmath82 and @xmath83 kev , which we barycentrically corrected . we folded the data on @xmath70 hz ( fig . [ fig : diffe ] ) , again including the frequency derivative in the calculations . as we can see , the same features are present in both bands ; the thermal and non - thermal emission are both pulsating and phase aligned . however , at higher energies the modulation is lower , indicating that the power law component is contaminated by the non - pulsating nebular emission . of the @xmath84 radio pulsars detected so far , only @xmath85 have also been found in the x - rays . these x - ray emitting pulsars represent a wide range of ages ( @xmath86 yrs ) , magnetic field strengths ( @xmath87 g ) , periods ( @xmath88 ms ) and spectral properties . in particular , only a sub - set of them are suitable for observing the thermal radiation from the neutron star surface , and therefore constrain the atmospheric chemical composition and the pulsar cooling history . thermal emission from the neutron star surface is not detectable in pulsars older than @xmath89 yr : standard cooling scenarios predict a sharp reduction in the surface temperature when surface photon emission overtakes the neutrino luminosity losses ( @xcite ) . in pulsars younger than @xmath90 yr strong non - thermal emission from the magnetosphere or the synchrotron nebula swamps the weaker thermal radiation and dominates the x - ray spectrum . only in middle - aged pulsars ( ages @xmath91 yr ) , is the non - thermal component much fainter ; hence for such objects the thermal radiation from the neutron star surface ( @xmath92 kev ) can dominate at soft x - ray / uv energies . to date , thermal emission has been detected in only very few radio pulsars , e.g. psr b0656 + 14 ( @xcite ) , psr b1055 - 52 ( @xcite ) , psr j0437 - 4715 ( @xcite ) , psr j0538 + 2817 ( @xcite ) , geminga ( @xcite ) , vela ( @xcite ) , and psr b1706 - 44 ( @xcite ) . the thermal emission detected above @xmath93 kev in the spectrum of the first four objects has been more plausibly interpreted as originating from a hot - polar cap . this agrees with the fact that these older sources should have a surface flux peaked in the uv and it is confirmed by a detection , in the brightest psr b0656 + 14 , of a further thermal component below @xmath94 kev ( @xcite ) . the younger vela pulsar ( @xmath95 yr ) is the only radio active source for which the thermal component observed in the soft x - rays is well explained by a magnetized cooling atmosphere ( @xcite ) . when this model is assumed instead of a blackbody , the inferred radius increases from @xmath96 km to @xmath97 km ( where @xmath98 is the distance in units of 294 pc ) and is therefore in agreement with a standard neutron star equation of state the only other neutron stars whose thermal component is better described by an atmospheric model , and for which this interpretation resolves all the inconsistencies which follow from the blackbody interpretation , are the radio - silent neutron stars 1e 1207 - 52 ( @xcite ) and rx j0822 - 4300 ( @xcite ) . here we present a further example . we have found that both a blackbody plus power - law and a magnetized atmospheric model plus power - law provide an excellent fit to the data , and are indistinguishable on a statistical basis . however , the latter has to be preferred on physical grounds , as we argue below , and , at a source distance of @xmath99 kpc , gives an emitting area of @xmath100 km , compatible with the size of a neutron star ( @xcite ) . our knowledge of neutron star interiors is still uncertain and accurate measurements of the neutron star surface temperature are particularly important to constrain the cooling models and provide information on the physics of the neutron star . roughly speaking , theoretical models predict a two - fold behavior of the cooling curves . in low - mass neutron stars neutrino emission is mainly due to a modified urca process and nucleon - nucleon bremsstrahlung . these are relatively weak mechanisms and produce _ slow cooling_. in stars of higher mass the neutrino emission is enhanced by a direct urca process ( or other mechanisms in exotic matter ) , therefore these stars cool down much faster ( _ fast cooling _ regime ) . to date ( see @xcite for a recent a discussion ) it has been realized that simple models which do not account for proton and neutron superfluidity fail in explaining the surface temperatures observed in many sources , unless objects such as e.g. vela , geminga , rx j1856 - 3754 do have exactly the critical mass that bounds the transition between the very different _ slow cooling _ and _ fast cooling _ regimes . this unlikely assumption can be avoided by including the effects of nucleon superfluidity . models with proton superfluidity included predict an intermediate region between fast cooling and slow cooling curves , which is expected to be populated by medium mass neutron stars ( roughly with @xmath101 between 1.4 and 1.65 @xmath102 ) . although the full picture only holds if , at the same time , neutron superfluidity is assumed to be rather weak , it is still interesting that many neutron stars ( as 1e 1207 - 52 , vela , rx j1856 - 3754 , psr 0656 + 14 ) have a surface temperature which falls in such a transition region . since the cooling curves in the transition region show a significant spread with the neutron star mass , if this scenario is correct we can select those curves which explain the observations and therefore attribute certain masses to the sources ( `` weighing '' neutron stars , @xcite ) . as we can see from the first two panels of figure 2 in @xcite , assuming an age of @xmath103 , the surface temperature of psr b1706 - 44 derived from the blackbody fit is even higher than the upper cooling curves i.e. those corresponding to the slow cooling regime . however , the surface temperature @xmath104 obtained by fitting with the magnetized model and @xmath105 km falls well within the above mentioned transition region of medium mass neutron stars . the mass of psr b1706 - 44 should then be @xmath106 or @xmath107 , depending on the kind of proton superfluidity assumed in the model ( 1p and 2p respectively ) . we note that interpreting the temperatures obtained from the spectral fits in the context of theoretical cooling curves relies on the true age of the pulsar being the same as the characteristic spin - down age , which may not be valid . the first detection of pulsations in the soft x - rays of psr b1706 - 44 by _ chandra _ was a broad modulation in the light curve , approximately in phase with the radio peak ( @xcite ) . its is known that the radio and gamma - ray pulses are not aligned , with the gamma - ray emission occurring @xmath108 cycles after the radio emission ( @xcite ) . the _ xmm - newton _ data allows us to resolve the substructure of the soft x - ray light curve , revealing the presence of two separate peaks . however , due to the lack of a contemporaneous radio observation of this glitching pulsar , we are unable to determine an absolute phase relationship between the x - ray , radio and gamma - ray pulse profiles . to investigate the origin of the pulsed emission in psr b1706 - 44 it is instructive to consider the timing behavior of vela , since both sources have similar spectral properties . the complex multiwavelength light curves of vela are the subject of continued interest and some evidence of correlations between the different energy bands have recently been established ( @xcite ) . unfortunately , in the case of vela , the key to deciphering pulsar emission mechanisms was the light curve in the @xmath109 kev _ rxte _ energy band ( obtained with a 92 ks and a @xmath110 ks pointing , @xcite ) , while a 132 ks _ rxte _ observation of psr b1706 - 44 only gave an upper limit on the pulsed emission in the @xmath111 kev band ( @xcite ) . the pulse profile of vela obtained with _ egret _ shows two distinct narrow pulses , at phase @xmath112 and @xmath113 , neither aligned with respect to the radio peak . considerable emission has been detected in the phase interval between the peaks ( @xcite ) . by contrast , in the _ egret _ band psr b1706 - 44 shows a broader modulation that extends over the full range of phases between @xmath114 , and is likely to consist of two broad peaks separated by @xmath115 ( @xcite ) . no gamma - ray emission is detected outside of the peaks and , as in vela , the _ egret _ pulses are out of phase with respect to the radio pulse . the difference in the gamma - ray light curves of the two sources is likely to be indicative of a different beaming geometry , with psr b1706 - 44 having greater alignment between the rotational and magnetic axes . this is also confirmed by radio - polarization studies ( @xcite ) . an alternative , but indistinguishable scenario is in terms of different viewing angles , in which case psr b1706 - 44 should be viewed more equator - on than vela . the soft x - ray profile of vela obtained with _ _ shows at least three peaks ( @xcite ) , at phases @xmath116 . the presence of several different pulses may be explained in terms of non - thermal radiation originating from more than one region in the pulsar magnetosphere ( the soft x - ray pulse profile has a total pulsed fraction of @xmath117 % ) . however , only the first two peaks have a plausible association with _ egret _ ( and _ rxte _ , see @xcite ) peaks at similar phases . although this association is more conclusive for the first peak , if it is the case , this suggests that the emission is probably caused by non - thermal radiation . on the other hand , radiation emergent from a stellar atmosphere is inherently anisotropic regardless of the local temperature and magnetic field , and , when combined with a non - uniform temperature profile at the neutron star surface , is capable of generating considerable pulsar modulation . also , pulsations are expected if the thermal flux originates from hot polar caps . in this respect , the fact that the third soft x - ray peak does not have a counterpart in the gamma - rays led to speculation that it may be thermal in origin ( @xcite ) . in these _ xmm - newton _ observations of psr b1706 - 44 we have found evidence for two peaks in the soft x - rays . whether there are similarities between these peaks and those in the soft x - ray light curve of vela , and the underlying emitting scenario , is unclear . we find that for psr b1706 - 44 the overall x - ray pulsation is much more prominent at softer energies , below @xmath118 kev , but unfortunately a comparison between the light curves in different energy bands does not allow us to establish unequivocally the thermal or non - thermal character of the two peaks ( see fig . [ fig : diffe ] ) . in fact , due to the modest spatial resolution of _ xmm - newton _ , the power - law component detected in the x - ray spectrum is probably highly contaminated by the non - pulsating nebula , which explains the overall decrease of the pulsed fraction at higher energies . however , we do observe the modulation in both bands and there is not a significant energy - dependent change in the phasing . as mentioned above , it has been only recently that deep multiwavelength observations of vela have successfully proven that the source has nearly phase - aligned pulse profiles and a spectral continuity from x - rays to gamma - rays ( @xcite ) . this is of great importance , since the connection between the pulse profiles in different bands was not proven before in vela - like pulsars , even at a phenomenological level . psr b1706 - 44 is considerably weaker than vela and it lies in the galactic center region where the diffuse gamma and x - ray radiation is more intense . however , our results indicate that the soft x - ray pulse profile of psr b1706 - 44 consists of two peaks , which may be similar to the vela light curve in the soft x - ray energies . to determine if psr b1706 - 44 is in fact `` vela - like '' not only spectrally , but also temporally , future spectrally resolved x - ray observations with high throughput in the @xmath109 kev band , with contemporaneous radio observations , are required . we thank the anonymous referee for helpful comments . we are grateful to v.e . zavlin for providing a grid of magnetized cooling atmosphere models . the authors thank gavin ramsey and sergey trudolyubov for discussions on timing issues . this work is based on observations obtained with _ xmm - newton _ , an esa science mission with instruments and contributions directly funded by esa member states and nasa . the authors acknowledge support from the institute of geophysics and planetary physics ( igpp ) program at lanl and nasa grants s-13776-g and nag5 - 7714 . becker , w. , brazier , k.t.s . , & trmper , j. , 1995 , , 298 , 528 becker , w. , & pavlov , g. , 2002 , ( astro - ph/0208356 ) bock , d.c .- j . , & gvaramadze , v.v . , 2002 , , 394 , 533 buccheri , r. , bennett , k. , bignami , g.f . , et al . , 1983 , , 128 , 245 cash , w. , 1979 , , 228 , 939 chadwick , p.m. , dickinson , m.r . , dipper , n.a . , et al . , 1998 , aph , 9 , 131 cordes , j.m . , & lazio , t.j.w . , 2002 , ( astro - ph/0207156 ) de jager , o.c . , 1991 , , 378 , 286 dodson , r. , & golap , k. , 2002 , , 334 , 1 finley , j.p . , srinivasan , r. , yoshitaka , s. , et al . , 1998 , , 493 , 884 frail , d.a . , goss , w.m . , & whiteoak , j.b.z . , 1994 , , 437 , 781 giacani , e.b . , frail , d.a . , goss , w.m . , & vieytes , m. , 2001 , , 121 , 3133 gotthelf , e.v . , halpern , j.p . , & dodson , r. , 2002 , , 567 , 125 guojun , q. , manchester , r.n . , lyne , a.g . , et al . , 1995 , , 274 , 572 halpern , j.p . , & wang , f.y - h . , 1997 , , 477 , 905 harding , a.k . , strickman , m.s . , gwinn , c. , et al . , 2002 , , 576 , 376 johnston , s. , lyne , a.g . , manchester , r.n . , et al . , 1992 , , 255 , 401 johnston , s. , manchester , r.n . , lyne , a.g . , et al . , 1995 , , 293 , 795 kaminker , a.d . , haensel , p. , & yakovlev , d.g . , 2001 , , 373 , l17 kanbach , g. , arzoumanian , z. , bertsch , d.l , et al . , 1994 , , 289 , 855 kifune , t. , tanimori , t. , ogio , s. , et al . , 1995 , , 438 , 91 koribalski , b. , johnston , s. , weisberg , j.h . , & wilson , w. , 1995 , , 441 , 756 lattimer , j.m . , & prakash , m. , 2001 , , 550 , 426 mardia , k.v . , 1972 , statistics of directional data ( london : academic ) mcadam , w.b osborne , j.l . , & parkinson , m.l . , 1993 , , 361 , 516 mcgowan , k.e . , kennea , j.a . , zane , s. , et al . , 2003 , , 591 , 380 mineo , t. , massaro , e. , cusumano , g. , & becker , w. , 2002 , , 392 , 181 nomoto , k. , & tsuruta , s. , 1987 , , 312 , 711 pavlov , g.g . , shibanov , yu.a . , & zavlin , v.e . , 1992 , , 253 , 193 pavlov , g.g . , sanwal , d. , garmire , g.p . , et al . , 2000 , aas meeting 196 , # 37.04 pavlov , g.g . , zavlin , v.e . , sanwal , d. , et al . , 2001 , , 552 , l129 pavlov , g.g . , zavlin , v.e . , & sanwal , d. , 2002 , proceedings of the 270 . we - heraeus seminar on _ `` neutron stars , pulsars and supernova remnants '' _ , eds . w. becker , h. lesh & j. trmper , p. 273 - 286 possenti , a. , mereghetti , s. , & colpi , m. , 1996 , , 313 , 565 ray , a. , harding , a.k . , & strikman , m. , 1999 , , 513 , 919 strickman , m.s . , harding , a.k . , & de jager , o.c . , 1999 , , 524 , 373 strder , l. , briel , u. , dennerl , k. , et al . , 2001 , , 365 , 18 taylor , j.h . , & cordes , j.m . , 1993 , , 411 , 674 thompson , d.j . , arzoumanian , z. , bertsch , d.l . , et al . , 1992 , , 359 , 615 thompson , d.j . , bailes , m. , bertsch , d.l . , et al . , 1996 , , 465 , 385 turner , m.j.l . , abbey , a. , arnaud , m. , et al . , 2001 , , 365 , 27 wang , n. , manchester , r.n . , pace , r.t . , et al . , 2000 , , 317 , 843 yakovlev , d.g . , gnedin , o.y . , kaminker , a.d . , et al . , 2002 , ( astro - ph/0306143 ) zane , s. , turolla , r. , & treves , a. , 2000 , , 537 , 387 zane , s. , haberl , f. , cropper , et al . , 2002 , , 344 , 345 zavlin , v.e . , pavlov , g.g . , & shibanov , yu.a . , 1996 , , 315 , 141 zavlin , v.e . , pavlov , g.g . , & trmper , j. , 1998 , , 331 , 821 zavlin , v.e . , trmper , j. , & pavlov , g.g . , 1999 , , 525 , 959 zavlin , v.e . , pavlov , g.g . , sanwal , d. , manchester , r.n . , trmper , j. , halpern , j.p . , becker , w. , 2002 , , 569 , 894 lcccccc bb & 5.5 ( fixed ) & & @xmath119 & @xmath120 & @xmath49 ( fixed ) & 4.88 [ 661 ] + bb & @xmath121 & & @xmath122 & @xmath123 & @xmath49 ( fixed ) & 2.92 [ 660 ] + pl & 5.5 ( fixed ) & @xmath124 & & & @xmath49 ( fixed ) & 1.82 [ 661 ] + pl & @xmath125 & @xmath126 & & & @xmath49 ( fixed ) & 1.17 [ 660 ] + bb+pl & 5.5 ( fixed ) & @xmath127 & @xmath128 & @xmath129 & @xmath49 ( fixed ) & 0.84 [ 659 ] + bb+pl & @xmath130 & @xmath131 & @xmath132 & @xmath133 & @xmath49 ( fixed ) & 0.84 [ 658 ] + atmos+pl & @xmath134 & @xmath135 & 10 ( fixed ) & @xmath136 & @xmath137 & 0.84 [ 658 ] + atmos+pl & @xmath138 & @xmath139 & 12 ( fixed ) & @xmath140 & @xmath141 & 0.84 [ 658 ] +

we report on the _ xmm - newton _ observations of the young , 102 ms pulsar psr b1706 - 44 . we have found that both a blackbody plus power - law and a magnetized atmospheric model plus power - law provide an excellent fit to the epic spectra . the two scenarios are therefore indistinguishable on a statistical basis , although we are inclined to prefer the latter on physical grounds . in this case , assuming a source distance of @xmath0 kpc , the size of the region responsible for the thermal emission is @xmath1 km , compatible with the surface of a neutron star . a comparison of the surface temperature of psr b1706 - 44 obtained from this fit with cooling curves favor a medium mass neutron star with @xmath2 or @xmath3 , depending on two different models of proton superfluidity in the interior . the large collecting area of _ xmm - newton _ allows us to resolve a substructure in the broad soft x - ray modulation detected by _ chandra _ , revealing the presence of two separate peaks with pulsed fractions of @xmath4 and @xmath5 , respectively .

it is widely believed that accretion disks around kerr black holes exist in many astrophysical environments , ranging from active galactic nuclei to some stellar binary systems @xcite . people usually assume that the inner boundary of a thin keplerian disk around a kerr black hole is located at the marginally stable circular orbit , inside which the centrifugal force is unable to stably balance the gravity of the central black hole @xcite . in the disk region , particles presumably move on nearly circular orbits with a very small inward radial velocity superposed on the circular motion , the gravity of the central black hole is approximately balanced by the centrifugal force . as disk particles reach the marginally stable circular orbit , the gravity of the central black hole becomes to dominate over the centrifugal force and the particles begin to nearly free - fall inwardly . the motion of fluid particles in the plunging region quickly becomes supersonic then the particles loose causal contact with the disk , as a result the torque at the inner boundary of the disk is approximately zero ( * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * and references therein ) . this is usually called the `` no - torque inner boundary condition '' of thin accretion disks . some recent studies on the magnetohydrodynamics ( mhd ) of accretion disks have challenged the `` no - torque inner boundary condition '' . magnetic fields have been demonstrated to be the most favorable agent for the viscous torque in an accretion disk transporting angular momentum outward ( * ? ? ? * and references therein ) . by considering the evolution of magnetic fields in the plunging region , krolik @xcite pointed out that magnetic fields can become dynamically important in the plunging region even though they are not so on the marginally stable circular orbit , and argued that the plunging material might exert a torque to the disk at the marginally stable circular orbit . with a simplified model , gammie @xcite solved maxwell s equations in the plunging region and estimated the torque on the marginally stable circular orbit . he demonstrated that the torque can be quite large and thus the radiation efficiency of the disk can be significantly larger than that for a standard accretion disk where the torque at the inner boundary is zero . furthermore , agol and krolik @xcite have investigated how a non - zero torque at the inner boundary affects the radiation efficiency of a disk . numerical simulations of mhd disks @xcite have greatly improved our understanding of disk accretion processes . these simulations show that the magneto - rotational instability effectively operates inside the disk and leads to accretion , though the accretion picture is much more complicated than that assumed in the standard theory of accretion disks . generally , the disk accretion is non - axisymmetric and strongly time - dependent . it is also found that , as disk material gets into the plunging region , the magnetic stress at the marginally stable circular orbit does not vanish but smoothly extends into the plunging region @xcite , though the effect is significantly reduced as the thickness of the disk goes down @xcite . furthermore , the specific angular momentum of particles in the plunging region does not remain constant , which implies that the magnetic field may be dynamically important in the plunging region @xcite . all these results are fascinating and encouraging . unfortunately , due to the limitation in space resolution and time integration , stationary and geometrically thin accretion disks are not accessible to the current 2-d and 3-d simulations . so it remains unclear how much insights we can get for stationary and geometrically thin accretion disks from these simulations @xcite . instead of small - scale and tangled magnetic fields in an accretion disk transporting angular momentum within the disk , a large - scale and ordered magnetic field connecting a black hole to its disk may also exist and play important roles in transportation of angular momentum and energy between the black hole and the disk @xcite . recent _ xmm - newton _ observations of some seyfert galaxies and galactic black hole candidates provide possible evidences for such a magnetic connection between a black hole and its disk @xcite . all these promote the importance of studying the evolution and the dynamical effects of magnetic fields around a kerr black hole . in this paper , we use a simple model to study the evolution of magnetic fields in the plunging region around a kerr black hole . we assume that around the black hole the spacetime metric is given by the kerr metric ; in the plunging region , which starts at the marginally stable circular orbit and ends at the horizon of the black hole , a stationary and axisymmetric plasma fluid flows inward along timelike geodesics in a small neighborhood of the equatorial plane . the plasma is perfectly conducting and a weak magnetic field is frozen to the plasma . the magnetic field and the velocity field have two components : radial and azimuthal . we will solve the two - dimensional maxwell s equations where the magnetic field depends on two variables : time and radius , and investigate the evolution of the magnetic field . this model is similar to that studied by gammie @xcite , but here we include the time variable . furthermore , we ignore the back - reaction of the magnetic field on the motion of the plasma fluid to make the model self - consistent , since if the dynamical effects of the magnetic field are important the strong electromagnetic force will make the fluid expand in the vertical direction . the ignorance of the back - reaction of the magnetic field will allow us to to analytically study the evolution of the magnetic field , but it will prevent us from quantitatively studying the dynamical effects of the magnetic field . however , we believe that the essential features of the evolution of the magnetic field in the plunging region is not sensitive to the details of the dynamical effects . to check the self - consistency of the model , we will estimate the dynamical effects of the magnetic field by considering the back - reaction of the magnetic field on the fluid motion . we will also discuss the self - consistent solutions to the coupled maxwell and dynamical equations , and look for implications to the `` no - torque inner boundary condition '' . the paper is organized as follows : in sec . [ sec2 ] we write down the maxwell s equations for an ideal mhd plasma . in sec . [ sec3 ] we write down the general forms of the magnetic field and the velocity field of the plasma fluid around a kerr black hole . in sec . [ sec4 ] we solve maxwell s equations with the approximations outlined above . in sec . [ sec5 ] , with the solutions obtained in sec . [ sec4 ] , we study the evolution of the magnetic field in the plunging region . in sec . [ sec6 ] , we check if magnetic fields can become dynamically important in the plunging region . in sec . [ sec7 ] we draw our conclusions . in the appendix we solve maxwell s equations in the disk region , since we want our solutions in the plunging region to be continuously joined to the solutions in the disk region . throughout the paper we use the geometrized units @xmath0 and the boyer - lindquist coordinates @xmath1 @xcite , except in the appendix where cylindrical coordinates are used . in a curved spacetime , maxwell s equations are @xmath2 } = 0\ ; , \hspace{0.5 cm } \label{max2}\end{aligned}\ ] ] where @xmath3 is the electromagnetic field tensor , @xmath4 is the current density 4-vector of electric charge . in eqs . ( [ max1 ] ) and ( [ max2 ] ) , @xmath5 denotes the covariant derivative operator that is compatible with metric : @xmath6 ; the square brackets `` [ ] '' denote antisymmetrization of a tensor . for an ideal mhd plasma fluid whose electric resistivity is zero , the electric field in the comoving frame is zero , i.e. @xmath7 where @xmath8 is the 4-velocity of the fluid . in other words , the magnetic field is frozen to the plasma fluid . then , the electromagnetic field tensor @xmath9 can be written as @xmath10 where @xmath11 is the magnetic field measured by an observer comoving with the fluid ( i.e. , having a 4-velocity @xmath8 ) , and @xmath12 is the totally antisymmetric tensor of the volume element that is associated with the metric @xmath13 . by definition , the magnetic field @xmath11 satisfies @xmath14 the corresponding electromagnetic stress - energy tensor is @xmath15 where @xmath16 . in the case of mhd , the electric current density @xmath4 is unknown , but it is only defined by eq . ( [ max1 ] ) . the fundamental variables are the electric field @xmath17 and the magnetic field @xmath11 . maxwell s equations are then reduced to eq . ( [ max2 ] ) . since in the comoving frame of an ideal mhd flow @xmath18 and @xmath11 is related to the electromagnetic tensor @xmath9 by eq . ( [ fab ] ) , the dual of eq . ( [ max2 ] ) gives @xcite @xmath19 } ) = 0 \;. \label{max3}\end{aligned}\ ] ] for an ideal mhd fluid eqs . ( [ max2 ] ) and ( [ max3 ] ) are equivalent . ( [ max3 ] ) can be expanded as @xmath20 the contraction of eq . ( [ max3a ] ) with @xmath21 leads to @xmath22 where @xmath23 is the acceleration of the fluid . in deriving eq . ( [ div ] ) we have used eq . ( [ bua ] ) and the identity @xmath24 . substituting eq . ( [ div ] ) into eq . ( [ max3a ] ) , we get @xmath25 the tensor @xmath26 is decomposed as @xcite @xmath27 where @xmath28 is the space - projection tensor , @xmath29 is the expansion , @xmath30 is the shear tensor , and @xmath31}$ ] is the vorticity tensor of the fluid . here the braces `` ( ) '' denote symmetrization of a tensor . it is easy to check that @xmath32 substituting eq . ( [ dab ] ) into eq . ( [ max3b ] ) , we obtain @xmath33 which shows that the evolution of the magnetic field is governed by the expansion , the shear , the vorticity , and the acceleration of the fluid . the contraction of eq . ( [ max3c ] ) with @xmath34 gives @xmath35 where we have used @xmath36 and eq . ( [ bua ] ) . we note that with the space - projection tensor @xmath37 , eq . ( [ div ] ) can be written as @xmath38 which says that the spatial divergence of the magnetic field is zero . in terms of differential forms , the tensor @xmath9 is a closed 2-form , then the maxwell equation ( [ max2 ] ) can be written as @xmath39 @xcite . using stokes theorem and eq . ( [ ezero ] ) , it can be shown that for a perfect conducting fluid the magnetic flux threading any 2-dimensional spatial surface @xmath40 is unchanged as the surface moves with the fluid @xmath41 this is the mathematical formulation for the statement that magnetic field lines are frozen to a perfectly conducting fluid . now let us assume that the background spacetime is the outside of a kerr black hole of mass @xmath42 and angular momentum @xmath43 , where @xmath44 . in boyer - lindquist coordinates , the kerr metric is @xcite @xmath45 where @xmath46 a kerr black hole usually has two event horizons : an inner event horizon and an outer event horizon , whose radii are given by the two roots of @xmath47 . , the kerr black hole becomes a schwarzschild black hole , then the inner event horizon disappears . when @xmath48 ( the case of an extreme kerr black hole ) , the inner event horizon coincides with outer event horizon . ] what is relevant in this paper is the outer event horizon , so whenever we talk about the `` event horizon '' we always mean the outer event horizon , whose radius is @xmath49 let us define an orthonormal tetrad attached to an observer comoving with the frame dragging of the kerr black hole , @xmath50 , by @xmath51\ ; , \hspace{4.3 cm } \nonumber \\ e_1^a \equiv \left(\frac{\delta}{\sigma}\right)^{1/2 } \left(\frac{\partial}{\partial r}\right)^a\ ; , \hspace{0.6 cm } e_2^a \equiv \frac{1}{\sigma^{1/2 } } \left(\frac{\partial}{\partial \theta}\right)^a\ ; , \hspace{0.6 cm } e_3^a \equiv \left(\frac{\sigma}{a}\right)^{1/2 } \frac{1 } { \sin\theta } \left(\frac{\partial}{\partial \phi}\right)^a\ ; , \label{ont2}\end{aligned}\ ] ] where @xmath52 which are respectively the lapse function and the frame dragging angular velocity . as @xmath53 , we have @xmath54 and @xmath55 , where @xmath56 is the angular velocity of the event horizon . then , the 4-velocity of the fluid , @xmath8 , can be decomposed as @xmath57 where @xmath58 are the components of the 3-velocity of the fluid relative to the observer comoving with the frame dragging , and @xmath59 is the lorentz factor . inserting eq . ( [ ont2 ] ) into eq . ( [ ua1 ] ) , we have @xmath60\ ; , \label{ua2}\end{aligned}\ ] ] where @xmath61 is the angular velocity of the fluid . the specific angular momentum of a fluid particle is @xmath62 obviously , @xmath63 when @xmath64 . thus , an observer comoving with the frame dragging of a kerr black hole has zero angular momentum @xcite . the specific energy of a fluid particle is @xmath65 in eq . ( [ sen ] ) , the term @xmath66 represents the coupling between the orbital angular momentum of the particle and the frame dragging of the kerr black hole . when the particle moves on a geodesic , @xmath67 and @xmath68 defined above are conserved @xcite . the magnetic field @xmath11 , which satisfies eq . ( [ bua ] ) , can be decomposed as @xmath69 from which we have @xmath70 inserting eq . ( [ ont2 ] ) into eq . ( [ ba1 ] ) , we have @xmath71 + b_{\hat{r } } \left(\frac{\delta}{\sigma } \right)^{1/2 } \left(\frac{\partial}{\partial r}\right)^a \nonumber\\ & & + \frac{b_{\hat{\theta}}}{\sigma^{1/2 } } \left(\frac { \partial}{\partial \theta}\right)^a + \frac{b_{\hat{\phi } } } { \sin\theta } \left(\frac{\sigma}{a}\right)^{1/2 } \left(\frac{\partial}{\partial \phi}\right)^a\ ; ; \label{ba2}\end{aligned}\ ] ] and , correspondingly @xmath72 the maxwell equations that we want to solve are given by eq . ( [ max3 ] ) . in terms of coordinate components , ( [ max3 ] ) takes a very simple form @xmath73 = 0 \ ; , \label{maxeq}\end{aligned}\ ] ] where @xmath74 , @xmath75 is the determinant of the metric tensor @xmath76 , @xmath77 and @xmath78 are respectively @xmath79 and @xmath80 because of the constraint eq . ( [ bua ] ) , among the four equations of ( [ maxeq ] ) only three are independent . since the background spacetime is stationary and axisymmetric , we can look for stationary and axisymmetric solutions with @xmath81 . however , since we are interested in the time evolution of magnetic fields , we will keep the @xmath82 terms on magnetic fields but adopt that @xmath83 . to simplify the problem , we further assume that in a small neighborhood of the equatorial plane ( i.e. , @xmath84 ) , @xmath85 ( i.e. , @xmath86 ) . this assumption , which has also been used by gammie @xcite , ensures that @xmath87 on the equatorial plane . we emphasize that , when @xmath88 , this assumption holds only if the fluid moves geodesically , which requires that the magnetic fields are weak and their dynamical effects are negligible . otherwise the electromagnetic force will make @xmath89 and @xmath90 non - zero except exactly on the equatorial plane . thus , hereafter we assume that fluid particles move on timelike geodesics in the plunging region . this assumption will be justified latter . now , let us focus on solutions on the equatorial plane ( @xmath91 ) . considering the fact that for the kerr metric @xmath92 on the equatorial plane , eq . ( [ maxeq ] ) is reduced to @xmath93 + \frac{\partial}{\partial r } \left[r^2 ( u^r b^\beta - u^\beta b^r)\right ] = 0 \;. \label{eq1}\end{aligned}\ ] ] for @xmath94 , eq . ( [ eq1 ] ) gives @xmath95 = 0 \ ; , \hspace{1 cm } \frac{\partial}{\partial r}\left[r^2 ( u^r b^t - u^t b^r)\right ] = 0 \ ; , \label{eq2}\end{aligned}\ ] ] respectively . the solution of eq . ( [ eq2 ] ) is @xmath96 where @xmath97 is a constant . with eq . ( [ flux ] ) , it can be checked that @xmath98 is the magnetic flux threading the 2-dimensional surface defined by @xmath99 , @xmath100 , @xmath101 and @xmath102 . for @xmath103 , eq . ( [ eq1 ] ) is automatically satisfied since @xmath104 in a small neighborhood of the equatorial plane everywhere and all the time . for @xmath105 , eq . ( [ eq1 ] ) gives @xmath106 + \frac{\partial}{\partial r } \left[r^2 ( u^r b^\phi - u^\phi b^r)\right ] = 0 \;. \label{eq3}\end{aligned}\ ] ] from the constraint eq . ( [ bua ] ) , we have @xmath107 substituting eq . ( [ btr ] ) for @xmath108 into eq . ( [ cons ] ) , we obtain @xmath109 \;. \label{bfr}\end{aligned}\ ] ] now , substitute eq . ( [ btr ] ) for @xmath108 and eq . ( [ bfr ] ) for @xmath110 into eq . ( [ eq3 ] ) , we obtain a first order partial differential equation @xmath111 where @xmath112 in deriving eq . ( [ eq4 ] ) we have used @xmath113 . ( [ eq4 ] ) simply says that @xmath114 is conserved along the worldline of a fluid particle : @xmath115 . let us define @xmath116 which is the coordinate time interval spent by a fluid particle to move from @xmath117 to @xmath118 . then eq . ( [ eq4 ] ) can be written as @xmath119 the solution of eq . ( [ eq5 ] ) is simply @xmath120 i.e. @xmath114 is a function of @xmath121 . ( [ psol ] ) gives a `` retarded '' solution to eq . ( [ eq5 ] ) : at any time @xmath122 the solution at radius @xmath118 is given by the solution at an earlier time @xmath123 at the radius @xmath124 . thus , a variation in the magnetic fields at any @xmath118 will propagate with the fluid motion ( fig . [ fig1 ] ) . the solution is unique if a suitable boundary or initial condition is imposed . for example , if a boundary condition is given on @xmath125 : @xmath126 , then the solution is @xmath127 . in order for the solution to exist for a region specified by @xmath128 in the @xmath129-space , the boundary function @xmath130 must be defined on the whole @xmath131-axis : @xmath132 . similarly , if a boundary condition is given on @xmath133 ( i.e. , @xmath134 ) : @xmath135 , then the solution is @xmath136 . in this case , in order for the solution to exist for a region specified by @xmath137 in the @xmath129-space , the boundary function @xmath138 must be defined on the whole @xmath122-axis : @xmath139 . we can also specify the boundary condition in another way @xmath140 then , the solution of eq . ( [ eq5 ] ) is @xmath141 i.e. , the value of @xmath114 in region @xmath142 ( region i ) is determined by the value of @xmath114 on the boundary @xmath143 ; the value of @xmath114 in region @xmath144 ( region ii ) is determined by the value of @xmath114 on the boundary @xmath145 ( see fig . [ fig2 ] ) . in order for the solutions to be smoothly matched on the diagonal line @xmath146 separating region i and region ii , @xmath130 and @xmath138 must be smoothly matched at @xmath147 : @xmath148 given the solution of @xmath114 in eq . ( [ psol ] ) , we can solve @xmath149 from eq . ( [ psi ] ) , then solve @xmath110 from eq . ( [ bfr ] ) . the results are @xmath150 \;. \label{fsol2}\end{aligned}\ ] ] using eq . ( [ btr ] ) , we obtain @xmath151 \;. \label{fsol2a}\end{aligned}\ ] ] note that @xmath108 , @xmath149 , and @xmath110 satisfy the constraint eq . ( [ cons ] ) , so among the three components only two are independent . from eqs . ( [ ualp ] ) , ( [ balp ] ) , ( [ fsol1 ] ) , ( [ fsol2 ] ) , and the fact that @xmath152 and @xmath153 , we can solve for @xmath154 and @xmath155 @xmath156 where eqs . ( [ sang ] ) and ( [ sen ] ) have been used to simplify the expression for @xmath155 . since we focus on the solutions on the equatorial plane , here and hereafter we set @xmath157 , and use @xmath158 , @xmath159 , and @xmath160 to refer their values at @xmath161 . note , in the solutions in eqs . ( [ fsol1 ] ) and ( [ fsol2 ] ) [ or , equivalently , eqs . ( [ br ] ) and ( [ bf ] ) ] all the dependence on time @xmath122 is contained in the function @xmath114 . since we assume that the fluid particles move on geodesics , the specific angular momentum @xmath67 and the specific energy @xmath68 are constants . if @xmath114 is also a constant , the combination @xmath162 is a constant , which we denote as @xmath163 . then , eqs . ( [ br ] ) and ( [ bf ] ) become @xmath164 which are stationary solutions of maxwell s equations . if the black hole is a schwarzschild black hole ( i.e. , the specific angular momentum @xmath165 ) , the stationary solutions are reduced to @xmath166 where @xmath167 , @xmath168 , and @xmath169 . the non - stationary solutions can be obtained by replacing @xmath163 with @xmath162 . we are interested in the evolution of magnetic fields in the plunging region in the equatorial plane between @xmath170 , the marginally stable circular orbit , and @xmath171 , the event horizon of the black hole . for direct circular orbits ( i.e. , corotating with @xmath172 ) around a kerr black hole , the radius of the marginally stable orbit is given by @xcite @xmath173^{1/2}\right\ } \ ; , \label{rms}\end{aligned}\ ] ] where @xmath174 ( @xmath175 ) if @xmath176 ( @xmath177 ) , and @xmath178 \;,\\ z_2 & \equiv & \left(z_1 ^ 2 + 3 \frac{a^2}{m^2}\right)^{1/2 } \;.\end{aligned}\ ] ] the angular velocity of a particle geodesically moving on the marginally stable circular orbit is @xmath179^{-1 } \;. \label{wr}\end{aligned}\ ] ] from eqs . ( [ angv ] ) , ( [ rms]-[wr ] ) , we can calculate the circular velocity on the marginally stable circular orbit by @xmath180 the corresponding lorentz factor by @xmath181 , specific angular momentum @xmath182 by eq . ( [ sang ] ) , and specific energy @xmath183 by eq . ( [ sen ] ) ( setting @xmath161 and @xmath170 ) . here and hereafter the subscript `` ms '' represents the values at @xmath184 . now let us choose the boundary radius @xmath185 , the conserved specific angular momentum and specific energy to be @xmath186 where @xmath187 . i.e. , we keep the specific energy the same as that on the marginally stable circular orbit , but decrease the specific angular momentum by a little amount . then , we can calculate the lorentz factor at @xmath170 , corresponding to the specific energy and the specific energy specified by eq . ( [ le0 ] ) , by @xmath188 and the corresponding boundary values of @xmath189 and @xmath190 at @xmath170 by @xmath191 from eqs . ( [ sang ] ) and ( [ sen ] ) , we can calculate @xmath192 , @xmath193 , @xmath189 , and @xmath190 at any @xmath118 by @xmath194^{1/2 } \;,\end{aligned}\ ] ] where @xmath68 and @xmath67 are given by eq . ( [ le0 ] ) . the parameter @xmath131 defined by eq . ( [ tau ] ) can then be calculated with @xmath195 with the above formulae at hands , we can calculate @xmath154 , @xmath155 , and @xmath196 at any @xmath118 by eqs . ( [ br ] ) and ( [ bf ] ) , giving the constant @xmath97 and the function @xmath197 . to determine @xmath97 and @xmath114 , we need to specify the boundary conditions for the magnetic field . we will consider stationary solutions and non - stationary solutions separately . it is straightforward to specify the boundary conditions for stationary solutions ( i.e. solutions with @xmath198 ) . to determine the solutions , we need only specify the values of @xmath154 and @xmath155 at @xmath170 : @xmath199 and @xmath200 . then , by eqs . ( [ br1 ] ) and ( [ bf1 ] ) , we have @xmath201 where @xmath202 , @xmath203 is the angular velocity corresponding to @xmath204 @xmath205 with the @xmath163 and @xmath114 determined above , we can calculate @xmath154 , @xmath155 , and @xmath196 at any @xmath118 by eqs . ( [ br1 ] ) and ( [ bf1 ] ) . since @xmath154 and @xmath155 linearly depend on @xmath199 and @xmath200 , it is sufficient to study the effects of @xmath199 and @xmath200 separately . the results for any linear combination of @xmath199 and @xmath200 are simply linear superpositions of the results for @xmath199 and @xmath200 separately . in fig . [ fig3 ] we show the evolution results of @xmath155 with the boundary condition @xmath206 and @xmath207 at @xmath170 , for different spinning status of the black hole and different values of @xmath208 that specify the kinetic boundary conditions of the fluid . all quantities are scaled to the mass of the black hole so we do not need to specify the value of @xmath42 . since @xmath154 does not depend on @xmath200 , @xmath154 is always zero . since @xmath209 and @xmath210 , @xmath155 evolves according to @xmath211 . though in the plunging region @xmath118 decreases , @xmath212 grows faster except at the neighbor of @xmath213 . so the evolution of @xmath155 is dominated by the variation of @xmath214 . thus , as fluid particles get into the plunging region , @xmath155 decreases quickly as clearly shown in fig . this radial expansion effect is not sensitive to the spin of the black hole , but very sensitive to the value of @xmath208 ( or , effectively , the initial value of @xmath214 ) . as @xmath208 decreases ( i.e. , @xmath215 decreases ) , the fluid expands more as it gets into the plunging region , so the value of @xmath155 decreases more . in fig . [ fig4 ] we show the evolution results of @xmath154 ( dashed lines ) and @xmath155 ( solid lines ) with the boundary condition @xmath216 and @xmath217 at @xmath170 . though at @xmath170 we have @xmath217 , in the plunging region @xmath155 becomes nonzero since @xmath155 depends on both @xmath199 and @xmath200 [ eqs . ( [ bf1 ] ) , ( [ c1 ] ) and ( [ c3 ] ) ] . this is the manifestation that the shear motion of the fluid in the plunging region generates @xmath155 from @xmath154 . the radial component of the magnetic field , @xmath154 , increases gradually as the fluid enters the plunging region , according to @xmath218 , and blows up on the black hole horizon where @xmath47 . the shear motion of the fluid does not amplify @xmath154 , which echos with the fact that @xmath154 is always zero if it is zero at @xmath170 ( fig . [ fig3 ] ) . since @xmath154 does not depend on @xmath214 , there is only one dashed line in each panel of fig . [ fig4 ] . in comparison , the toroidal component , @xmath155 , increases more quickly in the transition region , since the shear motion of the fluid magnifies @xmath155 . this is more prominent for small values of @xmath208 ( i.e. small @xmath219 ) , since @xmath220 [ eq . ( [ bf1 ] ) ] and @xmath190 is close to @xmath221 as the particles just leave the marginally stable circular orbit . for small values of @xmath208 , @xmath155 increases sharply as the fluid just gets into the plunging region , then decreases a little bit due to the radial expansion of the fluid . unlike @xmath154 , @xmath155 is always finite on the black hole horizon . we see that , @xmath154 and @xmath155 evolves in very different ways . from fig . [ fig4 ] we also see that the evolution of the magnetic field in the plunging region depends on the spin of the black hole , though not very sensitively . interestingly , the toroidal component and the poloidal component depend on the spin of the black hole in an opposite way . as the dimensionless spin parameter @xmath222 increases from zero to positive values , the shear amplification effect on the toroidal component of the magnetic field increases ( except for the case of @xmath223 for which the amplification effect is not prominent ) , while the amplification of the radial component caused by the compression in the azimuthal direction ( i.e. , the decrease in radius @xmath118 ) decreases . but , if @xmath222 decreases from zero to negative values , the shear amplification effect on the toroidal component of the magnetic field decreases , while the compression amplification of the radial component increases . the opposite dependence for positive and negative spins is probably due to the different coordinate distances from the marginally stable circular orbit to the black hole horizon for black holes of positive and negative spins . in fig . [ fig5 ] we show the evolution of @xmath224 [ defined by eq . ( [ bsq ] ) ] with the same boundary conditions in fig . as the fluid just enters the plunging region ( near the right ends of curves ) , @xmath224 sharply increases due to the small values of @xmath225 there , which is the manifestation of amplification effect caused by the shear rotation of the fluid . after that , i.e. , after the fluid obtains a large radial velocity ( the dashed lines in the figure ) , the amplification effect is reduced but the expansion effect becomes prominent [ see eq . ( [ evol ] ) ] . on the horizon of the black hole @xmath224 is always finite , so the boundary conditions on the horizon is satisfied ( * ? ? ? * and references therein ) . to specify the boundary conditions for non - stationary solutions is a little bit complicated . as discussed earlier , to determine the solutions in a region with @xmath226 and @xmath227 ( i.e. @xmath228 ) in the @xmath129-space , we need to specify the boundary conditions on the axes ( @xmath143 ) and ( @xmath229 ) , i.e. specify @xmath130 and @xmath138 [ see eq . ( [ bond ] ) ] . as an example , let us assume that @xmath230 , and @xmath217 on the axis ( @xmath143 ) . i.e. , the solution on the boundary @xmath133 is stationary for @xmath231 , and the @xmath232-component of the magnetic field in the plunging region is zero at @xmath125 . equivalently , we specify the boundary conditions as follows : @xmath233 then , if we apply eqs . ( [ br ] ) and ( [ bf ] ) to the boundary at @xmath170 and @xmath231 , we can solve for @xmath97 and @xmath234 . the results are @xmath235 \ ; , \label{c0s } \\ \psi_2 & = & - \gamma_0 b_{\hat{r}0 } a_{\rm ms}^{1/2 } ( \omega_0 - \omega_{\rm ms } v_{\hat{r}0}^2 ) \;. \label{psi2}\end{aligned}\ ] ] applying eq . ( [ bf ] ) to the boundary at @xmath125 and @xmath236 , and substituting eq . ( [ c0s ] ) for @xmath97 , we can solve for @xmath237 @xmath238 \ ; , \hspace{1 cm } \psi_1(r ) = \frac{(\omega - \omega v_{\hat{r}}^2 ) \gamma e c_0 } { ( \omega - \omega v_{\hat{r}}^2 ) \gamma l + \chi } \ ; , \label{psi1}\end{aligned}\ ] ] where @xmath239 is given by the inverse of @xmath240 . the values of @xmath154 on @xmath125 and @xmath241 is then determined by eq . ( [ br ] ) . with the @xmath237 and @xmath234 determined above , we can obtain @xmath114 by eq . ( [ phsol ] ) . this , together with the @xmath97 given by eq . ( [ c0s ] ) , allows us to calculate @xmath154 and @xmath155 with eqs . ( [ br ] ) and ( [ bf ] ) for any radius in the plunging region at any time @xmath128 . it is easy to check that @xmath242 , so the solutions are continuously matched on the line @xmath243 . since @xmath244 the solutions are smoothly matched on the line @xmath243 only in the limit @xmath245 since then we have @xmath246 at @xmath247 . if @xmath221 is not zero but small ( @xmath248 ) , the solutions are approximately smoothly matched on the line @xmath243 if @xmath249 is not large at @xmath170 . in figs [ fig6 ] [ fig8 ] we show the results for the non - stationary evolution of magnetic fields . each figure corresponds to a different spinning state of the black hole . in fig . [ fig6 ] @xmath250 , in fig . [ fig7 ] @xmath251 , while in fig . [ fig8 ] @xmath252 . the boundary conditions for the magnetic field are given by eqs . ( [ bond1 ] ) and ( [ bond2 ] ) . the kinetic boundary conditions of the fluid are given by eq . ( [ le0 ] ) , here we assume @xmath253 . in each figure , each panel corresponds to the state of the magnetic field at a particular moment . especially , the first ( left and up ) panel shows the initial ( @xmath125 ) state of the magnetic field [ i.e. eqs . ( [ bond1 ] ) and ( [ bond2 ] ) ] . in each panel , the thick dashed line shows @xmath154 , the thick solid line shows @xmath155 , and the thin lines show the corresponding stationary solutions . each curve starts from the marginally stable circular orbit ( right end ) and ends on the horizon of the black hole ( left end ) . from these figures we see that , though initially the magnetic field is deviated from the stationary state , it evolves to and finally saturates at the stationary state as time goes on . since we hold the magnetic field on the marginally stable circular orbit in a stationary state for @xmath231 [ eq . ( [ bond1 ] ) ] and the fluid moves from the marginally stable circular orbit toward the horizon of the black hole , the stationary state propagates from the marginally stable circular orbit toward the horizon of the black hole ( i.e. , from right to left in the figures ) . this is clear seen in figs . [ fig6 ] [ fig8 ] : the magnetic field at a radius closer to the right end of the curve gets into the stationary state earlier . in each panel , with a black dot we show the position of a particle that is at the marginally stable radius at @xmath125 , which clearly shows the propagation of the state of the magnetic field with the motion of the fluid . this suggests that the stationary state of the magnetic field in the plunging region is uniquely determined by the boundary conditions at the marginally stable circular orbit . the figures also show the dependence of the evolution of the magnetic field on the spin of the black hole . for a black hole with a negative @xmath254 a longer coordinate time is needed for most of the fluid in the plunging region to settle into the stationary state . this is apparently caused by the fact that a black hole with a negative @xmath254 has a larger coordinate radial distance from the marginally stable circular orbit to the horizon . in previous sections we have studied the evolution of magnetic fields in the plunging region and seen that magnetic fields can be amplified by the convergence ( the radial magnetic field ) and shear motion ( the toroidal magnetic field ) of the fluid . thus , we can ask if magnetic fields can become dynamically important in the plunging region assuming that their dynamical effects are negligible at the marginally stable circular orbit . we try to answer this question in this section . assuming that in the plunging region the gas pressure of the fluid is negligible , then the total stress - energy tensor of the fluid is @xmath255 where @xmath256 is the mass - energy density of the fluid matter measured by an observer comoving with the fluid , @xmath257 is the stress - energy tensor of the electromagnetic field as given by eq . ( [ tab ] ) . then , the equation of motion of the fluid is given by @xmath258 when maxwell s equations are satisfied we have @xmath259 @xcite . then , by eq . ( [ ezero ] ) , we have @xmath260 for a magnetic field frozen to the fluid . therefore , the contraction of @xmath21 with eq . ( [ dtab ] ) leads to the equation of continuity ) also holds if the motion of the fluid is locally adiabatic @xcite . ] @xmath261 where @xmath262 is the flux density vector of mass . the equation of continuity does not explicitly contain magnetic field variables . since @xmath263 and @xmath264 are killing vectors , eq . ( [ dtab ] ) leads to the conservation of angular momentum and energy @xmath265 where @xmath266 is the flux density vector of angular momentum , @xmath267 is the flux density vector of energy . here we look for stationary and axisymmetric solutions in the plunging region . then , with the assumptions adopted in this paper , i.e. @xmath268 , @xmath85 and @xmath269 , in the equatorial plane the equation of continuity is reduced to @xmath270 similarly , the equations of angular momentum and energy conservation are reduced to @xmath271\right\ } = 0 \ ; , \label{lcon1}\end{aligned}\ ] ] and @xmath272\right\ } = 0 \;. \label{econ1}\end{aligned}\ ] ] the solution of eq . ( [ cont2 ] ) is @xmath273 where @xmath274 is a constant to be determined by the boundary conditions , which measures the mass flux across radius @xmath118 . [ fig9 ] shows the variation of the mass density with radius in the plunging region , assuming that fluid particles move on geodesics . the kinetic boundary conditions are given by eq . ( [ le0 ] ) . in each panel ( corresponding to a different spinning state of the black hole ) , we show the ratio of @xmath275 corresponding to four different values of @xmath208 : @xmath276 , @xmath277 , @xmath278 and @xmath279 , where @xmath280 . we see that , the evolution of the mass density of the fluid sensitively depends on the value of @xmath208 ( or effectively , the value of @xmath281 ) . while for not very small @xmath208 ( e.g. , @xmath223 ) the variation of @xmath256 is not dramatic , for very small @xmath208 ( e.g. , @xmath282 ) the variation of @xmath256 is dramatic : as the fluid gets into the plunging region the mass density drops sharply . this is caused by the sharp increase in the ratio @xmath283 in the plunging region for small @xmath208 . from eqs . ( [ lcon1 ] ) and ( [ econ1 ] ) we see that the dynamical role played by the magnetic field is characterized by the following three dimensionless parameters @xmath284 the parameter @xmath285 measures the transfer of angular momentum and energy from the fluid particles to the magnetic field ; @xmath286 and @xmath287 measure the transportation of angular momentum and energy from one part to another by the magnetic tension in the rest frame of the fluid . to justify the assumption that the effects of the magnetic field on the dynamics of the fluid particles are negligible , we must require that @xmath288 in eq . ( [ eta ] ) , @xmath256 is given by eq . ( [ rhm ] ) , @xmath224 is calculated by eq . ( [ bsq ] ) ( setting @xmath289 ) , @xmath290 and @xmath291 are calculated by @xmath292 near the marginally stable circular orbit , we have @xmath293 , @xmath294 , and @xmath295 . therefore , from the definitions of @xmath286 and @xmath287 , we have @xmath296 at @xmath297 . from fig . [ fig9 ] we see that for small @xmath208 the mass density @xmath256 drops quickly as the fluid enters the plunging region . however , the evolution of @xmath224 in the plunging region sensitively depends on the orientation of the magnetic field on the marginally stable circular orbit . from fig . [ fig5 ] , the value of @xmath224 corresponding to an initially radial magnetic field increases as the fluid enters the plunging region , because of the convergence of the radial magnetic field and the shear amplification of the toroidal magnetic field . if the initial magnetic field is purely toroidal , on the other hand , from fig . [ fig3 ] we see that the magnetic field keeps purely toroidal in the plunging region and the dilution effect arising from the drop in the mass density makes @xmath155 ( and thus @xmath224 ) decrease in the plunging region . therefore , in the plunging region the variation of @xmath285 , @xmath286 , and @xmath287 sensitively depends on the boundary condition of the magnetic field on the marginally stable circular orbit . we can imagine that , if on the marginally stable circular orbit the magnetic field is purely radial , then in the plunging region @xmath224 increases so the ratio @xmath285 also increases as @xmath256 decreases . in such a case , @xmath285 may become close to or even greater than @xmath298 in the plunging region even if it is @xmath299 on the marginally stable circular orbit , then the dynamical effects of the magnetic field become important in the plunging region . on the other hand , if on the marginally stable circular orbit the magnetic field is purely toroidal , then in the plunging region @xmath224 decreases which causes @xmath285 to decrease also if @xmath224 decreases faster than @xmath256 . therefore , to correctly estimate the dynamical effects of magnetic fields in the plunging region , we must choose a sensible boundary condition for the magnetic field on the marginally stable circular orbit . the magnetic field on the marginally stable circular orbit is determined by the mhd processes in the disk region . in the disk region ( @xmath300 ) , particles move on nearly circular orbits with @xmath301 ) of turns around the central black hole . as a result , the magnetic field lines frozen to the fluid are wound up around the center an infinite number of times . therefore , in a stationary state , we expect that the magnetic field in the disk region is predominantly toroidal . in the appendix we solve maxwell s equations in the disk region and show that in the stationary state the magnetic field in the disk is likely to be parallel to the velocity field : @xmath302 . thus , in the stationary state , on the marginally stable circular orbit the magnetic field does not take any orientation but the one that satisfies the following condition @xmath303 on the marginally stable circular orbit , eq . ( [ pc ] ) implies that @xmath304 since @xmath305 . from eqs . ( [ fsol1 ] ) and ( [ fsol2 ] ) , we have @xmath306 then , eq . ( [ pc ] ) is satisfied if and only if @xmath307 on the marginally stable circular orbit . in the stationary state @xmath114 is a constant , so we have @xmath308 throughout the plunging region . thus , in the stationary state , eq . ( [ pc ] ) holds throughout the plunging region . ) [ or , equivalently , eq . ( [ pc ] ) ] is satisfied . however , we point out that this does not always mean that the corresponding electric field measured by an observer comoving with the frame dragging is zero , unless the black hole is a schwarzschild black hole . one can check that , for the solutions in eqs ( [ fsol1 ] ) and ( [ fsol2 ] ) , the corresponding electric field measured by an observer comoving with the frame dragging [ i.e. , with a 4-velocity @xmath309 given in eq . ( [ ont2 ] ) ] is @xmath310 . so , when @xmath307 , we have @xmath311 , which is zero only if @xmath312 ( then the frame - dragging frequency @xmath313 ) . ] inserting eqs . ( [ ualp ] ) and ( [ balp ] ) into eq . ( [ pc ] ) , we obtain @xmath314 we have calculated @xmath285 , @xmath286 , and @xmath287 in the plunging region , assuming that they all are @xmath299 at @xmath170 and the boundary condition ( [ psi0 ] ) is satisfied . at @xmath170 the toroidal magnetic field is assumed to be @xmath315 in units of @xmath316 , the corresponding radial magnetic field @xmath154 is then given by eq . ( [ pc1 ] ) . the parameter @xmath208 is taken to be @xmath276 , @xmath277 , @xmath278 and @xmath279 , alternatively . the results are shown in figs . [ fig10 ] and [ fig11 ] . from these figures we see that , in the plunging region the evolution of @xmath285 , @xmath286 and @xmath287 sensitively depends on the value of @xmath208 . for very small @xmath208 , @xmath285 , @xmath286 and @xmath287 quickly decrease as the fluid gets into the plunging region . this is caused by the fact that a very small @xmath208 ( i.e. , a very small @xmath317 ) corresponds to a very small @xmath318 according to the boundary condition ( [ pc ] ) , while the magnetic field in the plunging region is predominantly determined by @xmath318 instead of @xmath319 ( see figs . [ fig3 ] and [ fig4 ] ) . for a moderate @xmath208 ( e.g. , @xmath223 ) , @xmath285 , @xmath286 and @xmath287 may increase in the plunging region . however , even in this case , the conditions in eq . ( [ nc1 ] ) are always satisfied throughout the plunging region . if we insert the solutions of maxwell s equations that we obtained in sec . [ sec4 ] into eqs . ( [ lcon1 ] ) and ( [ econ1 ] ) , then apply eq . ( [ rhm ] ) , we obtain @xmath320 where @xmath321 when @xmath307 [ i.e. , eq . ( [ pc ] ) is satisfied in the plunging region ] , we have @xmath322 . then eq . ( [ de1 ] ) implies that @xmath68 keeps constant in the plunging region . however , since @xmath323 , by eq . ( [ dl1 ] ) @xmath67 varies in the plunging region . therefore , for the solutions satisfying the boundary condition ( [ pc ] ) [ or , equivalently , eq . ( [ psi0 ] ) ] , the specific energy of particles is conserved but the specific angular momentum changes . setting @xmath324 , @xmath325 , and @xmath307 , eqs . ( [ dl1 ] ) and ( [ de1 ] ) can be integrated to obtain @xmath326 where @xmath327 where @xmath328 and @xmath329 are constants measuring the angular momentum flux and the energy flux across radius @xmath118 , respectively . using @xmath330 , we can solve eqs . ( [ fl ] ) and ( [ fe ] ) for @xmath331 @xmath332 then , from @xmath333 and @xmath334 , we obtain the equation for @xmath335 @xmath336 ^ 2 \;. \label{wind}\end{aligned}\ ] ] though the factor @xmath337 appears in the denominators on the right - hand side of eq . ( [ wind ] ) , @xmath338 is not a singularity of the equation , since the factor @xmath339 disappears from the denominators if we expand the second term on the right - hand side of eq . ( [ wind ] ) , then combine with the first term . however , the factor @xmath340 $ ] represents a singularity of eq . ( [ wind ] ) at @xmath341 the differentiation of eq . ( [ wind ] ) with respect to @xmath335 gives rise to another singularity of the equation , which is at @xmath342 this singularity appears as one differentiates eq . ( [ wind ] ) with respect to @xmath118 to obtain a differential equation for @xmath335 . if we define the relativistic alfvn velocity by @xmath343 then we have @xmath344 and @xmath345 where @xmath346 . therefore , the singularities given by eqs . ( [ sing1 ] ) and ( [ sing2 ] ) are critical points related to the alfvn speed : the alfvn point [ eq . ( [ sing1 ] ) ] and the fast critical point [ eq . ( [ sing2 ] ) ] @xcite . since we have assumed that the gas pressure is zero , the slow critical point is at @xmath347 which is not relevant to us here . the alfvn point is not an x - type singularity and it does not impose any additional conditions on the solution for @xmath335 except setting the integral constant @xcite . therefore , what is really relevant here is the fast critical point given by eq . ( [ sing2 ] ) . for weak fields , we expect that the fast critical point is located at a radius close to the marginally stable radius , where the accretion flow transits from subsonic motion in the disk region to supersonic motion in the plunging region @xcite . any physical solutions must smoothly pass the critical points , which sets strict constraints on the integral constants . it is far beyond the scope of the current paper to fully explore the properties of critical points in detail . however , as a first order approximation , we can assume that @xmath348 , where @xmath349 is the radius at the fast critical point . then , from eq . ( [ sing2 ] ) we have the radial velocity of the fluid at @xmath184 @xmath350^{1/3 } \;. \label{ur0}\end{aligned}\ ] ] if we set @xmath170 in eq . ( [ wind ] ) , then substitute eq . ( [ ur0 ] ) into eq . ( [ wind ] ) , we can solve for the constant @xmath351 , as a function of @xmath352 and @xmath353 . therefore , for solutions that smoothly pass the critical points , among the three integral constants @xmath352 , @xmath353 , and @xmath351 , only two are independent . without explicitly solving the radial flow equation , i.e. eq . ( [ wind ] ) , we can also obtain some interesting results on the horizon of the black hole . since @xmath354 as @xmath53 , from eq . ( [ wind ] ) we have @xmath355 where @xmath356 . then , from eq . ( [ uf ] ) , we can obtain the specific angular momentum for particles at @xmath357 @xmath358 eq . ( [ fe ] ) says that , with the boundary condition given by eq . ( [ psi0 ] ) , the specific energy of fluid particles does not change in the plunging region . however , the specific angular momentum of fluid particles does change . the specific angular momentum for particles on the marginally stable circular orbit , @xmath359 , can be calculated from eq . ( [ uf ] ) by setting @xmath184 . then , by comparing @xmath360 to @xmath359 , we can see how much has changed in the specific angular momentum as particles move from the marginally stable circular orbit to the black hole horizon . we have calculated the ratio @xmath361 and presented the results in fig . [ fig12 ] . to make the solutions smoothly joined to those in the disk region , in our calculations we have adopted that the specific energy is equal to that of a particle moving on the marginally stable circular orbit . then , we have @xmath362 . from eqs . ( [ car ] ) and ( [ ur0 ] ) , on the marginally stable circular orbit there is a one - to - one correspondence between @xmath363 and @xmath364 therefore , we use @xmath285 at @xmath365 to specify the boundary value of the magnetic field , then @xmath352 is determined ( up to a sign which is not relevant here ) . then , with the above approach , @xmath351 is determined as a function of @xmath285 and @xmath353 . we have chosen @xmath285 to be @xmath276 , @xmath277 , and @xmath278 , alternatively . from fig . [ fig12 ] we see that , though @xmath67 is not constant in the plunging region , its variation is extremely small , if on the marginally stable circular orbit the dynamical effects of the magnetic field on the motion of the fluid particles is negligible ( then @xmath366 at @xmath170 ) . as @xmath222 increases from @xmath367 to @xmath298 , the variation in the specific angular momentum decreases , caused by the fact that the coordinate distance between the marginally stable circular orbit and the event horizon decreases with increasing @xmath222 . for a black hole with @xmath368 , we have @xmath369 if @xmath370 on @xmath170 . for the extreme case of @xmath371 , we have @xmath372 , i.e. the specific angular momentum does not change at all . for the extreme case of @xmath373 , which has the largest coordinate distance from the marginally stable circular orbit to the event horizon , the variation in @xmath67 is largest . however , even in this case , the variation in the specific angular momentum is also not big : @xmath374 if @xmath375 on @xmath170 . for the same models we have also calculated the ratios @xmath285 and @xmath376 at @xmath171 , and presented the results in fig . [ fig13 ] . because of eq . ( [ eta1 ] ) , the value of @xmath285 is given by the left - hand side of eq . ( [ car ] ) . the ratio @xmath376 can be calculated by @xmath377 from fig . [ fig13 ] we see that both @xmath378 ( solid curves ) and @xmath376 ( dashed curves ) are small at @xmath171 . for @xmath379 , both are @xmath380 if @xmath370 at @xmath170 . the ratios go down as @xmath222 increases . even for the extreme case of @xmath381 , the ratios are also not big at @xmath171 if @xmath370 at @xmath184 : @xmath382 and @xmath383 . eq . ( [ urh ] ) implies that @xmath384 is always finite . then , @xmath385 at @xmath386 since @xmath387 . then , eq . ( [ crr ] ) implies that @xmath388 also at @xmath171 since its right - hand side is finite . however , from eq . ( [ crur ] ) we have @xmath389 since @xmath390 and @xmath391 . since @xmath384 is finite , eqs . ( [ car ] ) and ( [ eta1 ] ) imply that @xmath285 is also finite at @xmath171 . then , since @xmath392 at @xmath171 , we must have @xmath393 eqs . ( [ crurh ] ) and ( [ crurh2 ] ) imply that fluid particles always supersonically fall into the black hole . the numerical results in fig . [ fig13 ] confirm this conclusion . from figs . [ fig10 ] [ fig13 ] , we can demonstrate that when the boundary condition ( [ pc ] ) [ or , equivalently , ( [ psi0 ] ) ] is satisfied , the dynamical effects of magnetic fields in the plunging region are unimportant if they are so on the marginally stable circular orbit . with a simple model we have studied the evolution of magnetic fields in the plunging region around a kerr black hole . the model contains the following assumptions : ( 1 ) the background spacetime is described by the kerr metric ; ( 2 ) the plasma is perfectly conducting so that the magnetic field is frozen to the plasma fluid ; ( 3 ) the kinematic approximation @xcite applies , i.e. the dynamical effects of the magnetic field on the fluid motion are negligible so that the plasma fluid flows along timelike geodesics toward the central black hole ; ( 4 ) in a small neighborhood of the equatorial plane ( i.e. , @xmath84 ) the magnetic field and the velocity field have only radial and azimuthal components . the assumption ( 4 ) implies that @xmath85 and @xmath394 on the equatorial plane . with above assumptions , we have exactly solved maxwell s equations for axisymmetric solutions ( i.e. @xmath83 ) . the solutions are given by eqs . ( [ fsol1 ] ) and ( [ fsol2 ] ) [ or , equivalently , eqs . ( [ br ] ) and ( [ bf ] ) ] , where @xmath97 is a constant measuring the magnetic flux through a circle in the equatorial plane , @xmath114 is a function determining the time - evolution of the magnetic field . both @xmath97 and @xmath114 are determined by the boundary conditions . the dependence of the solutions on the coordinate time @xmath122 is given by the function @xmath395 , where @xmath396 is defined by eq . ( [ tau ] ) . the function @xmath114 is uniquely determined by the initial and boundary conditions of the magnetic field . the general form of @xmath114 determines the retarded nature of the solutions : at any time the state of the magnetic field at a radius in the plunging region is determined by the state of the magnetic field at a larger radius and an earlier time ( figs . [ fig1 ] and [ fig2 ] ) . this suggests that the stationary state of the magnetic field in the plunging region is uniquely determined by the boundary conditions at the marginally stable circular orbit . examples for the evolution of magnetic fields in the plunging region are shown in figs . [ fig3 ] [ fig8 ] , for both the stationary case ( @xmath198 , figs . [ fig3 ] [ fig5 ] ) and the non - stationary case ( @xmath397 , figs . [ fig6 ] [ fig8 ] ) . from these figures we see that , the boundary value of the radial component of the magnetic field at the marginally stable circular orbit is more important in determining the strength of the magnetic field in the plunging region than the toroidal component . the initially toroidal component is attenuated in the plunging region by the radial expansion of the fluid ( fig . [ fig3 ] ) . the initially radial component is amplified by the azimuthal and vertical compression ( the convergence of the fluid ) , and a toroidal component is generated from the radial component by the shear motion of the fluid though the toroidal component is initially zero at the marginally stable circular orbit ( fig . [ fig4 ] ) . this leads to the amplification of the magnetic field in the plunging region ( fig . [ fig5 ] ) . the evolution of the magnetic field depends on the spinning state of the black hole , but is more sensitive to the initial value of the radial velocity on the marginally stable circular orbit [ or , equivalently the parameter @xmath208 defined by eq . ( [ le0 ] ) ] . the time - evolution of magnetic fields shown in figs . [ fig6 ] [ fig8 ] confirms our claim that the stationary state of the magnetic field in the plunging region is uniquely determined by the boundary conditions on the marginally stable circular orbit . if in the plunging region a magnetic field is initially deviated from the stationary solutions , it will evolve to and finally get saturated at the state given by the stationary solutions in a dynamical time scale determined by the free - fall motion of fluid particles in the plunging region . if the magnetic field on the marginally stable circular orbit is in a stationary state , the magnetic field in the plunging region will automatically settle into a stationary state . thus , the evolution of magnetic fields in the plunging region is very different from that in the disk region , in the latter case the state of the magnetic field is determined by the local mhd processes . the difference is caused by the following fact : in the disk region the fluid has a very small radial velocity so that local mhd processes have shorter time scales than the radial motion ; in the plunging region the fluid has a large radial velocity so that local mhd processes usually have longer time scales than the radial motion . in deriving the solutions we have assumed that the dynamical effects of the magnetic field in the plunging region are unimportant [ the kinematic approximation adopted in assumption ( 3 ) ] . to justify this assumption , we have studied the dynamical effects of magnetic fields in the stationary state in two ways . first , we estimate the dynamical effects of magnetic fields by considering the back - reaction : substituting the solutions of maxwell s equations we obtained , where we assumed that fluid particles move on geodesics , into the dynamical equations to check if the motion of the fluid is significantly affected by magnetic fields . in this way , we estimate the dynamical effects of magnetic fields on the motion of the fluid by calculating the parameters @xmath285 , @xmath286 and @xmath287 defined by eq . ( [ eta ] ) . the mass density of the fluid , which appears in the denominators of @xmath285 , @xmath286 and @xmath287 , drops quickly in the plunging region if @xmath305 on the marginally stable circular orbit ( fig . [ fig9 ] ) . however , the evolution of the magnetic field , which appears in the numerators of @xmath285 , @xmath286 and @xmath287 , sensitively depends on the orientation of the magnetic field on the marginally stable circular orbit , the latter is essentially determined by the mhd processes in the disk region . if we require that the solutions in the plunging region are smoothly joined to the solutions in the thin keplerian disk region , then it is reasonable to assume that on the marginally stable circular orbit , as well as in the disk region , the magnetic field is parallel to the velocity field : @xmath398 ( see the appendix ) . this implies that for the solutions in the plunging region we should have @xmath307 [ eq . ( [ psi0 ] ) ] , i.e. the orientation of the magnetic field follows the orientation of the velocity field of the fluid [ eq . ( [ pc ] ) ] . with such a boundary condition , we have calculated @xmath285 , @xmath286 and @xmath287 , assuming that particles move on geodesics . we see that , if they are @xmath299 on the marginally stable circular orbit , then they remain so in the plunging region ( figs . [ fig10 ] and [ fig11 ] ) . indeed , @xmath285 , @xmath286 and @xmath287 decrease in the plunging region for sufficiently small @xmath225 on the marginally stable circular orbit . second , we self - consistently solve the coupled maxwell and dynamical equations on the horizon of the black hole , check how much has changed in the specific angular momentum of fluid particles since they leave the marginally stable circular orbit . the specific energy of fluid particles is not changed by a magnetic field that satisfies @xmath307 [ see eq . ( [ fe ] ) ] . however , the specific angular momentum does change [ see eq . ( [ fl ] ) ] . we have calculated the specific angular momentum of fluid particles as they reach the horizon of the black hole , assuming that the fluid particles pass the fast critical point near the marginally stable circular orbit . the deviation from the specific angular momentum as the particles just leave the marginally stable circular orbit is shown in fig . [ fig12 ] . we see that , though the specific angular momentum of fluid particles is changed by the magnetic field , the effects are always small assuming that on the marginally stable circular orbit the dynamical effects of the magnetic field are not important . for example , for @xmath399 we have @xmath369 if @xmath400 on the marginally stable circular orbit . we have also calculated the ratios @xmath401 and @xmath402 on the horizon ( fig . [ fig13 ] ) , where @xmath403 is the radial component of the alfvn velocity . these two ratios are relevant to the critical points in the flow [ see eqs . ( [ crur ] ) , ( [ car ] ) , and ( [ eta1 ] ) ] and measure at what level the magnetic field affects the motion of the fluid . [ fig13 ] shows that their values on the horizon are quite small . both approaches ( back - reaction and self - consistent solutions ) confirm that the dynamical effects of the magnetic field are unimportant in the plunging region if they are so on the marginally stable circular orbit . our results differ from that of gammie @xcite , in which he claimed that in the plunging region the dynamical effects of magnetic fields can be important . the difference is caused by the fact that gammie used a boundary condition that is different from ours for the magnetic field . gammie assumed that @xmath404 , where @xmath405 is the angular velocity of the disk at the marginally stable circular orbit , while we assume that @xmath307 . gammie s boundary condition implies that @xmath406 at @xmath170 ( as clearly stated in his paper ) , which makes @xmath170 a singular point where @xmath407 ( to keep the mass flux @xmath408 nonzero ) . hence , gammie s solutions are not well - behaved at the marginally stable circular orbit . our boundary condition allows the solutions in the plunging region to be smoothly joined to the solutions in the disk region ( see the appendix ) , since in our solutions all physical quantities are finite at the marginally stable orbit . certainly , to precisely take care of the transition from the disk region to the plunging region , gas pressure and non - electromagnetic stress must be properly taken into account near the inner edge of the disk . as mentioned in the introduction , recently the `` no - torque inner boundary condition '' for thin accretion disks has been challenged by some authors ( including gammie ) based on their studies on the evolution of magnetic fields in the plunging region @xcite . the results in this paper suggest that in the plunging region the dynamical effects of the magnetic field are not important , if the solutions in the plunging region are smoothly joined to the the solutions in the thin keplerian disk region . thus , the argument against the `` no - torque inner boundary condition '' is not founded . finally , we note that in the paper we have neglected dissipative processes like magnetic reconnection and ohmic dissipation , and the evaporation effect arising from the magnetic buoyancy and mhd instabilities . these processes operate in disks to limit the amplification of magnetic fields ( * ? ? ? * ; * ? ? ? * ; * ? ? ? * and references therein ) . certainly it is conceivable that they can also operate in the plunging region to reduce the amplification effect of magnetic fields . if these processes are important , the results of this paper tend to overestimate the amplification of magnetic fields in the plunging region . then , the dynamical effects of magnetic fields should be weaker than that we have estimated without considering dissipative processes , which will strengthen our conclusion that the dynamical effects of magnetic fields are unimportant in the plunging region . in the disk region ( @xmath409 ) , viscosity plays an important role in the dynamics of disk particles . the viscosity transports angular momentum outward and dissipates energy , which leads to disk accretion . so , in the disk region , particles move on non - geodesic worldlines . in the vertical direction , the gravity of the black hole is balanced by the gradient of the total pressure (= gas pressure + radiation pressure + magnetic field pressure ) in the disk . therefore , in a small neighborhood of the disk central plane , disk particles are more likely moving on planes parallel to the equatorial plane , rather than moving radially as in the plunging region . to describe such a motion , cylindrical coordinates @xmath410 are more suitable , where @xmath122 and @xmath411 have the same meaning as those in the boyer - lindquist coordinates , to the first order @xmath118 also has the same meaning as that in the boyer - lindquist coordinates , but @xmath412 where @xmath413 . in the cylindrical coordinates , in a small neighborhood of the disk central plane the kerr metric can be written as @xcite @xmath414 where @xmath415 and @xmath160 are defined by eq . ( [ del ] ) with @xmath416 . the maxwell equations that we need to solve are again given by eq . ( [ maxeq ] ) , but now we have @xmath417 and @xmath418 . as in the case for the plunging region , we assume that the motion of fluid particles is stationary but the evolution of the magnetic field can depend on time , i.e. , we let @xmath419 but keep @xmath420 in the maxwell equations . furthermore , we assume that in a small neighborhood of the disk central plane @xmath421 so that @xmath422 on the equatorial plane . finally , we adopt @xmath423 because of the axisymmetry of the system . then , on the equatorial plane eq . ( [ maxeq ] ) is reduced to @xmath424 + \frac{\partial}{\partial r } \left[r ( u^r b^\beta - u^\beta b^r)\right ] = 0 \;. \label{eq1a}\end{aligned}\ ] ] eq . ( [ fsol1d ] ) can be written as @xmath428 assume that as @xmath429 the disk is keplerian so that @xmath430 and @xmath431 , and @xmath432 keeps finite , then we must have @xmath433 eq . ( [ limp ] ) has an important implication for stationary solutions . for stationary solutions we have @xmath434 , then by eq . ( [ limp ] ) we must have @xmath435 . so , for stationary solutions we have the retarded nature of the solution given by eq . ( [ psol ] ) . a time variation in the magnetic field at radius @xmath124 will lead to the same variation in the magnetic field at radius @xmath439 at a later time , assuming the fluid moves toward smaller radii ( i.e. toward the central black hole ) . the time delay is given by the time needed by a fluid particle moving from @xmath124 to @xmath440.,width=453 ] a sketch of the solution given by eq . ( [ phsol ] ) , corresponding to the boundary conditions given by eq . ( [ bond ] ) . the parameter @xmath131 is defined by eq . ( [ tau ] ) , which is the coordinate time needed by a particle moving from radius @xmath124 to radius @xmath118 . so , @xmath133 corresponds to @xmath117 , @xmath441 corresponds to the black hole horizon @xmath171 . the coordinate time is represented by @xmath122 . in this diagram , the worldlines of fluid particles are represented by straight lines @xmath442 . if the boundary conditions are imposed on the axes as shown in the diagram [ eq . ( [ bond ] ) ] , the solution in the region between the @xmath131-axis and the dashed line @xmath443 ( i.e. , region i ) is determined by the boundary condition on the @xmath131-axis , the solution in the region between the @xmath122-axis and the dashed line ( i.e. , region ii ) is determined by the boundary condition on the @xmath122-axis.,width=453 ] stationary evolution of an initially toroidal magnetic field in the plunging region around a kerr black hole . each panel corresponds to a different spinning state of the black hole , as indicated by the dimensionless parameter @xmath222 . each curve starts from the marginally stable circular orbit ( @xmath170 , right end ) , ends at the horizon of the black hole ( @xmath171 , left end ) . at @xmath170 , the magnetic field is purely toroidal and has a value @xmath200 , the fluid particles have specific energy identical to that of the marginally stable circular orbit , and specific angular momentum smaller than that of the marginally stable circular orbit by a tiny fraction @xmath208 [ eq . ( [ le0 ] ) ] . in each panel , each curve corresponds to different values of @xmath208 : @xmath276 , @xmath277 , @xmath278 and @xmath279 ( downward ) . the radial component of the magnetic field is always zero . , width=566 ] similar to fig . [ fig3 ] but the magnetic field is initially radial , i.e. @xmath216 and @xmath444 at @xmath170 . solid lines represent the toroidal component of the magnetic field , dashed lines represent the radial component . though initially the toroidal component of the magnetic field is zero , the shear motion of the fluid in the plunging region generates toroidal component from the radial component . the evolution of the radial component does not depend on the initial kinetic state of the fluid , so there is only one dashed line in each panel . the four solid lines representing the toroidal component of the magnetic field correspond to different values of @xmath208 : @xmath276 , @xmath277 , @xmath278 and @xmath279 ( upward).,width=566 ] stationary evolution of @xmath445 ( solid curves ) . the boundary conditions are the same as that in fig . the thin dashed line in each panel shows the absolute value of the radial velocity ( i.e. @xmath225 ) of the fluid , corresponding to @xmath282 . if we choose a different value of @xmath208 , the curve for the radial velocity will change slightly : the right end will change according to @xmath446 , the left end always approaches @xmath298 ( i.e. the speed of light ) . ( for brevity , @xmath447 $ ] and @xmath225 use the same scale as labeled on the left side of the box in each panel.),width=566 ] time evolution of magnetic fields around a schwarzschild black hole ( i.e. @xmath165 ) . each panel corresponds to a particular moment . thick solid lines show the toroidal component of the magnetic field ( i.e. , @xmath155 ) , thick dashed lines show the radial component ( i.e. , @xmath154 ) . the thin ( upper ) lines show the corresponding stationary state solutions . the fluid particles have specific energy identical to that of the marginally stable circular orbit , and specific angular momentum smaller than that of the marginally stable circular orbit by a factor of @xmath253 . the first ( left and up ) panel shows the initial and boundary conditions of the magnetic field : on @xmath170 ( right end ) , @xmath217 , @xmath216 for all time @xmath128 ; in the plunging region ( @xmath241 ) , @xmath217 at @xmath448 [ i.e. , eqs . ( [ bond1 ] ) and ( [ bond2 ] ) ; the corresponding @xmath154 at @xmath125 is automatically determined by the solutions ] . the figures show that , as time goes on , the magnetic field quickly approaches and saturates at the state given by the stationary solutions . the black dot in each panel shows the radial position of a fluid particle at each moment : initially the particle is at @xmath170.,width=566 ] stationary evolution of the mass density in the plunging region around a kerr black hole . each panel corresponds to a different spinning state of the black hole , as indicated by the values of @xmath222 . the right end of each curve corresponds to the marginally stable circular orbit ( @xmath449 ) . the left end of each curve corresponds to the horizon of the black hole ( @xmath171 ) . the kinetic boundary conditions are given by eq . ( [ le0 ] ) . in each panel the four curves correspond respectively to @xmath223 , @xmath277 , @xmath278 and @xmath279 ( downward ) . the mass density is in units of @xmath450 the mass density at the marginally stable circular orbit.,width=566 ] stationary evolution of the ratio @xmath451 in the plunging region . each panel corresponds to a different spinning state of the black hole , as indicated by the values of @xmath222 . the kinetic boundary conditions are given by eq . ( [ le0 ] ) . the boundary conditions for the magnetic field are : on @xmath452 we have @xmath315 in units of @xmath453 . the corresponding @xmath154 is calculated with eq . ( [ pc1 ] ) so that the magnetic field is always parallel to the velocity field . in each panel the four curves correspond respectively to @xmath223 , @xmath277 , @xmath278 and @xmath279 ( downward ) . each curve starts from the marginal stable circular orbit ( right end ) and ends at the horizon of the black hole ( left end).,width=566 ] the change in the specific angular momentum of fluid particles as they approach the event horizon of the black hole [ eq . ( [ dell ] ) ] , as a function of the spin of the black hole . it is assumed that the fast critical point is at @xmath456 . the magnetic field is parallel to the velocity field of the fluid , then the specific energy of fluid particles keeps constant in the plunging region , which we have assumed to be equal to the specific energy of a particle moving on the marginally stable circular orbit . the boundary value of the magnetic field at @xmath184 is specified by the ratio @xmath457 . the three curves corresponds to three different values of @xmath285 at @xmath170 : @xmath276 , @xmath277 , and @xmath278 ( downward ) . , width=566 ]

the evolution of magnetic fields frozen to a perfectly conducting plasma fluid around a kerr black hole is investigated . we focus on the plunging region between the black hole horizon and the marginally stable circular orbit in the equatorial plane , where the centrifugal force is unable to stably balance the gravitational force . adopting the kinematic approximation where the dynamical effects of magnetic fields on the fluid motion are ignored , we exactly solve maxwell s equations with the assumptions that the geodesic motion of the fluid is stationary and axisymmetric , the magnetic field has only radial and azimuthal components and depends only on time and radial coordinates . we show that the stationary state of the magnetic field in the plunging region is uniquely determined by the boundary conditions at the marginally stable circular orbit . if the magnetic field at the marginally stable circular orbit is in a stationary state , the magnetic field in the plunging region will quickly settle into a stationary state if it is not so initially , in a time determined by the dynamical time scale in the plunging region . the radial component of the magnetic field at the marginally stable circular orbit is more important than the toroidal component in determining the structure and evolution of the magnetic field in the plunging region . even if at the marginally stable circular orbit the toroidal component is zero , in the plunging region a toroidal component is quickly generated from the radial component by the shear motion of the fluid . finally , we discuss the dynamical effects of magnetic fields on the motion of the fluid in the plunging region . we show that the dynamical effects of magnetic fields are unimportant in the plunging region if they are negligible on the marginally stable circular orbit . this supports the `` no - torque inner boundary condition '' of thin disks , contrary to the claim in the recent literature .

in unified models for radio sources , bl lac objects are the low - power radio galaxies whose jets are the most highly beamed along our line of sight ( e.g. browne 1983 ; antonucci & ulvestad 1985 ; ulrich 1989 ; urry & padovani 1995 ) . their rapid variability , apparent superluminal motion , strong point - like emission in radio , optical and x - ray , and the detection of some sources in @xmath2-rays , are all explained if we are seeing emission from a relativistic jet closely aligned with our line of sight . low - power radio galaxies then represent the ` parent population ' of unaligned objects whose jets are less favourably aligned . these low - power radio galaxies are likely mostly to be fanaroff & riley ( 1974 ) class i ( fri ) objects , but the parent population may also include some transitional objects and low - excitation frii radio galaxies . an important test of this unified model is the degree to which the isotropic ( alignment - independent ) properties of bl lacs are similar to those of the parent population of radio galaxies . such tests have been made , on the whole successfully , by looking at the extended radio emission ( e.g. antonucci & ulvestad 1985 ; kollgaard et al . 1992 ; perlman & stocke 1993 , 1994 ) and properties of the host galaxies ( e.g. ulrich 1989 ; abraham , mchardy & crawford 1991 ; wurtz , stocke & yee 1996 ; falomo 1996 ) although there is some evidence that there are too few bl lacs associated with the most luminous host galaxies ( wurtz et al . 1996 ) . another isotropic indicator is the clustering environment . using two - point correlation analysis of optical fields , it has been shown that fri radio galaxies are normally found in groups or clusters of galaxies ( longair & seldner 1979 ; prestage & peacock 1988 ) and bl lacs seem also to inhabit groups or poor clusters ( pesce , falomo & treves 1995 ; smith , odea & baum 1995 ; wurtz , stocke & ellingson 1997 ) though it appears that , at least at low redshift , bl lacs are not often found in the dominant galaxies of rich clusters ( prestage & peacock 1988 ; owen , ledlow & keel 1996 ; wurtz et al.1997 ) ; for example , owen et al . ( 1996 ) find no bl lacs at the centres of abell clusters , a result inconsistent at the 95% confidence level with the numbers expected from the unified models of urry & padovani . clustering environment may also be investigated by x - ray observations . it has long been known that many objects close to the fri - frii luminosity boundary are associated with rich clusters having luminous x - ray haloes . recent observations with _ rosat _ have shown that more typical fri radio galaxies have extended thermal x - ray emission on scales characteristic of groups or poor clusters ( worrall & birkinshaw 1994 ) . this offers a new way to test the unification hypothesis ; such emission should be isotropic , and so we expect all bl lacs to have x - ray haloes comparable to those of fris . this test is difficult because it requires us to separate any extended x - ray emission from the bright unresolved emission of the bl lac nucleus . in this paper we describe such an analysis of _ rosat _ observations of the bl lac pks 0521@xmath1365 . pks 0521@xmath1365 is a well - studied bl lac with a redshift of 0.055 , comparable to the redshifts of the radio galaxies studied by worrall & birkinshaw ( 1994 ) . it is variously described in the literature as a blazar , a bl lac object , or an n - galaxy , and on multifrequency spectral index plots like those of sambruna , maraschi & urry ( 1996 ) is placed among radio - selected bl lacs . its host galaxy is easily detectable in the optical [ contributing @xmath3 per cent of the source luminosity at 5500 in an @xmath4 arcsec effective aperture ; falomo , scarpa & bersanelli ( 1994 ) ] and it exhibits strong , variable broad emission lines ( scarpa , falomo & pian 1995 ) . the host galaxy is a giant elliptical ( wurtz et al . pesce et al . ( 1995 ) suggest that the excess galaxy count around the object corresponds to a cluster of abell richness 0 or more ; they identify at least one , and up to four associated galaxies within 100 kpc . however , the cross - correlation analysis of wurtz et al . ( 1997 ) suggests a poorer cluster , with richness class @xmath5 . [ unbeamed ] in the radio , the source has a 408-mhz flux of 36.1 jy ( wright & otrupcek 1990 ) , corresponding to a power at that frequency of @xmath6 w hz@xmath7 sr@xmath7 ; this places it slightly above the nominal fri - frii luminosity boundary ( @xmath8 w hz@xmath7 sr@xmath7 at this frequency ) , though of course some of the 408-mhz emission is contributed by the core , presumed to be beamed . it exhibits a core - halo - hotspot morphology on arcsecond scales ( wardle , moore & angel 1984 ; ekers et al . 1989 ; see also section [ discuss ] ) , which , together with its comparatively high radio power , may suggest that it is an aligned version of a transitional fri - frii object . the prominent radio jet is also seen in optical synchrotron emission , extending about 6 arcsec from the nucleus ( e.g.keel 1986 , macchetto et al . no motion of the core components was detected in vlbi observations ( tingay et al . 1996 ) and this , together with the comparatively low ratios of nuclear to host - galaxy optical emission ( falomo et al . ) and radio core to extended radio flux ( antonucci & ulvestad 1985 ) , suggests a source that is only moderately relativistically boosted along the line of sight compared to the most extreme bl lacs . it was for this reason that we selected it as a suitable candidate for an x - ray search for extended emission with the _ rosat _ hri . pks 0521@xmath1365 has already been extensively observed at x - ray wavebands , with _ einstein _ ( worrall & wilkes 1990 ) , _ exosat _ ( sambruna et al . 1994 ) and the _ rosat _ pspc ( pian et al . 1996 ) , and was detected in @xmath2-rays by _ egret _ ( lin et al . 1996 ) , but none of these observations provides the resolution necessary to separate compact from extended structure . the source appears to have a concave spectrum in the x - ray waveband ; the energy index is @xmath9 in the soft band , flattening to @xmath10 at higher energies . throughout the paper we use a cosmology in which @xmath11 km s@xmath7 mpc@xmath7 , @xmath12 . at the distance of pks 0521@xmath1365 1 arcsec corresponds to 1.475 kpc . we observed pks 0521@xmath1365 with the _ rosat _ hri for 46.6 ks between 1995 feb 09 and 1995 feb 11 . the data were analysed using the iraf post - reduction off - line software ( pros ) . pks 0521@xmath1365 is detected at @xmath13 net counts in a 2.5-arcmin radius circle with background taken from an annulus between 2.5 and 3.5 arcmin . the source circle was chosen by expanding the radius until the background - subtracted counts enclosed stayed constant within the errors . a 25-arcsecond radius circle around a nearby point source ( 1 arcmin to the north ) was excluded from the analysis . the detected counts correspond to a rate of @xmath14 counts s@xmath7 . the source centroid is consistent to within a few arcseconds with the best available optical and radio positions . [ var ] a simple analysis , binning the data into 1-ks intervals , suggests that the source was not strongly variable during the 110 ks that contain our observations . to compare the source flux with the earlier _ rosat _ pspc observations , we re - analysed the 4.8 ks of pspc data from the _ rosat _ public archive , taking account of the nearby background sources . we detect the source at @xmath15 pspc counts in the same source region , or @xmath16 counts s@xmath7 . a spectral fit to a power - law model between 0.1 and 2.4 kev using the galactic @xmath17 ( @xmath18 @xmath19 ) of elvis , lockman & wilkes ( 1989 ) gives an energy spectral index of @xmath20 ( @xmath21 with 32 d.o.f . ; errors are @xmath22 for one interesting parameter ) , consistent with the results of pian et al.(1996 ) and sambruna ( 1997 ) ; the unabsorbed 1-kev flux density from the pspc data is then @xmath23 @xmath24jy , while the flux density from the hri data on the same assumptions is @xmath25 @xmath24jy . if @xmath17 is allowed to vary the best - fit spectral index is @xmath26 with @xmath27 @xmath19 ( @xmath28 with 31 d.o.f . ; errors @xmath22 for two interesting parameters ) . incorporating the spectral uncertainties , the flux in the 0.12.4 kev _ rosat _ band seems to be @xmath29 per cent lower in the hri data than it was at the epoch of the pspc observations ( 1992 aug 29 ) . we return to this point below . fig . [ contour ] ( left ) shows that the source is approximately symmetrical ; radial profiling is therefore appropriate . to search for extension in pks 0521@xmath1365 we first compared its radial profile with the standard empirical expression ( david et al . 1997 ) for the hri point response function ( prf ) . it will be seen from fig . [ radial ] ( left ) that the radial profile is not well fitted by a standard _ prf ( normalised to the central regions , which we expect to be dominated by the point source ) ; there is a deficit of counts at small radii and an excess of counts at large ones . it is well known that images from the hri can be badly affected by errors in the _ rosat _ aspect correction ; the effect is to produce spurious elongation ( ` smearing ' ) in the images of point sources , thus broadening the effective , azimuthally - averaged prf . such smearing is likely to be significant compared to the detector and instrument point response on scales of @xmath30 arcsec . as shown in fig . [ contour ] , the source appears elongated , in its central regions , in position angle roughly @xmath31 ; aspect smearing could therefore be responsible for some of the deviations from the nominal prf shown in fig . [ radial ] . the aspect errors are thought to arise because the star tracking system does not always calculate positions correctly , due to variations in the gains of the pixels of the star tracker ccd . the spacecraft wobble exacerbates this problem because it causes the guide stars to move across the tracking ccd on short timescales . methods for correcting aspect smearing ( e.g. morse 1994 ) therefore rely on binning the data according to the phase of the wobble ; so long as the satellite roll angle and the properties of the ccd are constant , the aspect error should be a function of the wobble phase only . to do this , we applied iraf / pros scripts provided by j. silverman and d. harris , based on suggestions by g. hasinger ( for more details see harris et al . we first established that all our data were taken at the same roll angle ; this , along with the relatively small number of closely spaced observation intervals , allowed us to analyse all the data together . the scripts bin the data as a function of wobble phase , utilising the fact that the relative wobble phase at any time is given simply by the spacecraft clock ( modulo clock resets , none of which occured during our observation ) . we chose to divide the data into 20 phase bins , giving approximately 500 counts per binned observation ; with 20 phase bins there are enough counts for centroiding in each bin , and examination of images derived from each bin showed them to be circular . a centroid was then found , using the _ detect _ suite of tasks within pros , for each individual bin , and the images were restacked so that the centroids aligned . the resulting image ( fig . [ contour ] , right ) appears less elliptical than the uncorrected image ( fig . [ contour ] , left ) and is considerably better fit , in its inner regions , by the nominal hri prf ( figure [ radial ] , right ) ; it appears that the stacking procedure has made the core slightly narrower than the nominal prf , which is of course derived from observations that may themselves have been affected by aspect uncertainties . we conclude that the inner regions of pks 0521@xmath1365 are well described as a point source . more importantly , even after dewobbling we still see an excess of counts over the prf at radii of 10 - 40 arcsec . there is no reason to doubt that this corresponds to real extended emission . to characterise the scale size of this extended emission we fit models consisting of a @xmath32-model ( sarazin 1986 ) and a point - source component , both convolved with the nominal prf , to the background - subtracted radial profile of the dewobbled image . to avoid large contributions to the fitting statistic due to the fact that the restacked data are narrower than the nominal prf , we sampled the radial profile more coarsely in its central regions for these fits . we performed fits for various different values of the parameter @xmath32 and the core radius . figure [ chi2 ] shows a plot of @xmath33 as a function of @xmath32 and core radius ; the best fit ( @xmath34 with 15 degrees of freedom ) has @xmath35 and a core radius of 8 arcsec , with a central normalisation of the @xmath32-model of @xmath36 counts arcsec@xmath37 . figure [ radialmod ] shows the best - fit model plotted with the data . fitting a point source alone to the data gives unacceptably high values of the fitting statistic ( @xmath38 ) . for comparison , we fitted a similar range of @xmath32-models to the archival pspc dataset . because of the broader intrinsic prf of the pspc , smearing is not considered to be a problem with this instrument . the best - fitting point source and @xmath32-model had a core radius of 35 arcsec with @xmath35 , as shown in fig . the data are better fit with this model than with a point - source model alone ( best - fit @xmath39 with 9 degrees of freedom ) , although the point - source fit is formally acceptable ( @xmath40 with 10 degrees of freedom ) . a large range of possible core radii are acceptable , and the hri best - fitting @xmath32 model is among those consistent with the pspc data within the 90 per cent confidence contour ; the normalisations of the models are also consistent . if an appropriately normalised @xmath32-model with core radius 8 arcsec and @xmath41 is subtracted from the pspc data the remaining emission is adequately modelled as a point source with no significant support for an additional extended component . we therefore conclude that the extended emission is well represented by the hri best - fit model . the best - fit point - source count rates in this model in the pspc and hri data are consistent to within a few per cent , suggesting that the inconsistency between the total pspc and hri count rates ( section [ var ] ) is a result of our not having taken into account the two spectral components corresponding to compact and extended emission . the @xmath32-model fits to the hri data described above imply that the extended x - ray emission has a core radius of 12 kpc and contributes @xmath42 counts to the total in a 2.5-arcmin radius circle . the corresponding 0.2 - 1.9 kev luminosity , assuming a raymond - smith model with @xmath43 kev and using galactic @xmath17 , is approximately @xmath44 w. this is considerably more luminous than the extended emission seen in most of the radio galaxies of worrall & birkinshaw ( 1994 ) , although the fwhm of 14 kpc is comparable to some of the smaller objects . the central cooling time , on the same temperature assumption , is @xmath45 yr , which , if the emission were thermal , would imply rapid cooling , with expected mass deposition rates of @xmath46 @xmath47 yr@xmath7 . the radius at which the cooling time is comparable to the hubble time would be @xmath48 kpc , larger than the core radius , which means that a simple single - temperature model of the extended emission is not self - consistent . cooling takes place for any reasonable choice of temperature of the x - ray gas . the extended x - ray component detected by the hri is comparable in size to the extended radio emission in pks 0521@xmath1365 which might be thought to suggest an explanation in terms of inverse - compton ( ic ) scattering by the energetic electrons in the radio lobes of photons from the cosmic microwave background radiation ( cmb ) and from the bl lac nucleus itself ( cf . brunetti , setti & comastri 1997 ) . [ overlay ] shows the x - ray image superposed on a 1.4-ghz vla radio map made from data supplied by g.g . since the flux on the shortest baselines of this radio observation ( made at bna configuration ) is equal to the single - dish parkes radio flux ( wright & otrupcek 1990 ) of 16.3 jy , we can be confident that the extended structure is well represented . the largest angular size ( las ) of the radio emission is about 50 arcsec , whereas the fwhm of the extended x - ray component is 10 arcsec , and fig . [ overlay ] shows that the radio emission is quite asymmetrically distributed around the nucleus , in contrast to the symmetry of the x - ray emission . however , we note that inverse - compton emission would come preferentially from the parts of the radio lobes closest to the active nucleus , so this in itself does not make an inverse - compton model implausible . since we do not know the unprojected geometry of the source , we model the extended radio - emitting component crudely as a uniform sphere centred around the nucleus . the flux density from the core is 2.7 jy [ consistent with the value of antonucci & ulvestad ( 1985 ) ] leaving 13.6 jy in extended emission . projection effects make it difficult to assess the true size ( and therefore volume ) of the lobes , but the flux expected from ic scattering from an object of a given synchrotron ( radio ) flux is only weakly dependent on the object s size , and so we assume a radius for the sphere of half the las . because the frequency of a scattered photon is raised by approximately the square of the lorentz factor ( @xmath2 ) of the scattering electron , and because the electron energy spectrum is dominated by electrons with @xmath49 [ i.e. electrons near the assumed low - energy cutoff in the energy distribution carilli et al . ( 1991 ) ] the most significant contribution in the x - ray band from the ic process is due to far - ir photons . we accordingly model the spectrum of the bl lac s nucleus in this region as a power law with spectral index 0.78 normalised to the measured 100-@xmath24 m flux of impey & negebauer ( 1988 ) . pian et al.(1996 ) argue that the doppler factor in the nucleus is of order unity , based on the relatively low radio core prominence discussed in section [ unbeamed ] and on models for the nuclear x - ray and @xmath2-ray emission , and we adopt this value in order to infer a luminosity from the observed flux ; we neglect the effects of possible circumnuclear obscuration , so that our source model is effectively an isotropically radiating central source surrounded by a uniform , spherically symmetrical electron distribution . using a modified version of the code of hardcastle , birkinshaw & worrall ( 1998a ) we find that the ic process in this model can not produce enough x - rays to account for the observed flux unless the lobes have magnetic field strength approximately an order of magnitude weaker than the value implied by equipartition of energy between magnetic field and relativistic electrons ( for electron filling factor unity ) ; this is relatively much weaker than fields that have been inferred in other sources from observations of x - ray emission attributed to inverse - compton or synchrotron - self - compton emission ( harris et al . 1994 ; feigelson et al . 1995 ; hardcastle et al . 1998a ; brunetti et al . 1998 ) . at equipartition , the x - rays from the ic process contribute at most 2 per cent of the observed flux , or less if the filling factor is lower or there is a significant contribution from relativistic protons . as a result of this simple analysis , we believe that it is unlikely that the extended x - rays seen around the source are due in large part to inverse - compton emission . instead , it seems more likely that they are indeed thermal emission from a rapidly cooling central region , and that pks 0521@xmath1365 inhabits a cooling flow ; this means that its environment is significantly different from those of the fris observed by worrall & birkinshaw ( 1994 ) , which tended to lie in less dense environments with much longer cooling times . this model is qualitatively consistent with the bright extended emission line region seen in pks 0521@xmath1365(boisson , cayatte & sol 1989 ) and with the strong polarization asymmetry seen in the radio observations . we attempt to fit the radial profile of the source with a cooling flow model based on that described by hardcastle , lawrence & worrall ( 1998b ) . in this version of the model , the outer regions of the source are fit with an isothermal @xmath32 model while the temperature and density inside the cooling radius ( @xmath50 ) are power law functions of radius . the electron density follows @xmath51 while the temperature is given by @xmath52 here @xmath53 is the core radius of the @xmath32 model , and @xmath54 is the inner limiting radius of the cooling flow , allowing us to avoid infinities at @xmath55 . @xmath56 , @xmath57 and @xmath58 are scale electron densities and @xmath59 , @xmath60 and @xmath61 scale temperatures ; @xmath61 corresponds to the temperature of the non - cooling gas . it is clear that matching temperatures and densities allows us to write all the scale factors in terms of @xmath58 and @xmath61 . the results are only weakly dependent on @xmath54 so long as it is small and we fix it at a value corresponding to 0.01 arcsec in what follows . the parameters @xmath62 and @xmath63 set the slope of the density and temperature distributions ; for a realistic cooling model we expect @xmath64 and @xmath65 , so that density increases and temperature decreases with decreasing radius . the ideal gas law implies that pressure goes as @xmath66 . if the atmosphere is required to be close to hydrostatic equilibrium , then matching mass as a function of radius inside and outside the cooling zone leads to a constraint on @xmath67 as a function of @xmath62 and @xmath63 : @xmath68 and if we assume some law for the radial dependence of pressure ( @xmath69 is consistent with observations of other cooling regions ) then the model has only five free parameters : @xmath32 , @xmath53 , @xmath62 , @xmath70 and @xmath58 , the last being a normalisation parameter that can be determined in the fit . in addition , such a model can only be physically realistic if the cooling time at @xmath50 is comparable to the system lifetime , or the hubble time : @xmath71 ( sarazin 1986 ) where @xmath72 is in years , @xmath73 in kev and @xmath74 in @xmath75 . we fit a range of representative cooling models to the data . as before , @xmath32 was chosen from a small number of possible values ( 0.5 , 0.667 , 0.9 ) while @xmath53 ranged from 11000 arcsec . we tried values of 1 , 1.5 , 2 , 5 and 10 kev for @xmath76 . @xmath62 was allowed to take the values 1.5 , 1.75 or 2.0 . no model consisting of a cooling flow alone was a good fit to the data . a number of models consisting of a cooling flow and a central point source were comparable in goodness of fit to the combination of @xmath32-model and point source discussed above , but most of these were ruled out by the constraint on cooling time . those that were not have small core radii , low external temperatures and steep temperature and density power laws : for example , the model with @xmath77 kev , @xmath78 , @xmath79 , @xmath80 arcsec is an acceptable fit to the hri radial profile ( @xmath81 ) . in this model just under 10 per cent of the total counts ( @xmath82 ) are assigned to the cooling flow component . the nominal cooling time at the core radius is @xmath83 years ; the mass deposition rate is approximately @xmath84 yr@xmath7 and the implied densities at 20 kpc correspond roughly with those required for pressure equilibrium with the cold gas inferred from observations of the extended emission - line region ( boisson et al . though the details of the model may not be correct , this shows that the extended emission of pks 0521@xmath1365 can plausibly be modelled as a cooling flow of this type . a strong nuclear source is still necessary ; the flux of the point - like component in the cooling - flow model is reduced by only @xmath85 per cent compared to that derived from the simple @xmath32-model fits of section [ analysis ] . we have carried out the first separation of nuclear and galaxy - scale extended x - ray emission in a bl lac object and found evidence that pks 0521@xmath1365 inhabits a dense and , presumably , rapidly cooling region of x - ray emission , a much more extreme environment than those found for ` typical ' fri radio galaxies ; we have shown that it can be modelled as a cooling flow in low - temperature cluster gas . if this result were extended to other bl lac objects , it would cause difficulties for models that seek to unify fris and bl lacs , and might imply some causal relationship between a dense and rapidly cooling atmosphere and the bl lac phenomenon . however , it may be that pks 0521@xmath1365 , with its high power and intermediate radio structure , is not representative of the bl lac class . further observations are planned both to verify the thermal nature of the galaxy - scale halo and to see whether pks 0521@xmath1365 is unusual among bl lacs in this respect . we are grateful to guy pooley for allowing us to use his vla observations of pks 0521@xmath1365 , and to the vla analysts for help in recovering the vla data from the archive . this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with nasa . the digitized sky surveys were produced at the space telescope science institute under u.s . government grant nag w-2166 . the national radio astronomy observatory is operated by associated universities inc . , under co - operative agreement with the national science foundation . this work was supported by nasa grant nag 5 - 2312 and pparc grant gr / k98582 . david l.p . , harnden f.r . , kearns k.e . , zombeck m.v . , harris d.e . , prestwich a. , primini f.a . , silverman j.d . , snowden s.l . , 1997 , rosat science data center report , available at url : @xmath86http://hea - www.harvard.edu / rosat / rsdc_www / hricalrep.html@xmath87 lin y.c . , bertsch d.l . , dingus b.l . , esposito j.a . , fichtle c.e . , hartman r.c . , hunter s.d . , kanback g. , kniffen d.a . , mayer - hasselwander h.a . , michelson p.f . , von montigny c. , mukherjee r. , nolan p.l . , scheid e. , sreekumar p. , thompson d.j . , 1995 , apj , 442 , 96 macchetto f. , albrecht r. , barbieri c. , blades j.c . , boksenberg a. , crane p. , deharveng j.m . , disney m.j . , jakobsen p. , kamperman t.m . , king i.r . , mackay c.d . , paresce f. , weigelt g. , baxter b. , greenfield p. , jedrzejewski r. , nota a. , sparks w.b . , 1991 , apj , 369 , l55 tingay s.j . , edwards p.g . , costa m.e . , lovell j.e.j . , mcculloch p.m. , jauncey d.l . , reynolds j.e . , tzioumis a.k . , migenes v. , gough r. , king e.a . , jones d.l . , preston r.a . , murphy d.w . , meier d.l . , van ommen t.d . , st . john m. , hoard d.w . , nicolson g.d . , wan t.-.s . , shen z.-.q . , 1996 , apj , 464 , 170

models that seek to unify bl lacs and low - power radio galaxies predict that the two types of object should show similar isotropically emitted x - ray emission . testing this is usually limited by difficulties in separating strong x - ray emission from a bl lac nucleus and surrounding low - surface brightness emission . in this paper we report rosat hri observations of the @xmath0 bl lac object pks 0521@xmath1365 . we are able to separate a luminous extended x - ray component from the bright nucleus . using a new radio map , we show that it is unlikely that the extended emission is due to inverse - compton scattering of photons from the active nucleus , and instead interpret it as thermal emission from dense , rapidly cooling gas . this is a more extreme environment than is found in typical fri radio galaxies , and may pose a problem for unified models . x - rays : galaxies galaxies : individual : pks 0521@xmath1365 galaxies : active

cepheid variables have been used to determine galactic distances for many decades ( e.g. @xcite , @xcite and references therein ) . the relationship between their period of pulsation and luminosity has been considered to be well defined at several times during its lifetime . however , the effect of the cepheid s chemical composition on the period - luminosity ( pl ) relation has never been adequately determined , leaving the relation open to systematic uncertainties that have not been fully investigated . proper calibration of this relation is essential not just to explain how metallicity affects the pulsations of cepheids , but to tie down the zero point of the extra - galactic distance scale and thus properly measure the hubble constant . metallicity and differential absorption will both change stellar colours @xcite . theoretical models ( e.g. ) show that the metal content of a cepheid will affect the derived distance modulus . if a star of an unknown metallicity is at an undetermined distance then it is difficult to untangle these two effects on the stars colours . ideally the pl relation would remove this problem as the distance to the star could be found without calculating the colour of the cepheid . the problem arises as the calibration of the pl relation does not take this effect into account . the distances that were used to calibrate the zero point of the relation were determined without considering how the reddening of the cepheids was affected by the combination of these two mechanisms , meaning that a systematic offset may have been introduced into the relation . there has been some investigation into the effect that metallicity has on the pl relation , beginning with @xcite . and used the cepheids detected in the eros gravitational lensing survey to show that discrepant values of @xmath7 could be reconciled by applying a metallicity correction to cepheid distance moduli . @xcite used the hubble space telescope to observe cepheids in m 101 , a nearby spiral galaxy with active star formation regions but low dust extinction properties . choosing a galaxy with low extinction means that there would be less reddening from dust ; any reddening of the colour of the cepheids could be explained as a metallicity effect . this study found that @xmath8 mag dex@xmath9 ( see also freedman et al . , 2001 ) . however , their conclusions were based on small numbers of cepheids , combined with the pl relations from @xcite . it is far from ideal to use such a small sample to quantify an effect that could mean that the entire extra - galactic distance scale is subject to a systematic offset . using a similar method but with a larger sample of cepheids , @xcite found @xmath10 mag dex@xmath9 . using a large sample of nearly 700 cepheids in 17 galaxies , @xcite had found a result compatible with these other works . most works use a sample of cepheids covering a range in metallicity and enforce consistency between distance moduli ; this is what @xcite and @xcite did for instance . an alternative approach used e.g. by @xcite is to assume that the distances to several galaxies are known a priori , from the tip of the red giant branch in their case , and find the correction needed to bring cepheid distances to these galaxies in agreement with that other standard candle . there are two possible solutions to the problem of the metallicity - extinction degeneracy . the method followed by @xcite is one possibility , the elimination of extinction from the equation . theoretically this method should work , but in practice it is not so simple . @xcite chose a galaxy with low extinction , not zero extinction . this implies that the extinction is not entirely eliminated , leaving the degeneracy that we are trying to overcome . however , this degeneracy could be removed by using the reddening - free wesenheit magnitude @xmath1 pl relation to calculate the distance . by eliminating extinction in this way , any discrepancies in the apparent distance modulus could be explained by the metallicity effect . the work presented here is based on a multi - epoch survey of m33 conducted with the wiyn 3.5 m telescope . m33 is a well studied galaxy , with observations of cepheids performed by hubble as early as 1926 @xcite . as it is nearby , many studies have determined the distance of m33 using different and independent methods , such as eclipsing binaries @xcite , the tip of the red giant branch ( trgb ) , masers @xcite and cepheids themselves @xcite . although some of these methods may not provide the same accuracy that a cepheid distance does , they allow us to have an independent starting point for our analysis of the pl relation of m33 . one of the reasons that m33 was chosen for this study is that it is well studied . its inclination angle is 53@xmath11 , and so any effect on distance modulus ( @xmath12 ) from the orientation will be extremely small . the effect on @xmath12 due to the orientation of the galaxy was found to be very small ( less than 0.01 mag ) . the metallicity gradient across the galaxy is well defined , so the chemical composition of the cepheids is relatively easy to determine just from their position in the galaxy . these properties inherent to m33 make it the ideal target for a study of this nature . the paper is organised as follows . section [ observations ] describes our observations . section [ data - reduction ] describes the reduction process , including the development of the photometry pipeline and the final calibration and astrometry . section [ results ] contains the final catalogues and assesses the accuracy of the photometry , with the final pl relations and distances in section [ pl - relation ] . finally , we examine the effect of metallicity on our results and derive a value for the metallicity correction parameter @xmath13 . the data set consists of 25 epochs of @xmath14 images of four fields of m33 obtained between 1998 and 2001 on the 3.5 m wiyn telescope . the first 20 epochs used the wiyn s2 kb imager , a single thinned tek / stis 2048x2048 ccd with 21 @xmath12 m pixels . for the final 5 epochs the new mini - mosaic ( mimo ) camera was used . the mini - mosaic consists of two thinned site 2048x4096 ccds with 15 @xmath12 m pixels . a single exposure covered @xmath15 arcmin for the imager , and @xmath16 arcmin for the mini - mosaic , with pixel scales of 0.195 and 0.141 arcsec pixel@xmath9 respectively . four fields were observed for the survey , two in the centre of m33 , and two in the outer , southern part of the galaxy . the locations of the fields are shown in figure [ fields - figure ] . the j2000.0 coordinates of the fields are shown in table [ field - table ] . all the fields were observed in the @xmath17 , @xmath18 and @xmath19 filters with both cameras . the position of the fields were chosen such that the metal - rich population at the centre of the galaxy could be compared with the more metal - poor population in the outer region . cepheids tend to be concentrated in the spiral arms , so the outer fields were located in the southern part of the galaxy to coincide with the clearest arm . the positions of the fields and the addition of the mini - mosaic camera meant that some objects were observed in either both inner or both outer fields , giving up to 38 observations for objects in the overlap regions . .j2000.0 coordinates of the four fields . m33 - 1 and m33 - 2 are situated in the centre of the galaxy . m33 - 3 and m33 - 4 are in the southern spiral arm . deprojected angular and linear distances of each field to the centre of the galaxy are given , assuming m33 is at a distance of 840 kpc , has an inclination angle of @xmath20 and the major axis has a position angle of @xmath21 . [ cols="<,^,^,^,^",options="header " , ] [ different - pls ] to quantify the effect of metallicity we use @xmath13 , as defined in equation [ gamma ] . we can substitute @xmath22 for @xmath23 mag , using the full sample distance moduli , and @xmath24 . this leads to a metallicity correction of @xmath25 mag dex@xmath9 . this is compatible with recent measurements , such as that of @xcite . our final distance measurement for m33 is calculated by assuming an lmc metallicity of @xmath26 @xcite . using @xmath25 we find that the metallicity corrected distance modulus to m33 is @xmath6 , consistent with recent trgb distance measurements . the individual distance moduli derived for each cepheid are shown in figure [ distances ] . the uncertainty on our result is significant . by using a larger cepheid sample the contribution from the distance moduli could be reduced . also , the sample only covers the two extremes of the metallicity gradient . we will investigate the effect further using the data from @xcite . their sample contains over 2000 cepheids spread over the entire galaxy , and hence the entire [ o / h ] range . this will allow us to fully investigate the effect over a wide range of [ o / h ] , rather than just fitting a straight line to two points , and the use of so many cepheids will bring the random uncertainty down considerably . we note that the largest contribution to the error budget is the uncertainty on the metallicity gradient of m33 . this could be reduced using a large sample of cepheids . recent work by @xcite has shown that the two periods of double mode cepheids can be used to calculate their metallicity . by using a large sample of these stars across the whole galaxy , in combination with observations of fundamental mode cepheids covering the same area it would be possible to calculate @xmath13 with a much higher degree of accuracy . we have produced a photometric catalogue in @xmath17,@xmath18 and @xmath19 for over 200,000 objects in two regions of m33 , identifying 167 fundamental mode cepheids . using the reddening free wesenheit lmc pl relations we have established distances to the two regions as @xmath27 mag and @xmath28 mag . the difference in these two distance moduli is believed to come from the effect of metallicity on the zero - point of the pl relation , and hence its effect on cepheid distances . we have quantified this effect by the parameter @xmath29}$ ] and find a value of @xmath25 mag dex@xmath9 , consistent with other recent estimates . for the range of metallicities discussed here it is appropriate to include @xmath13 in the pl relation . this leads to the numerical pl shown in equation [ altered - num - pl ] , assuming an lmc distance modulus of 18.4 mag . the metallicity of the region in question is denoted by @xmath30_{mes}$ ] . @xmath31_{lmc } - \log[o / h]_{mes } ) \end{split } \label{altered - num - pl}\ ] ] the analysis presented here assumes that _ only _ the zero - point of the pl relation changes with metallicity ; the slope does not significantly change between the mw and lmc metallicities . we have repeated the analysis using the mw slope of the @xmath1 pl relation from f07 and we found the same value for @xmath13 . this is an important result as it means that we can find reddening- and metallicity - corrected cepheid distances for any population from two colour ( @xmath18 and @xmath19 ) photometry without the time consuming task of calculating internal @xmath32 values for other galaxies . as noted above , the final result and particularly the error are significantly affected by the value of metallicity one uses for each region . we have used the recent results from . @xcite found a much shallower gradient however . using their values of metallicity in our two fields , we find @xmath33 mag dex@xmath9 , incompatible with other determinations of this parameter such a large value of @xmath13 would have dramatic effects on the distance scale . in particular , it would change the distance to the lmc by so much that the cepheid distances would become incompatible with all other measurements . the difference in distance moduli between the two fields is an argument in favour of a rather large metallicity gradient in m33 . however , the metallicity gradient of m33 is still subject to strong debate and , to quote @xcite , `` ... the expectation of homogeneous azimuthal metallicity in patchy star - forming spiral galaxies seems naive ... '' . our extension of the work of @xcite based on double - mode cepheids in m33 will shed light on this discrepancy by measuring the metallicity and its gradient using many more positions . hopefully this paper serves as an example of the need for better determinations of the metallicities of _ all _ galaxies where cepheids have been observed . it is also important to note the correlation between this work and the hubble key project @xcite ; although the m33 cepheids have shorter periods than the average in the hst project , the value of @xmath13 that we find is consistent with their work . we will further investigate the effect of metallicity by applying the methodology described above to the catalogues from @xcite . with the large number of cepheids we will be able to sample the whole [ o / h ] range of the galaxy and the whole @xmath34 parameter space , finally calibrating the effect that has plagued the cepheid pl relation and distance scale for many decades . we would like to profusely thank j .- beaulieu and j .- b . marquette for the m33 fundamental mode cepheid catalogues which have proved to be invaluable to this work . we are indebted to p. stetson and k. stanek for their assistance with creating and testing the photometry pipelines . we thank the referee , p. fouqu , for his swift and thorough report that significantly improved the content and presentation of this paper . vs is supported by an stfc studentship . alibert y. , baraffe i. , hauschildt p. allard f. , 1999 , a&a , 334 , 551 beaulieu j .- p . et al . , 1997 , a&a , 318 , l47 beaulieu j .- p . , buchler j. r. , marquette j .- b , hartman j. d. , schwarzenberg - czerny a. , 2006 , apj , 653 , l101 bell r. a. , rodgers a. w. , 1969 , mnras , 142 , 161 brooks r. s. , wilson c. d. , harris w. e. , 2004 , aj , 128 , 237 bonanos a. z. , et al . , 2006 , apj , 652 , 313 bono g. , caputo f. , fiorentino g. , marconi m. , musella i. , 2008 , apj , 684 , 102 brunthaler a. , reid m. j. , falcke h. , greenhill l. j. , henkel c. , 2005 , science , 307 , 1440 caputo f. , marconi m. , musella i. , santolamazza p. , 2000 , a&a , 359 , 1059 cardelli j. a. , clayton g. c. , mathis j. s. , 1989 , apj , 345 , 245 crockett n. r. , garnett d. r. , massey p. , jacoby g. , 2006 , apj , 637 , 741 galleti s. , bellazzini m. , ferraro f. r. , 2004 , a&a , 423 , 925 ferrarese l. , silbermann n. a. , mould j. r. , stetson p. b. , saha a. , freedman w. l. , kennicutt , r. c. , jr . , 2000 , pasp , 112 177 fernie j. d. , 1969 , pasp , 81 , 707 fouqu p. et al . , 2007 , a&a , 476 , 73 freedman w. f. , madore b. f. , 1990 , apj , 365 , 186 freedman w. f. , wilson c. d , madore b. f. , 1991 , apj , 372 , 455 freedman w. l. et al . , 2001 , apj , 553 , 47 hartman j. d. , bersier d. , stanek k. z , beaulieu j .- p . , kaluzny j. , marquette j .- b . , stetson p. b. , schwarzenberg - czerny a. , 2006 , mnras , 371 , 1405 hubble e. p. , 1926 , apj , 63 , 236 kennicutt r. c. , jr . , et al . , 1998 , apj , 498 , 181 kim m. , kim e. , lee m. g. , sarajedini a. , geisler d. , 2002 , aj , 123 , 244 kochanek c. s. , 1997 , apj , 491 , 13 kudritzki r. p. , urbaneja , m. a. , bresolin f. , przybilla n. , gieren w. , pietrzyski g. , 2008 , apj , 681 269 landolt a. u. , 1992 , aj , 104 , 340 mcconnachie a. w. , irwin m. j. , ferguson a. m. n. , ibata r. a. , lewis g. f. , tanvir n. , 2004 , mnras , 350 , 243 macri l. m. , stanek k. z , bersier d. , greenhill l. j. , reid m. j , 2006 , apj , 652 , 1133 magrini l. , vlchez j. m. , mampaso a. , corradi r. l. m. , leisy p. , 2007 , a&a , 470 , 865 massey p. , olsen k. a. g. , hodge p. w. , strong s. b. , jacoby g. h. , schlingman w. , smith r. c. , 2006 , aj , 131 , 2478 monet d. g. et al . , 2003 , aj , 125 , 984 ngeow c. , kanbur s. m. , neilson h. r. , nanthakumar a. , buonaccorsi j. , 2009 , apj , 693 , 691 pena m. , peimbert m. , torres - peimbert s. , ruiz m. t. , maza j. , 1995 , apj , 441 , 343 pietrzynski g. , gieren w. , fouqu p. , pont f. , 2002 , aj , 123 , 789 sakai s. , ferrarese l. , kennicutt jr . , r. c. , saha , a. , 2004 , apj , 608 , 42 sasselov d. d. et al . , 1997 , a&a , 324 , 471 simon n. r , lee a. s. , 1981 , apj , 248 , 291 stetson p. b. , 1987 , pasp , 99 , 191 stetson p. b. , 1990 , pasp , 102 , 932 stetson p. b. , 1994 , pasp , 106 , 250 stetson p. b. , 2005 , pasp , 117 , 563 szabados l. , 2003 , in gaia spectroscopy : science and technology , ed . u. munari , asp conf . , 298 , 237 udalski a. , soszynski i. , szymanski m. , kubiak m. , pietrzynski g. , wozniak p. , zebrun k. , 1999 , acta astron . , 49 , 223

we present the results from a multi - epoch survey of two regions of m33 using the 3.5 m wiyn telescope . the inner field is located close to the centre of the galaxy , with the outer region situated about @xmath0 kpc away in the southern spiral arm , allowing us to sample a large metallicity range . we have data for 167 fundamental mode cepheids in the two regions . the reddening - free wesenheit magnitude @xmath1 period - luminosity relations were used to establish the distance modulus of each region , with @xmath2 and @xmath3 . the apparent discrepancy between these two results can be explained by the significant metallicity gradient of the galaxy . we determine a value for the metallicity parameter of the period luminosity relation @xmath4 = @xmath5 , consistent with previous measurements . this leads to a metallicity corrected distance modulus to m33 of @xmath6 mag . catalogues cepheids galaxies : individual : m33 galaxies : distances and redshifts

multiple - input multiple - output ( mimo ) systems with large number ( e.g. , tens ) of transmit and receive antennas , referred to as ` large - mimo systems , ' are of interest because of the high capacities / spectral efficiencies theoretically predicted in these systems @xcite,@xcite . research in low - complexity receive processing ( e.g. , mimo detection ) techniques that can lead to practical realization of large - mimo systems is both nascent as well as promising . for e.g. , ntt docomo has already field demonstrated a @xmath7 v - blast system operating at 5 gbps data rate and 50 bps / hz spectral efficiency in 5 ghz band at a mobile speed of 10 km / hr @xcite . evolution of wifi standards ( evolution from ieee 802.11n to ieee 802.11ac to achieve multi - gigabit rate transmissions in 5 ghz band ) now considers @xmath8 mimo operation ; see @xmath8 mimo indoor channel sounding measurements at 5 ghz reported in @xcite for consideration in wifi standards . also , @xmath9 mimo channel sounding measurements at 5 ghz in indoor environments have been reported in @xcite . we note that , while rf / antenna technologies / measurements for large - mimo systems are getting matured , there is an increasing need to focus on low - complexity algorithms for detection in large - mimo systems to reap their high spectral efficiency benefits . in the above context , in our recent works , we have shown that certain algorithms from machine learning / artificial intelligence achieve near - optimal performance in large - mimo systems at low complexities @xcite-@xcite . in @xcite-@xcite , a local neighborhood search based algorithm , namely , a _ likelihood ascent search _ ( las ) algorithm , was proposed and shown to achieve close to maximum - likelihood ( ml ) performance in mimo systems with several tens of antennas ( e.g. , @xmath4 and @xmath10 mimo ) . subsequently , in @xcite,@xcite , another local search algorithm , namely , _ reactive tabu search _ ( rts ) algorithm , which performed better than the las algorithm through the use of a local minima exit strategy was presented . in @xcite , near - ml performance in a @xmath11 mimo system was demonstrated using a _ gibbs sampling _ based detection algorithm , where the symbols take values from @xmath3 . more recently , we , in @xcite , proposed a factor graph based _ belief propagation _ ( bp ) algorithm for large - mimo detection , where we adopted a gaussian approximation of the interference ( gai ) . the motivation for the present work arises from the following two observations on the rts and bp algorithms in @xcite,@xcite and @xcite : @xmath0 rts works for general @xmath1-qam . although rts was shown to achieve close to ml performance for 4-qam in large dimensions , significant performance improvement was still possible for higher - order qam ( e.g. , 16- and 64-qam ) . @xmath2 bp also was shown to achieve near - optimal performance for large dimensions , but only for @xmath3 alphabet . in this paper , we improve the large - mimo detection performance of higher - order qam signals by using a hybrid algorithm that employs rts and bp . in particular , we observed that when a detection error occurs at the rts output , the least significant bits ( lsb ) of the symbols are mostly in error . motivated by this observation , we propose to first reconstruct and cancel the interference due to bits other than the lsbs at the rts output and feed the interference cancelled received signal to the bp algorithm to improve the reliability of the lsbs . the output of the bp is then fed back to the rts for the next iteration . our simulation results show that the proposed rts - bp algorithm achieves better uncoded as well as coded ber performance compared to those achieved by rts in large - mimo systems with higher - order qam ( e.g. , rts - bp performs better by about 3.5 db at @xmath5 uncoded ber and by about 2.5 db at @xmath6 rate-3/4 turbo coded ber in @xmath4 v - blast with 64-qam ) at the same order of complexity as rts . the rest of this paper is organized as follows . in sec . [ sec2 ] , we introduce the rts and bp algorithms in @xcite,@xcite and @xcite and the motivation for the current work . the proposed hybrid rts - bp algorithm and its performance are presented in secs . [ sec3 ] and [ sec4 ] . conclusions are given in sec . consider a @xmath12 v - blast mimo system whose received signal vector , @xmath13 , is of the form @xmath14 where @xmath15 is the symbol vector transmitted , @xmath16 is the channel gain matrix , and @xmath17 is the noise vector whose entries are modeled as i.i.d @xmath18 . assuming rich scattering , we model the entries of @xmath19 as i.i.d @xmath20 . each element of @xmath21 is an @xmath1-pam or @xmath1-qam symbol . @xmath1-pam symbols take values from @xmath22 , where @xmath23 , and @xmath1-qam is nothing but two pams in quadrature . as in @xcite , we convert ( [ eqn1 ] ) into a real - valued system model , given by @xmath24 where @xmath25 , @xmath26 , @xmath27 , @xmath28 . for @xmath1-qam , @xmath29 $ ] can viewed to be from an underlying @xmath1-pam signal set , and so is @xmath30 $ ] . let @xmath31 denote the @xmath1-pam signal set from which @xmath32 takes values , @xmath33 . defining a @xmath34-dimensional signal space @xmath35 to be the cartesian product of @xmath36 to @xmath37 , the ml solution vector , @xmath38 , is given by @xmath39 whose complexity is exponential in @xmath40 . the rts algorithm in @xcite,@xcite is a low - complexity algorithm , which minimizes the ml metric in ( [ mldetection ] ) through a local neighborhood search . a detailed description of the rts algorithm for large - mimo detection is available in @xcite,@xcite . here , we present a brief summary of the key aspects of the algorithm , and its 16- and 64-qam performance that motivates the current work . the rts algorithm starts with an initial solution vector , defines a neighborhood around it ( i.e. , defines a set of neighboring vectors based on a neighborhood criteria ) , and moves to the best vector among the neighboring vectors ( even if the best neighboring vector is worse , in terms of likelihood , than the current solution vector ; this allows the algorithm to escape from local minima ) . this process is continued for a certain number of iterations , after which the algorithm is terminated and the best among the solution vectors in all the iterations is declared as the final solution vector . in defining the neighborhood of the solution vector in a given iteration , the algorithm attempts to avoid cycling by making the moves to solution vectors of the past few iterations as ` tabu ' ( i.e. , prohibits these moves ) , which ensures efficient search of the solution space . the number of these past iterations is parametrized as the ` tabu period . ' the search is referred to as fixed tabu search if the tabu period is kept constant . if the tabu period is dynamically changed ( e.g. , increase the tabu period if more repetitions of the solution vectors are observed in the search path ) , then the search is called reactive tabu search . we consider reactive tabu search because of its robustness ( choice of a good fixed tabu period can be tedious ) . the per - symbol complexity of rts for detection of v - blast signals is @xmath41 . figure [ fig1 ] shows the uncoded ber performance of rts using the algorithm parameters optimized through simulations for 4- , 16- , and 64-qam in a @xmath4 v - blast system . as lower bounds on the error performance in mimo , the siso awgn performance for 4- , 16- , and 64-qam are also plotted . it can be seen that , in the case of 4-qam , the rts performance is just about 0.5 db away from the siso awgn performance at @xmath5 ber . however , the gap between rts performance and siso awgn performance at @xmath5 ber widens for 16-qam and 64-qam ; the gap is 7.5 db for 16-qam and 16.5 db for 64-qam . this gap can be viewed as a potential indicator of the amount of improvement in performance possible further . a more appropriate indicator will be the gap between rts performance and the ml performance . since simulation of sphere decoding ( sd ) of @xmath4 v - blast with 16- and 64-qam ( 64 real dimensions ) is computationally intensive , we do not show the sd ( ml ) performance . nevertheless , the widening gap of rts performance from siso awgn performance for 16- and 64-qam seen in fig . [ fig1 ] motivated us to explore improved algorithms to achieve better performance than rts performance for higher - order qam . = 7.2 cm = 9.60 cm [ fig1 ] in @xcite , we presented a detection algorithm based on bp on factor graphs of mimo systems . in ( [ eqn2 ] ) , each entry of the vector @xmath42 is treated as a function node ( observation node ) , and each symbol , @xmath43 , as a variable node . a key ingredient in the bp algorithm in @xcite , which contributes to its low complexity , is the gaussian approximation of interference ( gai ) , where the interference plus noise term , @xmath44 , in is modeled as @xmath45 with @xmath46 , and @xmath47 , where @xmath48 is the @xmath49th element in @xmath50 . with @xmath32 s @xmath51 , the log - likelihood ratio ( llr ) of @xmath52 at observation node @xmath53 , denoted by @xmath54 , is @xmath55 the llr values computed at the observation nodes are passed to the variable nodes ( as shown in fig . [ fig2 ] ) . using these llrs , the variable nodes compute the probabilities @xmath56 and pass them back to the observation nodes ( fig . [ fig2 ] ) . this message passing is carried out for a certain number of iterations . at the end , @xmath52 is detected as @xmath57 it has been shown in @xcite that this bp algorithm with gai , like las and rts algorithms , exhibits ` large - system behavior , ' where the bit error performance improves with increasing number of dimensions . = 3.5 cm = 4.2 cm = 3.5 cm = 4.2 cm in fig . [ fig1 ] , the uncoded ber performance of this bp algorithm for 4-qam ( input data vector of size @xmath34 with elements from @xmath3 ) in @xmath4 v - blast is also plotted . we can see that the performance is almost the same as that of rts . in terms of complexity , the bp algorithm has the advantage of no need to compute an initial solution vector and @xmath58 , which is required in rts . the per - symbol complexity of the bp algorithm for detection in v - blast is @xmath59 . a limitation with this bp approach is that it is not for general @xmath1-qam . however , its good performance with @xmath3 alphabet at lower complexities than rts can be exploited to improve the higher - order qam performance of rts , as proposed in the following section . [ sec3 ] in this section , we highlight the rationale behind the hybrid rts - bp approach and present the proposed algorithm . _ why hybrid rts - bp ? _ the proposed hybrid rts - bp approach is motivated by the the following observation we made in our rts ber simulations . we observed that , at moderate to high snrs , when an rts output vector is in error , the least significant bits ( lsb ) of the data symbols are more likely to be in error than other bits . an analytical reasoning for this behavior can be given as follows . let @xmath60 be the transmit vector and @xmath61 be the corresponding output of the rts detector . let @xmath62 denote the @xmath1-pam alphabet that @xmath32 s take values from . consider the symbol - to - bit mapping , where we can write the value of each entry of @xmath61 as a linear combination of its constituent bits as @xmath63 where @xmath64 and @xmath65 . we note that the rts algorithm outputs a local minima as the solution vector . so , @xmath66 , being a local minima , satisfies the following conditions : @xmath67 where @xmath68 , and @xmath69 denotes the @xmath53th column of the identity matrix . defining @xmath70 , @xmath71 , and denoting the @xmath53th column of @xmath50 as @xmath72 , the conditions in ( [ localminima ] ) reduce to @xmath73 where @xmath74 denotes the @xmath49th element of @xmath75 . under moderate to high snr conditions , ignoring the noise , ( [ reducedcondition ] ) can be further reduced to @xmath76 where @xmath77 denotes the @xmath53th column of * f*. for rayleigh fading , @xmath78 is chi - square distributed with @xmath34 degrees of freedom with mean @xmath40 . approximating the distribution of @xmath74 to be normal with mean zero and variance @xmath79 for @xmath80 by central limit theorem , we can drop the @xmath81 in ( [ finalcondn1 ] ) . using the fact that the minimum value of @xmath82 is 2 , ( [ finalcondn1 ] ) can be simplified as @xmath83 where @xmath84 . also , if @xmath85 , by the normal approximation in the above @xmath86 now , the lhs in ( [ finalcondn2 ] ) being normal with variance proportional to @xmath87 and the rhs being positive , it can be seen that @xmath88 , @xmath89 take smaller values with higher probability . hence , the symbols of @xmath90 are nearest euclidean neighbors of their corresponding symbols of the global minima with high probability s and @xmath91 s take values from @xmath1-pam alphabet , @xmath91 is said to be the euclidean nearest neighbor of @xmath32 if @xmath92 . ] . now , because of the symbol - to - bit mapping in ( [ linearcomb ] ) , @xmath91 will differ from its nearest euclidean neighbors certainly in the lsb position , and may or may not differ in other bit positions . consequently , the lsbs of the symbols in the rts output @xmath61 are least reliable . the above observation then led us to consider improving the reliability of the lsbs of the rts output using the bp algorithm in @xcite , and iterate between rts and bp as follows . = 2.50 cm = 8.4 cm [ fig3 ] _ proposed hybrid rts - bp algorithm : _ figure [ fig3 ] shows the block schematic of the proposed hybrid rts - bp algorithm . the following four steps constitute the proposed algorithm . * _ step 1 : _ obtain @xmath61 using the rts algorithm . obtain the output bits @xmath93 , @xmath94 , @xmath95 , from @xmath61 and ( [ linearcomb ] ) . * _ step 2 : _ using the @xmath93 s from step 1 , reconstruct the interference from all bits other than the lsbs ( i.e. , interference from all bits other than @xmath96 s ) as @xmath97 where @xmath98^t$ ] . cancel the reconstructed interference in ( [ intf ] ) from * y * as @xmath99 * _ step 3 : _ run the bp - gai algorithm in sec . [ sec_bp ] on the vector @xmath100 in step 2 , and obtain an estimate of the lsbs . denote this lsb output vector from bp as @xmath101 . now , using @xmath101 from the bp output , and the @xmath102 , @xmath103 from the rts output in step 1 , reconstruct the symbol vector as @xmath104 * _ step 4 : _ repeat steps 1 to 3 using @xmath105 as the initial vector to the rts algorithm . the algorithm is stopped after a certain number of iterations between rts and bp . our simulations showed that two iterations between rts and bp are adequate to achieve good improvement ; more than two iterations resulted in only marginal improvement for the system parameters considered in the simulations . since the complexity of bp part of rts - bp is less than that of the rts part , the order of complexity of rts - bp is same as that of rts . = 7.3 cm = 9.8 cm [ fig4 ] [ sec4 ] in this section , we present the uncoded and coded ber performance of the proposed rts - bp algorithm evaluated through simulations . perfect knowledge of @xmath50 is assumed at the receiver . _ performance in large v - blast systems : _ figure [ fig4 ] shows the uncoded ber performance of @xmath4 v - blast with 16- and 64-qam . performance of both rts - bp as well as rts are shown . it can be seen that , at an uncoded ber of @xmath5 , rts - bp performs better than rts by about 3.6 db for 64-qam and by about 1.6 db for 16-qam . this illustrates the effectiveness of the proposed hybrid rts - bp approach . also , this improvement in uncoded ber is found to result in improved coded ber as well , as illustrated in fig . [ fig5 ] . in fig . [ fig5 ] , we have plotted the turbo coded ber of rts - bp and rts in @xmath4 v - blast with 64-qam for rate-1/2 ( 96 bps / hz ) and rate-3/4 ( 144 bps / hz ) turbo codes . it can be seen that , at a coded ber of @xmath6 , rts - bp performs better than rts by about 1.5 db at 96 bps / hz and by about 2.5 db at 144 bps / hz . = 7.3 cm = 9.8 cm [ fig5 ] _ performance in large non - orthogonal stbc mimo systems : _ we also evaluated the ber performance of large non - orthogonal stbc mimo systems with higher - order qam using rts - bp detection . figure [ fig6 ] shows the uncoded ber of @xmath106 and @xmath8 non - orthogonal stbc from cyclic division algebra @xcite for 16-qam . here again , we can see that rts - bp achieves better performance than rts . _ performance in frequency - selective large v - blast systems : _ we note that the performance plots in figs . [ fig4 ] to [ fig6 ] are for frequency - flat fading , which could be the fading scenario in mimo - ofdm systems where a frequency - selective fading channel is converted to frequency - flat channels on multiple subcarriers . rts - bp , rts , and las algorithms , being suited to work well in large dimensions , can be applied to equalize signals in frequency - selective channels in large - mimo systems . following the equivalent real - valued system model of the form in ( [ eqn2 ] ) for frequency - selective mimo systems developed in @xcite , we evaluated the performance of rts - bp , rts , and las algorithms in @xmath8 v - blast with 16-qam on a frequency selective channel with @xmath107 equal energy multipath components and @xmath108 symbols per frame . figure [ fig7 ] shows the superior performance of the rts - bp algorithm over the rts and las algorithms in this frequency - selective @xmath8 large - mimo system with 16-qam . = 7.3 cm = 10.0 cm [ fig6 ] = 7.3 cm = 10.0 cm [ fig7 ] [ sec5 ] we proposed a hybrid algorithm that exploited the good features of the rts and bp algorithms to achieve improved bit error performance and nearness to capacity performance for @xmath1-qam signals in large - mimo systems at practically affordable low complexities . we illustrated the performance gains of the proposed hybrid approach over the rts algorithm in flat - fading as well as frequency - selective fading for large v - blast as well as large non - orthogonal stbc mimo systems . we note ( e.g. , from the performance plots for 64-qam in figs . [ fig1 ] and [ fig5 ] ) that further improvement in performance beyond what is achieved by the proposed hybrid rts - bp algorithm could be possible . investigation of alternate detection strategies to achieve this possible improvement is a subject for further investigation .

low - complexity near - optimal detection of large - mimo signals has attracted recent research . recently , we proposed a local neighborhood search algorithm , namely _ reactive tabu search _ ( rts ) algorithm , as well as a factor - graph based _ belief propagation _ ( bp ) algorithm for low - complexity large - mimo detection . the motivation for the present work arises from the following two observations on the above two algorithms : @xmath0 rts works for general @xmath1-qam . although rts was shown to achieve close to optimal performance for 4-qam in large dimensions , significant performance improvement was still possible for higher - order qam ( e.g. , 16- and 64-qam ) . @xmath2 bp also was shown to achieve near - optimal performance for large dimensions , but only for @xmath3 alphabet . in this paper , we improve the large - mimo detection performance of higher - order qam signals by using a hybrid algorithm that employs rts and bp . in particular , motivated by the observation that when a detection error occurs at the rts output , the least significant bits ( lsb ) of the symbols are mostly in error , we propose to first reconstruct and cancel the interference due to bits other than lsbs at the rts output and feed the interference cancelled received signal to the bp algorithm to improve the reliability of the lsbs . the output of the bp is then fed back to rts for the next iteration . our simulation results show that in a @xmath4 v - blast system , the proposed rts - bp algorithm performs better than rts by about 3.5 db at @xmath5 uncoded ber and by about 2.5 db at @xmath6 rate-3/4 turbo coded ber with 64-qam at the same order of complexity as rts . we also illustrate the performance of large - mimo detection in frequency - selective fading channels .

the knowledge of the dynamics of disk galaxies is essential in order to understand their structure and history . unfortunately , disk galaxies are difficult systems to model dynamically , for several reasons . one of them is the presence of a large amount of interstellar dust , which obscures the light along the lines - of - sight . using extended radiative transfer models it is nowadays possible to recover quite accurately the three - dimensional light and dust distribution in disk galaxies ( kylafis & bahcall 1987 , xilouris et al . 1999 ) . but also the observed kinematics are affected by dust obscuration . indeed , each element along a line - of - sight carries its own kinematic information , and the projected kinematics are a weighted mean of all these contributions . we adopt the technique outlined in baes et al . ( 2000a , b ) in order to investigate in detail the effects of dust extinction on the mean projected velocity @xmath0 and the projected velocity dispersion @xmath1 . we adopt a galaxy model which consists of a double exponential disk and a de vaucouleurs bulge . we construct a dynamical model ( i.e. a potential and a phase - space distribution function ) for this galaxy . we choose a potential that gives rise to a flat rotation curve and represents a halo - disk structure ( batsleer & dejonghe 1994 ) . using the quadratic programming modelling procedure ( dejonghe 1989 ) we then construct a two - integral distribution function that is consistent with the light density . we add a double exponential dust disk to this model . finally , the dust - affected @xmath0 and @xmath1 can be calculated for various values of the inclination and optical depth . for galaxies which are face - on or moderately inclined , the effects of dust extinction on @xmath0 and @xmath1 are negligibly small . in the edge - on case , the dust - affected @xmath0-profile tends to apparent solid body rotation , as we only see the stars moving on the outer near edge of the disk . in meanwhile , the projected dispersion decreases drastically as a function of optical depth for the inner lines - of - light , as dust obscuration strongly reduces the contribution of the high random motions of the bulge stars . both effects are critically dependent on inclination , and they are already much weaker for galaxies which are only a few degrees from exactly edge - on ( see also bosma et al . from our results it is clear that the effects of dust obscuration on @xmath0 and @xmath1 are negligible for moderately inclined galaxies . hence it is quite safe to neglect dust extinction in the interpretation of projected kinematics . this leads us to propose the following strategy to construct dynamical models for disk galaxies . intermediately inclined disks are the best choice , as spectra at different position angles will then show different projections of the velocity ellipsoid . first , one should determine the three - dimensional light distribution of the galaxy , using deprojection techniques which take the dust into account . the accuracy of the results can be tested by comparing models in different wavebands with the galactic extinction curve ( xilouris et al . 1999 ) or by comparing the derived extinction profile with fir / submm emission ( alton et al . then , a set of potentials which are consistent with the rotation curve and the light distribution need to be determined . for each potential a three - integral model can be constructed . input for the fit should be the light density and the projected kinematics along ( at least ) both major and minor axes . the goodness of fit of the different models can then be used to constrain the set of possible potentials , which will reveal the mass distribution in the galaxy . the velocity field can then be analysed , in particular the behaviour of the velocity ellipsoid . this can shed a light on the mechanism responsible for the dynamical history of the disk ( jenkins & binney 1990 , gerssen et al . 1997 , 2000 ) . alton p.b . , xilouris e.m . , bianchi s. , davies j. , kylafis n. , 2000 , a&a , 356 , 795 baes m. , dejonghe h. , 2000 , mnras , 313 , 153 baes m. , dejonghe h. , de rijcke s. , 2000 , mnras , in press ( astro - ph/0006110 ) batsleer p. , dejonghe h. , 1994 , a&a , 287 , 43 bosma a. , byun y. , freeman k.c . , athanassoula e. , 1992 , apj , 400 , l21 dejonghe h. , 1989 , apj , 343 , 113 gerssen j. , kuijken k. , merrifield m.r . , 1997 , mnras , 288 , 618 gerssen j. , kuijken k. , merrifield m.r . , 2000 , astro - ph/0005454 jenkins a. , binney j. , 1990 , mnras , 245 , 305 kylafis n.d . , bahcall j.n . , 1987 , apj , 317 , 637 xilouris e.m . , byun y.i . , kylafis n.d . , paleologou e.v . , papamastorakis j. , 1999 , a&a , 344 , 868

disk galaxies contain a large amount of interstellar dust , which affects the projection of kinematic quantities . we investigate in detail the effects of dust extinction on the mean projected velocity and the projected velocity dispersion . we use our results to construct a general strategy to determine the dynamical structure of disk galaxies , with the aim to constrain their mass distribution and dynamical history .

the role of electron - phonon ( @xmath2-ph ) interaction in the high-@xmath3 superconductivity has received considerable attention since the discovery of oxygen isotope effect of bi@xmath0sr@xmath0cacu@xmath0o@xmath1 ( bi2212 ) in the angle - resolved photoemission spectra ( arpes)@xcite . the experimental data show that electronic band is shifted slightly with the @xmath4o/@xmath5o substitution , signifying the existence of @xmath2-ph interplay in cuprates . besides , theoretically clarifying this effect is of great significance , for the energy scale of shifts reported by the two groups in refs . @xcite and @xcite seem quite inconsistent with each other , and many questions still remain up to now . in order to have an insight into the isotope effect , in this work , we develop a new path - integral theory to calculate the photoemission spectra ( pes ) of cuprate superconductors , in which the electron - electron ( @xmath2-@xmath2 ) and @xmath2-ph correlations are treated on an equal footing . this theory is also applicable to other kind correlated materials . as an example , here , we also study the pes of boron - doped diamond ( bdd ) , which undertakes a semiconductor - metal phase transition on doping , and becomes a superconductor with the temperature decreases@xcite . the details of our theory will be presented in the next section , in connection with the study on isotope effect . calculation and discussion on pes of bdd are included in section 3 . a concluding remark can be found in the summary . in the cuo@xmath0 plane of cuprates , the electronic transfer is modulated by the vibration of oxygen atoms between the initial and final cu sites ( see in fig . 1 ) , resulting in an off - diagonal type @xmath2-ph coupling . in order to qualitatively clarify the isotope effect of bi2212 , we start from a half - filled hamiltonian including the @xmath2-@xmath2 repulsion and the above mentioned off - diagonal @xmath2-ph coupling ( @xmath6 and @xmath7 throughout this paper ) : @xmath8 where @xmath9 ( @xmath10 ) is the creation ( annihilation ) operator of an electron with spin @xmath11 at the cu site @xmath12 on a square lattice ( fig . the electrons hop between two nearest neighboring cu sites , denoted by @xmath13 , with a transfer energy @xmath14 . @xmath15 is the strength of coulomb repulsion between two electrons on the same cu site with opposite spins . the oxygen phonon is assumed to be of the einstein type with a frequency @xmath16 and a mass @xmath17 . @xmath18 ( @xmath19 ) is the mass change factor of phonon due to the isotope substitution . in the third term , @xmath20 is the dimensionless coordinate operator of the oxygen phonon locating between the nearest - neighboring cu sites @xmath12 and @xmath21 , and the sum denoted by @xmath22 just means a summation over all the phonon sites in the lattice . conduction plane in cuprates . the copper atom ( white circle ) is on the @xmath12-th site of a simple square lattice , where the electrons ( arrows ) reside . the oxygen atom ( black circle ) is located between the nearest - neighboring cu sites , and @xmath23 denotes its displacement from the equilibrium position . ] in the conduction plane of cuo@xmath0 , the electronic hopping integral @xmath24 can be expanded to the second order terms with respect to the phonon displacements @xmath20 as @xmath25 where @xmath26 is the bare hopping energy and @xmath27 the off - diagonal quadratic @xmath2-ph coupling constant . here we note the linear @xmath2-ph coupling does not occur owing to the lattice symmetry of present model . whereas the inter - site @xmath2-@xmath2 interaction is included in the screened values of @xmath26 and @xmath27 . in this section , we develop a path - integral theory for a model with both @xmath2-@xmath2 and @xmath2-ph interactions . by making use of the trotter s decoupling formula , the boltzmann operator is written as , @xmath28 applying the hubbard - stratonovitch transformation@xcite and the gaussian integral formula@xcite , we can decouple the two - body parts , so that the @xmath2-@xmath2 and @xmath2-ph correlated terms are replaced by a two - fold summation over the auxiliary spin and lattice configurations , which is the so - called path - integral . in this way , the boltzmann operator is rewritten into the path - integral form as , @xmath29 \right\ } \prod_l \left [ |x_q(l,\beta ) \rangle \langle x_q(l,0)| \right ] \right),\end{aligned}\ ] ] @xmath30 \left[a^{\dag}_{l \sigma}(\tau ) a_{l ' \sigma}(\tau ) + a^{\dag}_{l ' \sigma}(\tau ) a_{l \sigma}(\tau ) \right ] \nonumber\\ & & - \sqrt{u \over \delta } \sum_l x_m(l , \tau ) [ n_{l \uparrow}(\tau ) - n_{l \downarrow}(\tau ) ] , \\ \omega(x_q ) & \equiv & \sum_{\langle l , l ' \rangle } \left\ { { \lambda \over 2 \omega_0 } \left [ { \partial x_q(l , l',\tau ) \over \partial \tau } \right]^2 + { 1 \over 2 } \omega_0 x^2_q(l , l',\tau ) \right\}.\end{aligned}\ ] ] here , @xmath31 and @xmath32 correspond to the auxiliary spin and lattice field , respectively , @xmath33 symbolically denotes the integrals over the path @xmath34 synthesized by @xmath31 and @xmath32 , and @xmath35 is the eigenstate of phonon . @xmath36 is the time interval of the trotter s formula , @xmath37 , and @xmath38 is the absolute temperature . @xmath39 in eq . ( 4 ) is the time ordering operator . then the time evolution operator [ @xmath40 along a path @xmath34 is defined as @xmath41.\end{aligned}\ ] ] in terms of the boltzmann operator ( 4 ) and time evolution operator ( 7 ) , we define the free energy [ @xmath42 of the given path as @xmath43.\end{aligned}\ ] ] while , the partition function ( @xmath44 ) and total free energy ( @xmath45 ) are given as @xmath46 according to refs . @xcite and @xcite , we also define the one - body green s function [ @xmath47 on a path @xmath34 as @xmath48 where @xmath49 is the heisenberg representation of @xmath10 . it is really time - dependent and defined by @xmath50 meanwhile , the ordinary green s function [ @xmath51 can be obtained by the path - integral as @xmath52 this path - integral is evaluated by the quantum monte carlo ( qmc ) simulation method . if the qmc data of green s function @xmath53 is obtained , we can immediately calculate its fourier component [ @xmath54 as @xmath55 where @xmath56 is the momentum of the outgoing photo - electron . from this fourier component @xmath57 , we derive the momentum - specified spectral function [ @xmath58 by solving the integral equation @xmath59 we now present the qmc results on a 4@xmath604 square lattice , where @xmath26 is set as the unit of energy , and @xmath16=1.0 is used . for the qmc simulation , we impose a little large isotopic mass enhancement , @xmath61=1 and @xmath18=2 , to suppress the numerical error . in this calculation , we determine the binding energy @xmath62 by the moment analysis of the spectral function as @xmath63 . correspondingly , the isotope induced band shift is calculated by @xmath64 . with @xmath27 on a 4@xmath604 square lattice , when @xmath15=2.0 , @xmath65=10 , @xmath61=1 , @xmath18=2 . the filled circles are from qmc , and the solid curve from hfa as a guide for eyes . ( b ) the variations of @xmath66 with @xmath15 on a 4@xmath604 square lattice at @xmath65=10 , @xmath61=1 and @xmath18=2 . three different values of @xmath27 are used to show the @xmath27-dependence of @xmath66 . the discrete symbols are the results of qmc , and continuous curves by hfa as a reference . ] in fig . 2(a ) , we plot the ratio @xmath66 versus @xmath27 , at @xmath15=2.0 and @xmath65=10 , where @xmath67 is the band shift at the @xmath68 point of brillouin zone [ @xmath69=(0,0 ) ] , and @xmath70 is the isotopic change of phonon energy . the filled circles are calculated by qmc , and the solid curve by the mean - field theory with hartree - fork approximation ( hfa ) as a guide for eyes . here both theories figure out an increase of @xmath66 with @xmath27 , which means if the @xmath2-ph coupling is strong enough , a large band shift can be generated in a small cost of @xmath70 . in fig . 2(b ) , the ratio @xmath66 versus @xmath15 are shown for three different @xmath27 s , where the discrete symbols and continuous curves are the qmc and hfa results , respectively . one can see the ratio @xmath66 increases with @xmath27 . meanwhile , for a fixed @xmath27 , the ratio declines slightly as @xmath15 increases , indicating that the band shift is owing to the @xmath2-ph coupling , whereas the presence of @xmath15 partially reduces this effect . according to fig . 2 , the band shift thus can be regarded as a measure of the effective @xmath2-ph coupling strength in the system . it is well known that the pristine diamond is a big band gap insulator . lightly doped with boron , it shows a @xmath71-type character with an activation energy about 0.37 ev@xcite . recently , the research on bdd has become highly attractive since the remarkable discovery of superconductivity in this material@xcite . accompanied with the superconducting phase , a semiconductor - metal transition also occurs when the doping percentage is increased to certain level . in the normal metallic state , k. ishizaka _ et al_.@xcite declared the observation of a step - like multi - phonon satellite structure in the valence band pes , approximately distributed periodically in 0.150 ev below the fermi level , in addition to the emergence of a clear fermi edge . this periodic structure in pes reminds us of a similarity to the case of localized electron@xcite , wherein the coupling between electron and einstein phonons characterizes the spectra with discrete peaks at equal distance . moreover , the fermi edge and the step - like structure are observed together , probably originating from a co - existence of the two basic properties of electrons : itineracy and localization . in order to clarify this so - called co - existence theoretically , we apply our path - integral theory to the many - impurity - holstein ( mih ) model on a doped simple cubic lattice to derive its spectral density . the mih model includes the following two properties . one is the disorder of the system , that some atoms are replaced by dopant ones in a certain ratio . the other is the coupling between electrons and einstein phonons , being the simplest description of @xmath2-ph interactions . its hamiltonian is given as , @xmath72 where @xmath73 is the transfer energy . @xmath74 and @xmath75 are the creation and annihilation operators of electron with spin @xmath11 at site @xmath12 . electrons can hop only between the nearest neighboring sites expressed by @xmath76 . @xmath77 stands for the chemical potential of electrons and @xmath78 is the potential difference between after and before substitution at the doped sites labeled by @xmath79 . @xmath80 is the dimensionless coordinate operator for the phonon at site @xmath12 with frequency @xmath81 . @xmath82 denotes the _ e_-ph coupling constant . @xmath83 is the average electron number at site @xmath12 . to simplify the problem , we just consider the coupling at the doped sites hereafter , because in the pure diamond there is no evidence that the satellite structure appears , which suggests the coupling is important only after doping . by applying our path - integral theory to the disordered system , and following the formulation in section 2.2 , we obtain an electronic green s function similar to eq . 12 , @xmath84 in the numerical calculation , the path - integral of green s function in eq . 16 is also performed by the qmc simulation method as before . after averaging this site - dependent green s function over all the @xmath85 sites of the system , the photoemission spectral function [ @xmath86 can be reproduced through the analytic continuation as , @xmath87 finally , after imposing the fermi - dirac function @xmath88 $ ] , the pes intensity is given as @xmath89 , which can be compared with the experimental data . since we focus on the spectral region close to the fermi level , we use a simple cubic lattice of @xmath90 in real calculation without paying much attention to the detail of carbon valence band . as we just count the @xmath2-ph coupling at the doped sites , the phonon effect is not obvious in the whole system pes after averaging over all sites . for this reason , in the following , we will also present the pes of doped sites to illustrate the phonon effect . -ph coupling , by qmc . full ( dotted ) line is the spectrum of whole system ( doped sites ) . the inset zooms in the solid line near the fermi level . ] -ph coupling , by qmc . full ( dotted ) line is the spectrum of whole system ( doped sites ) . the inset zooms in the solid line near the fermi level . ] 3 shows the spectra for the simple cubic lattice at light ( 3.12% ) and heavy ( 12.5% ) doping rates , with a weak @xmath2-ph coupling of @xmath82=0.25 ev . from the whole system pes ( full lines ) , one can clearly see the emergence of a fermi edge with the increase of doping percentage . in the lightly doped case , the impurity levels are a little above the top of valence band , indicating the system is a semiconductor . while in the heavily doped case , the impurity band expands to overlap the top of valence band and closes the activation gap , leading to a semiconductor - metal phase transition . in the spectra of boron - doped sites ( dotted lines ) , one notices that , in addition to a broad hump due to the so - called fano effect@xcite , a satellite structure also appears in each case a little below the fermi level . we should note this side structure is just the quantum phonon peak due to the @xmath2-ph coupling . meanwhile , because of the fano effect , this phonon structure is modified by the impurity band to have a shoulder - like shape , rather than a @xmath91-like form as in the localized electron case of ref . @xcite . in fig . 4 , we show the spectra of a system with a strong @xmath2-ph coupling of @xmath82=0.50 ev , and a low doping rate of 3.12% . comparing with fig . 3(a ) , one clearly finds that , a large @xmath2-ph coupling at the doped sites greatly contributes to the expansion of impurity band , so that the fermi edge is clearly observed in the whole system pes ( full line ) . moreover , the second phonon side peak also appears due to the multi - phonon processes aroused by the strong @xmath2-ph coupling , corresponding to a step - like structure in the whole system pes , as affirmed in the inset . we develop a new path - integral theory to calculate the pes of correlated many - electron systems . the isotopic shift in the arpes of bi2212 is investigated by this theory based on a model including both @xmath2-@xmath2 and off - diagonal quadratic @xmath2-ph interactions . our calculation demonstrates that the band shift is primarily triggered by the @xmath2-ph coupling , while the presence of @xmath2-@xmath2 repulsion tends to suppress this effect . we also apply this theory to the mih model on a doped cubic lattice to clarify the spectral properties of bdd . it is clearly shown in the pes that a semiconductor - metal phase transition takes place due to the increases of @xmath2-ph coupling and dopant concentration . furthermore , the presence of fermi edge and phonon step - like structure indicates the co - existence of two basic characters of electron , itineracy and localization , in bdd .

a new path - integral theory is developed to calculate the photoemission spectra ( pes ) of correlated many - electron systems . the application to the study on bi@xmath0sr@xmath0cacu@xmath0o@xmath1 ( bi2212 ) and boron - doped diamond ( bdd ) is discussed in details . it is found that the isotopic shift in the angle - resolved photoemission spectra of bi2212 is due to the off - diagonal quadratic electron - phonon ( @xmath2-ph ) coupling , whereas the presence of electron - electron repulsion partially suppresses this effect . for the bdd , a semiconductor - metal phase transition , which is induced by increasing the @xmath2-ph coupling and dopant concentration , is reproduced by our theory . additionally , the presence of fermi edge and phonon step - like structure in pes is found to be due to a co - existence of itinerant and localized electronic states in bdd .

the peculiar electronic properties of graphene first observed experimentally by geim et al . @xcite and zhang et al . @xcite can accurately be described by massless two dimensional dirac fermion excitations ( for reviews on the physics of graphene see , e.g , refs . ) . owing to the proximity effect a superconductor can induce non - zero pair - potential in the graphene as well@xcite . such graphene - superconductor hybrid structures and in particular the andreev - reflection taking place at the graphene - superconductor interface was first studied theoretically by beenakker @xcite ( for a review on andreev reflection in graphene see ref . ) . soon after beenakker s pioneering work , supercurrent between two superconducting electrodes on top of a graphene monolayer has been observed experimentally by heersche _ et al _ , and later in refs . . in particular , the experimental results of ref . attest to the ballistic propagation of quasiparticles in graphene - superconductor hybrid structures , whereas the experiment of du _ _ et al__@xcite gave evidence that it was possible to fabricate transparent sg interfaces . these experiments have also sparked considerable theoretical interest in superconductor - graphene - superconductor ( sgs ) heterostructures . the short junction limit , where the coherence length @xmath0 ( here @xmath1 is the graphene fermi velocity and @xmath2 is the superconducting gap ) is smaller than the length @xmath3 of the junction , was first studied by titov and beenakker @xcite assuming ballistic graphene . in the opposite , long junction limit the density of states of the andreev levels was calculated first by titov , ossipov and beenakker @xcite . subsequently , numerous other theoretical works investigated the josephson current in sgs structures @xcite . the tunneling effect in sg structures has been studied in several works @xcite as well . other works in the field of graphene - superconductor heterostructures include studies on crossed andreev reflection in a graphene bipolar transistor @xcite , on @xmath4- and @xmath5-wave sg junctions @xcite and on ferromagnetic sg structures @xcite . very recently , using a phase - sensitive squid interferometry technique chialvo et al . has studied experimentally the current - phase relation ( c@xmath6r ) of graphene josephson junctions @xcite . in this work we calculate the josephson current in sgs structure as a function of the superconductor phase difference , the temperature and the length of the junction . in our theoretical treatment we adapted the method used by brouwer and beenakker for metallic chaotic josephson junctions @xcite . the approach allows to obtain results for finite temperature and is valid for junctions of arbitrary length . we note that this method has already been applied for calculating the persistent current through a _ n - p _ junction in graphene@xcite . wherever possible , we compare our results to previous ones derived for superconductor - normal conductor - superconductor ( sns ) junctions , where the normal conductor is a ballistic metal . the rest of the paper is organized in the following way : in the next section we introduce the theoretical approach that we use to obtain the josephson current . in section [ numerics ] we present and discuss the results of numerical calculations . finally , in section [ summary ] we give a brief summary . we consider a josephson junction in the @xmath7-@xmath8 plane . the normal graphene region ( g ) at @xmath9 separates the two superconducting regions formed by covering the graphene layer by two superconducting electrodes ( s ) in the regions @xmath10 and @xmath11 ( for the geometry see ref . ) . the width of the josephson junction along the @xmath8 axis is @xmath12 . owing to the valley degeneracy of the hamiltonian , the full dirac - boguliubov de gennes ( dbdg ) equations for graphene - superconductor systems decouple to two four by four , reduced hamiltonians that are related to each other by a unitary transformation ( see , e.g. , ref . ) . we now take the one corresponding to the valley @xmath13 . then the quasi particle excitations in the sgs systems are described by the reduced dbdg equations : @xmath14 \delta^*(x , y ) & \mu -h_0 \\ \end{array } \right ) \psi = \varepsilon \psi,\ ] ] where @xmath15 is the dirac hamiltonian . here @xmath16 and @xmath17 are pauli matrices , @xmath18 is the unit matrix and , @xmath19 is the chemical potential and @xmath20 is the excitation energy . the superconductor electrodes are doped by the potential @xmath21 ( here @xmath22 and constant , and @xmath23 is the heaviside function ) . the wave function @xmath24 is comprised of electron @xmath25 and hole @xmath26 wave functions which have opposite spin and valley indices . for the pair potential @xmath27 we assume a simple model : its magnitude @xmath2 is constant in the s regions , changes step - function - like at the sg interfaces ( so called rigid boundary condition , see ref . ) and is zero in the normal conducting region . similarly , we assume that the its phase is piecewise constant in the s regions . hence , the pair potential is given by @xmath28 for @xmath10 and @xmath29 for @xmath11 . band bending or other effects of the superconducting electrodes are neglected ( see e.g. ref . for the discussions of some of these effects in the case of normal conducting metal electrodes ) . the josephson current at finite temperature is given by@xcite @xmath30 , \label{eq : josephson - curr}\ ] ] where @xmath31 is the phase difference across the junction , @xmath32 is the density of states of the andreev levels . the factor of 4 accounts for the spin and valley degeneracy . as one can see from eq . ( [ eq : josephson - curr ] ) , a necessary step to calculate the josephson current is to obtain the density of states of andreev bound states in the sgs junction . to this end one can in principle proceed in the following way : one can write down a trial wave function in all three regions of the sgs structure . the boundary conditions at the two graphene - superconductor boundaries of the sgs junction then result in a secular equation @xmath33 whose solutions @xmath34 give the energies of the andreev bound states . for finite @xmath35 we obtained a @xmath36 determinant for the secular equation ( since this determinant is quite lengthy here we do not present its detailed form ) . once the energy levels @xmath34 of the sgs junction are known , the density of state is given by @xmath37 . however , the above outlined method is numerically quite cumbersome since one has to search for the zeros of the secular equation @xmath33 . to overcome this problem we now follow the method used by brouwer and beenakker in ref . . they rewrote the expression for the josephson current given in eq . ( [ eq : josephson - curr ] ) in a more convenient form . here we only summarize the main steps of the derivation . the secular equation can be written as @xmath38 , where @xmath39 is a function of @xmath40 but does not have zeros in the complex plane . thus it is easy to see that the density of states can be expressed as @xmath41 where @xmath42 is a positive infinitesimal . using the analytic properties of @xmath43 in the upper half of the complex @xmath40-plane and the fact that under the change @xmath44 the function @xmath43 goes over into its complex conjugate ( physically , this follows from the electron - hole symmetry ) , the @xmath40-integration can be extended from @xmath45 to @xmath46 . finally , after performing a partial integration the josephson current in eq . ( [ eq : josephson - curr ] ) can be rewritten as @xmath47 now closing the integration contour in the upper half of the complex @xmath40-plane and applying the residue theorem the josephson current becomes @xmath48 where @xmath49 are matsubara frequencies . note that in our model @xmath50 has no singularities for @xmath51 , thus the poles of the integrand in eq . ( [ josephson-2:eq ] ) come only from the hyperbolic tangent function . the main advantage of this result is that one does not need to obtain explicitly the solutions of the secular equation @xmath33 . moreover , this method immediately gives the finite temperature dependence of the josephson current . in the numerical calculations it turns out that the sum in eq . ( [ josephson-3:eq ] ) is rapidly convergent and usually one need to include only a finite number of terms . a similar result has been found for the persistent current through a _ n - p _ junction in graphene@xcite . we now consider the experimentally relevant case of highly doped superconductor electrodes , ie , the limit @xmath52 . by matching the wave functions at the graphene - superconductor boundaries of the sgs structure we found the same secular equation @xmath53 as that obtained by titov and beenakker using the transfer matrix method ( see eq . ( 14 ) in ref . ) . we used the ` infinite mass ' boundary conditions@xcite at @xmath54 and @xmath55 for which @xmath56 , where @xmath57 ( for @xmath58 the choice of the boundary conditions is irrelevant ) . for a given @xmath59 the solutions of the quantization condition @xmath53 give the andreev energy levels @xmath60 for @xmath61 . the secular equation @xmath53 is valid both in short and long junction limit@xcite . one can show that @xmath62 which is a necessary condition@xcite for writing the josephson current in the form of eq . ( [ josephson-2:eq ] ) . the current contribution from each propagating mode with transverse momentum @xmath63 can be calculated separately and the total current is the sum over these contributions : @xmath64 where @xmath65 is the number of propagating modes and the function @xmath66 determines the energy levels for a given transverse momentum @xmath67 . this equation is our starting point for calculating the supercurrent through a graphene based josephson junction . further analytical progress can be made in the short junction limit ( @xmath68 ) because the andreev levels @xmath60 can in that case be obtained in a closed form ( see eq . ( 16 ) in ref . ) . similarly , the summation over the matsubara frequencies in eq . ( [ josephson-4:eq ] ) can be performed analytically using the identity @xmath69= ( \pi \tanh ( \pi x/2)/(4 x)$ ] ( see ref . ) and we find @xmath70 here @xmath60 and @xmath71 can be found in ref . , while the temperature dependence of the superconductor gap @xmath72 for @xmath4-wave superconductors is given by @xmath73 where @xmath74 is the zero order modified bessel function , @xmath75 and @xmath76 is the euler s constant@xcite . for zero temperature from ( [ eq : short_aram_alt ] ) one can arrive at the same expression for the josephson current as that obtained by titov and beenakker ( ref . ) . for finite temperatures the summation over the transversal modes @xmath59 in eq . ( [ josephson-4:eq ] ) can not be evaluated analytically but numerically can easily be treated . we now present the results of numerical calculations for the josephson current using the most general formula given by eq . ( [ josephson-4:eq ] ) . ( color online ) the supercurrent ( in units of @xmath77 ) as a function of the phase difference @xmath31 . the parameters are as follows : in ( a ) and ( b ) [ short junction limit ] @xmath78 and @xmath79 ( black @xmath80 , red @xmath81 and blue @xmath82 , respectively ) . in ( c ) and ( d ) [ here @xmath83 we used @xmath84 and @xmath85 ( black @xmath80 , red @xmath81 and blue @xmath82 , respectively ) . in ( e ) and ( f ) we consider long junctions [ @xmath86 . in ( e ) @xmath87 ( black @xmath80 , red @xmath81 and blue @xmath82 , respectively ) . in ( f ) @xmath88 ( black @xmath80 , red @xmath81 and blue @xmath82 , respectively ) . the chemical potential is @xmath89 in ( a ) , ( c ) and ( e ) , whereas it is @xmath90 in ( b ) , ( d ) and ( f ) . the width of the junction is @xmath91 in all cases . ] figure [ fig : cprs ] shows the supercurrent as a function of phase @xmath31 for a number of interesting case : for short ( @xmath68 ) , intermediate ( @xmath92 ) and long ( @xmath93 ) junctions , assuming @xmath89 and for finite @xmath94 as well . one can make the following general observations : a ) the maximum current increases by increasing the doping ( @xmath94 value ) and by decreasing the temperature or the junction length ; b ) at higher temperatures the current shows a simple sinusoidal dependence on the phase in all of the cases while at low temperatures the position of the maxima of the currents are shifted to the right resulting in a skewness of the curves . following ref . we define the skewness by @xmath95 , where @xmath96 is the position of the maxima of the supercurrent . both the tendency to simple harmonic dependence for @xmath97 and a positive skewness ( i.e. @xmath98 ) are in line with previous calculations on josephson current in weak links which comprise a normal conducting metal or a tunnel barrier and assume that the pair potential changes abruptly at the normal - superconductor interface ( see e.g. ref . for a review ) . we find especially interesting the results shown in figs . [ fig : cprs](e ) and ( f ) . in the long junction limit for @xmath89 we find that the skewness is very small even at @xmath99 [ the curve denoted by black squares in fig . [ fig : cprs](e ) ] thus the c@xmath6r resembles a harmonic dependence . in contrast , still in the long junction limit but for @xmath90 and @xmath100 [ black squares in fig . [ fig : cprs](f ) ] we see that the current depends almost linearly on the phase and the curve resembles a saw - tooth . ( the transition to a saw - tooth - like dependence can already be seen in fig . [ fig : cprs](d ) where @xmath101 , c.f . fig . [ fig : cprs](b ) showing the short junction limit . ) it is interesting to note that the theoretical result for clean , long sns junctions at low temperature is a saw - toothed c@xmath6r @xcite . our numerics suggests that for sgs junctions in the same limit the saw - tooth is somewhat rounded - off and the slope of the curve is finite when @xmath102 . thus , the c@xmath6r in long , clean sgs junction seems to be closely resembling of , but not identical to the corresponding case in sns junctions . ( color online ) the skewness @xmath103 as a function of the critical current @xmath104 for different coherence lengths @xmath105 . the parameters are @xmath106 , @xmath107 ( black dots ) , @xmath108 , @xmath109 ( red squares ) and @xmath110 , @xmath111 ( blue triangles ) . the lines are guides to the eye . the ratio of the chemical potential and the superconducting gap was @xmath112 . the dotted line shows that for small critical currents the skewness depends linearly on @xmath104 . ] we calculated the skewness @xmath103 as a function the critical supercurrent @xmath104 ( the value of the current at @xmath96 ) and plotted the results for three different @xmath105 values , while keeping the junction length @xmath3 and width @xmath12 constant . the used @xmath113 values go from @xmath106 ( long junction limit ) to @xmath110 ( short junction limit ) . there are two important things to notice in fig . [ skewness_a : fig ] for small critical currents : a ) for a given junction length @xmath3 , as @xmath114 ( for higher temperatures ) the skewness @xmath103 also goes to zero , i.e. the c@xmath6r is approaching a simple sinusoidal form ; b ) @xmath103 depends linearly on @xmath104 for small critical currents , while at larger @xmath104 the dependence clearly deviates from a simple linear relation . the skewness has recently been measured in ref . and the case of @xmath106 , @xmath107 , shown by black dots in fig . [ skewness_a : fig ] , in principle corresponds to that of sample b in this experiment ( with estimated coherence length of @xmath115 nm @xcite ) . our calculations give a smaller slope than the measurements in ref . . according to our numerics , the larger slopes observed in this experiment would be attainable in the short junction limit . note however , that the exact slope would also depend on the value of the chemical potential which was not known , and importantly , the samples in the experiment of ref . are likely to have been in the quasi - diffusive limit , therefore we can not expect quantitative agreement with our ballistic theory . we have also calculated the temperature dependence of the critical current for short , intermediate and long junctions , taking two values of the chemical potential @xmath19 . the temperature dependence of the pair potential was taken into account using eq . ( [ gap_t : eq ] ) . the results are shown in fig . [ kt_depend : fig ] . ( color online ) the critical current @xmath104 as a function of @xmath116 . the parameters are @xmath78 in ( a ) , @xmath84 in ( b ) and @xmath117 in ( c ) . in ( b ) and ( c ) we used logarithmic scale . red @xmath81 denote the results for @xmath89 and black @xmath80 for @xmath118 . the width of the sample was fixed : @xmath91 . ] at this point it is interesting to make a quick sidestep and note that titov and beenakker ( ref . ) showed that for short junctions , at zero temperature and at the dirac point the c@xmath6r for ballistic graphene is formally identical to the classical result of kulik and omelyanchuk@xcite , which , however , assumes diffusive normal metal as a weak link . looking at the @xmath119 curves in the short junction limit [ fig . [ kt_depend : fig](a ) ] one can see that they are also qualitatively similar to the corresponding result of kulik and omelyanchuk@xcite [ c.f . fig . 7 in ref . ] . the close resemblance of certain properties the two types of josephson junctions therefore seems to extend to the temperature dependence of the critical current as well . in the opposite , long junction regime , for @xmath89 [ shown by red circles in fig . [ kt_depend : fig](c ) ] one can observe a short plateau in the current for small temperatures followed by an exponential decay . interestingly , a qualitatively very similar result has been obtained by gonzlez and perfetto in ref . , assuming tunnel coupling between the superconductor and the graphene and using a different formalism than ours . in the case of finite doping [ black squares in fig . [ kt_depend : fig](c ) ] an exponential decay of @xmath104 can be seen basically in the whole temperature range . the @xmath104 in clean sns junction exhibits the same qualitative dependence on @xmath120 ( ref . ) . experimentally , the @xmath119 relation was measured by du _ et al . _ ( ref . ) and by ojeda - aristizabal _ et al . _ ( ref . ) . again , quantitative comparison with our results is not possible because the graphene samples in the experiments were in the diffusive limit , but the observed dependence of @xmath104 on the temperature was qualitatively similar to the results shown in fig . [ kt_depend : fig](a ) . finally , we studied the length dependence of the critical current and the results are shown in fig . [ fig : l_depend ] . as a function of the junction length @xmath121 in logarithmic scale . in ( a ) we used @xmath89 . black @xmath80 and red @xmath81 denote the results of @xmath122 and @xmath123 calculations , respectively . the symbols in the inset of ( a ) show @xmath104 for @xmath122 , @xmath124 in double logarithmic plot , along with the fitted linear function ( solid line , see text ) . in ( b ) the chemical potential is @xmath125 . black @xmath80 and red @xmath81 denote the results of @xmath122 and @xmath126 calculations , respectively . the inset of ( b ) shows the @xmath122 calculations in linear scale for @xmath127 . the width of the junction was @xmath91 in all cases . [ fig : l_depend ] ] at the dirac point ( @xmath89 ) one can observe an exponential decay of @xmath104 for @xmath128 [ main panel of fig.[fig : l_depend](a ) ] . we could not see the @xmath129 dependence predicted in ref . . for @xmath130 and @xmath99 , however , we do find a power - law dependence @xmath131 [ see the inset of fig.[fig : l_depend](a ) ] and fitting the numerical results we obtained @xmath132 . this is remarkably close to the results of the self - consistent tight - binding calculations of black - schaffer and doniach ( ref . ) who found @xmath133 . considering now the case of doped graphene weak link , we found that for @xmath134 the @xmath3 dependence of @xmath104 can be well fitted by @xmath135 with @xmath136 . exponentially small critical current is also typical for clean sns junctions if @xmath3 is larger than the thermal coherence length ( for details see e.g. ref . ) . we note that in the case of @xmath124 and @xmath99 one can see oscillations in the @xmath104 vs @xmath3 curve [ shown in the inset of fig.[fig : l_depend](b ) ] whose study is left to a future work . in this work we calculated the josephson current in ballistic sgs structures . the most important assumption we made is that one can use rigid boundary conditions . i.e. that the change of the pair potential is step - function - like at the sg interfaces and that it does not depend on the supercurrent . we developed a general and numerically efficient approach to obtain the current - phase relation for arbitrary length of the junction as well as for finite doping and temperature . at low temperatures we have found that the current - phase relation differs from a simple harmonic dependence . in the case of short junctions and small critical currents the deviation of the current - phase curve from the sinusoidal form , the skewness , shows a linear dependence on the critical current , similarly to the observation of a recent experiment@xcite , though the slope of the curve did not match the experimental one . this is likely to be due to the fact that in the experiment the graphene sample was quasi - diffusive . in the long junction limit our results show that in clean sgs junctions the the current - phase relation transforms from the sinusoidal form at @xmath137 to a curve resembling saw - tooth at @xmath138 . in contrast to clean sns junctions however , our numerics suggests that the dependence is not exactly saw - toothed . we have also calculated the temperature and junction length dependence of the critical current . we have found similarities to both classical results for sns junctions and recent ones obtained for graphene but using a different formalism@xcite . in respect of these numerical calculations further theoretical progress is needed to unravel the relation between the sgs and sns results . since the fabrication of both ballistic graphene samples and transparent sg interfaces have already been demonstrated experimentally@xcite , we believe that our theoretical approach may be useful in the future to understand and analyze experimental data . in particular , the measurement of the c@xmath6r and the length and temperature dependence of @xmath104 in short junctions should be feasible . _ note added _ : during the peer - review process of the manuscript , a relevant preprint has appeared@xcite where the authors study the temperature dependence of the josephson current using self - consistent tight - binding numerical computations . their work is thus complementary to our and help to understand the scope of certain approximations , e.g. the rigid - boundary condition that we employed . _ acknowledgements : _ supported by the marie curie itn project nanoctm ( fp7-people - itn-2008 - 234970 ) and the hungarian science foundation otka under the contracts no . 75529 and no . 81492 . a. k. also acknowledges the support of epsrc . we acknowledge fruitful discussions with c. w. j. beenakker , e. perfetto , n. mason , d. j. van harlingen and c. chialvo . h. b. heersche , p. jarillo - herrero , j. b. oostinga1 , l. m. k. vandersypen , and a. f. morpurgo nature * 446 * , 56 ( 2007 ) ; h. b. heersche , p. jarillo - herrero , j. b. oostinga1 , l. m. k. vandersypen , and a. f. morpurgo , solid state commun . * 143 * , 72 ( 2007 ) ; h. b. heersche , p. jarillo - herrero , j. b. oostinga , l. m. k. vandersypen and a. f. morpurgo , eur . special topic * 148 * , 27 ( 2007 ) [ graphene week 2006 international conference , dresden , germany , 2006 ] .

we calculate the phase , the temperature and the junction length dependence of the supercurrent for ballistic graphene josephson - junctions . for low temperatures we find non - sinusoidal dependence of the supercurrent on the superconductor phase difference for both short and long junctions . the skewness , which characterizes the deviation of the current - phase relation from a simple sinusoidal one , shows a linear dependence on the critical current for small currents . we discuss the similarities and differences with respect to the classical theory of josephson junctions , where the weak link is formed by a diffusive or ballistic metal . the relation to other recent theoretical results on graphene josephson junctions is pointed out and the possible experimental relevance of our work is considered as well .

different cosmological probes seem to point toward a consensus scenario characterized by a universe which is experiencing an accelerated expansion . standard candles ( snia , e.g. @xcite ) , standard rulers ( clusters , cmb and bao , e.g. @xcite ) and the angular power spectrum of the cmb ( e.g. @xcite ) are used to probe the hubble expansion flow up to redshifts 1 - 2 ( by means of snia or clusters ) or up to the epoch of recombination ( through the cmb anisotropies ) . gamma ray bursts are observed up to very high redshifts : the farthest is grb080913c at z=6.7 @xcite . among grbs with known redshifts , 45% are at @xmath1 and 8% at @xmath2 . the high grb luminosity ( 10@xmath3 erg ) and their detection in the @xmath4ray band makes them attractive as a potential and complementary cosmological tool to constrain the cosmological models at @xmath1 . however , the problem is that grbs are all but standard candles @xcite : their isotropic equivalent energetics and luminosities span 3 - 4 orders of magnitudes . similarly to snia , it has been proposed to use correlations between various properties of the prompt emission @xcite and also of the afterglow emission @xcite to standardize grb energetics . * isotropic * energies ( luminosities ) can be computed for grbs with measured redshifts and well constrained spectral properties . the spectrum gives the bolometric fluence @xmath5 ( peak flux @xmath6 ) and then @xmath7 ( @xmath8 ) . these two quantities are strongly correlated with the rest frame peak energy of the @xmath9 spectrum @xcite . we have updated the sample of bursts with known @xmath10 and spectral parameters to jan 2009 ( the last being grb 090102 at z=1.547 ) . these are 97 grbs . the fit of both correlations with a powerlaw gives @xmath11@xmath12 and @xmath11@xmath13 . however , due to the large scatter of the data points @xmath14 is extremely large ( 492 and 612 for 95 degrees of freedom for the and correlations , respectively ) . the scatter of the data points is defined by their distance from the best fit line . by modeling the scatter distribution with a gaussian we find a logarithmic dispersion of @xmath150.23 dex and @xmath16=0.28 dex for the and correlation , respectively . these scatters are much larger than the statistical errors associated with the observables @xmath17 , @xmath18 , @xmath19 . there is then the possibility that a third variable is responsible for this large scatter . dispersion ( modeled as a gaussian ) of the data points around the best fit lines . red open symbols are the grbs with a jet break time measurement reported in the literature . ] the large dispersion of the and correlations prevents their use to standardize grb energetics . however , grbs are thought to be collimated sources . in the standard grb model the jetted outflow should produce a break in the afterglow light curve decay @xcite . the measure of the time of occurrence of this break @xmath20 allows to infer the jet opening angle @xmath21 ( under the standard afterglow model assumptions ) and to recover the * collimation corrected * energy @xmath22 . by collecting available estimates of jet break times from the literature we found @xcite and later confirmed @xcite that the collimation corrected energy is strongly correlated with . in fig.[fg1 ] the correlation is shown through the 29 bursts having a jet break in their optical afterglow light curves . we find @xmath23 with @xmath14=37.8 for 27 degrees of freedom assuming a wind like profile of the circumburst medium . this correlation has a dispersion @xmath24 which is consistent with the average statistical uncertainties on and and it is much smaller than the dispersion of the and correlations . this result also suggests that the dispersion of the correlation is due to the jet opening angle : by correcting for @xmath21 ( for each burst ) , the scatter of the correlation is reduced . due to its tightness the correlation can be used to constrain the cosmological parameters @xcite . the correlation , shown in fig.[fg1 ] , was found by assuming a cosmological model ( i.e. @xmath25 , @xmath26 ) . it would be a circular argument to use this particular correlation to constrain the cosmological parameters . we originally solved this problem by properly accounting for the dependence of the correlation from the cosmological parameters @xcite or by adopting a bayesian fitting method @xcite . fig.[fg2 ] shows the cosmological constraints obtained through the correlation with the most updated sample of 29 grb . these are compared with the constraints obtained with the sample of 19 grbs @xcite . also in this case we have applied the bayesian method that overcomes the circularity problem . the correlation is derived in the standard uniform jet scenario assuming a constant radiative efficiency and either a uniform @xcite or a wind @xcite circumburst density profile . @xcite discovered a completely empirical correlation @xmath27 between the three observables which are combined in the correlation . also the empirical correlation , due to its low scatter , can be used to derive constraints on the cosmological parameters @xcite . the mostly debated issue of the spectral energy correlations in general is that they are due to selection effects @xcite . it has also been claimed that the slope and normalization of the correlation evolve with redshift @xcite selection effects can be studied in the observational plane corresponding to these correlations : the and the planes , where @xmath5 and @xmath6 are the bolometic fluence and peak flux . we have studied @xcite two instrumental selection effects : the minimum flux required to trigger a burst ( trigger threshold ) and the minimum flux to properly analyze its prompt emission spectrum ( spectral threshold ) . the former has been claimed to bias the correlation @xcite . the sample of 76 bursts ( updated to sept . 2007 ) with measured redshifts is composed by grbs detected by different instruments . for this reason we modeled the trigger threshold and the spectral analysis threshold of the different detectors . we can exclude that the correlation is biased by the trigger threshold . also we can exclude that the spectral threshold is biasing the pre - swift sample . instead , the swift grb sample ( 27 events ) is biased by the spectral threshold . this is also due to the limited spectral energy range of the bat instrument on - board swift which limits the measure of in the 15 - 150 kev energy range . by considering sub - samples of bursts at different redshifts , we also exclude that the correlation slope or normalization change . recently @xcite added grbs without redshift in the and planes to compare the distribution of bursts with respect to the two selection effects . to this purpose a sample of 100 faint batse bursts , representative of a larger population of 1000 objects , was analyzed . by means of this complete , fluence - limited , grb sample , it was found that the fainter batse bursts have smaller than those of bright events . as a consequence , the of these bursts is correlated with the fluence , though with a slope flatter than that defined by bursts with z. selection effects , which are present , are not responsible for the existence of such a correlation . about six per cent of these bursts are surely outliers of the correlation , since they are inconsistent with it for any redshift . also correlates with the peak flux , with a slope similar to the correlation . in this case , there is only one sure outlier . the scatter of the correlation defined by the batse bursts of this sample is significantly smaller than the correlation of the same bursts , while for the bursts with known redshift the correlation is tighter than the one . once a very large number of bursts with and redshift will be available , we expect that the correlation will be similar to that currently found , whereas it is very likely that the correlation will become flatter and with a larger scatter . one of the main drawback of the correlation for cosmological use is that it still has few points : the original sample of 15 events @xcite has only doubled since 2004 . with the launch of swift @xcite several jet breaks were expected to be measured , especially in the x ray band . jet breaks should be achromatic because they are produced by a geometric effect . all the jet breaks used to compute in the pre - swift era were , instead , obtained from the optical light curves . swift revealed a complex x ray afterglow light curve : the early afterglow is often characterized by a steep decay followed by a shallow phase lasting thousands of seconds @xcite . a characteristic break time is that ending the x ray shallow phase . @xcite showed that this time is not a jet break . a possible interpretation is that the shallow phase is produced by a long lasting central engine activity @xcite as also supported by the presence of strong precursors , post - cursors , and x - ray flares in a sizable fraction of bursts . often the x - ray and the optical afterglow light curves do not track one another , suggesting that they are two different emission components . we selected a sample of 33 gamma ray bursts ( grbs ) detected by swift , with known redshift and optical extinction at the host frame @xcite . the de absorbed and k corrected x ray and optical rest frame light curves are modelled as the sum of two components : emission from the forward shock due to the interaction of a fireball with the circum - burst medium and an additional component , treated in a completely phenomenological way . the latter can be identified , among other possibilities , as `` late prompt '' emission produced by a long lived central engine with mechanisms similar to those responsible for the production of the `` standard '' early prompt radiation . we find a good agreement with the data , despite of their complexity and diversity . our approach allows us to interpret the behaviour of the optical and x - ray afterglows in a coherent way , by a relatively simple scenario . within this context it is possible to explain why sometimes no jet break is observed ; why , even if a jet break is observed , it is often chromatic ; why the steepening after the jet break time is often shallower than predicted . the use of the correlation for cosmology requires to measure the redshift , the prompt emission and @xmath20 . the latter is the most critical observable : the jet break is typically observed at 1 - 2 days after the trigger and the light curve needs to be sampled at much later epochs in order to infer @xmath20 when the afterglow can be very dim , also because swift detects higher redshift bursts than before @xcite . therefore , it is interesting to explore if other correlations can be employed to standardize grb energetics . recent attempts @xcite tried to use the and correlations . in this cases , however , one has to take into account that these correlations are affected by a dispersion which is much larger than the statistical uncertainty on the data points . the scatter of these correlations is due to three terms . one is the statistical uncertainty in the measurements of the parameters @xmath28 . a second contribution to the scatter comes from systematic errors @xmath29 which could also have a physical origin but is difficult to model . a third contribution to the scatter could finally be due to the cosmological model @xmath30 : in the real cosmology its contribution should be minimized . in the and correlations the last two terms are dominating the scatter . vs @xmath31 are normalized to the average value of the statistical error associate to the variable they are assigned to . top panel : the extra - scatter term is assigned to and the results similar to those of @xcite are found . a minimum ( though weak ) is found for @xmath32 . the solid curve is found with the likelihood function of eq.1 , the dotted and dashed curved are found with the symmetric likelihood function of @xcite . the dot - dashed line is found with the @xmath33 fitting method . bottom panel : the extra - scatter term is assigned to . this is also the most obvious assumption as it is to depend on the cosmological parameters . no minimum is found in this case of the extra - scatter term for any value of @xmath31.,scaledwidth=80.0% ] @xcite ( @xcite ) proposed the use of the ( ) correlation to constrain the cosmological parameters . what is appealing is the possibility to use a correlation defined by a large grb sample ( much larger than that defining the correlation ) and to use only on prompt emission observables ( and or ) . however , in both correlations the non statistical scatter @xmath34 need to be modeled . these terms ( combined ) are treated as a free parameter , i.e. the extra scatter @xmath35 which is assumed to have a gaussian distribution equal for all the data points . to test the possibility of using the correlation for cosmology , @xcite fit the correlation in different `` cosmologies '' ( a flat universe is assumed ) and derive , in function of @xmath31 , the best fit values of the free parameters , i.e. the slope @xmath36 and the normalization @xmath37 of the correlation and the extra scatter term @xmath35 . the correlation is fitted with the likelihood function : @xmath38=\frac{1}{2}\sum_{i } \log[{1\over 2\pi(m^2\sigma_{x , i}^2+\sigma_{y , i}^2+m^2\sigma_{x}^2+\sigma_{y}^2 } ] + & \\ - \frac{(y_{i}-mx_{i}-q)^2}{m^2\sigma_{x , i}^2+\sigma_{y , i}^2+m^2\sigma_{x}^2+\sigma_{y}^2 } & \end{aligned}\ ] ] @xcite assume y= , x = and set the extra scatter term @xmath41 , i.e. they give the ( free ) extra scatter only to . they find that the extra scatter @xmath42 shows a minimum corresponding @xmath43 . therefore , they apply the standard procedure to derive constraints on the cosmological models through the correlation . however , in the ( ) correlations it is ( ) that depends on the cosmological parameters ( through the luminosity distance @xmath44 ) . therefore , the extra - scatter term should be assigned to x= , i.e. @xmath45 . for this reason we repeated the same test on the and correlations and we do not find any minimum of the extra - scatter @xmath46 when it is assigned to which is actually the variable that depends on the cosmological parameters . we verified our results also by ( a ) adopting the symmetric likelihood function of @xcite ( having the term @xmath47 in the numerator of the first term of eq.1 ) ; ( b ) fitting with the least square method ; ( c ) inverting the order of the fitting variables ( i.e. setting y = and x= ) . similar results are also found for the correlation . our results are shown in fig.[fg3 ] . in the correlation ( fig.[fg1 ] ) the scatter of the data points is already consistent with the statistical errors associated with and . the only residual scatter is due to the cosmological model . this is why , without assumptions on the nature and `` normality '' of the unknown extra scatter term , the correlation is preferable to standardize the grb energetics . the correlation also proves that most of the scatter of the correlation is due to the jet opening angle , which is different from burst to burst . this shows ( as also recently demonstrated by @xcite ) that the and correlations can not be used straightforwardly to constrain the cosmological parameters due to the unknown nature of the extra scatter they are affected by . 9 astier , p. , guy , j. , regnault , n. , et al . , _ a&a _ , 2006 , 447 , 31 percival , w. j. , cole , s. , eisenstein , d. j. , et al . , _ mnras _ , 2007 , 381 , 1053 lewis , a. , _ phrvd _ , 2008 , 78 , 3002 greiner , j. , kruehler , t. , fynbo , j. p. u. , et al . , _ apj subm . ( arxiv:0810.2314 ) _ , 2009 bloom , j. s. , frail , d. a. , kulkarni , s. r. , _ apj _ , 2003 , 549 , 674 firmani , c. , ghisellini , g. , avila - reese , v. , et al . , _ mnras _ , 2006 , 370 , 185 firmani , c. , avila - reese , v. , ghisellini , g. , et al . ; _ mnras _ , 2006 , 372 , l28 ghirlanda , g. , ghisellini , g. , lazzati , d. , _ apj _ , 2004 , 616 , 331 ghirlanda , g. , ghisellini , g. , lazzati , d. , et al . , _ apj _ , 2004 , 613 , l13 liang , e. and zhang , b. , _ apj _ , 2005 , 633 , 611 schaefer , b. , _ apj _ , 2007 , 660 , 16 amati , l. , frontera , f. , tavani , m. , et al . , _ a&a _ , 2002 , 390 , 81 yonetoku , d. , murakami , t. , nakamura , t. , et al . , _ apj _ , 2004 , 609 , 935 rhoads , j. , e. , _ apj _ , 1997 , 487 , l1 nava , l. ; ghisellini , g. ; ghirlanda , g. , et al . , _ a&a _ , 2006 , 450 , 471 ghirlanda , g. ; ghisellini , g. ; firmani , c. , _ njph _ , 8 , 123 riess , a. g. , strolger , l .- g . , tonry , j. , et al . , _ apj _ , 2004 , 607 , 665 butler , n. r. , kocevski , d. , bloom , j. s. , _ aipc _ , 1000 , 16 li , li - xin , _ mnras _ , 2007 , 379 , l55 ghirlanda , g. , nava , l. , ghisellini , g. , et al . , _ mnras _ , 2008 , 387 , 319 nava , l. , ghirlanda , g. , ghisellini , g. , et al . , _ mnras _ , 2008 , 391 , 639 gehrels , n. , chincarini , g. , giommi , p. , et al . , _ apj _ , 2004 , 611 , 1005 burrows , d. n. , romano , p. , falcone , a. , et al . , _ sci _ , 2005 , 309 , 1833 nava , l. , ghisellini , g. , ghirlanda , g. , et al . , _ mnras _ , 2007 , 377 , 1464 ghisellini , g. , ghirlanda , g. , nava , l. , et al . , _ apj _ , 2007 , 658 , l75 ghisellini , g. , nardini , m. , ghirlanda , g. , et al . , _ mnras in press , arxiv0811.1038 _ , 2009 berger , e. , kulkarni , s. r. , fox , d. b. , et al . , _ apj _ , 2005 , 634 , 501 amati , l. , guidorzi , c. , frontera , f. , et al . , _ mnras _ , 2008 , 391 , 577 kodama , y. , yonetoku , d. , murakami , t. , et al . , _ mnras _ , 2008 , 391 , l1 reichart , d. e. , _ apj _ , 2001 , 553 , 235 basilakos , s. & perivolaropoulos , l. , _ mnras _ , 2008 , 391 , 411

several interesting correlations among gamma ray bursts ( grb ) prompt and afterglow properties have been found in the recent years . some of these correlations have been proposed also to standardize grb energetics to use them as standard candles in constraining the expansion history of the universe up to @xmath0 . however , given the still unexplained nature of most of these correlations , only the less scattered correlations can be used for constraining the cosmological parameters . the updated correlation is presented . caveats of alternative methods of standardizing grb energetics are discussed . address = inaf - osservatorio astronomico di brera . via e. bianchi 46 , i-23807 merate ( lc ) , italy

was discovered as part of the hamburg / eso quasar survey and was first identified as a gravitational lens candidate by wisotzki et al . ( @xcite ) . it consists of two lensed images of a radio - quiet quasar at @xmath1= 2.319 that are separated by @xmath23.2 . the lensing galaxy was discovered from ground based near - ir ( courbin , lidman & magain , @xcite ) and hst optical observations ( remy et al . @xcite ; hereafter r98 ) and is 1 from the brighter quasar image ( component a ) . from a spectrophotometric monitoring program that lasted several years , wisotzki et al . ( @xcite ; hereafter w98 ) measured a time delay of 0.73 years between the two components . to convert this into an estimate of the hubble constant , one needs to determine the geometry of the system ( astrometry and lens and source redshifts ) and to model the mass distribution of the lens . several mass models for the lens have been published ( w98 ; r98 ; courbin et al . @xcite ; hereafter c2000 ; lehar et al . @xcite ) and precise astrometry from hst images is available . since the source redshift is known , the remaining unknown is the redshift of the lens . in their discovery paper , wisotzki et al . ( @xcite ) noted that the continuum of component a was considerably harder than that of component b , and that the broad emission lines of the two components were identical , only scaled by a factor of 2.8 . this was still the case during the spectrophotometric monitoring carried out by wisotzki et al . ( @xcite ) and in ir spectra covering the 1 to 2.5 @xmath3 m range ( c2000 ) . this has been interpreted as microlensing of the continuum emitting region of the quasar in component a , which is twice as close to the lensing galaxy as component b. presumably , the broad line region is too large to be affected . it may be possible to use this information to gain insight into the geometries and sizes of the continuum emitting and broad line regions ; however , detailed predictions require the lens redshift . from high resolution spectroscopic observations ( smette et al . @xcite ; lopez et al . @xcite ) , several metallic absorption line systems have been detected : @xmath4 and @xmath5 . the systems at @xmath6 , @xmath7 and @xmath8 contain lines that are mostly detected in component a , the component that is closest to the lens . at one time or another , each system has been individually proposed to be the lens . despite its importance , the redshift of the lensing galaxy has proved elusive . apart from the many unpublished attempts to measure it , the redshift of the lens has been estimated by indirect means . from ir and optical photometry , c2000 gave a redshift in the range @xmath9 to @xmath10 while from the time delay and a model of the lens , w98 estimated @xmath11 . from the position of the lens on the fundamental plane , kochanek et al . ( @xcite ) derived @xmath12 . although model dependent , the two latter estimates prove to be very close to the truth . in this paper we describe the successful measurement of the lens redshift , @xmath13 , using the spectral deconvolution method described by courbin et al . ( @xcite ) . the observations were taken with fors2 on kueyen ( vlt / ut2 ) at the cerro paranal observatory on april 1st , 2000 , which also happened to be the first scheduled observing night for this telescope , and consisted of three 1080 second exposures with the g200i grism and the high resolution collimator . this gives a scale of 0.1 per pixel in the spatial direction and a scale of approximately 4 per pixel in the spectral direction . the observing conditions were clear and the external seeing varied between 0.5 and 0.9 . we used the movable slits in the fors2 focal plane to select targets and set the slit width . one slit was placed on and aligned along the two quasar components , three slits were placed on field stars that were used to determine the psf required for the deconvolution ( see below ) and to correct for telluric absorption . used in this mode , the slits of fors2 have a fixed length . this is less flexible than punched or laser designed masks , but it was adequate for our experiment . the slit width was set to 1 , which is larger than the fwhm of the images . for the deconvolution to work well , it is important that the targets are not too heavily occulted by the slit edges , since this can lead to differential occultation between the psf stars and the quasar images and hence to a less accurate deconvolution . this can be minimized by carefully centroiding the stars in the slits and by having slits that are significantly wider than the psf . additionally , preparing the mask from an image that was taken with the same instrument , as was done in this case , also minimises differential alignment errors . instrumental signatures were removed in the standard manner using iraf . the bias was removed by subtracting a constant from each frame and the pixel - to - pixel sensitivities were normalised with screen flats . the arc frames were used to correct for slit distortion and to set the wavelength scale . all spectra were rebinned to the same linear wavelength scale . the position of the night sky lines and the absorption features in component a were used to check the accuracy of the wavelength calibration , which was found to be around @xmath14 . the lensing galaxy is four to five magnitudes fainter than component a of and only one arc second away . even with excellent seeing , the spectrum of the lens is heavily contaminated by that of the quasar and needs to be extracted with sophisticated techniques . the technique implemented by courbin et al . ( @xcite ) spatially deconvolves spectra and decomposes them into point - sources and extended components . it is therefore very well suited to the present problem : extracting the spectrum of a faint extended lens galaxy close to a bright quasar spectrum . as is the case with image deconvolution , a reference psf is necessary . out of the three psf stars that were available , two were unfortunately too faint . in order to build the psf spectrum , we therefore used only one psf star ( which was slightly fainter than the quasar itself ) in combination with the spectrum of the bright quasar image itself . only half of the spatial profile of the quasar was actually used , the one unaffected by light contamination from the lensing galaxy . the deconvolved spectrum has a pixel size of 0.05 while the undeconvolved data have a pixel size of 0.1 . the deconvolved spectrum is displayed in fig . [ 2d_spec ] . as the psf stars were observed at exactly the same time and under the same conditions as the lensed quasar , they could also be used to remove telluric features . the spectra of the psf stars were combined and then divided directly into the spectra of the lens and the two components of the quasar . before division , the continuum of the psf stars was normalised with a high - order function and obvious absorption features , such as @xmath15 and the calcium triplet at 8600 , were removed . as we do not know the spectral type of the psf stars , weaker features may be present . ideally , one should choose hot , featureless stars to make such a correction . despite this , the correction works rather well . the 1-d spectrum of the lens , extracted directly from the 2-d deconvolved image ( extended component only ) , is shown in fig . [ lens ] . an unsmoothed version and a smoothed version shifted by 0.03 units on the vertical scale are shown in the lower half of the figure . also shown , but reduced by a factor of 20 , is the spectrum of component a. the calcium h and k absorption lines of the lensing galaxy are clearly detected and the g - band is marginally detected . the redshift was measured by cross correlating the lens spectrum with the elliptical template from the kinney - calzetti spectral atlas ( kinney et al . @xcite ) , which has been plotted ( third curve from the bottom ) with a shift of 0.05 units in fig . for this we used the iraf task rvsao v2.0 ( kurtz and mink 1998 ) . the redshift is @xmath0 , where the error incorporates the internal error reported by the cross - correlation routine and the error in the wavelength calibration . an alternative measure of the accuracy is given by the r - statistic of tonry and davis ( @xcite ) , which is the signal - to - noise ratio of the main peak in the cross correlation . we find that @xmath16 , which indicates that the redshift is reliable ( kurtz & mink @xcite ) . the redshift is in good agreement with the redshift of the multiple absorption line system at a redshift of @xmath6 ( smette et al . @xcite ; lopez , @xcite ) . this absorption line system is made up of many sub - components spread over a range of redshifts ( @xmath17 ) . although it is not possible to say if the absorption line system and the lens are the same object , the two systems are probably dynamically associated . as the quality of the psf was not optimal , the quasar spectra from the deconvolved images are noisier than what one would calculate from the photon noise only . therefore , the extraction of the quasar spectra was not done from the deconvolved spectrum but with narrow apertures from the undeconvolved spectrum . since the galaxy is much fainter than either component of the quasar and since it is not spatially coincident , the contamination of the quasar spectra by the lensing galaxy is negligible . the spectra of the two components of the quasar are shown in fig . [ quasar ] . in this figure we plot the data before and after the telluric features have been removed . the a and b atmospheric bands are removed very well and the feii absorption lines at @xmath18 become visible . additionally , the mgii emission line , which is almost obscured by the strong telluric feature at 9300 , is recovered . as in w98 and c2000 , we also plot the difference between the spectrum of component a and a scaled version of the spectrum of component b , that is @xmath19 , where , @xmath20 , is set to 2.8 . the difference is plotted in fig . [ diff ] , where we plot it before and after telluric correction . apart from a slight excess at 6390 and the metallic absorption lines in component a , the broad lines of the quasar disappear and only the excess continuum from component a remains . the strongest absorbers are now clear and are marked with the vertical dotted lines . the wavelengths of the main absorption lines and the redshifts of the corresponding absorption line systems are given in smette et al . ( @xcite ) . the aim of our observations were modest : to measure the redshift of the lensing galaxy in , one of the ever growing list of lenses with known time delays . this was possible with only one hour of vlt time during average weather conditions . the prospects of measuring the lens redshifts in other systems are therefore very good , at least for double systems , where aligning the slit simultaneously on the quasar images and on the lens is straightforward . with the time delay and the lens and source redshifts known and several mass models published , it is relatively straight forward to derive the hubble constant . w98 present two models to describe the lens : a singular isothermal sphere with external shear and a singular isothermal ellipsoid without external shear . for the former ( the model presented by r98 is very similar ) , one derives @xmath21 km / s / mpc and for the latter , one derives @xmath22 km / s / mpc . we have set @xmath23 . c2000 also present two models : a singular ellipsoid without external shear and a two component model which consists of a singular isothermal ellipsoid to represent the lensing galaxy and a more extended component representing a galaxy cluster centered on the lensing galaxy . the first model leads to @xmath24 km / s / mpc , but the second leads to values for the hubble constant which are a factor of 1.5 to 2 lower . is somewhat unusual in that the lensing galaxy is closest to the brighter component . it has been pointed out by several authors that the lens probably does not consist of a single galaxy ( w98 , r98 , c2000 ; lehar et al . additional extended mass is required to match all the observational constraints : the position of the two images relative to the lens , the magnification ratio of the images , and the lens orientation and ellipticity . deep images in the near - ir may be able to reveal the distribution of this additional mass . as the lens appears to be associated with the absorption system seen in the quasar spectrum at @xmath6 , one could use the velocity spread in this system to constrain the lens velocity dispersion . however , this assumes that the absorption line system and the lens are the same object or , at the very least , in the same potential well . without further supporting evidence that links the absorption line system to lens , it is premature to use the velocity spread as a constraint on the mass models . the lens redshift estimates of w98 and kochanek et al . ( @xcite ) have proven to be accurate , whereas the estimate of c2000 from ir and optical photometry is less so . the optical photometry is in good agreement with the vlt spectrum ; however , if the lens is an elliptical galaxy , then the ir points are over - estimated by 0.3 magnitudes . the cause of this discrepancy requires further investigation . it is a pleasure to thank thomas szeifert for his expert support at the telescope . frdric courbin acknowledges financial support from the chilean grant fondecyt/3990024 . additional support from the european southern observatory , through cnrs / conicyt grant 8730 `` mirages gravitationnels avec le vlt : distribution de matire noire et contraintes cosmologiques''as well as through the tmr lensnet network ( erbfmrxct97 - 0172 ) , is also gratefully acknowledged .

using fors2 , mounted on kueyen ( ut2 of the vlt ) , we have obtained the redshift of the lensing galaxy in the gravitational lens system . we measure @xmath0 , in good agreement with previous estimates based on the time delay and the position of the lens on the fundamental plane . it also coincides with one of the metal line absorption systems that are seen in high resolution spectra of .

stars form out of gas in the densest areas of molecular clouds . early studies of molecular cloud structure ( e.g. @xcite ) concluded that a division into discrete clumps of emission was the best description . within such clumps , star formation occurs inside denser cores @xcite . a growing consensus now maintains that molecular clouds have a scale - free structure governed by turbulence ( e.g.@xcite ) , with clumps only an arbitrary categorization . we may reconcile these viewpoints to some extent ; the clump population has a power - law spectrum of mass implying the overall cloud has a similar distribution ( e.g. @xcite ) . equally , the self - similarity of molecular clouds must break down where gravitational collapse becomes important . this paper focusses on the decomposition of molecular clouds into _ clumps _ of emission . the utility of such a description depends on whether the located clumps accurately represent star - forming cores . clumps are typically located in either dust continuum or spectral - line datasets , which both have their own advantages and drawbacks . continuum observations select high - density cores , which tend to be self - gravitating , but associate more mass to clumps than exists in reality , since objects may be superposed along the line of sight ( e.g. @xcite ) . clump - finding analyses on spectral - line cubes , with their added velocity dimension , should enable clumps to be more accurately assigned but are subject to larger uncertainties from the details of the radiative transfer than continuum emission . ultimately , high signal - to - noise ratios are required in both cases for clumps to converge towards the underlying core population @xcite . nevertheless , many insights into star formation from clump populations remain compelling . @xcite first pointed out that the mass distribution of compact continuum clumps ( the clump mass distribution ; hereafter cmd ) is remarkably similar to the initial mass function of stars ( imf ) . at high masses , the slope is consistent with a salpeter power law ( @xmath7 , @xcite ) , considerably steeper than the power law for co clumps ( @xmath8 , e.g. @xcite ) . at lower masses , the cmd slope becomes shallower , around the point that samples start to suffer from incompleteness . this suggests that the mass of a star is established at its earliest phases ( see also @xcite ) and would seem to rule out models where the shape of the imf is formed later , through e.g. dynamical interactions @xcite . however , the mapping of the cmd on to the imf is not straightforward and many evolutionary schemes would fit the present data ( e.g. @xcite ) . many wide - field surveys are about to or have just begun , which will locate and characterize thousands of star - forming cores from continuum data , e.g. the james clerk maxwell telescope ( jcmt ) and _ herschel _ gould belt surveys @xcite . one of the key science drivers for these projects is the measurement of the cmd to high precision at low masses , into the brown - dwarf regime ( @xmath9m@xmath6 ) . thus , an examination of different source - finding techniques is particularly timely . many different methods have been deployed in the literature to locate clumps in molecular - line and continuum data , from identifications by eye to automated algorithms . however , few studies _ compare _ the sources located with different techniques ( a notable exception is @xcite ) . in this paper , we closely compare the populations of continuum clumps found in the perseus molecular cloud ( hereafter simply perseus ) using the two most popular automated algorithms , clfind @xcite and gaussclumps@xcite , to highlight their differences and biases . we re - examine scuba 850 data , presented originally by @xcite , in the four clusters of cores where we have complementary harp spectral - line data @xcite , namely : ngc 1333 , ic348/hh211 ( ic348 for short ) , l1448 and l1455 . although only investigating a sub - set of the scuba data limits our sample size , these are the sources whose kinematics we will investigate subsequently and any cmds from the spectral - line data will be directly comparable ( curtis & richer , in prep . ) . the regions selected still encompass the majority of the scuba cores identified by @xcite , 58 out of 103 ; so we can extrapolate any conclusions to the entire cloud population . this work is divided into two principal parts . in section [ sec : clumppopulation ] , we identify clumps using the two algorithms before matching them to a catalogue of scuba cores , classified as protostellar or starless on the basis of their seds by @xcite ( hereafter ) . second , sections [ sec : properties ] and [ sec : cmd ] present an analysis of the physical properties of the cores before we summarize our conclusions in section [ sec : summary ] . many different terminologies have been used to describe the hierarchical structure in molecular clouds ; we follow @xcite . within molecular clouds , individual over - densities are termed _ clumps _ , these are the objects identified by automated algorithms and do not necessarily go on to form stars . clumps may contain _ cores _ , which are the direct precursors of individual or multiple stars . every clump that does not contain an embedded object is referred to as _ starless_. of these starless cores , only a subset , the _ prestellar _ cores ( formerly pre - protostellar cores , @xcite ) will be gravitationally bound and thus go on to form stars . we extracted fully calibrated and reduced scuba 850 maps across the four regions in perseus we observed with harp from the data presented by @xcite , where we refer the reader for details of the observations and processing . in short , the data were taken during 20 nights between 1999 and 2003 . the sky opacity measured at 225ghz varied from @xmath10 to 0.080 , with most of the data taken in good conditions ( @xmath11 ) . the beam size is 14arcsec ( 0.017pc at 250pc , our assumed distance to perseus ) and the maps are sampled on a 3arcsec grid . the rms noise level varies across the maps from typical values of @xmath12mjybeam@xmath13 in ngc 1333 and ic348 to higher values , 32 and 46mjybeam@xmath13 in l1448 and l1455 . hereafter , to be consistent with the @xcite studies and therefore allowing a direct comparison of our source catalogues , we take the noise to be @xmath14mjybeam@xmath13 , the typical value for the _ entire _ cloud . ( sub)millimetre continuum mapping selects young , cold cores ( 520k ) at the earliest stages of star formation . these largely self - gravitating cores are intermediate in properties between co clumps and infrared young stellar objects ( see @xcite for a review ) . the dust s thermal emission is optically thin nearly everywhere at these wavelengths , accurately tracing density throughout the cloud . 1 . ` clumpfind ' or clfind @xcite , which contours the data at evenly - spaced levels , starting close to the peak value . all distinct contiguous areas are considered above each contour level in turn . if an area has no pixels previously assigned to a clump , it is marked as a new clump . alternatively if there is a pre - existing clump assignment , that clump is extended to the next lowest contour level . when two or more different assignments exist in a area , the pixels are distributed using a friend - of - friends algorithm . gaussclumps @xcite , which uses a least - squares fitting procedure to break - down the data into a number of gaussian - shaped clumps , in a fashion similar to the clean algorithms used in interferometric mapping . clfind is a partitioning method with each pixel only allocated to a single clump , whereas gaussclumps permits any number of clumps to overlap at a single spatial position . this means gaussclumps is better at handling blended sources , although its output clumps necessarily conform to a strict gaussian profile as opposed to the arbitrary shapes from clfind . the way each algorithm find clumps of emission is controlled by a number of tuning parameters . our overall aim in this paper is to compare the populations of clumps found _ independently _ by clfind and gaussclumps . we therefore try to find the same objects with both algorithms , namely clumps larger than the beamsize with peaks greater than @xmath15 , but do not try to make the algorithms find exactly the same identifications clump - to - clump . in so doing , we follow the majority of previous studies and select parameters for each algorithm that are considered optimal for rejecting spurious detections . in an alternative approach , one could try to produce clump catalogues which were as similar as possible . however , the standard algorithm outputs we supply are quite different because of the very different way clumps are found with the two algorithms . given that we can never definitively know the true , underlying clump population , then selecting one algorithm population to make more similar to the other would be completely arbitrary _ a priori_. indeed , it is only through detailed analysis , such as the one presented in this paper , or monte - carlo simulations that such a decision regarding algorithm selection could be made . the most important parameters to tune clfind s output are the value of the lowest contour level ( @xmath16 ) and the contour level spacing ( @xmath17 ) . @xmath18 of 2@xmath19 is considered optimal for rejecting clumps otherwise fitted to noise spikes @xcite . @xmath20 essentially sets the clump peak threshold ( @xmath21 ) . after some experimentation we settled on the optimum parameters to find plausible clumps : @xmath18 and @xmath16 both @xmath22 , so clumps must have peaks @xmath23 . additionally , the clumps had to be larger than the beam size and contain at least 7 pixels . we find 85 clumps in this manner which are depicted across the four regions in fig . [ fig : clfind_clumps ] . the full catalogue of clfind detections is listed in tab . [ tab : clfind_detections ] . for the gaussclumps algorithm the most sensitive tuning parameter was the threshold , the minimum value at which the routine would attempt a model fit ( @xmath24 ) . as for clfind , clumps with widths smaller than the beam size ( 14arcsec ) were rejected . in total , 122 clumps were located in the four regions as shown in fig . [ fig : gauss_clumps ] . tab.[tab : gclumps_detections ] provides the full the list of gaussclumps identifications . although we identify clumps in the same data with the same algorithm , find fewer clumps ( the total number of clumps found in this work and is summarized in tab . [ table : numberclumps_allalgs ] ) . use the idl implementation of clfind on the scuba maps with an unsharp mask applied to remove structure on spatial scales @xmath25arcmin . the most plausible explanation for this discrepancy is the different clump threshold , although there are other differences . in our catalogue , we include all clfind detections found with @xmath26 and @xmath27 . this ensures the detections all have peaks @xmath28 . however , only select their clfind detections ( with @xmath29 and @xmath30 ) that have peaks @xmath31 . additionally , they select less bright sources , with peaks @xmath32 that also have an identification by eye in their original paper @xcite or are bolocam sources @xcite . + @xmath33 peak clump flux position . + @xmath34 clump effective radius , the geometric mean of the clump major and minor ` sizes ' ( @xmath35 ) . each size is formed from the standard deviation of the clump co - ordinates about the centroid position , weighted by the pixel data values . + @xmath36 clump peak flux . + @xmath37 clump total flux . + @xmath38 clump mass derived from the total flux assuming optically thin , isothermal dust at 12 k ( see section [ sec : dustmasses ] ) .

we present a new analysis of the properties of star - forming cores in the perseus molecular cloud , identified in scuba 850data originally presented by @xcite . our goal is to determine which core properties can be robustly identified and which depend on the extraction technique . four regions in the cloud are examined : ngc 1333 , ic348/hh211 , l1448 and l1455 . we identify clumps of dust emission using two popular automated algorithms , clfind and gaussclumps , finding 85 and 122 clumps in total respectively . using the catalogues of @xcite , we separate these clumps into starless , class 0 and class i cores . some trends are true for both populations : clumps become increasingly elongated over time ; clumps are consistent with constant surface brightness objects ( i.e. @xmath0 ) , with an average brightness @xmath110 times larger than the surrounding molecular cloud ; the clump mass distribution ( cmd ) resembles the stellar intial mass function , with a slope @xmath2 for clfind and @xmath3 for gaussclumps , which straddle the salpeter value ( @xmath4 ) . the mass at which the slope shallows ( similar for both algorithms at @xmath5m@xmath6 ) implies a star - forming efficiency of between 10 and 20per cent . other trends reported elsewhere depend critically on the clump - finding technique : we find protostellar clumps are both smaller ( for gaussclumps ) and larger ( for clfind ) than their starless counterparts ; the functional form , best - fitting to the cmd , is different for the two algorithms . the gaussclumps cmd is best - fitted with a log - normal distribution , whereas a broken power law is best for clfind ; the reported lack of massive starless cores in previous studies ( e.g.@xcite ) can be seen in the clfind but not the gaussclumps data . our approach , exploiting two extraction techniques , highlights similarities and differences between the clump populations , illustrating the caution that must be exercised when comparing results from different studies and interpreting the properties of samples of continuum cores . [ firstpage ] submillimetre dust : extinction stars : formation stars : evolution ism : clouds ism : individual : perseus .

[ [ an - intriguing - discrepancy . ] ] an intriguing discrepancy . + + + + + + + + + + + + + + + + + + + + + + + + + + there is an intriguing and longrunning discrepancy in categorical logic between the way conjunction is coupled to implication in cartesian closed categories , and the way existential quantification is coupled to universal quantification in hyperdoctrines . in a cartesian closed category @xmath1 , every object @xmath2 induces an adjunction @xmath3 where the implication functor @xmath4 is right adjoint to the conjunction functor @xmath5 this categorical situation should be compared with the way quantification is handled in a hyperdoctrine . recall that a hyperdoctrine in the sense of lawvere is first of all a ( pseudo ) functor @xmath6 from a base category @xmath7 to the category @xmath8 of small categories and functors . the intuition behind this definition is that every object @xmath2 of the category @xmath7 is assigned a `` category of predicates '' noted @xmath9 , and every morphism @xmath10 of @xmath7 induces a functor @xmath11 called `` substitution '' along @xmath12 . the leading example of a hyperdoctrine is the `` subset hyperdoctrine '' with basis the category @xmath13 of sets and functions , equipped with the powerset functor @xmath14 which transports every set @xmath2 to the set @xmath15 of subsets of @xmath2 ordered by inclusion . note that the ordered set @xmath9 is seen here as the ordered category where two subsets @xmath16 are related by a morphism @xmath17 precisely when @xmath18 . the substitution functor along a function @xmath10 is defined by transporting every subset @xmath19 to its inverse image @xmath20 the definition of a hyperdoctrine then additionally asks for a pair of functors @xmath21 called `` existential quantification '' and `` universal quantification '' along @xmath12 , which are respectively left and right adjoint to the substitution functor : @xmath22 in the case of the subset hyperdoctrine , the functors @xmath23 and @xmath24 transport a subset @xmath25 to the following subsets of @xmath26 : @xmath27 the difference between ( [ equation / ccc - adjunction ] ) and ( [ equation / hyperdoctrine - adjunction ] ) is especially notable if one thinks of dependent type theory , where existential quantification provides a dependent form of conjunction , and universal quantification a dependent form of implication . it is thus puzzling to see conjunction and implication directly coupled by an adjunction in ( [ equation / ccc - adjunction ] ) while they form in ( [ equation / hyperdoctrine - adjunction ] ) a `` mnage trois '' with the substitution functor @xmath28 as intermediate . in the present introduction , we explain how to reconcile the two points of view in the specific subset hyperdoctrine on @xmath13 . the choice of this hyperdoctrine is mainly pedagogical : we find clarifying to explain some of our ideas in this familiar example . however , as we will see , the ideas developed in this introduction lift very smoothly to the more sophisticated situation when one replaces @xmath13 by the cartesian closed category @xmath29 of small categories and functors , and where the `` category of predicates '' @xmath9 over a small category @xmath2 is defined as the contravariant presheaf category @xmath30.\ ] ] lawvere introduced this example in his original article on hyperdoctrines @xcite , and also considered its restriction to presheaves over groupoids ( @xmath31 ) in his article describing a treatment of equality in hyperdoctrines , which relied on a `` frobenius reciprocity '' condition and certain beck - chevalley conditions @xcite . the presheaf hyperdoctrine is an important example despite the fact that it does not in general satisfy these conditions , and indeed , lawvere even writes that this fact < < should not be taken as indicative of a lack of vitality [ ... ] or even of a lack of a satisfactory theory of equality > > for the presheaf hyperdoctrine , but rather < < that we have probably been too naive in defining equality in a manner too closely suggested by the classical conception > > @xcite . we will come back to this important point later in the introduction . [ [ from - functions - to - relations . ] ] from functions to relations . + + + + + + + + + + + + + + + + + + + + + + + + + + + + our procedure to reconcile ( [ equation / ccc - adjunction ] ) and ( [ equation / hyperdoctrine - adjunction ] ) is inspired by linear logic and the shift from the cartesian closed category @xmath32 to the symmetric monoidal closed ( and in fact , compact closed ) category @xmath33 which underlies its discovery by girard @xcite . in particular , we will make a great usage of the two `` embedding '' functors @xmath34 which transport a set @xmath2 to itself , and a function @xmath10 to the binary relations @xmath35 where @xmath36 notation : we write @xmath37 for a binary relation @xmath38 which defines a morphism @xmath39 in the category @xmath33 . these two faithful ( but not full ) functors @xmath40 and @xmath41 transport the category @xmath32 and its opposite category @xmath42 in the same category @xmath33 . [ [ bifibrations . ] ] bifibrations . + + + + + + + + + + + + + another important ingredient and source of inspiration for our work is the notion of _ bifibration _ which we like to see as a particular instance of _ type refinement system _ in the terminology of @xcite . a bifibration may be defined as a functor @xmath43 which is at the same time a fibration and an opfibration . following the principles and notations of type refinement systems , an object @xmath44 of @xmath45 is said to `` refine '' an object @xmath2 of @xmath7 ( written @xmath46 ) if @xmath47 , while a `` derivation '' of a typing judgment @xmath48 ( where @xmath49 , @xmath46 , and @xmath50 ) is defined as a morphism @xmath51 in the category @xmath45 whose image by the functor @xmath52 is the morphism @xmath12 . the definition of a bifibration then asserts that the functor ( or `` refinement system '' ) @xmath52 is equipped with operations for pushing or pulling an object of @xmath45 along a morphism of @xmath7 @xmath53 such that there is a one - to - one correspondence of derivations , @xmath54 written here as invertible inference rules in the proof - theoretic style of refinement systems ( see @xcite for details ) . notably , this structure is sufficient to derive inference rules @xmath55 as well as isomorphisms @xmath56 @xmath57 and a three - way correspondence of derivations : @xmath58 this argument establishes that any ( cloven ) bifibration @xmath59 determines a pair of ( pseudo ) functors @xmath60 as well as a family of adjunctions @xmath61 relating the corresponding functors between the `` fibre '' categories @xmath62 and @xmath63 ( defined as subcategories of @xmath45 containing only those morphisms that project by @xmath52 to identity morphisms in @xmath7 ) . [ [ a - subset - bifibration - over - sets - and - relations . ] ] a subset bifibration over sets and relations . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + putting these two sources of inspiration together : linear logic and bifibrations , we construct a `` subset bifibration '' @xmath64 where the category @xmath65 has objects the pairs @xmath66 consisting of a set @xmath2 together with a subset @xmath25 ; and morphisms @xmath67 defined as the binary relations @xmath37 satisfying the property @xmath68 the functor @xmath52 transports every object @xmath66 to the first component @xmath2 , and every morphism @xmath69 to the underlying relation @xmath37 . the category @xmath65 may be seen as a category of `` pointed objects '' in @xmath33 , since an object @xmath66 is the same thing as a relation @xmath70 , with morphisms defined using the 2-categorical structure of @xmath33 provided by inclusion of binary relations . the fiber or category of predicates @xmath71 associated to a set @xmath2 by the functor @xmath52 is simply the set @xmath72 of subsets of @xmath2 ordered by inclusion . an important point is that the functor @xmath52 just defined is a bifibration . given a binary relation @xmath73 the two functors @xmath74 are defined in the following way : @xmath75 for all subsets @xmath25 and @xmath19 . an easy computation shows that @xmath76 and @xmath77 define a pair of adjoint functors @xmath78 because @xmath79 is equivalent to @xmath80 for every @xmath25 and @xmath19 . from this , we conclude that the functor @xmath81 is a bifibration . the associated fibre functor @xmath82 transports every set @xmath2 to the set @xmath72 of subsets of @xmath2 ordered by inclusion . [ [ one - hyperdoctrine - decomposed - into - two - bifibrations . ] ] one hyperdoctrine decomposed into two bifibrations . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the construction of the subset bifibration @xmath81 on sets and relations leads us to a new way to think about existential and universal quantification in the subset hyperdoctrine @xmath14 on the category @xmath13 . indeed , given a hyperdoctrine @xmath6 it is always possible to `` decorrelate '' the pair of adjunctions ( [ equation / hyperdoctrine - adjunction ] ) by defining a pair of ( pseudo ) functors @xmath83 where @xmath84 and where @xmath85 transports every object @xmath86 to the category @xmath9 and every morphism @xmath10 to the functor @xmath87 the key observation here is that the left - hand side adjunction @xmath88 of the hyperdoctrine @xmath52 ensures that @xmath89 determines a bifibration with basis the category @xmath7 , while the right - hand side adjunction @xmath90 ensures that @xmath85 determines a bifibration with basis the opposite category @xmath91 . in the case of the subset hyperdoctrine on @xmath13 , one obtains in this way two ( pseudo ) functors @xmath92 which provide an alternative and equivalent formulation of the original subset hyperdoctrine @xmath14 on the category @xmath32 . in particular , @xmath89 and @xmath85 determine a pair of bifibrations @xmath93 where the categories @xmath94 and @xmath95 have the same objects defined as pairs @xmath66 consisting of a set @xmath2 and of a subset @xmath25 , while the morphisms @xmath96 are defined as the functions @xmath10 satisfying the property @xmath97 in the case of @xmath94 and as the functions @xmath98 satisfying the property @xmath99 in the case of @xmath95 . note the change of orientation in the definition of the morphisms of @xmath94 and of @xmath95 . as expected , the functors @xmath100 and @xmath101 transport every such morphism @xmath102 to the underlying morphism @xmath10 in the category @xmath32 for the functor @xmath100 and in the category @xmath42 for the functor @xmath101 . [ [ putting - everything - back - together . ] ] putting everything back together . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + a quite extraordinary and instructive phenomenon appears at this point : the two bifibrations @xmath100 and @xmath101 and thus the hyperdoctrine @xmath14 on @xmath32 may be recovered from the bifibration @xmath81 and the two embedding functors : @xmath103 the reason is that , for every function @xmath10 , the following equations hold : @xmath104 for all @xmath25 . from this follows that @xmath105 by uniqueness of a left or of a right adjoint , these two equations together with ( [ equation / hyperdoctrine - adjunction ] ) and ( [ equation / rel - adjunction ] ) imply the series of equalities : @xmath106 the resulting picture reconciles ( [ equation / ccc - adjunction ] ) and ( [ equation / hyperdoctrine - adjunction ] ) since the original series of adjunctions of a hyperdoctrine ( [ equation / hyperdoctrine - adjunction ] ) is replaced by a pair of adjunctions @xmath107 living in two different bifibrations @xmath100 and @xmath101 , together with an equality between the two functors @xmath108 and @xmath109 . an interesting outcome of our decomposition of the subset hyperdoctrine @xmath14 over @xmath32 is that the existential quantification is entirely handled by the bifibration @xmath100 while the universal quantification is entirely handled by the bifibration @xmath101 . the decomposition reveals moreover that the substitution functor @xmath28 of the subset hyperdoctrine is not primitive , since it is the `` superposition '' of the two equal functors @xmath108 and @xmath109 . formally speaking , recall that every bifibration @xmath110 may be `` pulled back '' along a functor @xmath111 in order to define a bifibration @xmath112 on the category @xmath1 : @xmath113^ { } \ar[dddd]_-{q } \ar@{}[ddddrrrrrrrrrr]|-{pullback } & & & & & & & & & & { \mathscr{e}}\ar[dddd]^{p } \\ \\ & & & & & & & & & & \\ \\ { \mathscr{c}}\ar[rrrrrrrrrr]^{f } & & & & & & & & & & { \mathscr{b}}}\ ] ] whose fibre functor @xmath114 is simply obtained by precomposing the fibre functor @xmath14 of the bifibration @xmath52 with the functor @xmath115 : @xmath116 in other words , the category of predicates @xmath117 associated to an object @xmath2 of the category @xmath1 coincides with the category of predicates @xmath118 of its image by the functor @xmath119 . in the discussion above , we have just established that [ thm / reconstruction ] the two bifibrations @xmath120 are equal to the bifibration @xmath64 pulled back along the embedding functors @xmath121 and @xmath122 . this means that there exists a pair of pullback diagrams @xmath123 \ar[dddd]_-{{{p}^{\oplus } } } \ar@{}[ddddrrrrrr]|-{pullback } & & & & & & { { \mathbf{rel}_{\bullet } } } \ar@{}[ddddrrrrrr]|-{pullback } \ar[dddd]^{p } & & & & & & { { { { \mathbf{subset } } } } ^{\ominus } } \ar[llllll ] \ar[dddd]^-{{{p}^{\ominus } } } \\ \\ & & & & & & & & & & & & \\ \\ { { \mathbf{set } } } \ar[rrrrrr]^-{{{{\mathrm{emb}}}^{\oplus } } } & & & & & & { { \mathbf{rel } } } & & & & & & { { \mathbf{set } } } ^{\,op } \ar[llllll]_-{{{{\mathrm{emb}}}^{\ominus } } } } \ ] ] which enable us to derive the two bifibrations @xmath100 and @xmath101 and thus the subset hyperdoctrine @xmath14 on the category @xmath32 of sets and functions , from the subset bifibration @xmath52 on the category @xmath33 of sets and relations . [ [ a - monoidal - closed - refinement - system . ] ] a monoidal closed refinement system . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + now that we have given theoretical precedence to the subset bifibration @xmath81 over the hyperdoctrine @xmath14 on @xmath32 , there remains to study the properties of this bifibration @xmath52 more closely . in our work on refinement systems , we have advocated the fundamental role played by the interaction between the adjunctions of a monoidal closed refinement system which would be a bifibration at the same time : @xmath124 where @xmath44 is a refinement and @xmath10 a morphism of the basis category . recall that by `` ( symmetric ) monoidal closed refinement system '' , we simply mean a functor @xmath110 where the categories @xmath45 and @xmath7 are ( symmetric ) monoidal closed and where the functor @xmath52 preserves the ( symmetric ) monoidal closed structure of @xmath45 up to coherent isomorphisms . a primary observation is that the refinement system @xmath125 is symmetric monoidal closed with tensor product and implication of the category @xmath65 defined as @xmath126 where the subsets @xmath127 and @xmath128 of @xmath129 are defined as @xmath130 note that the implication @xmath131 in @xmath65 is transported by the functor @xmath52 to the set @xmath132 which plays the role of internal hom @xmath133 in the compact closed category @xmath134 . it is very instructive to study how the adjunctions @xmath135 coming from the bifibrational and monoidal closed structure of @xmath52 interact , for @xmath136 an object of @xmath65 and @xmath137 a morphism of @xmath33 . for instance , one has the equality @xmath138 for all subsets @xmath25 and @xmath19 and relations @xmath139 , @xmath140 . on the other hand , the canonical inclusion @xmath141 is not an equality in general , for subsets @xmath25 and @xmath19 and relations @xmath142 , @xmath143 . consider for instance the case where @xmath144 , where the two subsets @xmath145 are not equal for general subsets @xmath146 and @xmath147 . [ [ monoidal - closed - categories - as - chiralities . ] ] monoidal closed categories as chiralities . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the observation that @xmath65 is a symmetric monoidal closed category leads us to the idea of reformulating it as a `` symmetric monoidal closed chirality '' in the sense of @xcite . recall that : a symmetric monoidal closed chirality @xmath148 is a pair of symmetric monoidal categories @xmath149 equipped with a symmetric monoidal equivalence @xmath150 ^ -{(-)^{\ast } } & & & & & & & & \ar@<.5ex>[llllllll]^-{^{\ast}(- ) } & ( { \mathscr{b}},{\varovee},{\mathbf{false}})^{\,op(0,1)}}\ ] ] where the exponent @xmath151 means that the orientation of the tensor product @xmath152 ( of dimension 0 ) and of the morphisms ( of dimension 1 ) have been reversed ; together with two ( pseudo)actions @xmath153 together with two natural bijections : @xmath154 for @xmath155 and @xmath156 , satisfying moreover two coherence diagram , see @xcite for details . every symmetric monoidal closed category @xmath157 may be equivalently formulated as the symmetric monoidal closed chirality defined by the pair of opposite categories : @xmath158 the advantage of this formulation is that the intuitionistic implication of the monoidal closed category @xmath159 may be `` decomposed '' in just the same way as in classical logic or in linear logic : @xmath160 where the operation @xmath161 implements an involutive negation , and where the notation @xmath152 reflects the fact that the tensor product of @xmath7 should be understood as a disjunction . [ [ bifibrations - as - chiralities . ] ] bifibrations as chiralities . + + + + + + + + + + + + + + + + + + + + + + + + + + + + one main contribution of the paper is to observe that the notion of `` chirality '' may be very elegantly adapted to the notion of bifibration . a bifibration chirality @xmath162 is a pair of opfibrations @xmath52 and @xmath163 @xmath164 together with a pair of equivalences @xmath165 ^ -{(-)^{\ast } } & & & & & & \ar@<.5ex>[llllll]^-{^{\ast}(- ) } & { \mathscr{f}}^{{\,op } } } \quad\quad\quad \xymatrix @-1.9pc { { \mathscr{b}}&\ar@<.5ex>[rrrrrr]^-{(-)^{\ast } } & & & & & & \ar@<.5ex>[llllll]^-{^{\ast}(- ) } & { \mathscr{c}}^{{\,op}}}\ ] ] inducing an equivalence of refinement systems : @xmath166_-{p } \ar@<.5ex>[rrrrrrrr]^-{(-)^{\ast } } & & & & & & & & \ar@<.5ex>[llllllll]^-{^{\ast}(- ) } { \mathscr{f}}^{{\,op}}\ar[ddddd]^-{q^{\,op } } \\ \\ \\ \\ \\ { \mathscr{b}}\ar@<.5ex>[rrrrrrrr]^-{(-)^{\ast } } & & & & & & & & \ar@<.5ex>[llllllll]^-{^{\ast}(- ) } { \mathscr{c}}^{{\,op}}}\ ] ] in a bifibration chirality @xmath162 , the opfibration @xmath52 is automatically a fibration , where the pullback @xmath167 can be computed as @xmath168 equation ( [ equation / notpushnot - intro ] ) follows from the fact that the pushforward @xmath169 in the opfibration @xmath163 is a pullback in the fibration @xmath170 , and that equivalences of refinement systems preserve pullbacks . we can also derive it more explicitly in proof - theoretic style , from the invertible inferences @xmath171 the subset bifibration @xmath172 can be formulated as a bifibration chirality @xmath162 as follows . define @xmath173 to be the category whose objects are subsets @xmath25 , @xmath19 and whose morphisms @xmath174 are binary relations @xmath175 satisfying the property @xmath176 the category @xmath173 comes equipped with an evident forgetful functor @xmath177 which defines an opfibration . we obtain in this way the bifibration chirality : @xmath178_-{p } \ar@<.5ex>[rrrrrrrr]^-{(-)^{\ast } } & & & & & & & & \ar@<.5ex>[llllllll]^-{^{\ast}(- ) } { { \mathbf{rel}_{\circ}^{\,op } } } \ar[ddddd]^-{q^{\,op } } \\ \\ \\ \\ \\ { { \mathbf{rel } } } \ar@<.5ex>[rrrrrrrr]^-{(-)^{\ast } } & & & & & & & & \ar@<.5ex>[llllllll]^-{^{\ast}(- ) } { { \mathbf{rel } } } ^{{\,op}}}\ ] ] where the functor @xmath161 transports a set @xmath2 and a subset @xmath25 to themselves , and reverses a binary relation @xmath37 in the expected way : @xmath179 and similarly for @xmath180 . one obtains the equations @xmath181 where @xmath37 in both equations and @xmath182 in @xmath52 for equation ( [ eqn : a ] ) , while @xmath182 in @xmath163 for equation ( [ eqn : b ] ) . note that the universal quantifier @xmath77 of equation ( [ eqn : a ] ) is computed in @xmath52 while the universal quantifier @xmath77 of equation ( [ eqn : b ] ) is computed in @xmath163 . not only that , the category @xmath173 defines together with @xmath65 a symmetric monoidal chirality @xmath183 putting together the bifibration chirality @xmath162 with the symmetric monoidal closed chirality @xmath148 , we may for instance rewrite equation ( [ equation / exists - vs - tensor ] ) as the following pair of dual equations : @xmath184 where @xmath139 , @xmath140 and @xmath185 , @xmath182 in @xmath52 for equation ( [ eqn : c ] ) while @xmath142 , @xmath143 and @xmath185 , @xmath182 in @xmath163 for equation ( [ eqn : d ] ) . this pair of dual formulas is fundamental : in particular , it has the remarkable property of unifying equation ( [ equation / exists - vs - tensor ] ) with the other equation @xmath186 valid in @xmath81 and more generally in any bifibration @xmath110 which is at the same time a symmetric monoidal closed refinement system ( see prop 2.4 in @xcite ) . to that purpose , one needs to replace the tensor product @xmath187 in formula ( [ eqn : d ] ) by the action of @xmath163 on @xmath52 written ( on purpose ) with the same notation @xmath188 . understood in this alternative way , the formula ( [ eqn : d ] ) is not equivalent anymore to equation ( [ equation / exists - vs - tensor ] ) but to equation ( [ equation / forall - vs - multimap ] ) where the refinement @xmath128 and change - of - basis morphism @xmath189 @xmath190 are decomposed in the same way as we did in ( [ equation / chiral - decomposition ] ) for the implication formula , using the formalism of monoidal closed chiralities : @xmath191 where @xmath139 and @xmath143 and @xmath185 in @xmath163 , @xmath182 in @xmath52 . on the other hand , we have seen in ( [ equation / inclusion - not - equality ] ) that we have two canonical morphisms which are not invertible in general : @xmath192 where @xmath142 , @xmath143 , @xmath185 , @xmath182 in @xmath52 in equation ( [ eqn : e ] ) and where @xmath139 , @xmath140 , @xmath185 , @xmath182 in @xmath163 in equation ( [ eqn : f ] ) . one main achievement of our approach is to recover the dualities of linear logic in categorical situations such as the subset hyperdoctrine on @xmath32 or the presheaf hyperdoctrine on @xmath8 , which are traditionally seen as intuitionistic . we will see in particular ( [ section / comparison - with - linear - logic ] ) that the formulas ( [ eqn : a])([eqn : f ] ) are bifibrational generalisations of familiar distributivity laws of linear logic . [ [ the - identity - predicate - in - the - subset - bifibration . ] ] the identity predicate in the subset bifibration . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as defined by lawvere , a hyperdoctrine is a pseudofunctor @xmath193 from a cartesian closed category @xmath7 whose fibers @xmath9 are themselves cartesian closed categories , and such that every substitution functor @xmath28 has a left adjoint @xmath23 and a right adjoint @xmath24 . given such a hyperdoctrine @xmath193 , lawvere suggested to define the identity predicate @xmath194 associated to an object @xmath86 as the terminal object @xmath195 existentially quantified along the diagonal map @xmath196 , as follows : @xmath197 in the case of the subset hyperdoctrine @xmath14 , another construction of the identity predicate is possible , starting this time from the subset bifibration @xmath81 on sets and relations . consider the binary relation @xmath198 obtained by currying the identity relation @xmath199 , where @xmath200 is the singleton set . then , define the identity predicate @xmath201 as the singleton subset @xmath202 existentially quantified along @xmath203 : @xmath204 the identity predicates @xmath205 and @xmath201 coincide in the case of the subset hyperdoctrine , but we will see ( [ section / identity - type ] ) that they differ in the case of the presheaf hyperdoctrine , and that @xmath201 appears to be the appropriate definition in that case . [ [ plan - of - the - paper . ] ] plan of the paper . + + + + + + + + + + + + + + + + + + after this long and detailed introduction , we explain in [ section / presheaf - bifibration ] how to adapt smoothly all the results established here for the subset hyperdoctrine on @xmath32 to the presheaf hyperdoctrine on @xmath8 . we then come back to the question of identity in [ section / identity - type ] , explaining how the definition @xmath201 lifts naturally to the presheaf hyperdoctrine and more generally to any monoidal closed bifibration . in [ section / string - diagrams ] we introduce a string diagram notation for presheaves ( highly reminiscent of c. s. peirce s `` existential graphs '' for predicate logic ) which is derived from the decomposition of monoidal closed bifibrations as monoidal closed bifibration chiralities . finally , in [ section / comparison - with - linear - logic ] we explain in what sense the formulas ( [ eqn : a])([eqn : f ] ) extend to bifibrations some familiar distributivity principles of linear logic . [ [ related - works . ] ] related works . + + + + + + + + + + + + + + the literature contains several different answers to the question of what exactly it means to combine `` linear logic with bifibrations '' . the approach that we develop here as well as in prior work @xcite is to consider a functor which is both ( symmetric ) monoidal closed and a bifibration , with these two structures provided independently ( but generating a rich interaction ) . the same approach is taken in hasegawa s work @xcite on logical predicates for models of linear logic , as well as in katsumata s work on logical predicates for computational effects @xcite . both build on hermida s thesis @xcite which considered a notion of `` fibred - ccc '' , although a subtle difference with hermida s work is that the latter is phrased in terms of _ fibred adjunctions _ @xcite , meaning that the functors associated to the cartesian closed structure are explicitly required to preserve cartesian morphisms . that idea can also be seen as the background for birkedal , mgelberg , and petersen s work on linear abadi - plotkin logic @xcite , as well as shulman s definition @xcite of `` monoidal bifibration '' that asks for the tensor product operation @xmath206 of the total category to preserve both cartesian and opcartesian morphisms . our perspective is that when such preservation properties hold , they should rather be seen as a consequence of an underlying adjunction between refinement systems ( * ? ? ? most importantly , the requirement that the tensor product operation preserves cartesian morphisms is violated for the key models introduced in this paper , and in general we only have the non - invertible principle ( [ eqn : e ] ) . in this section , we explain how to adapt to the presheaf hyperdoctrine @xmath14 on @xmath8 everything which was established in the introduction for the subset hyperdoctrine @xmath14 on @xmath32 . the first step is to replace the category @xmath33 by the bicategory @xmath207 ( introduced by bnabou @xcite ) whose objects are small categories @xmath208 , 1-cells @xmath175 are _ distributors _ defined as functors @xmath209 and 2-cells are natural transformations between distributors . the identity 1-cell @xmath210 is defined as the hom functor @xmath211 , which we denote @xmath212 , and the composition of two distributors @xmath175 and @xmath213 is defined using the coend formula @xmath214 the category @xmath65 is then replaced by the category @xmath215 whose objects are the distributors @xmath216 , @xmath217 ( i.e. , contravariant presheaves ) , and whose morphisms @xmath174 are pairs @xmath218 consisting of a distributor @xmath175 and of a natural transformation @xmath219 , which may be depicted as @xmath220_r\ar[dr]^s & \\ a \ar[rr]_m & \utwocell<\omit>{\alpha } & b}}\parbox{\wd0}{\box0}\endgroup}}\ ] ] similarly , the category @xmath173 is replaced by the category @xmath221 whose objects are the distributors @xmath222 , @xmath223 ( i.e. , covariant presheaves ) , and whose morphisms @xmath174 are pairs @xmath218 consisting of a distributor @xmath175 together with a natural transformation @xmath224 , which may be depicted as @xmath225^{s } & \utwocell<\omit>{\alpha } & a\ar[ul]_{r}\ar[ll]^{m}}}\parbox{\wd0}{\box0}\endgroup}}\ ] ] note that @xmath215 and @xmath221 are bicategories just like @xmath207 , but we prefer to consider them as categories for simplicity . the two obvious functors @xmath226 are opfibrations , and they define together a bifibration chirality , @xmath227_-{p } \ar@<.5ex>[rrrrrrrr]^-{(-)^{\ast } } & & & & & & & & \ar@<.5ex>[llllllll]^-{^{\ast}(- ) } { { \mathbf{dist}}_\circ } ^{{\,op}}\ar[ddddd]^-{q^{\,op } } \\ \\ \\ \\ \\ { \mathbf{dist}}\ar@<.5ex>[rrrrrrrr]^-{(-)^{\ast } } & & & & & & & & \ar@<.5ex>[llllllll]^-{^{\ast}(- ) } { \mathbf{dist}}^{{\,op}}}\ ] ] where the equivalence between @xmath215 and @xmath228 transports every small category @xmath2 to its opposite category @xmath229 , every distributor @xmath37 to the opposite distributor @xmath230 defined as the functor @xmath231 and every contravariant presheaf @xmath70 to the covariant presheaf @xmath232 . the equivalence between the refinement systems @xmath52 and @xmath170 follows from the bijective correspondence of 2-cells in @xmath207 : @xmath220_r\ar[dr]^s & \\ a \ar[rr]_m & \utwocell<\omit > { } & b}}\parbox{\wd0}{\box0}\endgroup } } \quad\leftrightarrow\quad \scalebox{1}{{\begingroup\setbox0=\hbox{\xymatrix{&1 & \\ a^{{\,op}}\ar[ur]^{r^ * } & \utwocell<\omit > { } & b^{{\,op}}\ar[ul]_{s^*}\ar[ll]^{m^*}}}\parbox{\wd0}{\box0}\endgroup}}\ ] ] the induced bifibrational structure on @xmath233 may be explicitly defined using coends and ends , as categorical analogues of the corresponding formulas for @xmath234 : @xmath235 the following property is fundamental : the refinement system @xmath236 is symmetric monoidal closed with tensor product and implication of the category @xmath215 defined as @xmath237 where @xmath70 and @xmath238 are contravariant presheaves . note that the implication @xmath128 in @xmath215 is transported by @xmath52 to the implication @xmath133 defined as @xmath239 in the compact closed bicategory @xmath207 . from this follows that the category @xmath221 together with the category @xmath215 defines a symmetric monoidal closed chirality @xmath240 which thus satisfies formulas ( [ eqn : a])([eqn : f ] ) stated in the introduction . now , let us recall that in lawvere s presheaf hyperdoctrine @xmath241\ ] ] the substitution operation @xmath242 \quad { \stackrel{}{\longrightarrow } } \quad [ a^{{\,op } } , { { \mathbf{set } } } ] \ ] ] along a functor @xmath243 is defined by precomposition @xmath244 while the quantifiers @xmath245 \quad { \stackrel{}{\longrightarrow } } \quad [ b^{{\,op } } , { { \mathbf{set } } } ] \ ] ] may be defined as coends / ends : @xmath246 in essentially the same way as we saw earlier for the subset hyperdoctrine over sets , the presheaf hyperdoctrine can be decomposed into a pair of bifibrations @xmath247 where : * @xmath248 has as objects pairs @xmath66 consisting of a category @xmath2 together with contravariant presheaf @xmath249 , and morphisms @xmath250 consisting of a pair of a functor @xmath243 together with a natural transformation @xmath251 ; * @xmath252 has as objects pairs @xmath66 consisting of a category @xmath2 together with a covariant presheaf @xmath253 , and morphisms @xmath250 consisting of a pair of a functor @xmath254 together with a natural transformation @xmath255 ; * @xmath100 and @xmath101 are the evident projections . moreover , there are a pair of embedding functors @xmath256 acting as the identity on objects and sending a functor @xmath243 to the respective distributors @xmath257 with the property that [ thm / reconstruction - presheaves ] the bifibrations @xmath258 and @xmath259 are pullbacks of @xmath260 along the functors @xmath40 and @xmath41 , respectively . once again , this theorem implies that the triple adjunction @xmath261 of the presheaf hyperdoctrine may be reduced to a pair of adjunctions @xmath262 of the bifibration @xmath260 . the other important logical ingredient in lawvere s original definition of a hyperdoctrine is the cartesian closed structure of each category of predicates @xmath9 . here we note that the well - known ccc structure on presheaf categories may be further decomposed using the monoidal closed structure of the presheaf bifibration on distributors , beginning with the following elementary observation ( recalled from @xcite ) : [ prop : dayconstruction ] if @xmath263 is a monoidal closed refinement system which is also a bifibration , then every monoid @xmath264 in the basis determines a monoidal closed structure on the fiber @xmath62 , where the tensor and implication are defined for all @xmath265 by @xmath266 and the tensor unit is defined by @xmath267 . every category is a comonoid @xmath268 in @xmath8 , and is hence transported by the functor @xmath41 to a monoid : @xmath269 the fiber of @xmath2 in @xmath260 is thus automatically endowed with a monoidal closed structure by , @xmath270 @xmath271 and it is straightforward to verify using equations ( [ equation/(co)end - formulas ] ) and ( [ equation / pshsmc ] ) that this monoidal closed structure @xmath272 is isomorphic to the usual ccc structure on the presheaf category @xmath273 $ ] . we now turn to lawvere s abstract definition of the identity predicate @xmath274 in an arbitrary hyperdoctrine @xcite . in the presheaf hyperdoctrine this definition yields @xmath275 which does not seem to give an appropriate notion of identity ( any pair of objects are `` equated '' so long as they can be completed to a cospan ) , even in the case when the category @xmath2 is a groupoid . lawvere remarked that a more natural choice of generalized `` identity predicate '' on a category @xmath2 within the presheaf hyperdoctrine is the functor @xmath276 . our first observation is that this version of the identity predicate may be easily defined as a pushforward in the symmetric monoidal closed refinement system @xmath260 by @xmath277 and that more generally we can recover the presheaf associated to a distributor @xmath175 by the formula @xmath278 indeed , this abstract recipe allows us to define the `` graph '' of a morphism @xmath49 in a monoidal closed category @xmath7 with respect to any monoidal closed bifibration @xmath110 @xmath279 where @xmath280 is the monoidal unit of @xmath45 . we then have [ thm : pushpull - canonical ] let @xmath281 be a monoidal closed refinement system which is also a bifibration , and suppose given refinements @xmath46 and @xmath50 in @xmath45 and a morphism @xmath49 in @xmath7 . then there are isomorphisms @xmath282 where @xmath283 is the left evaluation map , and where @xmath284 is the right currying of @xmath285 . _ both formulas follow easily from distributivity properties discussed in the introduction together with the axioms of monoidal closed categories : @xmath286 equation ( [ equation / push - canonical ] ) may be compared with lawvere s equation @xcite @xmath287 where @xmath288 and @xmath289 are the projection maps and where the graph @xmath290 is defined by substitution along @xmath291 into @xmath292 . lawvere proved that this equation holds for any hyperdoctrine satisfying frobenius reciprocity and a beck - chevalley condition , but he also explicitly observed that those conditions are violated by the presheaf hyperdoctrine . on the other hand , equation ( [ equation / pull - canonical ] ) may be thought of as an abstract generalization of yoneda s lemma . indeed , one can consider the formula in the bifibration @xmath293 , which is also a cartesian closed refinement system @xcite . taking @xmath294 to be an object of the category @xmath2 , the two sides of ( [ equation / pull - canonical ] ) expand to @xmath295 which is the precise statement of the yoneda lemma . the prolific american logician charles sanders peirce developed during the late 19th and early 20th centuries a system for representing logical deductions as certain topological surgeries on diagrams he called `` existential graphs . '' a key element of peirce s graphical calculus was the idea of identifying the subject of two predicates by joining them with an arc . for example , the diagram expresses that there is something which is both black and a bird ( such as a crow ) . a second key element of existential graphs was the use of an enclosing curve ( which peirce called a `` cut '' or `` sep '' ) to negate a proposition . thus the diagram expresses that there does not exist a man who is not mortal , i.e. , that every man is mortal . similarly , the diagram expresses that there is some particular bird that every woman s daughter loves ( the most popular bird at the park ) . geraldine brady and todd trimble gave a categorical interpretation of existential graphs @xcite within joyal and street s framework of _ string diagrams _ for monoidal categories @xcite . their analysis began with a categorical axiomatization of classical first - order logic in terms of _ boolean hyperdoctrines _ , that is , hyperdoctrines with boolean algebra fibers and satisfying the beck - chevalley condition . they then went on to describe how to interpret the predicates of any such hyperdoctrine as 1-cells in an appropriate compact closed bicategory of boolean - valued relations . in this part of the paper we reveal a surprising development of brady and trimble s work , by explaining how the logical structure of the refinement system @xmath260 in combination with its chiral opposite @xmath296 leads in a relatively straightforward way to a string diagram calculus for presheaves that is remarkably reminiscent of peirce s existential graphs this despite the fact that our `` predicates '' and `` relations '' are by no means boolean - valued ! we will assume that the reader already has some familiarity with string diagrams for monoidal categories in general , and with the standard conventions for compact closed categories ( otherwise , the reader is encouraged to read @xcite for a quick introduction ) . for example , following those conventions , the diagram @xmath297 represents the distributor @xmath298 obtained by composing distributors @xmath175 , @xmath299 , and @xmath300 in the indicated way . observe that here we read the diagram from top - to - bottom and left - to - right , while we place labels indicating the underlying categories to the left - hand side of each oriented wire . more abstractly , in topological terms these conventions can be said to rely on the assumption that the surface in which the diagram of the distributor is embedded ( in this case , the page ) is equipped with an orientation . to represent a contravariant presheaf @xmath249 seen as a refinement @xmath46 in @xmath260 , we lay it out just as one would an ordinary distributor @xmath216 , but framed by a light blue background to indicate that we view it as an object of @xmath215 . therefore , the operations of taking the tensor product of presheaves or the pushforward along a distributor , @xmath301 which can be defined respectively in terms of horizontal and vertical composition of 1-cells in @xmath207 , are displayed like so : @xmath302 let us note here that there is a topological interpretation of the refinement relation , in the sense that _ a diagram embedded in a surface refines its boundary . _ as for implication between presheaves or the pullback along a distributor , @xmath303 we base our conventions on the equations @xmath304 and draw the following diagrams : @xmath305 here the dualization operation @xmath306 and its inverse @xmath307 are being represented as `` functorial boxes '' @xcite , which take the mirror image of the boundary wires across the box from a blue region to a red region and vice versa , while the action @xmath308 is represented by gluing a blue diagram inside a dark red region . one subtlety is that since the dualization operations reverse the orientation of the tensor product , we must therefore read horizontal juxtaposition in the red region ( corresponding to the action @xmath309 ) running right - to - left rather than left - to - right . also , it is worth noting that these conventions ensure that inside any red region there is always exactly one boxed ( i.e. , negated ) blue region , which can be seen as a sort of intuitionistic restriction on peirce s system . we next deduce some equations on the diagrams that are implied by the axioms of a monoidal closed bifibration chirality . equations between purely positive formulas such as @xmath310 are geometrically manifest using these conventions , just as the axioms of monoidal categories are geometrically manifest using ordinary string diagrams . on the other hand , since dualization is an involutive operation , it is also always possible to add or remove an annulus around the diagram of a contravariant or covariant presheaf without changing its meaning : @xmath311 seen in this way , the important distributivity law @xmath312 simply removes an annulus , pushes one component ( @xmath137 ) outside the blue region , and places the annulus back in another location : @xmath313 finally , the two formulas ( [ equation / push - canonical ] ) and ( [ equation / pull - canonical ] ) derived for the identity predicates defined in [ section / identity - type ] have the following geometric interpretation : @xmath314 @xmath315 besides capturing isomorphism , one can also express natural transformations between presheaves as certain diagrammatic moves or `` surgeries '' . the unit and counit of the two families of adjunctions @xmath316 yield directed versions of rule ( [ equation / doublesep ] ) , @xmath317 @xmath318 where rule ( [ equation / evalcoeval ] ) reduces to ( [ equation / doublesep ] ) in the case that @xmath319 , and ( [ equation / unitcounit ] ) to ( [ equation / doublesep ] ) in the case that @xmath320 . one main benefit of our approach based on chiralities is that it enables us to recover the dualities of classical logic in categorical situations like the subset hyperdoctrine @xmath14 on @xmath32 , or the presheaf hyperdoctrine @xmath14 on @xmath8 , which are traditionally seen as intuitionistic . we thus find instructive to explain in what sense the six principles ( [ eqn : a])([eqn : f ] ) asserted for the monoidal closed bifibration chirality @xmath321 generalise well - known principles of linear logic . to that purpose , we consider a @xmath322-autonomous category @xmath323 with finite products and coproducts and we construct two categories @xmath324 and @xmath325 whose objects @xmath208 are the finite sets seen as discrete categories ; and whose morphisms @xmath37 are @xmath323-valued matrices defined as functors @xmath326 . composition of @xmath37 and @xmath327 in @xmath324 is defined as @xmath328 whereas composition of @xmath37 and @xmath327 in @xmath325 is defined as @xmath329 @xmath330 is the category whose objects are the @xmath323-valued matrices @xmath70 , @xmath238 , which may be alternatively seen as families @xmath331 or @xmath332 of objects of @xmath323 ; and whose morphisms @xmath17 are the pairs @xmath218 consisting of a matrix @xmath37 and of a natural transformation @xmath333 , which may be alternatively seen as a family of morphisms living in the category @xmath323 @xmath334 indexed by the pairs @xmath335 . similarly , @xmath336 is the category whose objects are matrices @xmath222 , @xmath223 ; and whose morphisms @xmath174 are the pairs @xmath218 consisting of a matrix @xmath175 together with a natural transformation @xmath337 , which may be alternatively seen as families of morphisms living in the category @xmath323 @xmath338 indexed by the pairs @xmath335 . the forgetful functors @xmath339 and @xmath340 define a bifibration chirality @xmath341_-{p } \ar@<.5ex>[rrrrrrrr]^-{(-)^{\ast } } & & & & & & & & \ar@<.5ex>[llllllll]^-{^{\ast}(- ) } { { \mathbf{mat}_{\circ } } } ( { \mathscr{v}})^{\,op}\ar[ddddd]^-{q^{\,op } } \\ \\ \\ \\ \\ { { \mathbf{mat}_{\oplus } } } ( { \mathscr{v } } ) \ar@<.5ex>[rrrrrrrr]^-{(-)^{\ast } } & & & & & & & & \ar@<.5ex>[llllllll]^-{^{\ast}(- ) } { { \mathbf{mat}_{\ominus } } } ( { \mathscr{v}})^{\,op } } \ ] ] where a finite set @xmath2 in @xmath330 is transported by @xmath342 to itself : @xmath343 ; where a morphism @xmath37 of @xmath344 is transported to the morphism @xmath345 in @xmath346 obtained by flipping the inputs @xmath2 and @xmath26 and by applying pointwise negation in @xmath323 : @xmath347 and where the object @xmath348 in @xmath349 is defined by pointwise negation in @xmath323 : @xmath350 the fact that @xmath162 define a bifibration chirality follows from the existence of the natural bijection @xmath351 in the @xmath322-autonomous category @xmath323 . given a @xmath323-valued matrix @xmath37 , the existential quantification of @xmath70 along @xmath137 and the universal quantification of @xmath352 along @xmath37 in the bifibration @xmath52 are given by the formulas @xmath353 @xmath354 the bifibration chirality @xmath162 is also monoidal closed , with conjunction and disjunction defined pointwise : @xmath355 where @xmath185 and @xmath182 . in this specific monoidal closed bifibration chirality @xmath162 , the formulas ( [ eqn : a])([eqn : f ] ) enable us to recover familiar principles of linear logic : jean bnabou . distributors at work . notes from a course at tu darmstadt in june 2000 , taken by thomas streicher . lars birkedal , rasmus e. mgelberg , rasmus l. petersen . category - theoretic models of linear abadi & plotkin logic . _ theory and applications of categories _ , 20:7 , 2008 , 116151 . geraldine brady and todd h. trimble . a string diagram calculus for predicate logic and c. s. peirce s system beta . 1998 november 8 ; revised 2000 june 22 . unpublished manuscript . masahito hasegawa . categorical glueing and logical predicates for models of linear logic . technical report rims-1223 , research institute for mathematical sciences , kyoto university , 1999 . paul - andr mellis and noam zeilberger . functors are type refinement systems . popl 2015 . paul - andr mellis and noam zeilberger . an isbell duality theorem for type refinement systems . july 31 , 2015 .

combining insights from the study of type refinement systems and of monoidal closed chiralities , we show how to reconstruct lawvere s hyperdoctrine of presheaves using a full and faithful embedding into a monoidal closed bifibration living now over the compact closed category of small categories and distributors . besides revealing dualities which are not immediately apparent in the traditional presentation of the presheaf hyperdoctrine , this reconstruction leads us to an axiomatic treatment of directed equality predicates ( modelled by @xmath0 presheaves ) , realizing a vision initially set out by lawvere ( 1970 ) . it also leads to a simple calculus of string diagrams ( representing presheaves ) that is highly reminiscent of c. s. peirce s existential graphs for predicate logic , refining an earlier interpretation of existential graphs in terms of boolean hyperdoctrines by brady and trimble . finally , we illustrate how this work extends to a bifibrational setting a number of fundamental ideas of linear logic .

one of the most pervasive tendencies of humans is putting things in ranking order . in human societies these tendencies are reflected in their social interactions and networks being hierarchical in many respects . hierarchies and ranks emerge due to individuals subjective perceptions that some other individuals are in some respect better . then a relevant research question is whether or not the formation and structure of hierarchies in human societies can be understood by making the assumption that the dominant driving force of people in social interactions is to enhance their own `` value '' or `` status '' relative to others . we call this assumption `` better than - hypothesis '' ( bth ) and note that it is closely related to the thinking of the school of individual psychology founded by adler in the early 1900s @xcite , which , while starting with the assumption that human individuals universally strive for `` superiority '' over others , emphasizes inferiority avoidance as a motive for many human actions . further studies of this kind of individuals status - seeking behaviour , especially concerning consumer behaviour and economics , include the canonical references by veblen @xcite , duesenberry @xcite and packard @xcite ( see also refs @xcite ) . in addition there is a closely related sociological model called social dominance theory @xcite , which proposes that the construction and preservation of social hierarchies is one of the main motivations of humans in their social interactions and networks . however , the most relevant observational facts concerning bth come from the field of experimental economics , especially from the results of experiments on the so - called `` ultimatum game '' @xcite , where the human players have been shown to reject too unequal distributions of money . the concept of _ inequity aversion _ , that is the observed social phenomenon of humans preferring equal treatment in their societies , is often invoked to explain these observations . recently some models featuring inequity aversion have been proposed in refs . @xcite . all of these models , although from different fields of study , have something to do with the relative standings between different human individuals and groups , and so they could all be considered to emerge from or be based on a single principle such as bth . it is this generality which makes bth an intriguing and interesting object of study . there are even some studies on economic data , such as @xcite , that suggest a link between relative social standings and human well - being , and considerations of social status have measurable effects on brain functions , as shown in e.g. @xcite . these studies imply that bth could well be something fundamental to human nature . the competition for a better hierarchical position among humans can be intense and sometimes even violent . however , humans have other characteristics including egalitarianism as well as striving for fairness . these traits could be interpreted in the context of bth by remarking that people need to live in societies and make diverse social bonds , which in turn would contribute to their social status . this means that the members of society when they make decisions , need to take the feelings of others into account . hence the behavioral patterns of individuals in social networks should then be characterised by sensitivity to the status of the other individuals in the network . this sensitivity manifests itself as inequity aversion and treating others fairly . to find out what in this context are the plausible and relevant mechanisms of human sociality driving societal level community formation we will focus on improving the bth - based approach by using the frame of agent - based models and studying the emergence of social norms in such social systems , following the tradition presented in refs . @xcite . in this study we use an agent - based network model applying bth - based approach to simulate social interactions dependent on societal values and rank , to get insight to their global effects on the structure of society . we find that in such a model society with a given constant ranking system the social network forms a degree hierarchy on top of the ranking system under bth , such that the agents degrees tend to increase , the further away their rank is from the average . the structure of the paper is as follows . in section [ model ] we motivate the basics of bth using the simple and well - researched ultimatum game as an example , and in section [ modelv1 ] we show how the findings from this can be utilised as a part of agent - based models . in section [ nc ] we present the numerical results of the simulations from the model , and in section [ meanfield ] we analyse them . the two final sections discuss the possible interpretations of the results and present the conclusions . in this section we describe the theoretical basis for our model . we start by analysing the ultimatum game first proposed in @xcite , as it allows us to derive a basic form for the social gain function in our model . the ultimatum game is a game with two players , where one player has the task to make a proposal to the other player about how a given sum of money should be divided between them . the second player then gets to choose if the proposal is acceptable or not ; if it is , the money is divided as proposed . if not , neither player gets anything . experiments show that humans playing this game normally do not accept deals that are perceived to be unfair , i.e. in situations in which the proposer gets too large a share of the money ( see , e.g. refs . this is a classic problem in the mainstream economics , where humans are assumed to be rational and , therefore , accept something rather than nothing . we implement bth in the ultimatum game by interpreting the money used in a deal as a way of comparing the status between the one who accepts the proposal ( called from now on the accepter ) and its proposer . we denote the change of `` status '' of the accepter as @xmath0 , which takes into account it s own monetary gain , and the gain in relation to the proposer . therefore , the simplest expression for @xmath0 is , @xmath1\\ & -[r_a(t_0 ) - r_p(t_0 ) ] , \end{aligned } \label{perusta}\ ] ] where @xmath2 and @xmath3 stand for the monetary reserves ( in the context of the game ) of the accepter and proposer , respectively , at time @xmath4 , with @xmath5 being the time before the deal and @xmath6 the time after the deal . in terms of economic theory , @xmath0 would be called the accepter s change of utility , which is ordinarily assumed to consist of the term @xmath7 or the absolute payoff of the accepter . the additional terms @xmath8-[r_a(t_0 ) - r_p(t_0)]$ ] that stem from the bth measuring the change in relative standings of the accepter and proposer . the actual bth utility function for the accepter in the ultimatum game takes the form @xmath9.\ ] ] according to eq . ( [ perusta ] ) , the accepter will refuse the deal for @xmath10 , and will accept it for @xmath11 , with @xmath12 being the borderline case . should the deal be rejected , @xmath13 and @xmath14 , and , consequently , @xmath15 . if we denote by @xmath16 the total amount of money to be shared and by @xmath17 the actual amount money that the proposer has reserved for the accepter , then in the case where the transaction does take place we have @xmath18 and @xmath19 . if we further assume that @xmath20 ( i.e. the players start on equal footing , which may very well be the case in the context of the game at least ) , it follows from eq.([perusta ] ) that the smallest offer that the accepter expects from the proposer is one third of the maximum @xmath16 , i.e. the condition @xmath21 must hold for the proposal to be acceptable . previous literature shows that the minimum offers that people are usually willing to accept are around @xmath22 of a given quantity @xcite , in close agreement with the calculation above . moreover , we note that if the term @xmath7 in eq . ( [ perusta ] ) is neglected , then the accepter will never settle for less than half of the total amount . next we use eq . ( [ perusta ] ) to illustrate how the bth - based approach can be implemented in the context of agent based social simulations . the trick is to generalise this equation to the cases of many players , and with multiple and different kinds of items being exchanged . if there are @xmath23 players ( denoted by @xmath24 ) , the change of status @xmath25 of the individual @xmath26 may be written as follows @xmath27 \nonumber \\ & & - \sum_{j \neq i } [ r_i(t_0 ) - r_j(t_0 ) ] . \label{glin}\end{aligned}\ ] ] in the case that players have several ways to measure their status , we may introduce normalisation factors to compare the relative value of the exchanged items , and write the change of status as , @xmath28 \nonumber \\ & & - \bigg . \sum_{j \neq i } \left [ r^{\alpha}_i(t_0 ) - r^{\alpha}_j(t_0 ) \right ] \bigg\ } \label{ydelta}\end{aligned}\ ] ] where the index @xmath29 runs over the various items determining the status , and @xmath30 is the normalisation factor of each item , which may vary from one player to another . the utility function associated with eq . ( [ ydelta ] ) reads then @xmath31 \bigg\}.\ ] ] in this section we present an agent - based model of a ranked social system of @xmath23 agents , in which the agents exchange their views of the ranking system itself . to each agent @xmath26 we assign a parameter @xmath32 to describe the rank of the agent , and a state variable @xmath33 to denote the opinion of the agent @xmath26 of the social value attached to parameter @xmath34 . the social value is then a relative quantity in the minds of the agents , and they value each other in either ascending or descending order according to the `` ranking parameter '' , we call @xmath34 , and @xmath33 determines which order a given agent @xmath26 prefers and how strongly . generally speaking , the sign of @xmath33 represents the chosen order , @xmath35 for descending and @xmath36 for ascending order , while its magnitude represents the strength of conviction : with @xmath37 the agent can be said to support equality of all the agents irrespective of the ranking parameter , with @xmath38 the agent thinks that @xmath34 should directly define the hierarchy of the society , and for the case @xmath39 or @xmath40 correspondinly downplaying or emphasizing the significance of @xmath34 , respectively . here , we adopt the maximum value for @xmath41 to be @xmath42 . to put the relation of the social value and the opinion parameter into more precise terms , we adopt the following expression for the term we call `` ranking pressure '' : @xmath43 where the summation is over the whole network . now , we define the social value of the agent @xmath26 in the eyes of agent @xmath44 as @xmath45 , and assume that the agents take into account the views of their neighbours in addition to their own when evaluating their total social value @xmath46 . thus , we write @xmath47 where @xmath48 denotes agents that are one step away from agent @xmath26 . the first term on the right could be considered as the agent s `` self esteem '' and the second the `` social value '' given to it by its first neighbours . it should be noted at this point that in defining @xmath46 in terms of @xmath49 we have assumed that @xmath32 does not confer direct advantages or disadvantages for the agents . therefore , the most natural interpretation for @xmath32 is that it represents the ownership of pure status symbols or veblen goods , i.e. goods that are only , or mostly , desirable due to their status - enhancing properties , such as luxury items . if @xmath32 would give some advantages or disadvantages for the agents , eq . ( [ totsosval ] ) would have to be revised accordingly . the agents in our simulations attempt to gain as much social value as possible , both in absolute and relative terms , and do this either by changing their opinion variables or adjusting their relations to other agents . the system we use here is purely reactive , with agents reacting to the changes in their social environment in accordance with bth . the social gain function could then be written as @xmath50 where @xmath5 and @xmath6 are the two consecutive time steps . the sign of this function determines the direction of the changes in @xmath33 . the decision making method employed by the agents is thus a simple hill climbing algorithm : at a given time step @xmath4 , first an agent observes the quantity @xmath51 , and then changes its variable @xmath52 , which leads to a recurrence relation of the form @xmath53 where @xmath54 $ ] and @xmath55 is a small increment . in the spirit of simulated annealing techniques , the magnitude of the change is larger at the beginning of the dynamics and falls linearly with time to a minimum value , the maximum and minimum values being @xmath56 and @xmath57 , respectively , and the time period to reach the minimum is 1000 time steps . in general , the links between the agents in the social network may change in time for which purpose we use the following rewiring scheme of ref . the social network of the agents is initially random , but will change periodically , i.e. at every @xmath58 time steps of the dynamical eq . ( [ dm ] ) . given the definition of the total social value of an agent in eq . ( [ totsosval ] ) , the gain function eq.([altdel ] ) can be used to calculate the loss or gain in total social status when forming or breaking new social bonds . in this study , we take any positive gain as sufficient to justify the rearrangement of social relations between the agents . when agent @xmath26 considers cutting an existing bond with agent @xmath44 , the gain function has the form @xmath59 where @xmath60 is the current number of neighbours of agent @xmath26 . similarly , when agent @xmath26 considers forming a new bond with agent @xmath44 , the social gain function reads @xmath61 since any positive change indicated by the functions above leads to rewiring , it is the sign of these functions that determines whether or not links between agents are broken or created . for instance , if @xmath62 , the link between agents @xmath26 and @xmath44 will be cut , and preserved if @xmath63 . in the same vein , a link between agents @xmath26 and @xmath44 will be created if @xmath64 , and not created otherwise . it should be noted that when forming links the opinions of both agents are taken into account : the relation formation only succeeds if @xmath65 and @xmath66 are both positive . the agents will form all the relationships they can in a rewiring cycle . the numerical simulations of the models of social system described in the previous sections are performed as follows . first , the initial state of the system is set at random with the agents given a relatively small initial opinion @xmath33 between @xmath67 and @xmath68 , a ranking parameter @xmath32 between @xmath69 and maximum value of @xmath70 and initial connections to other agents , with initial average degree of @xmath71 . the opinion and ranking parameters are chosen using a random number generator , which returns a flat distribution . the dynamics are then run for @xmath72 time steps , which is , according to our test runs , sufficient for the general structure of the network to settle . however , the dynamics of the opinion variables do not have a set stopping point , so they may experience fluctuations even when such fluctuations do not have an effect on the network structure anymore . to obtain reliable statistics , the same simulations are repeated @xmath73 times with random initial values , and averages are calculated from these repeated tests for the quantities under study . the rewiring timescale @xmath58 is fixed to @xmath73 in our simulations , since this value lies in the range where communities are formed in the opinion formation model of ref . the main parameter whose effect is studied here is the number of simulated agents , @xmath23 . the main objective of this research is to study the structure of the social networks created under bth assumption in the case of a rigid ranking system , which we perform using the model explained in section [ modelv1 ] . the most interesting properties of the system are then associated with assortativity , or the tendency of agents with high degrees connecting to other highly connected agents , and homophily , or the inclination of similar agents forming connections between each other . in the context of this study , homophily refers to agents with similar ranking parameters forming connections with each other . the averaged numerical results extracted from the simulations consist then of the standard network properties , i.e. degree @xmath74 the shortest path @xmath75 , the average clustering coefficient @xmath76 , the mean number of second neighbours @xmath77 , susceptibility @xmath78 and average assortativity coefficient @xmath79 , and a homophily coefficient @xmath80 . susceptibility here refers to average cluster size , which is calculated as the second moment of the number of @xmath81 sized clusters , @xmath82 : @xmath83 as customary in percolation theory , the largest connected component of the network is not counted in calculating @xmath81 . for the assortativity coefficient we use the definition given in @xcite , and the homophily coefficient is defined using pearson s product moment coefficient , which measures the goodness of a linear fit to a given data . for a sample it can be defined as @xmath84 where @xmath85 and @xmath86 are vectors containing the value parameters of agents linked by link @xmath26 , @xmath87 and @xmath88 are the mean values of these vectors , respectively , and @xmath89 is the total number of links . more specifically , if agents @xmath29 and @xmath90 are connected by link @xmath26 , then @xmath91 and @xmath92 . the links are indexed as follows : the links involving the first agent are given the first indices , then follow the links involving the second agent but not the first , and so on , without repeating links that have already been indexed . it should be noted , however , that @xmath93 only measures linear correlation between the ranking parameters of linked agents , it does not indicate how steep these trends are . to check whether the system is truly homophilic , then , one needs to make a linear fit to the data : the closer the obtained linear coefficients are to @xmath68 , the greater the homophily . = 0.85 the average network properties of the system , with graphs illustrating the behaviour of the system are shown in fig . [ fig : np100r ] as functions of the population size , which is varied between @xmath94 and @xmath95 . the main observations that can be made about the graphs in fig.[fig : np100r](a ) are that at lower population levels they show a tendency of breaking apart into many subcomponents of different sizes , while for larger population sizes they tend to consist of a single large component and possibly some smaller separate clusters . a noteworthy fact about these clusters is that they consist of agents with similar values of the ranking pressure @xmath49 , which means that the network exhibits homophily in this case . the largest clusters are found at extreme values , and they become smaller when one approaches @xmath69 , which also corresponds to average ranking parameters . in the high population case the picture becomes more complicated due to the emergence of clusters that contain agents with opposing opinions as well . these new clusters tend to be less connected than the previously described homogeneous ones , and they tend to connect to the large subgraphs , thus forming a single giant graph . a closer look reveals that these agents generally have opinion variables and ranking pressures with opposite signs and are depicted as triangles in fig . [ fig : np100r ] and named ` contrarians' from now on . a naive analysis would indicate that the agents with positive ranking pressures should always support the ascending hierarchy , and the agents with negative ranking pressures should always support the descending hierarchy . however , the contrarian agents exhibit opposite preferences . the reason why this behaviour is status - wise profitable can be found by looking into the connections of the contrarian agents , details of which are shown in figs . [ fig:1bis ] and [ fig : scatter2 ] . as it turns out , most of the connections they form are to other similar agents but with opposite `` polarity '' to theirs , i.e. the contrarian agent with negative ranking pressure forms connections mostly with contrarian agents of positive ranking pressure , and vice versa . an important quantity in the model is the social value @xmath46 , which is a product of the opinion and ranking pressure , as it is seen in eq . [ totsosval ] . if @xmath33 and @xmath49 have opposite signs then the `` self esteem '' part of @xmath46 is negative , which in itself does not mean that the agent can not develop a contrarian opinion , since the second sum in the equation could be positive because it depends on the opinions of the neighbours @xmath44 . this allows the contrarian to be able to make connections with agents of the same or opposite ranking pressure . additionally , by looking at all connections among contrarians we find that they mostly have @xmath49 of the same sign as @xmath96 , as illustrated in the example of fig . [ fig:1bis ] . = 0.85 this situation can be status - wise beneficial to all parties involved , since the small penalty to an agent s self - esteem is more than compensated by the respect that the agent will gain in this case from other agents . the fact that the agents could find this strategy using as primitive an intelligence setup as hill climbing is astounding . another interesting thing about the contrarians is that they appear mostly as connections between clusters that are defined as communities of `` normal '' agents . the various kinds of behaviour exhibited by the social networks have a marked effect on the network properties also shown in fig . [ fig : np100r](b ) . the most obvious is the gradually rising normalised maximum cluster size ( @xmath97 ) , which is about 40% of total population size for @xmath98 , and over 95% for @xmath99 . from the figure it seems that the maximum cluster size reaches 50% of the population size for approximately @xmath100 , after which point we may assume that the contrarian behavioural patterns start to become progressively more pronounced . the susceptibility @xmath101 at first rises pretty much linearly , which is not too surprising because of the tendency of network to break into smaller subgraphs at low population sizes . however , once the population size reaches about @xmath102 , the susceptibility starts to decay , most likely due to the main component of the network becoming more prominent , with a decay pattern that is almost piecewise linear itself , apart from fairly large fluctuations . while fairly high throughout , the homophily @xmath103 and clustering coefficients @xmath104 gradually fall as functions of population size , almost certainly due to the proliferation of contrarians . there are no great changes in other network properties , as in the average assortativity coefficient ( @xmath105 ) , although some faint systematic tendencies can be discerned , a slight rising of the average path length ( @xmath106 ) , as well as slightly decreasing average number of clusters ( @xmath107 ) . the rest of the properties per agent are nearly constant , a slight rise of the average number number of second neighbours ( @xmath108 ) , and a just perceptible decrease of the average cluster size ( @xmath109 ) and average degree ( @xmath110 ) . a way to illustrate the homophily of the system is to make a scatterplot of the ranking parameters of linked agents . as it is seen in fig.[fig : scatter2 ] , the correlation turns out to be very homophilous , as the ranking parameters of linked agents correspond very closely to one other . the emergence of the contrarians is also clearly seen : for @xmath98 , the percentage of contrarians in 10 realisations is 5.4% , while for @xmath111 is 13.8% , for @xmath112 is 24.9% and for one realisation in in a network of 1000 agents is 38.5% . there is also a clear decreasing linear trend due to the contrarians . the rising trend is without doubt caused by the normal agents , who tend to associate with agents of similar rank , and the decreasing trend is likewise due to the contrarians . both trends have a similar tendency to form square - like patterns along the diagonals , with each `` square '' corresponding to some of the many visible communities of the graphs . = 0.85 + as explained above , it is necessary to check whether the correspondence of the rankings is truly homophilic . in table [ tab:1 ] we show the value of the homophily coefficient of normal and contrarian agents for networks of various sizes . the data were taken from 10 different numerical realisations in each case . observe that normal agents have values very near one , and contrarians are around one half . also in the table we show the slope of the regression of @xmath32 vs. @xmath113 . normal agents are very close to one , indicating high degree of homophily , and the contrarians are negative and around 0.6 , indicating that they mostly form connection with agents that have opposite signs of the ranking pressure , and are much less homophilic . .homophily measurement from 10 numerical realisations of networks with different sizes . [ cols="<,^,^,^,^,^,^,^,^ " , ] from condition([expla4 ] ) it follows that the largest community in @xmath114 comprises of those agents with ranking parameters over @xmath115 , which means that the community will have approximately @xmath116 members , when one takes into account the fact that the ranking parameters are uniformly distributed . the second largest community , likewise , consists of those agents with ranking parameters between @xmath117 and @xmath118 , and has about @xmath119 members . the nth ( @xmath120 ) largest group will have ranking parameters between @xmath121 and @xmath122 , and have @xmath123 members . while ranking parameters naturally vary from agent to agent within the communities , the average value of the ranking parameters of each group @xmath124 falls approximately to the middle point of each ranking range due to the uniform distribution of the parameters : @xmath125 from figs . [ fig : scatter2 ] and [ fig : p1000 ] we see that only a maximum of four to five of these communities exist in practice at any one time , so we limit our approximation to these groups . by assuming the groups to be fully connected , as they seem to be in the graphs , and approximating sums of the products of the degrees and ranking parameters with the products of their average values we get @xmath126 where @xmath127 denotes the average ranking parameter and @xmath128 the number of members of the @xmath129th largest group . substituting the values given in table [ t:1 ] we get @xmath130 which in turn can be inserted into ( [ cgsimple ] ) : @xmath131 finally , using the definition of @xmath49 the condition ( [ cgsimple2 ] ) can be written in the form @xmath132 where @xmath133 . from condition ( [ cgsimple3 ] ) we can see that the probability of agent @xmath26 following the conventional wisdom diminishes with rising @xmath23 and decreasing @xmath60 , which is what we saw happening in our simulations judging from the results shown in the previous section . to take an example , for the @xmath99 case the largest community of agents with positive ranking pressures comprises of about @xmath134 agents . this means , according to the inequality ( [ cgsimple3 ] ) , that the agents with @xmath135 could definitively be expected to always choose to have positive opinion variables . from fig . [ fig : scatter2 ] we can tell that the real threshold is closer to @xmath136 , which is , however , in remarkably good agreement with the approximate value of @xmath137 when one takes into the consideration the fact that the appearance of the contrarians themselves was not taken into account in the derivation of ( [ cgsimple3 ] ) , and that for @xmath99 they are already very prominent . if one were to derive the condition equivalent to ( [ cgsimple3 ] ) with contrarian strategies taken into account , one would need to consider the effect that the contrarians have on their neighbours total social value @xmath138 . thus , condition ( [ cgsimple3 ] ) will most likely not hold for networks with larger @xmath23 . the last question we need to address as regards to the contrarians is the fact that they are often embedded in the groups of normal agents . so why would it be status - wise beneficial for an agent using normal strategy to retain , let alone form , a link with a contrarian agent ? let us consider a situation where an agent pursues contrarian strategies in a group of normal agents . returning to inequality ( [ expla1 ] ) , we see that it is acceptable for a normal agent @xmath26 with @xmath139 and maximal @xmath140 to form a link with a contrarian agent @xmath44 with @xmath141 and minimal @xmath142 if @xmath143 and we assume that all the neighbours of agents @xmath26 and @xmath44 also have the maximal opinion parameters . the striking fact about this relation is that it is always fulfilled in this case , meaning that @xmath26 would always find formation of links with contrarians acceptable . from the point of view of agent @xmath44 , however , linking to @xmath26 is only acceptable if @xmath144 which leads to the very same result as before . i.e. @xmath44 will not form a link with @xmath26 unless @xmath145 , if @xmath26 and @xmath44 are to belong to the same group . thus the contrarian agents would behave and be treated as normal agents when forming relations , which is surprising considering that their contribution to the total social value of other agents is negative . the latter fact is demonstrated in the simulations with some of the most counterintuitive behaviours of the model , namely , relations between agents being first broken and immediately reinstated . let us use eq . ( [ cutgainij ] ) to determine , whether the agent @xmath26 from the previous calculations would benefit from cutting the link with agent @xmath44 , even when expression ( [ cnrel ] ) says that @xmath26 would also form a link with @xmath44 in the event that such a link did not exist . with the previously stated assumptions , we arrive to the following condition @xmath146 for the link to be cut . from ( [ cncut ] ) we see that unless @xmath147 , agent @xmath26 will cut its ties with the contrarians . since @xmath148 for the largest groups , the inequality ( [ cncut ] ) is likely to be true most of the time , leading to links between agents @xmath26 and @xmath44 being cut and immediately reformed repeatedly , since inequality ( [ cnrel ] ) also holds . as suggested above , we have observed this behavioural pattern in our simulations , and to some extent it can be observed in fig . [ fig:1bis ] , in which it is seen that only the connections between contrarian and normal agents @xmath26 and @xmath44 can the signs of @xmath33 and @xmath149 being of opposite signs . it should again be stressed , however , that the calculations above do not take into account the existence of more than one contrarian . having more contrarians in the system allows them to form links between each other , which has a sizeable effect on the overall structure of the network . in summary , it could be said that while conditions ( [ expla3 ] ) and ( [ cnrel2 ] ) provide surprisingly well fitting approximations as to how a given agent chooses to link with other agents , the conditions ( [ cgsimple3 ] ) , ( [ cnrel ] ) and ( [ cncut ] ) ( though pointing to the right direction ) only give vague qualitative explanations for the behaviour of contrarians and can not be expected to yield precise numerical predictions . the interpretation of the ranking parameter @xmath32 serves as the key to find possible parallels between our model and the real world . as it describes a single property of an agent , the links between the agents only correspond to exchanges of opinion on whether the agents with larger @xmath32 are `` better '' than the agents with smaller @xmath32 , or vice versa . agents could be considered as being embedded a larger social context , and in this context they could , in principle , have other social connections . in this case , the results presented in the previous section are best interpreted in terms of echo chambers , which means that agents prefer such a social hierarchy in which they have better relative rank , and seek to communicate their opinion to others . the agents whose ranking parameters are further away from the average , are more vocal in broadcasting their views and gather supporters , since they rank highly in their chosen hierarchy and , therefore , would benefit from their hierarchy becoming more widely accepted . on the other , the agents with average ranking parameters are much more reluctant to take part in the conversation at all , since they do not rank highly in either of the hierarchies . then the end result for small system sizes is that agents divide themselves according to their ranking pressure into two or more distinct communities supporting opposing hierarchies , in which the agents with similar rankings lump together and refuse to communicate with those that disagree . it is the shutting out of the opposing point of view that makes this system s behaviour reminiscent of echo chambers found in reality . however , with increasing system size the agents develop more nuanced positions on their preferred hierarchies due to mounting social competition , as is seen in the emergence of the contrarians . there is , however , an alternative way to interpret the ranking parameter . it could be taken to represent an aggregate of all the social properties of an agent , thereby representing its total standing in the societal status measures . in this case the connections could represent the totality of the agents social interactions , and the opinion variables the agents attitude to the ( current ) state society at large . with this interpretation the rupture between the different communities observed for smaller system sizes would actually represent a real disintegration of the society . this might have implications concerning early human migrations , as they could easily have been influenced by social pressures as well as material needs . if the environmental pressures define a minimum group size necessary for a comfortable life for a tribe , and this minimum is smaller than the limit at which the tribe is forced to be adopting more advanced strategies to enhance social stability , as exemplified by the contrarians in our simulations , the tribe may well split , with splinter groups migrating elsewhere . other than the different economic games , bth can also shed some light into the well known paradox of value , also known as diamond - water paradox , which refers to the fact that diamonds are far more valued in monetary terms than water , even though water is necessary for life and diamonds not . from the bth view point the solution to this paradox is obvious : water , being necessary requirement for life , has to be available in sufficient quantities to all living humans , which means that owning water or its source does not set an individual apart from others , that is , an individual can not really compare favorably to others on grounds of having water . diamonds , on the other hand , are relatively rare , and thus can not be owned by everyone . therefore , an individual possessing diamonds is compared favourably to others , and so diamonds acquire a relatively high value in comparison to water in the minds of humans , in a very similar manner with which the veblen goods become valuable . then bth , in a sense , contains in itself a natural definition of value , although further work is needed to determine how exactly this status - value relates to other forms of value , such as value derived from usefulness or necessity . in section [ model ] we only analysed the behaviour of the accepter , since this is straightforward in comparison to predicting the behaviour of the proposer . the experiments on the ultimatum game often find that the proposers tend to offer fair shares to accepters , which is easily explained in the context of bth by the desire of the proposer to have the proposal accepted : the proposers only offer shares that they would accept themselves , and in this way eq.([accepted ] ) also restricts the proposers offers , although it can not tell the exact amount of money offered . to be able to give a better estimate for the offers one would need to study the learning processes that shape the proposers experience on how uneven treatment people are usually willing to tolerate . this is , however , outside the scope of this paper . the behaviour of dictators in the dictator game @xcite is somewhat more difficult to analyse using bth . the dictator game is similar to the ultimatum game , the only difference being that the other player does not even get to make a choice , and only receives what the first player , or dictator , endows . it has been observed that @xcite in this game the dictators tend to be rather generous , which is difficult but not impossible to explain in the context of bth , if one takes into account the effect of reputation and other `` social goods '' . the nature of such influence on the behaviour of the dictator will be studied in a later work . however , there are some indications that bth could very well be applied to the dictator game when all the social effects are taken into account . it has been reported @xcite that when the rules of the dictator game are modified so that instead of giving money to the other players , the dictator gets to take some or all of the money given to the other players ( thus turning the game into a `` taking game '' ) , the dictator s behaviour changes from egalitarian to self serving , i.e. taking often the majority or even all of the available money . from the bth point of view , the dictator s observed behaviour change can potentially be explained in terms of social norms . in the ordinary dictator game the dictator may still feel bound by the usual norms of the society , while in the `` taking game '' it is encouraged to go against these norms . this sets the `` taker '' apart from the other player in particular , and other members of the society in general . hence the dictator feels `` better '' than the others when breaking the norms with impunity , and act on this feeling by taking money from the other players . the fact that bth can possibly lead to formation of norms as well as rebellion against these norms is well worth of further studies . relating to the known results of the ultimatum game , we have formulated a hypothesis explaining the observed behaviour of humans in terms of superiority maximization , or `` better than''-hypothesis , and presented a simple agent - based model to implement this hypothesis . the model describes agents with constant ranking parameters and raises the question whether the agents with larger ranks are `` better '' than agents with smaller ranks or the other way around . we have found that the social system produced by our model , features homophily , meaning that agents forming social ties with other agents with similar ranking parameters , and assortativity , describing the tendency of highly / lowly connected agents forming links with other highly / lowly connected agents . in addition we find community formation , both in terms of there being communities with opposing opinions and in terms of the communities with the same opinion fracturing into smaller ones according to their ranking parameters . furthermore , we have observed the formation of a hierarchy , in the sense of a connectivity hierarchy being formed on top of the one defined by the ranking parameters , with the agents with extreme ranking parameters presenting higher connectivity than the agents with average ranking parameters . moreover , we have found that the resulting social networks tend to be disconnected for small system sizes , but mostly connected for larger system sizes . this fact may have some relevance for research of early human migrations , hinting of the effects of social pressure in shaping the social network . j.e.s . acknowledges financial support from niilo helander s foundation , g.i . acknowledges a visiting fellowship from the aalto science institute , and k.k . acknowledges financial support by the academy of finland research project ( cosdyn ) no . 276439 and eu horizon 2020 fet open ria project ( ibsen ) no . r.a.b . wants to thank aalto university for kind hospitality during the development of this work . rab acknowledges financial support from conacyt through project 799616 . we acknowledge the computational resources provided by the aalto science - it project . fagundes , m. s. , ossowski , s. , cerquides , j. & noriega , p. ( 2016 ) design and evaluation of norm - aware agents based on normative markov decision processes . _ international journal of approximate reasoning _ , * 78 * , 3361 . j. , h. , boyd , r. , bowles , s. , camerer , c. , fehr , e. & gintis , h. ( 2004 ) _ foundations of human sociality : economic experiments and ethnographic evidence from fifteen small - scale societies_. oxford university press

in human societies , people s willingness to compete and strive for better social status as well as being envious of those perceived in some way superior lead to social structures that are intrinsically hierarchical . here we propose an agent - based , network model to mimic the ranking behaviour of individuals and its possible repercussions in human society . the main ingredient of the model is the assumption that the relevant feature of social interactions is each individual s keenness to maximise his or her status relative to others . the social networks produced by the model are homophilous and assortative , as frequently observed in human communities and most of the network properties seem quite independent of its size . however , it is seen that for small number of agents the resulting network consists of disjoint weakly connected communities while being highly assortative and homophilic . on the other hand larger networks turn out to be more cohesive with larger communities but less homophilic . we find that the reason for these changes is that larger network size allows agents to use new strategies for maximizing their social status allowing for more diverse links between them . community formation , opinion formation , social hierarchy

recent advances in nanoscience are expanding the limits of phononic thermal transport , both in the low conductance and high conductance side @xcite . the highest thermal conductance of a particular material is achieved , when the phonons do not scatter at all ; this is the ballistic transport limit . at room temperature the ballistic limit is quite hard to achieve , but at low temperatures ( a few k and below ) phonon - phonon and phonon - impurity scattering become so weak that ballistic transport can be observed @xcite , and the emitted power has the typical stefan - boltzmann form @xcite @xmath1 summing over the different phonon modes with speeds of sound @xmath2 and radiator emissivities @xmath3 . in addition to bulk samples , ballistic phonon transport was also observed for thin amorphous sin@xmath4 membranes recently @xcite . at low temperatures , the dominant thermal phonon wavelength becomes comparable to the membrane thickness ( for a 1 m @xmath5 membrane this happens at @xmath6100mk ) leading to an effectively 2d phonon gas . we have recently considered ballistic and diffusive thermal conduction in thin membranes @xcite , spanning the transition from the 3d to the fully 2d limit , using the correct low - frequency modes known from elasticity theory ( lamb modes ) . here we explicitly discuss the dependence of ballistic thermal conduction on materials parameters . the simple question we would like to answer is : what kind of material is the best ( worst ) ballistic thermal conductor for any thickness of the membrane , and can we even define a material - only dependent thermal conductivity that does not depend on the sample dimensions ? for the more typical cases where phonon transport is not ballistic , but limited by different scattering mechanisms in the sample , it is possible to define a material dependent local thermal conductivity @xmath7 , which depends on the specific heat @xmath8 , average speed of sound @xmath9 and phonon mean free path @xmath10 as @xmath11 . for 3d crystals , @xmath8 is proportional to @xmath12 , but because @xmath10 scales as @xmath13 , where @xmath14 for phonon - electron , impurity , and phonon - phonon scattering , respectively , @xmath7 always grows with higher speed of sound @xcite . this is the intuitive result that high speed of sound materials are better thermal conductors ( like diamond ) . however , for the case of ballistic conduction this is not true anymore . the thermal conductance @xmath0 ( one can not define a local @xmath7 anymore ) is calculated from the total power @xmath15 that is radiated from an object as @xmath16 . @xmath0 actually has a dependence @xmath17 , as can be seen from the stefan - boltzmann law , eq . ( [ eqn_stefan_boltzmann_law ] ) . the material with the _ lowest _ speed of sound is the best thermal conductor ! in the following we will investigate what is the situation for thin membranes . in isotropic 3d bulk systems there are three independent phonon modes , two transversally and one longitudinally polarized , with sound velocities @xmath18 and @xmath19 , respectively . in the presence of boundaries , the bulk phonon modes couple to each other and form a new set of eigenmodes , which in the case of a free standing membrane are horizontal shear modes ( @xmath20 ) and symmetric ( @xmath21 ) and antisymmetric ( @xmath22 ) lamb modes @xcite . the frequency @xmath23 for the @xmath20 modes is simply @xmath24 , where @xmath25 is the wave vector component parallel to membrane surfaces , @xmath26 is the membrane thickness and the integer @xmath27 is the branch number . the dispersion relations of the @xmath21 and @xmath22 modes can not be given in a closed form , but have to be calculated numerically @xcite . the lowest three branches , dominant for thin membranes at low temperatures , have the low frequency expressions @xmath28 , @xmath29 and @xmath30 , with the effective sound velocity @xmath31 of the @xmath21 mode and the effective mass @xmath32^{-1}$ ] of the @xmath22 mode `` particle '' @xcite . this lowest @xmath22 mode has a quadratic dispersion instead of the usual phonon - like linear one , and is mostly responsible for the non - trivial behavior of thin membranes at low temperatures . figure [ fig_fig1 ] shows a schematic of the system in consideration . a thin metal film of perimeter @xmath33 heats the membrane directly below @xcite . to simplify the discussion , we assume that the hot phonons from the heater have a thermal distribution and are radiated from its perimeter into the membrane , similar to black body radiation . we further assume that no radiation is backscattered . with these assumptions the total heat flow out of the detector is @xmath34 where @xmath35 is the bose - einstein distribution and @xmath36 and @xmath27 are the mode and branch indices @xcite . if enough branches are used , the 3d to 2d transition can be computed from this expression . if the membrane is thin and temperature low ( @xmath37 ) , only the lowest branch of each mode is occupied ( @xmath27=0 ) , and we are fully in the 2d limit , in which case we get from eq . ( [ eqn_p_of_t ] ) @xmath38\,.\label{eqn_p_2d}\ ] ] note that the effective mass of the lowest @xmath22 mode depends on the membrane thickness and hence in the 2d limit @xmath39 . in the 3d limit @xmath40 , the dominant phonon wavelength is much smaller than @xmath26 , leading to decoupling of the longitudinal and transversal modes and @xmath41 as expected , @xmath42 . this means that at a fixed temperature the radiated power will first decrease with decreasing @xmath26 , then reach a global minimum and will increase again , if we decrease @xmath26 further . the minimum is approximately at the 2d-3d crossover thickness @xmath43 . ( a ) ballistic conductance @xmath0 and ( b ) ballistic conductance per unit area @xmath44 as function of membrane thickness @xmath26 . the dash - dotted line in ( a ) is @xmath45 . ] ( a ) ballistic conductance @xmath0 and ( b ) ballistic conductance per unit area @xmath44 as function of membrane thickness @xmath26 . the dash - dotted line in ( a ) is @xmath45 . ] using the 300 lowest modes , we have numerically computed the ballistic phonon conductance @xmath0 for a range of parameters that spans the transition from 2d to 3d . [ fig_fig2 ] ( a ) shows the computed @xmath0 for @xmath46=100 mk and @xmath33=1.6 m as a function of the membrane thickness @xmath26 for four different materials : pb ( @xmath47 m / s , @xmath48 m / s ) , al ( @xmath49 m / s , @xmath50 m / s ) , sin ( @xmath51 m / s , @xmath52 m / s ) , and poly - diamond ( @xmath53 m / s , @xmath54 m / s ) . the values of @xmath9 are calculated values for the isotropic case , based on measured single crystal elastic constants @xcite . all materials have a physically relevant critical thickness , ranging from @xmath55 nm for pb to @xmath56 nm in diamond , below which @xmath0 increases with decreasing thickness . the material with highest @xmath0 is pb , as it has the lowest speeds of sound , and diamond is the poorest conductor . in the 2d limit , where @xmath57 , the differences between the materials are not as strong as in the 3d limit . one could say that all materials start to resemble each other in the ultrathin range @xmath58 nm . on the other hand , in the 3d limit , @xmath0 increases linearly with @xmath26 , the common sense result . therefore , we could try to define a materials parameter @xmath44 , the conductance per unit emission area , for all thicknesses . this is plotted in fig . [ fig_fig2 ] ( b ) . as we see , in the 3d limit @xmath44 is not a function of @xmath26 , and is truly a materials parameter . however , when the membrane becomes thinner , @xmath44 starts increasing strongly . this means that thin membranes are actually more effective thermal conductors , per unit emission area . this increase in ballistic `` conductivity '' is striking : for diamond it can be four orders of magnitude from @xmath59 m to @xmath60 nm . therefore , a way of increasing the ballistic thermal conductance in a given volume would be to use a structure composed of many parallel thin membranes of a material with low speed of sound . logarithmic derivative of the ballistic conductance @xmath61 as function of temperature . in the 3d limit @xmath62 , while in the 2d limit @xmath63 . ] logarithmic derivative of the ballistic conductance @xmath61 as function of temperature . in the 3d limit @xmath62 , while in the 2d limit @xmath63 . ] the dependence of @xmath0 on temperature is plotted in fig . [ fig_fig3 ] for @xmath64 nm . in the 3d limit , @xmath0 is naturally proportional to @xmath65 . in the 2d limit however , the power of the temperature dependence does not just drop by one , but instead we have @xmath66 , and the crossover temperature is between 50 .. 400 mk for @xmath64 nm . to give a better picture of how the power of the temperature dependence of @xmath0 changes with temperature , we plot the logarithmic derivative @xmath67 of @xmath0 in fig . [ fig_fig4 ] . for a given temperature , ballistic thermal conduction is maximized by choosing a material with low speed of sound , and by using several parallel thin membranes , instead of a bulk slab of the same total thickness . a bulk slab of diamond of thickness 30 @xmath68 m has the same conductance as a 2 nm thick lead membrane . the lead membrane is thus about 15 000 more effective conductor . applications in cooling ultrasensitive thermal detectors can be envisioned . this work was supported by the academy of finland project nos . 118665 and 118231 . 10 kim w , wang r and majumdar a 2007 _ nanotoday _ * 2 * 40 von gutfeld r j and nethercot a h 1964 _ phys . * 12 * 641 klitsner t , vancleve j e , fisher h e and pohl r o 1988 _ phys . _ b * 38 * 7576 wolfe j _ imaging phonons _ ( cambridge : cambridge university press ) hoevers h f c , ridder m l , germeau a , brujin m p , de korte p a j and wiegerink r j 2005 _ appl . phys . lett . _ * 86 * 251903 khn t and maasilta i j 2006 _ nucl . instrum . methods phys . res . _ a * 559 * 724 khn t and maasilta i j 2007 _ preprint _ cond - mat/0702542 khn t , anghel d v , galperin y m and manninen m _ phys . b in press _ arxiv:0705.1936v1 auld b a 1990 _ acoustic fields and waves in solids _ vol ii 2nd ed ( malabar : robert e. krieger ) khn t , anghel d v , pekola j p , manninen m and galperin y m 2004 _ phys . rev . _ b * 70 * , 125425 note that the prefactor is @xmath69 instead of the factor @xmath70 that was mistakenly given in ref . simmons g and wang h 1971 _ single crystal elastic constants and calculated aggregate properties : a handbook _ 2nd ed ( cambridge : mit press )

at low temperatures , phonon scattering can become so weak that phonon transport becomes ballistic . we calculate the ballistic phonon conductance @xmath0 for membranes using elasticity theory , considering the transition from three to two dimensions . we discuss the temperature and thickness dependence and especially concentrate on the issue of material parameters . for all membrane thicknesses , the best conductors have , counter - intuitively , the _ lowest _ speed of sound .

working out a quantitative description of the properties of dense strongly interacting matter produced in ultrarelativistic heavy ion collisions presents one of the most fascinating problems in high energy physics . the simplest ( albeit not unique ) way of putting the experimental data from rhic @xcite and lhc @xcite into a coherent framework is to describe the essential physics of these collisions as a hydrodynamical expansion of primordial quark - gluon matter that , after a short transient period , reaches sufficient level of local equilibration allowing the usage of hydrodynamics . the features of the experimentally observed energy flow , in particular the presence of a strong elliptic flow , suggest early equilibration of the initially produced matter and small shear viscosity of the expanding fluid , see e.g. the discussion in @xcite and @xcite devoted to rhic and lhc results respectively . can stylized features of primordial quark - gluon matter , in particular its anomalously low viscosity , be described within a weakly coupled theory , i.e. as a plasma composed of quasiparticles with the quantum numbers of quarks and gluons ? to address this question let us recall that extensive experimental studies of `` ordinary '' electromagnetic plasma has demonstrated , see e.g. @xcite , that it is practically never observed in the state of textbook thermal equilibrium . realistic description of the properties of experimentally observed qed plasma is possible only through taking into account the presence , in addition to thermal excitations , of randomly excited fields . the resulting state was termed _ turbulent plasma_. collective properties of turbulent plasmas are markedly different from those of the ordinary equilibrium plasmas . in particular , they are characterized by anomalously low shear viscosity and conductivity , dominant effects of coherent nonlinear structures on transport properties . thus it is natural to consider turbulent qcd plasma as a natural candidate for describing the primordial quark - gluon matter in the weak coupling regime . calculation of shear viscosity of turbulent qgp performed in @xcite has indeed demonstrated that its shear viscosity is anomalously small . in the present paper we focus on studying the leading turbulent contributions to polarization properties of turbulent relativistic plasma . for simplicity we restrict our consideration to the abelian case . the non - abelian generalization is briefly described in section [ conc ] . a weakly turbulent plasma is described as perturbation of an equilibrated system of ( quasi-)particles by weak turbulent fields @xmath0 . in the collisionless vlasov approximation , the plasma properties are defined by the following system of equations ( @xmath1 is a regular non - turbulent field ) : @xmath2f(p , x , q)=0 \nonumber\\ & & \partial^{\mu}\left ( f^r_{\mu \nu } + f^t_{\mu \nu } \right)= j_{\nu}(x ) = e \sum_{q , s}\int dp\ , p_{\nu}\ , q\ , f(p , x , q ) . \label{kinetic+maxw}\end{aligned}\ ] ] the stochastic ensemble of turbulent fields is assumed to be gaussian and characterized by the following correlators : @xmath3 in the present study we use the following parametrization of the two - point correlator @xmath4 @xcite : @xmath5\ ] ] turbulent polarization arises as a ( linear ) response to a regular perturbation that depends on turbulent fields . it is fully characterized by the polarization tensor @xmath6 defined as a variational derivative of the averaged induced current @xmath7 over the regular gauge potential @xmath8 : @xmath9 let us rewrite the kinetic equation in ( [ kinetic+maxw ] ) in the following condensed form @xmath10 where @xmath11 is a distribution function characterizing the original non - turbulent plasma and introduce the following systematic expansion in the turbulent and regular fields : @xmath12 where powers of @xmath13 count those of @xmath14 and powers of @xmath15 count those of @xmath16 . turbulent polarization is described by contributions of the first order in the regular and the second in the turbulent fields . the lowest nontrivial contribution to the induced current ( [ incur ] ) is thus given by @xmath17 . we have @xmath18 where @xmath19 generic expression for the polarization tensor taking into account turbulent effects can be written as @xmath20 where @xmath21 . both longitudinal and transverse components can be presented as a sum of hard thermal loops ( htl ) contributions and the gradient expansion in the turbulent scale @xmath22 : @xmath23 \label{gradexp } \nonumber\end{aligned}\ ] ] @xmath24 and the standard htl contribution @xmath25 , \nonumber \\ & & \pi_{t}^{\mathrm htl } ( \omega,{\left| \mathbf { k } \right|})= m^2_d \dfrac{x^2}{2 } \left[1+\dfrac{1}{2 x } \ ; ( 1-x^2 ) \ ; l(x ) \right ] \nonumber \\ & & l(x ) \equiv \ln\left|\dfrac{1+x}{1-x}\right|-\imath\pi\theta(1-x ) ; \;\;\ ; m^2_d = e^2 t^2/3 . \label{htl}\end{aligned}\ ] ] the computation of turbulent polarization was carried out to second order in the gradient expansion @xcite . in what follows we restrict ourselves to discussing the leading contribution to the imaginary part of the polarization function corresponding to the turbulent modification of landau damping in ( [ htl ] ) : @xmath26 the functions @xmath27 and @xmath28 are shown in fig . [ pct1 ] . $ ] ( solid lines ) and @xmath29 $ ] ( dashed lines ) . left : transverse response ; right : longitudinal response.,title="fig:",scaledwidth=45.0% ] $ ] ( solid lines ) and @xmath29 $ ] ( dashed lines ) . left : transverse response ; right : longitudinal response.,title="fig:",scaledwidth=45.0% ] the conclusions following from fig . [ pct1 ] can be formulated as follows : 1 . * timelike domain . * from fig . [ pct1 ] we see that the sign of the imaginary part of the turbulent contribution to the polarization operator in the timelike domain @xmath30 is negative and corresponds to turbulent damping of timelike collective excitations . this refers to both transverse and longitudinal modes . as the htl contribution in this domain is absent , this turbulent damping is a universal phenomenon present for all @xmath31 such that @xmath32 and all values of the parameters involved ( @xmath22 , @xmath33 , @xmath34 ) . the turbulent damping leads to an attenuation of the propagation of collective excitations at some characteristic distance . * spacelike domain . * the situation in the spacelike domain @xmath35 is more diverse . in contrast with the timelike domain the gradient expansion for the imaginary part of the polarization tensor starts from the negative htl contribution corresponding to landau damping . as seen from fig . [ pct1 ] the imaginary parts of turbulent contributions to the longitudinal polarization tensor are negative and are thus amplifying the landau damping . the most interesting contributions come from the turbulent contributions to the transverse polarization tensor . we see that the electric contribution @xmath36 $ ] in the spacelike domain is positive at all @xmath37 while the magnetic contribution @xmath38 $ ] is negative for @xmath39 and positive for @xmath40 . this means that the turbulent plasma becomes unstable for sufficiently strong turbulent fields . let us briefly discuss some relevant points : * 1 . * the above presented results are obtained in the framework of a perturbative expansion based on two crucial assumptions . first , one assumes slow temporal evolution of the distribution function due to particle interaction with turbulent fields thus neglecting the corresponding @xmath41 contributions . second , changes in the distribution function are treated as small . this , in turn , means that turbulent fields should be small enough . in this sense the reliable results refer to small modifications of landau damping but , as the onset of turbulent instability takes place for parametrically large fields , this result should be considered as a qualitative indication . the observed instability can be termed `` secondary '' because the turbulent fields themselves result from some `` first level '' instabilities . the origin of the effect is in turbulent stochastic inhomogeneity and thus similar to stochastic transition radiation , which vanishes in the limit @xmath42 ) @xcite . the non - abelian generalization of the above - described results for the imaginary part of the polarization tensor leads to identical expressions , the only difference being in trivial color factors , just as in the htl case . i. arsene , et al . ( brahms ) , _ nucl . phys . _ * a757 * ( 2005 ) , 1 + b.b . back , et al . ( phobos ) , _ nucl . * a757 * ( 2005 ) , 28 + j. adams , et al . ( star ) , _ nucl . phys . _ * a757 * ( 2005 ) , 102 + k. adcox , et al . ( phenix ) , _ nucl . phys . _ * a757 * ( 2005 ) , 184 v.v . tamoykin , _ astrophysics and space science _ ( 1972 ) , 120 + m.r . kirakosyan , a.v . leonidov , `` stochastic jet quenching in high energy nuclear collisions '' , arxiv:0810.5442 [ hep - ph ] + m.r . kirakosyan , a.v . leonidov , `` energy loss in stochastic abelian medium '' , proc . quarks 2008 , zagorsk , russia , arxiv:0809.2179 [ hep - ph ]

polarization properties of turbulent stochastically inhomogeneous ultrarelativistic qed plasma are studied . it is shown that the sign of nonlinear turbulent landau damping corresponds to an instability of the spacelike modes and , for sufficiently large turbulent fields , to an actual instability of a system .

single layer graphene is a zero - gap semiconductor whose low energy quasiparticles obey the massless dirac equation @xcite . at charge neutrality , the fermi surface reduces to two inequivalent fermi points , forming a non - analyticity in the density of states , which can be viewed as a rather simple quantum critical point @xcite . on top of that , however , as a consequence of the linear dispersion of the 2d quasiparticles coulomb interactions are unusually strong . they are only marginally irrelevant under renormalization , flowing only logarithmically to zero with decreasing temperature @xmath0 , see @xcite . this is reflected , e.g. , in the inelastic scattering rate being proportional to @xmath1 ( @xmath2 ) , where @xmath3 is the ( slowly running ) dimensionless `` fine structure constant '' characterizing the strength of coulomb interactions , where @xmath4 is the dielectric constant of the adjacent medium and @xmath5 is the fermi velocity of the linearly dispersing quasiparticles . this large scattering rate nearly saturates a heisenberg uncertainty principle for quasiparticles @xcite , according to which the scattering rate is conjectured never to exceed significantly the thermal energy scale . indeed , upon approaching @xmath6 one expects a chiral symmetry breaking quantum phase transition towards an insulator @xcite with very different low energy excitations . due to the strong marginal interactions , the neutrality point of graphene is very similar to quantum critical points of more complex , strongly coupled materials @xcite . in the quantum critical window , i.e. , at small chemical potential of the carriers , @xmath7 , the latter form an interacting `` hot '' electron - hole plasma with rather unusual transport properties which we discuss below . at finite carrier density , an estimate of the inelastic scattering rate in random phase and born approximation leads to ^-1_inel ~max ( t,|| ) , where @xmath8 denotes the renormalized strength of coulomb interactions @xmath9 at a given energy scale @xmath10 $ ] , whereby @xmath11 is a uv cutoff . at finite @xmath12 , the scattering rate decreases rather quickly according to the familiar law @xmath13 , independent of the interaction strength in the ultraviolet . the quantum - critical window is clearly distinguished by its strong inelastic scattering rate @xmath14 , which has several interesting consequences . as was first pointed out in the context of the superfluid - insulator quantum phase transition @xcite the particle - hole symmetric point @xmath15 exhibits a finite collision - dominated conductivity , even in the absence of impurities . indeed , the application of an external electrical field induces counter propagating particle and hole currents , and thus no net momentum . the latter is usually the source of infinite current response unless the momentum decays due to impurities . however , in neutral graphene one finds a disorder - independent conductivity which is solely due to electron - hole friction . scaling arguments based on the drude formula , the thermal density of carriers @xmath16 , the inelastic scattering rate and a @xmath0-dependent `` effective mass '' @xmath17 suggest a conductivity which grows logarithmically with @xmath0 ( = 0)~ = . this is indeed confirmed by a microscopic calculation based on the semiclassical boltzmann equation , which becomes asymptotically exact for @xmath18 where the coupling is @xmath19 , yielding the prefactor @xmath20 @xcite . for the same reason as the electrical conductivity remains finite at particle - hole symmetry , the thermal conductivity @xmath4 diverges at @xmath15 . for the case of relativistically invariant systems this has been shown by vojta _ et al . _ @xmath4 describes the heat current response to a thermal gradient in the absence of an electrical current . usually , the latter forbids the excitation of a finite , non - decaying momentum , and this ensures a finite heat current response . at particle - hole symmetry , however , the momentum created by a thermal gradient does not carry a net current and is thus not affected by the boundary condition . it follows that within the bulk of a sample a thermal gradient can not be sustained at @xmath15 ( see ref . @xcite for a discussion of @xmath4 in a sample coupled to leads ) . for graphene , both relativistic hydrodynamics @xcite and boltzmann theory yield the leading divergence ( 0)= ( ) ^2 , @xmath21 and @xmath22 being the pressure , energy density and charge density of the fluid , respectively . this relation can be interpreted as a relativistic wiedemann - franz - like relation between @xmath23 and @xmath4 . a further consequence of the strong coulomb coupling in graphene , and more generally , of quantum criticality , is the anomalously low value of the shear viscosity @xmath24 . its ratio to the entropy density , @xmath25 is the crucial parameter in the navier - stokes equation which controls the occurrence of turbulence via the reynolds number = , where @xmath26 is a typical length and @xmath27 a typical velocity scale of the electronic current flow . the tendency towards electronic turbulence is stronger the larger is @xmath28 . full - fledged turbulence might require @xmath29 in 2d , but interesting , complex flow is already expected at experimentally accessible values @xmath30 @xcite . viscosity having the units of @xmath31 with @xmath32 a density the ratio has units of @xmath33 . for massless fermions or bosons , the coefficient of proportionality is essentially the mean free path divided by the thermal de broglie wavelength . this ratio is usually large , but becomes of order @xmath34 when the scattering rate tends to saturate heisenberg s uncertainty relation . for certain strongly coupled relativistic liquids the low value @xmath35 was obtained via the ads - cft correspondence . interestingly , a similarly low value is found for graphene when the weak coupling result @xmath36 is extrapolated to values of order @xmath6 . @xcite at charge neutrality , @xmath37 is thus a rather direct measure for the dimensionless interaction strength @xmath38 ( equal to @xmath9 at @xmath15 ) , and can serve as an experimental indicator for the latter . at finite carrier density , @xmath39 , the ratio @xmath25 behaves however as @xmath40 , and thus does not directly inform about the strength of the dimensionless coupling @xmath38 . a better measure is instead furnished by the estimate ( ) ^2 . one may speculate that in this case @xmath24 is `` softly '' bounded from below by the restriction @xmath41 . while the quasiparticles of graphene have relativistic kinematics and the non - interacting theory enjoys lorentz - invariance , the instantaneous coulomb interactions obviously break this symmetry . however , one may expect that at hydrodynamic frequency and length scales , the relativistic invariance re - emerges . this is indeed the case , since the role of interactions mainly consists in establishing local equilibrium , while the relaxation towards global equilibrium is governed by the conservation laws of charge , momentum and energy . the latter essentially remain those of a relativistic fluid , thus fixing the structure of all thermoelectric response functions at small frequencies and wavevectors @xcite . it can be shown @xcite that the specificity of the interactions enters the low - energy hydrodynamic description only via the values of the two dissipative transport coefficients @xmath23 and @xmath24 discussed above for @xmath15 . a third coefficient , the bulk viscosity , is very small in graphene due to its near scale invariance . a microscopic analysis @xcite shows that a hydrodynamic description still applies , even when translational invariance and thus momentum conservation is weakly broken by a small magnetic field and dilute disorder . the necessary condition is that the scattering rates associated with those perturbations are significantly smaller than @xmath42 . an interesting prediction of the resulting magneto - hydrodynamics is the presence of a collective cyclotron resonance @xcite . its origin is a collective circulating motion of the electron - hole plasma with a frequency proportional to its net charge density @xmath22 . in contrast to systems with galilean invariance with a single type of charge carrier , for which kohn s theorem ensures a sharp , undamped cyclotron resonance , the cyclotron motion in graphene is damped due to electron - hole friction . our investigations of interaction effects in graphene were to a large extent stimulated by very similar physical phenomena in strongly coupled critical systems and ultrarelativistic matter such as the quark - gluon plasma . @xcite indeed , it is interesting to confront and compare the properties of graphene with results obtained for relativistic supersymmetric @xmath43 gauge theories in @xmath44 dimensions , with which one hopes to catch some of the physically relevant features of the quark gluon plasma . these gauge theories are strongly coupled field theories with an emerging conformal symmetry in the infrared . the ads@xmath45-cft@xmath46 correspondence stipulates that this theory is exactly dual to an einstein - maxwell gravity theory in @xmath47 dimensions , which in the large @xmath48 limit becomes exactly solvable . among others , this allows one to compute transport coefficients of the gauge theory @xcite = ( ) ^1/2 n^3/2 ; s , n^3/2t^2;=. these results can be physically interpreted as reflecting properties of @xmath49 effective degrees of freedom @xcite , whose strong interactions result in an inelastic scattering time which essentially saturates the heisenberg - like bound @xmath50 . the ads - cft correspondence also confirmed the predictions of relativistic magnetohydrodynamics , including the cyclotron resonance , as discussed above . graphene exhibits several peculiar transport properties which it shares with other quantum critical systems as well as with ultrarelativistic matter . the crucial ingredients behind these phenomena are anomalously strong coulomb interactions in 2d and the relativistic kinematics . experimental verifications at moderately high @xmath0 should soon be within reach due to the availability of clean suspended graphene .

electrons and holes in clean , charge - neutral graphene behave like a strongly coupled relativistic liquid . the thermo - electric transport properties of the interacting dirac quasiparticles are rather special , being constrained by an emergent lorentz covariance at hydrodynamic frequency scales . at small carrier density and high temperatures , graphene exhibits signatures of a quantum critical system with an inelastic scattering rate set only by temperature , a conductivity with a nearly universal value , solely due to electron - hole friction , and a very low viscosity . in this regime one finds pronounced deviations from standard fermi liquid behavior . these results , obtained by boltzmann transport theory at weak electron - electron coupling , are fully consistent with the predictions of relativistic hydrodynamics . interestingly , very analogous behavior is found in certain strongly coupled relativistic liquids , which can be analyzed exactly via the ads - cft correspondence , and which had helped identifying and establishing the peculiar properties of graphene .

many high energy scattering experiments probe hadron structure very close to the light - cone . hence correlation functions probed in such experiments have a particularly simple physical interpretation in the light - front ( lf ) framework ( or infinite momentum frame ) , i.e. in a hamiltonian framework where @xmath1 is ` time ' . for example , parton distribution functions measured in deep - inelastic scattering experiments provide information about the lf momentum distribution of quarks in the target hadron . off - forward parton distributions , probed in deeply virtual compton scattering , have the interpretation of matrix elements of the lf momentum density operator between states of unequal momenta . furthermore , the pion distribution amplitude , relevant for example for the asymptotic pion form factor , can be related to the lf wavefunction for the pion in the @xmath2 fock component when @xmath3 and @xmath4 have the same @xmath5 position . even though these important observables have their most physical interpretation on the lf , it is , at least in principle , possible to calculate them in any approach to qcd . however , what distinguishes the lf framework from all other formulation of qcd is that the above observables have a very direct and physical connection to the microscopic degrees of freedom in terms of which the hamiltonian is constructed . because of this unique feature , it should be much easier in this framework to gain a physical understanding between experiment and phenomenology on the one hand and the underlying qcd dynamics on the other . other motivations to study qcd formulated on the lf derive from the fact that the complexity of the vacuum seemingly shifts from the states to the operators in this framework @xcite . this results in a separation between the physics of the vacuum and the parton structure of hadrons which implies for example that a constituent picture of hadrons has a chance to make sense @xcite . of course , just like in any other approach to qcd , it is necessary to regularize both uv and ir divergences before one can even attempt to perform nonperturbative calculations . the transverse lattice @xcite is an attempt to combine advantages of the lf and lattice formulations of qcd . in this approach to lf - qcd the time and one space direction ( say @xmath6 ) are kept continuous , while the two ` transverse ' directions @xmath7 are discretized . keeping the time and @xmath6 directions continuous has the advantage of preserving manifest boost invariance for boosts in the @xmath6 direction . furthermore , since @xmath8 also remain continuous , this formulation still allows a canonical lf hamiltonian approach . on the other hand , working on a position space lattice in the transverse direction allows one to introduce a gauge invariant cutoff on @xmath5 momenta in a manner that is similar to euclidean or hamiltonian lattice gauge theory . in summary , the lf formulation has the advantage of utilizing degrees of freedom that are very physical since many high - energy scattering observables ( such as deep - inelastic scattering cross sections ) have very simple and intuitive interpretations as equal lf - time ( @xmath9 ) correlation functions . using a gauge invariant ( position space- ) lattice cutoff in the @xmath5 direction within the lf framework has the advantage of being able to avoid the notorious @xmath10 divergences from the gauge field in lf - gauge which plague many other hamiltonian lf approaches to qcd @xcite . the hybrid treatment ( continuous versus discrete ) of the longitudinal / transverse directions implies an analogous hybrid treatment of the longitudinal versus transverse gauge field : the longitudinal gauge field degrees of freedom are the non - compact @xmath11 while the transverse gauge degrees of freedom are compact link - fields . each of these degrees of freedom depend on two continuous ( @xmath12 ) and two discrete ( @xmath13 ) space - time variables , i.e. from a formal point of view the canonical transverse lattice formulation is equivalent to a large number of coupled @xmath14 dimensional gauge theories ( the longitudinal gauge fields at each @xmath13 ) coupled to nonlinear @xmath15 model degrees of freedom ( the link fields ) @xcite . for a variety of reasons it is advantageous to work with transverse gauge degrees of freedom that are general matrix fields rather than @xmath16 . first of all , we would like to work at a cutoff scale which is small ( in momentum scale ) since only then do we have a chance to find low lying hadrons that are simple ( i.e. contain only few constituents ) . if one wants to work on a very coarse lattice , it is useful to introduce smeared or averaged degrees of freedom . upon averaging over neighboring ` chains ' of @xmath17 fields one obtains degrees of freedom which still transform in the same way as the original @xmath17 degrees of freedom under gauge transformations but are general matrix degrees of freedom no longer obeying @xmath18 and @xmath19 . the price that one has to pay for introducing these smeared degrees of freedom are more complicated interactions . the advantage is that low lying hadrons can be described in a fock expansion ( this has been confirmed by calculations of the static quark - antiquark potential @xcite and glueball spectra @xcite ) . another important advantage of this ` color - dielectric ' approach is that it is much easier to construct a fock expansion of states out of general linear matrix fields than out of fields that are subject to non - linear @xmath17 constraints . in the color - dielectric approach the complexity is shifted from the states to the hamiltonian : in principle , there exists an exact prescription for the transformation from one set of degrees of freedom ( here @xmath20 s ) to blocked degrees of freedom @xmath21 e^-s_eff.(m ) = e^-s_can.(u ) ( m- _ av_i u_i ) . the problem with this prescription is that @xmath22 is not only very difficult to determine directly , but in general also contains arbitrarily complicated interactions . a much more practical approach towards determining the effective interaction among the link fields nonperturbatively is the use of lorentz invariance . this strategy has been used in a systematic study of glueball masses in ref . @xcite , where more details can be found regarding the effective interaction . one starts by making the most general ansatz for the effective interaction which is invariant under those symmetries of qcd that are not broken by the @xmath5 lattice . this still leaves an infinite number of possible terms and for practical reasons , only terms up to fourth order in the fields and only local ( in the @xmath5 direction ) terms have been included in the ref . the coefficients of the remaining terms are then fitted to maximize lorentz covariance for physical observables , such as the @xmath23 potential ( rotational invariance ! ) and covariance of the glueball dispersion relation . it should be emphasized that these are first principle calculations in the sense that the only phenomenological input parameter is the overall mass scale ( which can for example be taken to be the lowest glueball mass or the string tension ) . the only other input that is used is the requirement of lorentz invariance . the numerical results from refs . @xcite within this approach are very encouraging : * with only a few parameters , approximate lorentz invariance could be achieved for a relatively large number of glueball dispersion relations simultaneously @xcite as well as for the @xmath23 potential * the glueball spectrum that was obtained numerically on the @xmath5 lattice for @xmath24 is consistent with euclidean monte carlo lattice gauge theory calculations performed at finite @xmath25 and extrapolated to @xmath24 . for further details on these very interesting results , the reader is referred to refs . @xcite and references therein . in this paper , we are applying this program to mesons on the transverse lattice within the femtoworm approximation @xcite . this hamiltonian has already been analyzed in ref . @xcite for mesons with @xmath26 . in this work , as well as in a related independent paper @xcite , the work from ref . @xcite is generalized to mesons with nonzero transverse momenta . this generalization allows one to address the issue of lorentz invariance of meson spectra . there are significant differences in the general approach between our work and ref . first of all , while we employ continuous basis function techniques to solve the longitudinal dynamics , refs . @xcite use discrete light - cone quantization ( dlcq ) . secondly , there is also a difference in the procedure : both this work and ref . @xcite use a longitudinal gauge coupling taken from pure glue calculations , while ref . @xcite refits this parameter . the latter difference has important consequences which we discuss in detail in section iii . our results confirm some of the results presented in ref . @xcite , which serves as a non - trivial test for both approaches . however , the main motivation for our work lies in a different direction . our goal was to understand to what extent the requirement of imposing lorentz invariance sufficiently constrains the parameters of the hamiltonian first studied in ref . @xcite and also to investigate whether this hamiltonian and the femtoworm approximation are able to yield lorentz invariant results . the paper is organized as follows . we briefly describe the hamiltonian , the approximations used to solve it and some of it s subtle properties in section ii . in section iii , we provide a perturbative analysis of the @xmath27 splitting . the results from this section will be of importance for understanding the non - perturbative numerical results presented in section iv . this is followed by some comments and comparisons of this work with another concurrent work in section v and a brief summary in section vi . the transverse lattice hamiltonian for wilson fermions that forms the basis of our work and ref . @xcite was first constructed in ref . @xcite , and we refer the reader to these works for more details . here we restrict ourselves to a more qualitative description of the degrees of freedom that enter the hamiltonian and we follow up later with more specifics when we discuss the integral equations describing mesons in the femtoworm approximation . the degrees of freedom that enter the hamiltonian are quark and antiquark creation operators at each site and link - fields ( which , in the spirit of the color - dielectric approach @xcite , we take as general matrix fields ) on the transverse links connecting the sites . for simplicity , we work in the large @xmath25 limit . the first approximation that we use is a light front tamm - dancoff truncation of the fock state expansion upto and including the 3-particle sector . upon reminding oneself that the fermion degrees of freedom ( quarks and antiquarks ) always occupy the lattice ( transverse ) sites and the gauge degrees of freedom ( link fields ) reside on the lattice spacings , it is easy to understand the following two direct consequences of the aforementioned truncation : * the 2-particle states consist of a quark and an antiquark sitting on the same lattice site . * the 3-particle states consist of a quark and an antiquark separated by * at most * one link field . since link fields always reside on the lattice spacings as demanded by gauge invariance , the 3-particle fock space is therefore spanned by all different configurations of a quark and an antiquark separated by one link field . this dumb - bell shaped configuration of the 3-sector naturally requires one to introduce an index to specify it s transverse orientation . we do so as shown in fig . [ fig : orient ] . in keeping with the truncated fock space , one can therefore write the meson state vector as : meson= c_2q|q+ c_3q|qgthe total angular momentum of the fermions in the 2-sector is contributed wholly by their respective spins since there is no contribution from orbital angular momentum when they are both on the same site . on the other hand , the total angular momentum of the 3-particle configuration is the sum of both their spins as well as the orbital angular momentum typical of dumb - bell shaped systems . to characterize a 2-particle state , one needs the spins of the two quarks along with their momenta , both longitudinal as well as transverse . additionally , a 3-particle state will require the direction of it s configuration ( see fig . [ fig : orient ] ) in the discretized _ xy _ directions . the hamiltonian , however , remains invariant when translated in multiples of the transverse lattice spacing . consequently , the transverse momenta of the mesons are conserved ( modulo @xmath28 ) . this allows one to solve the light front hamiltonian equation , @xmath29 , for fixed values of the transverse momentum , @xmath30 . for a given value of @xmath30 we therefore make the following ansatz : q|q&=&__e^i_.__0 ^1dx b^_s_1(x , _ ) + & & d^_s_2(1-x,_)_s_1,s_2(x)0[eq:2.2 ] q|qg&=&__e^i_.(_+ ) + & & _ 0 ^ 1dx_0 ^ 1-xdyb^_s_1(x,_)d^_s_2(y,_+_i ) + & & a^(1-x - y)_s_1,s_2(x , y,_i,)0[eq:2.3 ] where the spin degrees of freedom , @xmath31 and the orbital degrees of freedom , @xmath32 . in the above , @xmath33 and @xmath34 are the wavefunction amplitudes in the 2-particle and 3-particle sectors respectively . it is a straight forward though tedious procedure to show that the qcd hamiltonian with the above state vectors in the large @xmath35 limit and under the light front gauge yields the following integral equations in terms of the wavefunction amplitudes in the 2-particle and 3-particle sectors respectively : h_22_s_1,s_2(x)&=&(+ ) _ s_1,s_2(x ) + & & -^1_0dx [ eq:5 ] h_33&&_s_1,s_2(x , y,_i)=(++ ) + & & _ s_1,s_2(x , y,_i ) -^1-y_0dx + & & -^1-x_0dy + [ eq:6 ] in eq . ( [ eq:5 ] ) , the first two terms on the right hand side are the kinetic energies of the quark and the anti - quark while the third term describes the coulomb interaction between them . in eq . ( [ eq:6 ] ) , the first three terms on the right hand side are the kinetic energies of the quark , anti - quark and the link fields , the fourth term describes the coulomb interaction between the quark(antiquark ) and the link field and the fifth term describes the coulomb interaction between the antiquark(quark ) and the link field . it should be noted that a coulomb interaction term between the quark and antiquark in the 3-particle sector does not appear since it is disallowed under the large @xmath35 limit . three remarks are in order at this point . first , @xmath36 and @xmath37 , i.e. the terms in the hamiltonian that are diagonal in fock space are independent of the quark spin and the orientation of the link . second , the @xmath38 s in eqs . ( [ eq:5 ] ) and ( [ eq:6 ] ) still depend implicitly on @xmath30 because although @xmath30 is conserved , @xmath30 still enters the integral equations as a parameter . finally , eqs . ( [ eq:5 ] ) and ( [ eq:6 ] ) have been rescaled in a manner such that @xmath39 . the natural question that arises at this point is : how do the 2-sector and the 3-sector mix with each other . this brings us to the issue of propagation of the mesons on the lattice . qualitatively , the way this takes place is as follows : a @xmath3(@xmath4 ) hops to a neighboring site with the emission of a link field ( 2-sector @xmath40 3-sector ) . the @xmath4(@xmath3 ) sitting on the original site then jumps to join the @xmath3(@xmath4 ) by absorbing the link field ( 3-sector @xmath40 2-sector ) . since this propagation resembles that of an inchworm on a scale of a femtometer , we were inspired to called this the _ femtoworm approximation_. the hopping of a quark or an antiquark from one site to the next can cause it to either flip its spin or not . accordingly , we introduce two distinct couplings between the fermion and the link degrees of freedom . in this work , we label the spin - flip coupling as @xmath41 and the non spin - flip coupling as @xmath42 . the corresponding interaction terms are presented in details later in this section . the natural question that now arises from the above is : is it possible for a link field that is emitted by a quark(antiquark ) through the @xmath42 coupling to be absorbed by an antiquark(quark ) through the @xmath41 term , in other words , ` mixed ' hopping ? the answer to this is ` no ' . to understand the hopping interactions better one needs to analyze the rotational properties of the hamiltonian more closely . for @xmath43 , the @xmath5 lattice hamiltonian is invariant under rotations around the @xmath44 by multiples of @xmath45 , giving rise to conservation of total angular momentum modulo @xmath46 . since @xmath47 in the 2 particle sector and @xmath48 can only assume the values @xmath49 , @xmath50 , and @xmath51 , @xmath48 in the 2 particle sector is conserved . since mixed hopping would change @xmath48 by one unit , this means that the sum of all contributions from mixed hopping must add up to zero . the observation that helicity conservation ( modulo 4 ) for @xmath26 prohibits interference between the two hopping terms was already made in ref . the interaction of the 2-particle sector with the 3-particle sector will therefore have four different possibilities : * a quark(or an antiquark ) in the 2-sector emits a link through the @xmath41 term and flips it s spin . * a quark(or an antiquark ) in the 2-sector emits a link through the @xmath42 term and does not flip it s spin . of course , the interaction to go from the 3-sector to the 2-sector has the opposite effect , i.e , * a quark(or an antiquark ) in the 3-sector absorbs a link through the @xmath41 term and flips it s spin . * a quark(or an antiquark ) in the 3-sector absorbs a link through the @xmath42 term and does not flip it s spin . in order to understand the spin dynamics due to the hopping terms as described above we present below the interaction hamiltonian in all its glory . spin flip hopping in the @xmath52-direction : h_x^flip&= & + & -&i(- ) _ _ spin flip hopping in the @xmath53-direction : h_y^flip&= & + & -&(- ) _ _ hopping without spin flip : h^no - flip&= & + & -&(+ ) _ _ implicit in the above expressions are link field creation and annihilation operators on the link connecting the initial and final state . the transition matrix elements are obtained by sandwiching the above interactions between the 2-particle and the 3-particle states as written in eqs . ( [ eq:2.2 ] ) and ( [ eq:2.3 ] ) and are provided in the table in appendix b. the schematic structure of the hamiltonian therefore looks like the following : h= + . [ eq : ham ] where @xmath54 and @xmath55 involve all the transition matrix elements . in order to evaluate all the matrix elements , we expanded the wavefunction amplitudes using a basis of continuous functions as follows : ( x)=_k=0^kx^+k(1-x)^(x , y)=_k=0^_l=0 ^ -ka_klx^+ky^+ l(1-x - y)^+ where the @xmath56 in the eq . 2.10 and that in eq . 2.11 need not be equal . in the table ( see appendix b ) , @xmath57(@xmath58 ) are the overlap integrals when a quark(antiquark ) in the 2-sector goes to the 3-sector by emitting a link field through the @xmath42 hopping term ( the spin non - flip term ) and @xmath59(@xmath60 ) are the overlap integrals when a quark(antiquark ) in the 2-sector goes to the 3-sector by emitting a link field through the @xmath41 hopping term ( the spin - flip term ) . see appendix a for more details of these overlap integrals . a factor @xmath61 was introduced in the integrand of the transition matrix elements to regularize the infrared ( small @xmath62 ) divergence that arises from the self energy of quarks in the case when the quark that emits a link field also absorbs it . of course , all physical observables will have to be independent of the value of @xmath63 as will be demonstrated in section iv . the result of all the above is the following matrix equation for each @xmath64 and @xmath65 : h&&= + & & + .[eq : ham2 ] where @xmath66 and @xmath67 are the norm matrices , @xmath68 and @xmath69 are the matrix elements in the pure 2-particle and 3-particle sectors , respectively and finally @xmath70 and @xmath71 are the transition matrix elements . the above matrix equation is then converted into an eigenvalue equation that yields eigenvectors in terms of the @xmath72 s and @xmath73 s . another significant result of this approximation that is worth mentioning at this point is that unlike the low lying mesons , the higher excited states are not expected to show the correct dispersion . the reason for this is that since the higher states are very excited , the quark and the antiquark will be physically farther apart . this would mean that these mesons will mostly or always remain in the @xmath74 ( where @xmath20 is the link field ) state ( see fig . [ fig : prob ] ) and for it to propagate , it will need to go to a @xmath75 state or states with larger number of link fields . but since our approximation does not allow states with more than one link field , these excited states simply can not propagate . in order to gain a qualitative understanding about the interplay between different parameters in the @xmath5 lattice hamiltonian , it is instructive to study a simple model , where one treats the admixture of the 3-particle fock component to the @xmath28 and @xmath76 as a perturbation . to @xmath77 order , i.e. when the coupling between 2 and 3 particle fock component is turned off , there is no spin dependence of the interactions and the @xmath28 and @xmath76 are degenerate . likewise , there is no ` hopping ' ( i.e. @xmath5 propagation ) of mesons and thus energies are independent of @xmath78 giving rise to an infinite @xmath5 lattice spacing ( in physical units ) . in the next order we treat the coupling between 2 and 3 particle fock components as a perturbation ( note that interactions which are diagonal in the particle number , such as the confining interaction in the longitudinal direction are still treated non - perturbatively ) . as described in section ii , there are two interactions that mix fock sectors : hopping due to the @xmath42-term ( without helicity flip ) and hopping due to the vector coupling @xmath41 ( with helicity flip ) . starting from a basis of t hooft eigenstates which are plane waves in the @xmath5 direction and where the @xmath2 in the 2 particle fock component carry spins @xmath79 , @xmath80 , @xmath81 , and @xmath82 respectively , one thus finds for the energy in second order perturbation theory to be : h = m_0 ^ 2 & - & m_1,r^2 ( cccc + & 0 & 0 & 0 + 0 & + & 0 & 0 + 0 & 0 & + & 0 + 0 & 0 & 0 & + ) + & + & m_1,v^2 ( cccc 0 & 0 & 0 & - + 0 & 0 & + & 0 + 0 & + & 0 & 0 + - & 0 & 0 & 0 ) . [ eq : m2 ] + here @xmath83 and @xmath84 are some second order perturbation theory expressions involving matrix elements between 2 and 3 particle states that are eigenstates of the diagonal parts of the hamiltonian ( kinetic + coulomb ) , and @xmath85 . several general and important features can be read off from this result . first of all , and most importantly , eq . ( [ eq : m2 ] ) shows that the r - term gives rise to a dispersion relation with the same @xmath5 speed of light for the @xmath28 and the @xmath76 s , while the vector interaction breaks that symmetry . this observation already indicates that it may be desirable to keep the @xmath86-term much larger than the spin - flip term . we will elaborate on this point below . at @xmath87 , the eigenstates of the above hamiltonian are the @xmath88 i.e. @xmath89 and @xmath90 , with @xmath91 , the @xmath92 and the @xmath93 for nonzero @xmath5 momenta , there will in general be mixing among the @xmath94 and the @xmath95 , but not among the other states since helicity in the 2-particle fock sector is still conserved modulo 2 . expanding around @xmath43 , and denoting @xmath96 one finds to @xmath97 the following eigenstates and eigenvalues . c|c|c & m^2(0 ) & m^2(*k*_^2)-m^2(0 ) + & & + - & |m^2 -m_1,v^2 & m_1,r^2+m_1,v^2 + & & + + & |m^2+m_1,v^2 & m_1,r^2-m_1,v^2 + & & + - & |m^2 & m_1,r^2 -m_1,v^2 + & & + + & |m^2 & m_1,r^2 + m_1,v^2 [ eq : disp ] . ( [ eq : disp ] ) illustrates a fundamental dilemma that hampers any attempt to fully restore lorentz invariance within the femtoworm approximation : @xmath84 not only governs the splitting between the @xmath28 and the ( h=0 ) @xmath76 but is also responsible for violations of lorentz invariance among the different helicity states : if one determines the @xmath5 lattice spacing in physical units for each meson separately , by demanding that the @xmath5 speed of light equals @xmath51 , one finds for example . |_&= & m_1,r^2 + m_1,v^2 + . |__0 & = & m_1,r^2 - m_1,v^2 i.e. increasing the @xmath27 splitting is typically accompanied by an increase in lorentz invariance violation . |_= m__0 ^ 2 - m_^2 . for the @xmath88 the breaking is of a similar scale , plus one also observes an anisotropy in the dispersion relation on the same scale . therefore , in order to avoid a large breaking of lorentz invariance , it will be necessary that m_r,1 ^ 2 m__0 ^ 2 - m_^2 . if one keeps the @xmath27 splitting fixed at its physical value then there are two ways to achieve this condition . one possibility is to simply increase the yukawa coupling that appears in the @xmath86-term . this increase of the @xmath86-term tends to decrease the @xmath5 lattice spacing for both @xmath28 and @xmath76 s ( see fig . [ fig : m_r ] ) and in order to achieve satisfactory lorentz invariance ( in the sense of uniform @xmath5 lattice spacings ) one needs to make the lattice spacing smaller than the compton wavelength of the @xmath76 meson . however , one can not make the @xmath86-term coupling arbitrarily large because at some point there occurs an instability ( tachyonic @xmath98 ! ) . such instabilities for large coupling are common in the lf formulation of models with yukawa coupling and might be related to a phase transition ( similar to the phase transition in @xmath99 theory that occurs as the coupling is increased ) . fortunately , there exists another possibility to make these matrix elements large , without increasing the yukawa couplings . this derives from the fact that the hopping interactions are proportional to @xmath100 , where @xmath52 ( @xmath101 ) are the momenta of the active quark before(after ) the hopping . because of the singularity as @xmath52 , @xmath102 , matrix elements of the hopping terms are greatly enhanced if the unperturbed wave functions are large near @xmath103 and @xmath104 . since the unperturbed wave functions in the 2 particle fock component vanish like @xmath105 near @xmath103 , where @xmath106 , matrix elements of the hopping interaction become very large when one makes @xmath107 very large . therefore , the larger one chooses @xmath108 , the more one restores lorentz invariance of the @xmath28 and @xmath76 dispersion relations because one can keep the @xmath28-@xmath76 splitting fixed while decreasing the coupling of the spin flip interaction . at the same time , keeping the @xmath86-term interaction fixed one increases the dominance of the @xmath86-term contribution in @xmath109 and thus not only reduces the lattice spacing in physical units , but also obtains dispersion relations for the @xmath28 and the @xmath76 s that look more and more similar as demanded by lorentz invariance . unfortunately , we are not completely free to pick whatever value of @xmath110 we like because @xmath111 and @xmath112 are largely fixed by the center of mass in the @xmath28-@xmath76 system as well as by fitting the physical string tension in the pure glue sector . the above analysis may also shed some light on the shape of the pion distribution amplitude in ref . @xcite , which is much more flat ( actually , it even exhibits a slight double - hump shape ) compared to our results ( see next section ) . the reason is that the numerical value for @xmath112 in ref . @xcite was _ not _ taken from the pure glue calculation , but was rather used as a free parameter , which was also varied in order to maximize lorentz covariance . although such a _ first principle _ approach would normally be desirable , minimizing lorentz covariance violations for the @xmath28 and @xmath76 dispersion relations drives @xmath113 . when the longitudinal gauge coupling @xmath112 is very large compared to other scales in the problem , the wavefunction tends to become very flat a phenomenon familiar from the t hooft equation . of course , since a dlcq cutoff was employed in ref . @xcite , @xmath112 never really become infinite , but through the extrapolation procedure the near flat shape for the pion distribution amplitude emerged . in a more recent paper @xcite , where dalley adopted a procedure similar to ours and @xmath112 was kept fixed , a pion distribution amplitude is obtained that is much more consistent with ours . one can not expect this hamiltonian to describe the higher excited states because of the truncation of the fock space upto the 3-particle sector . this is demonstrated in fig . [ fig : prob ] where we show typical results for the probability to find each state to be in the 3-particle sector . one finds that whereas the low - lying states have a fairly large probability to be in the 2-particle sector , it is quite the opposite for the higher states . we therefore expect that excited states are much more strongly affected by the ommission of 4 and more particle states and consider in the following only the @xmath28 and @xmath76 states in our study of lorentz covariance . the parameters in the hamiltonian are : * longitudinal gauge coupling * @xmath86-term coupling ( hopping without spin flip ) * spin - flip coupling ( hopping with spin flip ) * kinetic mass ( 2 particle sector ) * kinetic mass ( 3 particle sector ) . the observables that we studied showed little dependence on this parameter , so we kept it fixed at a constituent mass . * link field mass . similar to kinetic mass in 3 particle sector . furthermore , demanding lorentz invariance in the pure glue sector , one finds a renormalized trajectory . this makes sense since the @xmath5 lattice scale is unphysical . we keep the link field mass fixed at a value which yields relatively small @xmath5 lattice spacing . in the spirit of the color - dielectric approach @xcite , we would like to use only chiral symmetry ( small pion mass ) , lorentz invariance and one length scale ( e.g. physical @xmath76 mass or string tension ) as the only input into our calculations . however , as explained in the previous section , if one uses as input only the @xmath28 and @xmath76 masses , demanding exact lorentz invariance within the framework of the femtoworm approximation would drive @xmath114 to infinity and would thus give rise to string tensions that are inconsistent with phenomenology . likewise , as input only the @xmath28 mass together with lorentz invariance would give rise to a degenerate @xmath28-@xmath76 system . it is thus clear that , as long as one does not go beyond a truncation of the fock space above 3 particles ( @xmath3 , @xmath4 and one link quantum ) , a first principle calculation ( where only one mass scale is used as input ) will not be possible . in the following , we will therefore use two phenomenological scales as input parameters : the physical value of the @xmath76 mass , as well as the physical string tension . although the dependence of physical parameters on the input parameters is in general rather complex and non - perturbative , one can understand at a qualitative level how the input parameters influence the relevant physical scales : the physical string tension in the longitudinal direction determines @xmath112 in physical units . modulo mass renormalization due to the yukawa couplings , the quark mass in the 2-particle fock component and the gauge coupling are strongly constrained by fitting the physical string tension ( which determines @xmath112 and the center of mass of the @xmath28-@xmath76 system . this leaves us with only the @xmath86-term and the helicity flip hopping term couplings as parameters to vary . the helicity flip term is not only responsible for @xmath28-@xmath76 splitting within our approximations but also for violations of lorentz invariance ( different @xmath5 lattice spacings in physical units for different mesons ) . the observables that we studied , including lorentz invariance , show rather little sensitivity to the precise values of the 3-particle sector quark masses ( fig . [ fig:3sector ] ) , which we thus keep at a value corresponding to a constituent mass ( about half the @xmath76 mass ) . this leaves us with 4 free parameters . using a value of @xmath115 @xmath116 in physical units , as determined from the string tension in ref . @xcite ( which is larger than the one previously used @xcite ) , we were able to produce the physical @xmath28-@xmath76 splitting with only a relatively small spin - flip coupling . this allows us to chose the @xmath86-term large enough so that @xmath86-term hopping dominates over spin - flip hopping and therefore violations of lorentz invariance ( as measured by comparing quadratic terms in the dispersion relation of mesons in the @xmath28-@xmath76 sector ) are only on the order of @xmath117 . the average @xmath5 lattice spacing ( from the dependence of the energy on @xmath118 ) is found to be @xmath119 . for the @xmath28 wave function , we find a shape ( see fig . [ fig : pion ] ) that is very close to the asymptotic shape @xmath120 . this is surprising if one considers that the lattice spacing is still relatively large and hence the momentum scale is still very low . the @xmath76 meson distribution functions are shown in figs . [ fig : rho0 ] and [ fig : rho1 ] . notice that the momentum dependence of the @xmath28 dispersion relation is stronger than the one for the @xmath76 and for the parameters used in our calculations this even leads to a level crossing and @xmath76 have opposite c - parity . ] the reason for this behavior is very simple . as pointed out in ref . @xcite , for @xmath26 spin - flip hopping is attractive for the @xmath28 and repulsive for the @xmath121 , while the opposite is true at the end of the brillouin zone . more precisely , for @xmath122 , the hopping term in the @xmath52 direction is attractive for the @xmath28 and repulsive for the @xmath123 with the opposite being true for @xmath124 . an analogous statement holds for the @xmath53-direction . if there were no @xmath86-term present , the dispersion would arise solely from the spin flip term and the @xmath121 in the corner of the brillouin zone would be degenerate with the @xmath28 at @xmath26 a phenomenon which is nothing but a manifestation of species doubling at the hadronic level . of course , since we have included an @xmath86-term , the @xmath123 in the corner of the brillouin zone is no longer degenerate with the @xmath28 at @xmath26 , but the level crossing remains ( see fig . [ fig : bzonep ] ) . similarly one can understand the degeneracy of the @xmath28 and @xmath123 at the edge of the brillouin zone @xmath125 . at that point , the x - hopping term is repulsive for the @xmath28 and attractive for the @xmath123 . the opposite is true for the @xmath53-hopping term and the two contributions cancel one another leaving the @xmath28 and the @xmath123 degenerate . from the normalization of the @xmath28 distribution function we were able to extract a numerical estimate for the @xmath28 decay constant @xmath126 in terms of the @xmath127 helicity component of the @xmath28 wavefunction in the two particle fock component @xmath128 f_= _ 0 ^ 1 dx _ + -(x ) 300 mev , [ eq : fpi ] which is much larger than the experimental value @xmath129 . we expect that including higher fock components will reduce the size of the two particle fock component and hence also @xmath130 . the parton distribution function for the @xmath28 is shown in fig . [ fig : pdf ] . the fact that @xmath131 goes to zero at small @xmath52 is an artifact of our @xmath63-cutoff and we expect a nonzero value for @xmath132 in the femtoworm approximation as @xmath133 . finally , we also calculated the electro - magnetic form factor for the @xmath28 for purely transverse momentum transfers ( where overlap formulas can be used ) and we find an rms radius for the @xmath28 of @xmath134 . obviously , since our lattice spacing is only about @xmath135 and since we do not let the quark - antiquark separation to exceed one lattice unit , one should not expect very good results for this observable . since the rms - radius is already poor , displaying the form factor is not very meaningful . as mentioned in section ii , physical observables should be independent of the infrared cut - off , @xmath63 ( as long as the cutoff is small ) . we verified this cut - off independence for a variety of observables . as an example , we demonstrate in fig . [ fig : epsilon ] the cutoff independence of the dispersion relations for light mesons . the dashed line is for @xmath63 = 0.05 and the continuous line is for @xmath63=0.1 . we feel it is worthwhile to compare our results with those in a concurrent paper @xcite that appeared as we were completing the manuscript for this paper . there are a few noteworthy technical differences between the two . firstly , the treatment of the longitudinal dynamics are different ( dlcq in ref . @xcite while we employed continuous basis functions ) . secondly , the pure glue string tension are different ( @xmath136 vs. @xmath137 ) otherwise , both this work and ref . @xcite were based on the formalism outlined in ref . @xcite . we used continuous basis functions because there are some unresolved issues regarding the convergence of dlcq calculations in terms of the dlcq resolution @xcite . however , because the numerical approximations in both works are very crude , these issues are not yet very important and therefore the main difference in this context is more in calculational details . because both works used the femtoworm approximation , the rms radius of the @xmath28 is rather low compared to the experimental value . the lattice spacing , which defines the maximal @xmath2-separation , is slightly larger in ref . consequently , the resulting rms radius in ref . @xcite , though still much smaller than the experimental result , is slightly larger than ours . here we should also point out that the algebraic expression used in ref . @xcite for calculating @xmath130 is incorrect . with our normalization , the result for @xmath130 in ref . @xcite increases by a factor of three @xcite and becomes much more similar to ours . since both works determine the parameters in the effective theory by imposing lorentz symmetry as a nonperturbative renormalization condition , we feel that it is important to have a handle on the violation of lorentz invariance ( or lack thereof ) within the femtoworm approximation . that this violation is inevitable was first pointed out in ref . @xcite ( in which some results from this work were already presented ) and has been further reiterated in section iii of this paper . since it is impossible to achieve both lorentz invariance and the correct @xmath27 splitting simultaneously , one , therefore , needs to have a quantitative idea of the degree to which lorentz invariance can be achieved while simlutaneously yielding the correct @xmath27 splitting . although ref . @xcite provides several @xmath138 plots , it is not quantitatively clear from these plots how large the violations of lorentz invariance are for the @xmath27 dispersion relations at the minimum of @xmath138 . it is therefore not possible for us to compare the two works with resepect to the degree to which lorentz symmetry is violated . here we should also point out an interesting connection between this work and ref . @xcite on the one hand and ref . @xcite on the other : although the spirit of ref . @xcite was more phenomenological , the final results ( parton distribution functions , distribution amplitudes ) obtained in ref . @xcite were very similar to both ref . @xcite and this work . another difference between our results and those in ref . @xcite is that our @xmath28 distribution amplitude is closer to the ` asymptotic ' shape than the one in ref . @xcite a behavior which is also reflected in the parton distribution function . however , this difference is not really significant and it may just be due to the differences in the handling of the longitudinal dynamics which may favor slightly different sets of parameters . we presented results from a study of light mesons in qcd formulated on a transverse lattice using light - front quantization . in transverse lattice qcd the gauge field in the ( continuous ) longitudinal directions is described by a non - compact vector potential , and in the ( discrete ) transverse directions by a non - compact link field . our approach was based on the color - dielectric formulation @xcite , where one assumes that the fields have been `` smeared '' over a finite area and thus are no longer subject to a strict @xmath139 constraint . the smeared degrees of freedom are subject to effective interactions which reflect the dynamics of the underlying microscopic degrees of freedom . although this effective interaction is not known _ a priori _ , it is highly constrained because it must still be gauge invariant and it should still give rise to lorentz invariant observables . in this work we investigated transverse lattice qcd with fermions . the starting point was an ansatz for the effective interactions which contained interaction terms up to fourth order in the fields which are still invariant under those symmetries not broken by the transverse lattice . the goal of this work was to determine the coefficients in this ansatz by demanding enhanced lorentz symmetry from physical observables . as input parameters we used the physical masses of the @xmath28 and the @xmath76 as well as the string tension . what we found is that the splitting between the @xmath28 and @xmath76 mesons strongly constrains the quark helicity flip interaction term in the lf - hamiltonian . meson propagation in the @xmath5 direction involves helicity flip and nonflip hopping terms for the quarks . once the nonflip hopping term was constrained by fitting the observed @xmath27 splitting , the only remaining parameter on which the physical lattice spacing ( as determined from the meson dispersion relation ) depends very strongly is the helicity nonflip term . even though we scanned the parameter space thoroughly , we were only able to find matching values for @xmath140 for the @xmath28 and @xmath76 mesons within about a 10% - 20% error . similar lorentz invariance violations were found in the splitting between @xmath76 mesons of helicity zero and one . we believe that both these errors are due to the truncation of the fock space ( which may contain an infinite number of particles ) as well as the ansatz for the effective interaction ( which may contain an infinite number of terms ) . in fact , given these rather crude approximations , it is actually remarkable that the violations of lorentz invariance in the light meson sector are only about 10 - 20% . the model seems very promising though it displays some basic limitations as far as restoration of full lorentz invariance is concerned . this problem was first identified in one of our earlier papers @xcite but subsequent new understanding @xcite about the couplings helped isolate the problem and reduce it significantly for the low - lying mesons . for the higher mesons , the femtoworm approximation is not expected to work unless more link field degrees of freedom are introduced , a task that is possible but computationally non - trivial . while we were completing this work , similar studies of mesons on the transverse lattice appeared @xcite , which were based on dlcq rather than basis function techniques for handling the longitudinal dynamics . however , there is another important difference between our work and ref . @xcite : we took the numerical value for the longitudinal gauge coupling @xmath141 that was obtained in pure glue studies on the @xmath5 lattice @xcite , while ref . @xcite allowed @xmath141 to ` float ' in addition to the other parameters in order to restore lorentz symmetry . as we explained in section iii , increasing @xmath141 , as compared to other dimensionful parameters , tends to decrease violations of lorentz symmetry in the @xmath142 dispersion relations . therefore , the results presented in ref . @xcite where extrapolated from calculations performed with values of @xmath141 that were much larger than ours ( and larger than @xmath141 used in the pure glue calculation in ref . this difference in _ procedure _ explains to some extent the difference in results between this work and refs . @xcite : most importantly , if one increases @xmath141 in physical units , while keeping the physical pion mass fixed , tends to decrease all other dimensionful parameters . the pion distribution amplitudes presented in ref . @xcite , where @xmath141 was allowed to float completely freely , although slightly ` double humped ' , is actually numerically consistent with an almost completely flat distribution since it was obtained by extrapolation in the dlcq parameter and there is some uncertainty in this procedure , and this is consistent with the above observation regarding @xmath141 . in ref . @xcite , @xmath141 was no longer allowed to float and the results for the distribution amplitude are much more similar to ours . * acknowledgments * m.b . would like to thank s. dalley for many interesting and clarifying discussions about the @xmath5 lattice . this work was supported by a grant from doe ( fg03 - 95er40965 ) and through jefferson lab by contract de - ac05 - 84er40150 under which the southeastern universities research association ( sura ) operates the thomas jefferson national accelerator facility . to understand the overlap integrals , let us look at two specific elements of the table ( see appendix b ) , namely , the first two elements of the first row . for the first element , clearly there is no spin flip and the hopping is along the @xmath53-direction . there are two possibilities through which this transition can take place . pictorially , they are as shown in fig . [ fig : non - flip ] . notice that both diagrams yield the same 3-sector lattice configuration which is shown to the right . these two diagrams correspond to the terms containing @xmath143 and @xmath144 in the the spin non - flip interaction hamiltonian . when these two terms are sandwiched between @xmath145 and @xmath146 from eqns . 2.2 and 2.3 , one obtains the first matrix element of the table . in particular , @xmath57 and @xmath58 are obtained when the integrals over the quark and the antiquark momenta are carried out in the above matrix element as follows : f_q^r = dx dy ^*(x , y ) & & ( + ) + & & ( 1-y ) f_|q^r = dx dy ^*(x , y ) & & ( + ) + & & ( x ) similarly , let us look at the second element of the first row of the table . clearly , this is a spin flip interaction and the hopping takes place in the @xmath53-direction . pictorially , this is shown in fig . [ fig : flip ] . this diagram corresponds to the term in @xmath147 containing @xmath148 . the equivalent 3-sector lattice configuration is once again shown to the right . the corresponding matrix element is therefore obtained by sandwiching this term of the hamiltonian between @xmath145 and @xmath146 and in particular @xmath60 is obtained when the integrals over the quark and the antiquark momenta are carried out in the above matrix element as follows : f_q = dx dy ^*(x , y ) & & ( - ) + & & ( 1-y ) f_|q = dx dy ^*(x , y ) & & ( - ) + & & ( x ) . @xmath149 m. burkardt , advances nucl . phys . * 23*,1 ( 1996 ) . k. wilson et al . , phys . rev . d*49 * , 6720 ( 1994 ) . bardeen , r.b . pearson , and e. rabinovici , phys . * , 1037 ( 1980 ) m. burkardt , invited talk given at `` 11th international light - cone school and workshop '' , eds . c. ji and d .- p . min , hep - th/9908195 . p. griffin , proc . to _ theory of hadrons and light - front qcd _ , ed . s. glazek , world scientific , singapore , 1994 . m. burkardt and b. klindworth , phys . d*55 * , 1001 ( 1997 ) s. dalley and b. vande sande , phys . rev . d*59 * , ( 1999 ) ; phys . lett.*82 * , 1088 ( 1999 ) ; nucl . suppl.)*83 * , 116 ( 2000 ) ; phys . d*62*,(2000 ) m. burkardt and h. el - khozondar , phys . rev . d*60 * , ( 1999 ) s. dalley , nucl . suppl.)*90 * , 227 ( 2000 ) s.k . seal and m. burkardt , b ( proc . suppl . ) 90 , 233(2000 ) s. dalley , hep - ph/0101318 . m. burkardt and s. seal , hep - ph/0101338 . m. burkardt , phys . rev . d * 57 * , 1136 ( 1998 ) . m. burkardt and s. seal , invited talk at 5th international workshop on particle physics phenomenology , chi - pen , taitung , taiwan , 8 - 11 nov 2000 ; hep - ph/0101338 . s. dalley , private communications .

we present results from a study of light meson spectra and structure obtained within the framework of light - front qcd formulated on a transverse lattice . we discuss how imposing lorentz covariance conditions on meson dispersion relations allows determination of parameters in the transverse lattice hamiltonian . the pion distribution amplitude obtained in this framework is rather close to its asymptotic shape . #

any quantum theory of gravity will have to propose a prescription to deal with the unavoidable @xcite nonlinearities that plague ctcs . this requires some sort of modification of the dynamical equations of motions of quantum mechanics that are always linear . deutsch in his seminal paper @xcite proposed one such prescription , based on a self - consistency condition referred to the state of the systems inside the ctc . deutsch s theory has recently been critiqued by several authors as exhibiting self - contradictory features @xcite . by contrast , although any quantum theory of time travel quantum mechanics is likely to yield strange and counter - intuitive results , p - ctcs appear to be less pathological @xcite . they are based on a different self - consistent condition that states that self - contradictory events do not happen ( novikov principle @xcite ) . pegg points out that this can arise because of destructive interference of self - contradictory histories @xcite . here we further compare deutsch s and post - selected closed timelike curves , and give an in - depth analysis of the latter , showing how they can be naturally obtained in the path - integral formulation of quantum theory and deriving the equations of motions that describe the interactions with ctcs . as noted , in addition to general - relativistic ctcs , our proposed theory can also be seen as a theoretical elaboration of wheeler s assertion to feynman that ` an electron is a positron moving backward in time ' @xcite . in particular , any quantum theory which allows the nonlinear process of postselection supports time travel even in the absence of general - relativistic closed timelike curves . the mechanism of p - ctcs @xcite can be summarized by saying that they behave exactly as if the initial state of the system in the p - ctc were in a maximal entangled state ( entangled with an external purification space ) and the final state were post - selected to be in the same entangled state . when the probability amplitude for the transition between these two states is null , we postulate that the related event does not happen ( so that the novikov principle @xcite is enforced ) . by contrast , deutsch s ctcs are based on imposing the consistency condition @xmath4 , \labell{conscon}\;\end{aligned}\ ] ] where @xmath3 is the state of the system inside the closed timelike curve , @xmath5 is the state of the system outside ( i.e. of the chronology - respecting part of spacetime ) , @xmath6 is the unitary transformation that is responsible for eventual interactions among the two systems , and where the trace is performed over the chronology - respecting system . the existence of a state @xmath7 that satisfies is ensured by the fact that any completely - positive map of the form @xmath8=\mbox{tr}_a[u(\rho\otimes\rho_a)u^\dag]$ ] always has at least one fixed point @xmath7 ( or , equivalently , one eigenvector @xmath7 with eigenvalue one ) . if more than one state @xmath3 satisfies the consistency condition , deutsch separately postulates a `` maximum entropy rule '' , requesting that the maximum entropy one must be chosen . note that deutsch s formulation assumes that the state exiting the ctc in the past is completely uncorrelated with the chronology - preserving variables at that time : the time - traveler s ` memories ' of events in the future are no longer valid . = .55 the primary conceptual difference between deutsch s ctcs and p - ctcs lies in the self - consistency condition imposed . consider a measurement that can be made either on the state of the system as it enters the ctc , or on the state as it emerges from the ctc . deutsch demands that these two measurements yield the same statistics for the ctc state alone : that is , the density matrix of the system as it enters the ctc is the same as the density matrix of the system as it exits the ctc . by contrast , we demand that these two measurements yield the same statistics for the ctc state _ together with its correlations with any chronology preserving variables_. it is this demand that closed timelike curves respect both statistics for the time - traveling state together with its correlations with other variables that distinguishes p - ctcs from deutsch s ctcs . the fact that p - ctcs respect correlations effectively enforces the novikov principle @xcite , and , as will be seen below , makes p - ctcs consistent with path - integral approaches to ctcs . the connection between p - ctcs and teleportation @xcite is illustrated ( see fig . [ f : teleport ] ) with the following simple example that employs qubits ( extensions to higher dimensional systems are straightforward ) . suppose that the initial bell state is @xmath9 ( but any maximally entangled bell state will equivalently work ) , and suppose that the initial state of the system entering the ctc is @xmath10 . then the joint state of the three systems ( system 1 entering the ctc , system 2 emerging from the ctc , and system 3 , its purification ) is given by @xmath11 . these three systems are denoted by the three vertical lines of fig . [ f : teleport]b . it is immediate to see that this state can be also written as @xmath12 where @xmath13 and @xmath14 are the four states in a bell basis for qubit systems and @xmath15s are the three pauli matrices . is equivalent to eq . ( 5 ) of ref . @xcite , where the extension to higher dimensional systems is presented ( the extension to infinite dimensional systems is presented in @xcite ) . it is immediate to see that , if the system 1 entering the ctc together with the purification system 3 are post - selected to be in the same bell state @xmath16 as the initial one , then only the first term of eq . survives . apart from an inconsequential minus sign , this implies that the system 2 emerging from the ctc is in the state @xmath17 , which is exactly the same state of the system that has entered ( rather , will enter ) the ctc . it seems that , based on what is currently known on these two approaches , we can not conclusively choose p - ctcs over deutsch s , or _ vice versa_. both arise from reasonable physical assumptions and both are consistent with different approaches to reconciling quantum mechanics with closed timelike curves in general relativity . a final decision on which of the two is `` actually the case '' may have to be postponed to when a full quantum theory of gravity is derived ( which would allow to calculate from first principles what happens in a ctc ) or when a ctc is discovered that can be tested experimentally . however , because of the huge recent interest on ctcs in physics and in computer science ( e.g. see @xcite ) , it is important to point out that there are reasonable alternatives to the leading theory in the field . we also point out that our post - selection based description of ctcs seems to be less pathological than deutsch s : for example p - ctcs have less computational power and do not require to separately postulate a `` maximum entropy rule '' @xcite . therefore , they are in some sense preferable , at least from an occam s razor perspective . independent of such questions of aesthetic preference , as we will now show , p - ctcs are consistent with previous path integral formulations of closed timelike curves , whereas deutsch s ctcs are not . path integrals @xcite allow one to calculate the transition amplitude for going from an initial state @xmath18 to a final state @xmath19 as an integral over paths of the action , i.e. @xmath20,\mbox { where } s=\int_0^{\tau } dtl(x,\dot x ) \labell{feyn}\;,\end{aligned}\ ] ] and where @xmath21 is the lagrangian and @xmath22 is the action , @xmath23 and @xmath24 are the position representations of @xmath18 and @xmath19 respectively ( i.e. @xmath25 ) , and the paths in the integration over paths all start in @xmath0 and end in @xmath26 . of course in this form it is suited only to describing the dynamics of a particle in space ( or a collection of particles ) . it will be extended to other systems in the next section . in order to add a ctc , we first divide the spacetime into two parts , @xmath27 \labell{feyn2}\;,\end{aligned}\ ] ] the `` conventional '' strategy to deal with ctcs using path integrals is to send the system @xmath28 to a prior time unchanged ( i.e. with the same values of @xmath29 ) , while the other system ( the chronology - respecting one ) evolves normally . this is enforced by imposing periodic boundary conditions on the ctc boundaries . namely , the probability amplitude for the chronology - respecting system is @xmath30 \labell{feyn3}\;,\end{aligned}\ ] ] where the @xmath31-function ensures that the initial and final boundary conditions in the ctc system are the same . note that we have removed @xmath32 and @xmath33 , but we are coherently adding all possible initial and final conditions ( through the @xmath34 and @xmath35 integrals ) . this implies that it is not possible to assign a definite state to the system inside a ctc : all possible states of the system ( except possibly forbidden ones ) are compatible with such boundary conditions . note also that the boundary conditions of eq . have previously appeared in the literature ( e.g. see @xcite and , in the classical context , in the seminal paper @xcite ) . 1to show that eq . is the same formula that one obtains using post - selected teleportation , we have to calculate @xmath36 , where @xmath37 is a maximally entangled epr state @xcite and where the hamiltonian acts only on the system and on the first of the two hilbert spaces of @xmath38 . use eq . for the system and for the first hilbert space of @xmath38 to obtain @xmath39 = \int^{\infty}_{-\infty } dx dx^{\prime } dy dy^{\prime } i ( x ) f^{\ast } ( y)\ ; \delta(x'-y ' ) \int^{y , y^{\prime}}_{x , x^{\prime } } { \cal d } x(t ) \exp \left [ \frac{i}{\hbar } s \right ] \labell{tp}\;,\end{aligned}\ ] ] where we have used the position representation @xmath40 . is clearly equal to eq . since @xmath41 . note that this result is independent of the particular form of the epr state @xmath38 as long as it is maximally entangled in position ( and hence in momentum ) . all the above discussion holds for initial and final pure states . however , the extension to mixed states in the path - integral formulation is straightforward : one only needs to employ appropriate purification spaces @xcite . the formulas then reduce to the previous ones . here we briefly comment on the two - state vector formalism of quantum mechanics @xcite . it is based on post - selection of the final state and on renormalizing the resulting transition amplitudes : it is a time - symmetrical formulation of quantum mechanics in which not only the initial state , but also the final state is specified . as such , it shares many properties with our post - selection based treatment of ctcs . in particular , in both theories it is impossible to assign a definite quantum state at each time : in the two - state formalism the unitary evolution forward in time from the initial state might give a different mid - time state with respect to the unitary evolution backward in time from the final state . analogously , in a p - ctc , it is impossible to assign a definite state to the ctc system at any time , given the cyclicity of time there . this is evident , for example , from eq . : in the ctc system no state is assigned , only periodic boundary conditions . another aspect that the two - state formalism and p - ctcs share is the nonlinear renormalization of the states and probabilities . in both cases this arises because of the post - selection . in addition to the two - state formalism , our approach can also be related to weak values @xcite , since we might be performing measurements between when the system emerges from the ctc and when it re - enters it . considerations analogous to the ones presented above apply . it would be a mistake , however , to think that the theory of post - selected closed timelike curves in some sense requires or even singles out the weak value theory . although the two are compatible with each other , the theory of p - ctcs is essentially a ` free - standing ' theory that does not give preference to one interpretation of quantum mechanics over another . the formula was derived in the path - integral formulation of quantum mechanics , but it can be easily extended to generic quantum evolution . we start by recalling the usual kraus decomposition of a generic quantum evolution ( that can describe the evolution of both isolated and open systems ) . it is given by @xmath42=\mbox{tr}_e[u(\rho\otimes|e\rangle\langle e|)u^\dag]= \sum_i\langle i|u|e\rangle\:\rho\:\langle e|u^\dag|i\rangle\nonumber\\ & & = \sum_ib_i\rho b_i^\dag \labell{cp}\;,\end{aligned}\ ] ] where @xmath43 is the initial state of the environment ( or , equivalently , of a putative abstract purification space ) , @xmath6 is the unitary operator governing the interaction between system initially in the state @xmath7 and environment , and @xmath44 is the kraus operator . in contrast , the nonlinear evolution of our post - selected teleportation scheme is given by @xmath45=\mbox{tr}_{ee ' } \big[(u\otimes\openone_{e ' } ) \left(\rho\otimes|\psi\rangle\langle\psi|\right ) \left(u^\dag\otimes\openone_{e'}\right)\nonumber\\&&\times \left(\openone\otimes|\psi\rangle\langle\psi|\right)\big]= \sum_{l , j}\langle l|u|l\rangle\:\rho\:\langle j|u^\dag|j\rangle= c\rho c^\dag \labell{nl},\end{aligned}\ ] ] where @xmath46tr@xmath47 $ ] and @xmath48 ( or any other maximally entangled state , which would give the same result ) . obviously , the evolution in is nonlinear ( because of the post - selection ) , so one has to renormalize the final state : @xmath49\to { \cal n}[\rho]/$]tr@xmath50 $ ] . in other words , according to our approach , a chronology - respecting system in a state @xmath7 that interacts with a ctc using a unitary @xmath6 will undergo the transformation @xmath51=\frac{c\;\rho\;c^\dag } { \mbox{tr}[c\;\rho\;c^\dag ] } \labell{evol}\;,\end{aligned}\ ] ] where we suppose that the evolution does not happen if @xmath46tr@xmath47=0 $ ] . the comparison with is instructive : there the non - unitarity comes from the inaccessibility of the environment . analogously , in the non - unitarity comes from the fact that , after the ctc is closed , for the chronology - respecting system it will be forever inaccessible . the nonlinearity of is more difficult to interpret , but is connected with the periodic boundary conditions in the ctc . note that this general evolution equation is consistent with previous derivations based on path integrals . for example , it is equivalent to eq . ( 4.6 ) of ref . @xcite by hartle . however , in contrast to here , the actual form of the evolution operators @xmath28 is not provided there . as a further example , consider ref . @xcite , where politzer derives a path integral approach of ctcs for qubits , using grassmann fields . ( 5 ) is compatible with eq . . he also derives a nonunitary evolution that is consistent with eq . in the case in which the initial state is pure . in particular , this implies that , also in the general qudit case , our post - selected teleportation approach gives the same result one would obtain from a specific path - integral formulation . in addition , it has been pointed out many times before ( e.g. see @xcite ) that when quantum fields inside a ctc interact with external fields , linearity and unitarity is lost . it is also worth to notice that there have been various proposals to restore unitarity by modifying the structure of quantum mechanics itself or by postulating an inaccessible purification space that is added to uphold unitarity @xcite . the evolution coming from our approach is to be compared with deutsch s evolution , @xmath52=\mbox{tr}_{ctc}[u(\rho_{ctc}\otimes\rho)u^\dag],\mbox { where } \nonumber\\&&\rho_{ctc}=\mbox{tr}_a[u(\rho_{ctc}\otimes\rho)u^\dag ] \labell{consconds}\;\end{aligned}\ ] ] satisfies the consistency condition . the direct comparison of eqs . and highlights the differences in the general prescription for the dynamics of ctcs of these two approaches . even though the results presented in this section are directly applicable only to general finite - dimensional systems , the extension to systems living in infinite - dimensional separable hilbert spaces seems conceptually straightforward , although mathematically involved . in his path - integral formulation of ctcs , hartle notes that ctcs might necessitate abandoning not only unitarity and linearity , but even the familiar hilbert space formulation of quantum mechanics @xcite indeed , the fact that the state of a system at a given time can be written as the tensor product states of subsystems relies crucially on the fact that operators corresponding to spacelike separated regions of spacetime commute with each other . when ctcs are introduced , the notion of ` spacelike ' separation becomes muddied . the formulation of closed timelike curves in terms of p - ctcs shows , however , that the hilbert space structure of quantum mechanics can be retained . although the theory of p - ctcs was developed to address the question of quantum mechanics in general - relativistic closed timelike curves , it also allows us to address the possibility of time travel in other contexts . essentially , any quantum theory that allows the nonlinear process of projection onto some particular state , such as the entangled states of p - ctcs , allows time travel even when no spacetime closed timelike curve exists . indeed , the mechanism for such time travel closely follows wheeler s famous telephone call above . non - general - relativistic p - ctcs can be implemented by the creation of and projection onto entangled particle - antiparticle pairs . such a mechanism is just what is used in our experimental tests of p - ctcs @xcite : although projection is a non - linear process that can not be implemented deterministically in ordinary quantum mechanics , it can easily be implemented in a probabilistic fashion . consequently , the effect of p - ctcs can be tested simply by performing quantum teleportation experiments , and by post - selecting only the results that correspond to the desired entangled - state output . if it turns out that the linearity of quantum mechanics is only approximate , and that projection onto particular states does in fact occur for example , at the singularities of black holes @xcite then it might be possible to implement time travel even in the absence of a general - relativistic closed timelike curve . the formalism of p - ctcs shows that such quantum time travel can be thought of as a kind of quantum tunneling backwards in time , which can take place even in the absence of a classical path from future to past . it has been long known that nonlinear quantum mechanics potentially allows the rapid solution of hard problems such as np - complete problems @xcite . the nonlinearities in the quantum mechanics of closed timelike curves is no exception @xcite . aaronson and watrous have shown quantum computation with deutsch s closed timelike curves allows the solution of any problem in pspace , the set of problems that can be solved using polynomial space resources @xcite . similarly , aaronson has shown that quantum computation combined with post - selection allows the solution of any problem in the computational class pp , probabilistic polynomial time(where a probabilistic polynomial turing machine accepts with probability @xmath53 if and only if the answer is `` yes . '' ) . quantum computation with post - selection explicitly allows p - ctcs , and p - ctcs in turn allow the performance of any desired post - selected quantum computation . accordingly , quantum computation with p - ctcs can solve any problem in pp , including np - complete problems . since the class pp is thought to be strictly contained in pspace , quantum computation with p - ctcs is apparently strictly less powerful than quantum computation with deutsch s ctcs . in the case of quantum computating with deutschian ctcs , bennett _ et al . _ @xcite have questioned whether the notion of programming a quantum computer even makes sense . @xcite notes that in deutsch s closed timelike curves , the nonlinearity introduces ambiguities in the definition of state preparation : as is well - known in nonlinear quantum theories , the result of sending _ either _ the state @xmath10 through a closed - timelike curve _ or _ the state @xmath54 is no longer equivalent to sending the mixed state @xmath55 through the curve . the problem with computation arises because , as is clear from our grandfather - paradox circuit @xcite , deutsch s closed timelike curves typically break the correlation between chronology preserving variables and the components of a mixed state that enters the curve : the component that enters the ctc as @xmath56 can exit the curve as @xmath57 , even if the overall mixed state exiting the curve is the same as the one that enters . consequently , bennett _ et al . _ argue , the programmer who is using a deutschian closed timelike curve as part of her quantum computer typically finds the output of the curve is completely decorrelated from the problem she would like to solve : the curve emits random states . in contrast , because p - ctcs are formulated explicitly to retain correlations with chronology preserving curves , quantum computation using p - ctcs do not suffer from state - preparation ambiguity . that is not so say that p - ctcs are computationally innocuous : their nonlinear nature typically renormalizes the probability of states in an input superposition , yielding to strange and counter - intuitive effects . for example , any ctc can be used to compress any computation to depth one , as shown in fig . [ f : comput ] . indeed , it is exactly the ability of nonlinear quantum mechanics to renormalize probabilities from their conventional values that gives rise to the amplification of small components of quantum superpositions that allows the solution of hard problems . not least of the counter - intuitive effects of p - ctcs is that they could still solve hard computational problems with ease ! the ` excessive ' computational power of p - ctcs is effectively an argument for why the types of nonlinearities that give rise to p - ctcs , if they exist , should only be found under highly exceptional circumstances such as general - relativistic closed timelike curves or black - hole singularities . this paper reviewed quantum mechanical theories for time travel , focusing on the theory of p - ctcs @xcite . our purpose in presenting this work is to make precise the similarities and differences between varying quantum theories of time travel . we summarize our findings here . we have extensively argued that p - ctcs are physically inequivalent to deutsch s ctcs . in sec . [ s : pathint ] we showed that p - ctcs are compatible with the path - integral formulation of quantum mechanics . this formulation is at the basis of most of the previous analysis of quantum descriptions of closed time - like curves , since it is particularly suited to calculations of quantum mechanics in curved space time . p - ctcs are reminiscent of , and consistent with , the two - state - vector and weak - value formulation of quantum mechanics . it is important to note , however , that p - ctcs do not in any sense require such a formulation . then , in sec . [ s : general ] we extended our analysis to general systems where the path - integral formulation may not always be possible and derived a simple prescription for the calculation of the ctc dynamics , namely eq . . in this way we have performed a complete characterization of p - ctc in the most commonly employed frameworks for quantum mechanics , with the exception of algebraic methods ( e.g. see @xcite ) . in sec . [ s : generalrel ] we have argued that , as wheeler s picture of positrons as electrons moving backwards in time suggests , p - ctcs might also allow time travel in spacetimes without general - relativistic closed timelike curves . if nature somehow provides the nonlinear dynamics afforded by final - state projection , then it is possible for particles ( and , in principle , people ) to tunnel from the future to the past . our hope in elaborating the theory of p - ctcs is that this theory may prove useful in formulating a quantum theory of gravity , by providing new insight on one of the most perplexing consequences of general relativity , i.e. , the possibility of time - travel . k. gdel , rev . * 21 * , 447 ( 1949 ) . bonnor , _ j. phys . * 13 * , 2121 ( 1980 ) . gott , phys . lett . * 66 * , 1126 ( 1991 ) . morris , k.s . thorne , u. yurtsever , _ phys . * 61 * , 1446 ( 1988 ) . politzer , phys . d * 49 * , 3981 ( 1994 ) . boulware , phys . d * 46 * , 4421 ( 1992 ) . hartle , phys . d * 49 * , 6543 ( 1994 ) . van stockum , proc . a * 57 * , 135 ( 1937 ) . politzer , phys . d * 46 * , 4470 ( 1992 ) . d. deutsch , _ phys . d _ * 44 * , 3197 ( 1991 ) . a. einstein in _ albert einstein , philosopher - scientist : the library of living philosophers volume vii _ , p.a . schilpp ( ed . ) ( open court , 3rd edition , 1998 ) . c. lanczos , _ z. phys . _ * 21 * , 73 ( 1924 ) . s. hawking , _ phys . rev . d _ * 46 * , 603 ( 1992 ) . s. deser , r. jackiw , g.t hooft , _ phys . * 68 * , 267 ( 1992 ) . carroll , e. farhi , a.h . guth , k.d . olum , _ phys . d _ * 50 * , 6190 ( 1994 ) . r. feynman , in _ nobel lectures , physics 1963 - 1970 , _ elsevier publishing company , amsterdam , 1972 . s. lloyd , l. maccone , r. garcia - patron , v. giovannetti , y. shikano , s. pirandola , l.a . rozema , a. darabi , y. soudagar , l.k . shalm , and a.m. steinberg , arxiv:1005.2219 ( 2010 ) . g. t. horowitz and j. maldacena , _ jhep _ * 0402 * , 8 ( 2004 ) . pegg , in `` time s arrows , quantum measurement and superluminal behaviour '' , edited by d.mugnai et al ( consiglio nazionale delle richerche , roma , 2001 ) p.113 , arxiv : quant - ph/0506141 ( 2005 ) . bennett and b. schumacher , lecture at qupon conference vienna , austria may 2005 http://www.research.ibm.com/people/b/bennetc/quponbshort.pdf ; c.h . bennett and b. schumacher , lecture at tata institute for fundamental research , mumbai , india , february 2002 ; archived aug 2003 at http://web.archive.org/web/*/http://qpip-server.tcs.tifr.res.in/ qpip / html / courses / bennett / tifr5.pdf t.c . ralph , arxiv : quant - ph/0510038v1 ( 2005 ) . g. svetlichny , arxiv:0902.4898 ( 2009 ) . g. chiribella , g. m. dariano , p. perinotti , b. valiron , arxiv:0912.0195 ( 2009 ) . brukner , j .- w . pan , c. simon , g. weihs , a. zeilinger , phys . rev . a * 67 * , 034304 ( 2003 ) . n.d . birrell , p.c.w . davies , _ quantum fields in curved space _ ( cambridge univ . press , cambridge , 1982 ) . j. friedman , m.s . morris , i.d . novikov , f. echeverria , g. klinkhammer , k.s . thorne , and u. yurtsever , phys . d * 42 * , 1915 ( 1990 ) . friedman , n.j . papastamatiou , j.z . simon , phys . d * 46 * , 4456 ( 1992 ) . friedman , m.s . morris , phys . * 66 * , 401 ( 1991 ) . gonzlez - daz , phys . d * 58 * , 124011 ( 1998 ) . ralph , phys . a * 76 * , 012336 ( 2007 ) . ralph , g.j . milburn , and t. downes , phys . a * 79 * , 022121 ( 2009 ) . bennett c.h . , leung d. , smith g. , smolin j.a . * 103 * , 170502 ( 2009 ) . ralph , c.r . myers , arxiv:1003.1987 ( 2010 ) . bennett , g. brassard , c. crpeau , r. jozsa , a. peres , w.k . wootters phys . . lett . * 70 * , 1895 ( 1993 ) . braunstein , h.j . kimble , phys . rev . lett . * 80 * , 869 ( 1998 ) . d. bacon , phys . a * 70 * , 032309 ( 2004 ) . brun , j. harrington , m.m . wilde phys . * 102 * , 210402 ( 2009 ) . s. aaronson , j. watrous , proc . a * 465 * , 631 ( 2009 ) ; arxiv:0808.2669 . s. aaronson , proc . a * 461 * , 3473 ( 2005 ) ; arxiv : quant - ph/0412187 . brun , found . phys . lett . * 16 * , 245 ( 2003 ) . feynman , and a.r . hibbs , _ quantum mechanics and path integrals _ , ( mcgraw - hill , new york , 1965 ) . r. p. feynman , rev . phys . * 20 * , 367 ( 1948 ) . f. echeverria , g. klinkhammer , k.s . thorne , phys . d * 44 * , 1077 ( 1991 ) . a. einstein , b. podolsky , n. rosen , phys . rev . * 47 * , 777 ( 1935 ) . y. aharonov , l. vaidman , in _ time in quantum mechanics _ , edited by muga j.g . , sala mayato r. , egusquizai . l. eds . ( berlin , heidelberg : springer ) p 399 ; arxiv : quant - ph/0105101 ( 2001 ) . y. shikano and a. hosoya , j. phys . a : math . theor . * 43 * , 025304 ( 2010 ) . g. chiribella , g.m . dariano , p. perinotti , phys . a * 80 * , 022339 ( 2009 ) . y. aharonov , p. g. bergmann , and j. l. lebowitz , phys . rev . * 134 * , b1410 ( 1964 ) . aharonov y. , albert d.z . and vaidman l. , phys . rev . 60 * 1351 ( 1988 ) . cassidy , phys . d * 52 * , 5676 ( 1995 ) . a. anderson , phys . d * 51 * , 5707 ( 1995 ) . fewster , c. g. wells , phys . d * 52 * , 5773 ( 1995 ) . abrams , d. , and s. lloyd _ phys . _ , * 81 * , 3992 ( 1998 ) . u. yurtsever , class . quantum grav . * 11 * , 999 ( 1994 ) .

this paper discusses the quantum mechanics of closed timelike curves ( ctcs ) and of other potential methods for time travel . we analyze a specific proposal for such quantum time travel , the quantum description of ctcs based on post - selected teleportation ( p - ctcs ) . we compare the theory of p - ctcs to previously proposed quantum theories of time travel : the theory is physically inequivalent to deutsch s theory of ctcs , but it is consistent with path - integral approaches ( which are the best suited for analyzing quantum field theory in curved spacetime ) . we derive the dynamical equations that a chronology - respecting system interacting with a ctc will experience . we discuss the possibility of time travel in the absence of general relativistic closed timelike curves , and investigate the implications of p - ctcs for enhancing the power of computation . einstein s theory of general relativity allows the existence of closed timelike curves , paths through spacetime that , if followed , allow a time traveler whether human being or elementary particle to interact with her former self . the possibility of such closed timelike curves ( ctcs ) was pointed out by kurt gdel @xcite , and a variety of spacetimes containing closed timelike curves have been proposed @xcite . reconciling closed timelike curves with quantum mechanics is a difficult problem that has been addressed repeatedly , for example , using path integral techniques @xcite . this paper explores a particular version of closed timelike curves based on combining quantum teleportation with post - selection . the resulting post - selected closed timelike curves ( p - ctcs ) provide a self - consistent picture of the quantum mechanics of time - travel . p - ctcs offer a theory of closed timelike curves that is physically inequivalent to other hilbert - space based theories , e.g. , that of deutsch @xcite . as in all versions of time travel , closed timelike curves embody apparent paradoxes , such as the grandfather paradox , in which the time traveller inadvertently or on purpose performs an action that causes her future self not to exist . einstein ( a good friend of gdel ) was himself seriously disturbed by the discovery of ctcs @xcite . because the theory of p - ctcs rely on post - selection , they provide self - consistent resolutions to such paradoxes : anything that happens in a p - ctc can also happen in conventional quantum mechanics with some probability . similarly , the post - selected nature of p - ctcs allows the predictions and retrodictions of the theory to be tested experimentally , even in the absence of an actual general - relativistic closed timelike curve . time travel is a subject that has fascinated human beings for thousands of years . in the hindu epic , the mahabarata , for example , king revaita accepts an invitation to visit brahma s palace . although he stays for only a few days , when he returns to earth he finds that many eons have passed . the japanese fisherman in the folk tale urashima taro , having saved a sea turtle , is invited to the palace of the sea - king ; upon returning home discovers on the beach a crumbling monument , centuries old , memorializing him . the gaelic hero finn mccool suffers a similar fate . these stories also dwell on the dangers of time travel . urashima taro is given a magic box and told not to open it . finn receives the gift of a magic horse and told not to dismount . when , inevitably , taro opens the box , and finn s toe touches the ground , they instantaneously age and crumble into dust . these tales involve time travel to the future . perhaps because of the various paradoxes to which it gives rise , the concept of travel to the past is a more recent invention . starting in the late eighteenth century , a few narratives take a stab at time travel to the past , the best known being charles dickens s _ a christmas carol , _ and mark twain s _ a connecticut yankee in king arthur s court . _ the contemporary notion of time travel , together with all its attendant paradoxes , did not come into being until h.g . wells masterpiece , _ the time machine _ , which is also the first book to propose an actual device that can be used to travel back and forth in time . as frequently happens , scientific theories of time travel lagged behind the fictional versions . although einstein s theory of general relativity implicitly allows travel to the past , it took several decades before gdel proposed an explicit space - time geometry containing closed timelike curves ( ctcs ) . the gdel universe consists of a cloud of swirling dust , of sufficient gravitational power to support closed timelike curves . later , it was realized that closed timelike curves are a generic feature of highly curved , rotating spacetimes : the kerr solution for a rotating black hole contains closed timelike curves within the black hole horizon ; and massive rapidly rotating cylinders typically are associated with closed timelike curves @xcite . the topic of closed timelike curves in general relativity continues to inspire debate : hawking s chronology protection postulate , for example , suggests that the conditions needed to create closed timelike curves can not arise in any physically realizable spacetime @xcite . for example , while gott showed that cosmic string geometries can contain closed timelike curves @xcite , deser _ et al . _ showed that physical cosmic strings can not create ctcs from scratch @xcite . at bottom , the behavior of matter is governed by the laws of quantum mechanics . considerable effort has gone into constructing quantum mechanical theories for closed timelike curves . the initial efforts to construct such theories involved path integral formulations of quantum mechanics . hartle and politzer pointed out that in the presence of closed timelike curves , the ordinary correspondence between the path - integral formulation of quantum mechanics and the formulation in terms of unitary evolution of states in hilbert space breaks down @xcite . morris _ et al . _ explored the quantum prescriptions needed to construct closed timelike curves in the presence of wormholes , bits of spacetime geometry that , like the handle of a coffee cup , ` break off ' from the main body of the universe and rejoin it in the the past @xcite . meanwhile , deutsch formulated a theory of closed timelike curves in the context of hilbert space , by postulating self - consistency conditions for the states that enter and exit the closed timelike curve @xcite . general relativistic closed timelike curves provide one potential mechanism for time travel , but they need not provide the only one . quantum mechanics supports a variety of counter - intuitive phenomena which might allow time travel even in the absence of a closed timelike curve in the geometry of spacetime . one of the best - known versions of non - general relativistic quantum versions of time travel comes from wheeler , as described by feynman in his nobel prize lecture @xcite : i received a telephone call one day at the graduate college at princeton from professor wheeler , in which he said , `` feynman , i know why all electrons have the same charge and the same mass . '' `` why ? '' `` because , they are all the same electron ! '' and , then he explained on the telephone , `` suppose that the world lines which we were ordinarily considering before in time and space - instead of only going up in time were a tremendous knot , and then , when we cut through the knot , by the plane corresponding to a fixed time , we would see many , many world lines and that would represent many electrons , except for one thing . if in one section this is an ordinary electron world line , in the section in which it reversed itself and is coming back from the future we have the wrong sign to the proper time - to the proper four velocities - and that s equivalent to changing the sign of the charge , and , therefore , that part of a path would act like a positron . '' as we will see , post - selected closed timelike curves make up a precise physical theory which instantiates wheeler s whimsical idea . the purpose of the current paper is to provide a unifying description of closed timelike curves in quantum mechanics . we start from the prescription that time travel effectively represents a communication channel from the future to the past . quantum time travel , then , should be described by a quantum communication channel to the past . a well - known quantum communication channel is given by quantum teleportation , in which shared entanglement combined with quantum measurement and classical communication allows quantum states to be transported between sender and receiver . we show that if quantum teleportation is combined with post - selection , then the result is a quantum channel to the past . the entanglement occurs between the forward- and backward- going parts of the curve , and post - selection replaces the quantum measurement and obviates the need for classical communication , allowing time travel to take place . the resulting theory allows a description both of the quantum mechanics of general relativistic closed timelike curves , and of wheeler - like quantum time travel in ordinary spacetime . as described in previous work @xcite , the notion that entanglement and projection can give rise to closed timelike curves to has arisen independently in a variety of contexts . this combination lies at the heart of the horowitz - maldacena model for information escape from black holes @xcite , and gottesmann and preskill note in passing that this mechanism might be used for time travel @xcite . pegg explored the use of a related mechanism for ` probabilistic time machines ' @xcite . bennett and schumacher have explored similar notions in unpublished work @xcite . ralph suggests using teleportation for time traveling , although in a different setting , namely , displacing the entangled resource in time @xcite . svetlichny describes experimental techniques for investigating quantum travel based on entanglement and projection @xcite . chiribella _ eet al . _ cosider this mechanism while analyzing extensions to the quantum computational model @xcite . brukner _ et al . _ have analyzed probabilistic teleportation ( where only the cases in which the bell measurement yields the desired result are retained ) as a computational resource in @xcite . the outline of the paper follows . in sec . [ s : pctcs ] we describe p - ctcs and deutsch s mechanism in detail , emphasizing the differences between the two approaches . then , in sec . [ s : pathint ] we relate p - ctcs to the path - integral formulation of quantum mechanics . this formulation is particularly suited for the description of quantum field theory in curved spacetime @xcite , and has been used before to provide quantum descriptions of closed timelike curves @xcite . our proposal is consistent with these path - integral approaches . in particular , the path - integral description of fermions using grassmann fields given by politzer @xcite yields a dynamical description which coincides with ours for systems of quantum bits . other descriptions , such as hartle s @xcite , are more difficult to compare as they do not provide an explicit prescription to calculate the details of the dynamics of the interaction with systems inside closed timelike curves . in any case , their general framework is consistent with our derivations . by contrast , deutsch s ctcs are not compatible with the politzer path - integral approach , and are analyzed by him on a different footing @xcite . indeed , suppose that the path integral is performed over classical paths which agree both at the entrance to and at the exit from the ctc , so that @xmath0-in , @xmath1-in are the same as @xmath0-out , @xmath1-out . similarly , in the grassmann case , suppose that spin - up along the @xmath2 axis at the entrance emerges as spin - up along the @xmath2 axis at the exit . then , the quantum version of the ctc must exhibit the same perfect correlation between input and output . but , as the grandfather paradox experiment @xcite shows , deutsch s ctcs need not exhibit such correlations : spin - up in is mapped to spin - down out ( although the overall quantum state remains the same ) . by contrast , p - ctcs exhibit perfect correlation between in- and out- versions of all variables . note that a quantum - field theoretical justification of deutsch s solution is proposed in @xcite and is based on introducing additional hilbert subspaces for particles and fields along the geodesic : observables at different points along the geodesic commute because they act on different hilbert spaces . the path - integral formulation also shows that using p - ctcs it is impossible to assign a well defined state to the system in the ctc . this is a natural requirement ( or , at least , a desirable property ) , given the cyclicity of time there . in contrast , deutsch s consistency condition is explicitly built to provide a prescription for a definite quantum state @xmath3 of the system in the ctc . in sec . [ s : general ] we go beyond the path - integral formulation and provide the dynamical evolution formulas in the context of generic quantum mechanics ( the hilbert - space formulation ) . namely , we treat the ctc as a generic quantum transformation , where the transformed system emerges at a previous time `` after '' eventually interacting with some chronology - respecting systems . in this framework we obtain the explicit prescription of how to calculate the nonlinear evolution of the state of the system in the chronology - respecting part of the spacetime . this nonlinearity is exactly of the form that previous investigations ( e.g. hartle s @xcite ) have predicted . in sec . [ s : generalrel ] we consider time travel situations that are independent from general - relativistic ctcs . we then conclude in sec . [ s : comp ] with considerations on the computational power of the different models of ctcs .

energy spectra and colour - colour diagrams ( cd ) are often used to study neutron - star low - mass x - ray binaries ( ns - lmxbs ; e.g. , * ? ? ? the evolution of the energy spectrum and the tracks on the cd are thought to be driven by variations of mass accretion rate , and reflect changes in the configuration of the accretion flow ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . in the low - hard state the accretion rate is low , the disc is truncated at large radii ( * ? ? ? * ; * ? ? ? * ; * ? ? ? * see however lin et al . 2007 ) and the energy spectrum is dominated by a hard / comptonised power - law component . when the source accretion rate increases , the disc truncation radius decreases and eventually reaches the last stable orbit . in the high - soft state the accretion rate is high and the energy spectrum is dominated by a soft component , possibly a combination of the accretion disc and the neutron star . the characteristic frequencies ( e.g. , quasi - periodic oscillations , qpos ) in the power density spectra ( pds ) of these systems also change with the source luminosity and inferred mass accretion rate ( e.g. , * ? ? ? * ; * ? ? ? kilohertz ( khz ) qpos have been detected in many ns - lmxbs ( for a review see * ? ? ? * and references therein ) . the upper khz qpo ( from the pair of qpos the one at the highest frequency ) in these systems has been interpreted in terms of characteristic frequencies ( e.g. , the keplerian frequency ) in a geometrically thin accretion disc @xcite . in this scenario , changes of the upper khz qpo frequency reflect changes of the inner disc radius , driven by mass accretion rate . indeed , the frequency of the upper khz qpos is strongly correlated with the hard colour of the source @xcite . several models have been proposed to explain the lower khz qpo in these systems . @xcite suggested a lense - thirring precession model , in which the frequencies of the qpos are associated with the fundamental frequencies of geodesic motion of clumps of gas around the compact object . in the relativistic resonance model @xcite , the khz qpos appear at frequencies that correspond to a coupling between two oscillations modes of the accretion disc . in the beat - frequency model @xcite , the lower khz qpo originates from the interaction between the spin frequency of the ns and material orbiting at the inner edge of the accretion disc . none of these models , however , have so far been able to fully explain all the properties of khz qpos ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? 4u 163653 is a ns - lmxb that shows regular state transitions with a cycle of @xmath340 days ( e.g. , * ? ? ? * ) , making it an excellent source to study correlations between its spectral and timing properties . the full range of spectral states ( low / hard state , high / soft state , transitional state ) has been observed in this source @xcite . a pair of khz qpos were discovered by @xcite and @xcite . the upper khz qpo has been observed in different states . its central frequency shows a clear correlation with the hard colour of the source @xcite . the lower khz - qpo in 4u 163652 is only detected over a narrow range of hard colour values @xcite . the emission mechanism of the lower khz - qpo is still unclear ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) we analysed the broadband energy spectra of 4u 163653 to investigate the evolution of the different spectral and timing components as a function of the spectral state of the source . a comparison the different continuum components in the energy spectrum with the properties of the khz qpos at the same state may provide an important clue to understand the origin of the khz qpos and the evolution of the accretion flow geometry . in [ data ] we describe the observations , data reduction and analysis methods , and in [ result ] we present the results on the temporal and spectral analysis of these data . finally , in [ discussion ] we discuss our findings and summarise our conclusions . we analysed the whole archival data ( 1576 observations ) from the rossi x - ray timing explorer ( _ rxte _ ) proportional counter array ( pca ; * ? ? ? * ) and the high - energy x - ray timing experiment ( hexte ; * ? ? ? * ) of the ns - lmxb 4u 163653 . we reduced the data using the heasoft package version 6.13 . we extracted pca spectra from the proportional counter unit number 2 ( pcu-2 ) only , since this was the best - calibrated detector and the only one which was always on in all the observations . to extract the spectra of the source we first examined the light curves to identify and remove x - ray bursts from the data . for the hexte data we generated the spectra using cluster b only , since after january 2006 cluster a stopped rocking and could no longer measure the background . for each observation we extracted one pca and hexte x - ray spectrum , respectively . the pca and hexte background spectra were extracted using the standard rxte tools pcabackest and hxtback , respectively . we built instrument response files for the pca and hexte data using pcarsp and hxtrsp , respectively . for each observation we computed fourier power density spectra ( pds ) in the @xmath4 kev band every 16 s from event - mode data . for this we binned the light curves to 1/4096 s , corresponding to a nyquist frequency of 2048 hz . before computing the fourier transform we removed detector dropouts , but we did not subtract the background or applied any dead - time correction before calculating the pds . finally we calculated an average pds per observation normalised as in @xcite . we finally used the procedures described in @xcite to detect and fit the qpos in each pds . we detected khz qpos in 581 out of 1576 observations . we detected the lower khz qpo in 403 out of those 583 observations . , width=316 ] we used the standard-2 data ( 16-s time - resolution and 129 channels covering the full @xmath4 kev pca band ) to calculate x - ray colours of the source ( see * ? ? ? * for details ) . we defined hard and soft colours as the @xmath5 kev and @xmath6 kev count rate ratios , respectively . we show the cd of all observations of 4u 1636@xmath153 in figure [ fig : ccd ] , with one point per rxte observation . in that figure we parameterised the position of the source in the cd by the length of the solid curve @xmath7 ( see , e.g. * ) , fixing the values of @xmath8 and @xmath9 at the top - right and the bottom - left vertex of the cd , respectively . for the spectral analysis of 4u 163653 , we used the package xspec v12.7 @xcite . we fitted the pca and hexte spectra simultaneously in the 3.0@xmath125.0 and 25.0@xmath1180.0 kev range , respectively . we included the effect of interstellar absorption using the component phabs with cross - sections of @xcite and solar abundances from @xcite , and we fixed the column density to @xmath10 @xmath11 @xcite . we added a multiplicative factor to the model to account for calibration uncertainties between the pca and hexte . we set this factor to unity for the pca and left it free for the hexte spectra . many models have been proposed to fit the spectra of accreting ns - lmxbs in the past ( e.g. * ? ? ? * ; * ? ? ? most models include at least two components : a soft thermal component that represents the emission from the ns surface ( or boundary layer ) and the accretion disc , and a hard comptonised component that represents the emission from a corona of hot electrons . in the spectral fitting of 4u 163653 we used the same continuum model as in @xcite , who fit _ xmm - newton _ data down to 0.8 kev . we used a multi - colour disc blackbody ( diskbb in xspec ) to fit the thermal emission from the disc , a single - temperature black body ( bb , bbodyrad in xspec ) to fit the thermal emission from the ns surface ( or boundary layer ) , and a comptonisation model , nthcomp , to fit the comptonised component @xcite . the parameters of bbodyrad and diskbb are the blackbody temperature , @xmath12 , the temperature at the inner disc radius , @xmath13 , and their normalisations , respectively . the parameters of nthcomp are the asymptotic power - law photon index , @xmath2 , the electron temperature of the corona , @xmath14 , the seed photon temperature , @xmath15 , and the normalisation . the seed photons can be from either the ns surface and boundary layer or the accretion disc . @xcite analysed the spectra of six observations of 4u 163653 taken with _ xmm - newton _ and _ rxte _ simultaneously . they tried both the bbodyrad or the diskbb component as the source of seed photons for nthcomp , and concluded that diskbb was the best option . therefore , in this work we chose diskbb as the source of the soft seed photons and we linked the seed photons temperature in nthcomp to the temperature of the diskbb component . in the fits , the value of @xmath2 was constrained to be larger or equal to 1.1 @xcite and the value of @xmath14 was restricted to be larger or equal to 2.5 kev @xcite from the initial fits , there were always relatively broad residuals between 6 and 7 kev in the _ rxte_/pca spectra . these residuals likely reflect the presence of an iron line in the data . the iron line has also been confirmed with _ xmm - newton _ and _ suzaku _ observations @xcite . we then added a gaussian component with a variable width to the spectrum to fit this line . due to the low spectral resolution of the pca ( @xmath31 kev at @xmath16 kev ) , we can not constrain the central value of the line very well . so we fixed the line at 6.5 kev in all our fittings . fitting simultaneously data taken from _ xmm - newton _ and _ rxte _ , @xcite found that the temperature of bbodyrad was @xmath17 kev , and remained more or less constant as the source moved across the cd . they also found that the temperature at the inner edge of the accretion disc ranged from @xmath30.2 kev to @xmath30.8 kev when the source moved from the low / hard to the high / soft state . since the pca only extends down to 3 kev , @xmath12 and @xmath13 are difficult to constrain simultaneously . therefore , we initially left @xmath13 free and , following the results of @xcite , we fixed @xmath18 to 1.7 kev . we then repeated the analysis fixing @xmath12 to each of these values : 1.5 , 1.6 , 1.8 , 1.9 and 2.0 kev . from the previous analysis we found that @xmath13 , linked to @xmath15 in nthcomp , was not well constrained during the fits . we therefore interpolated the value of @xmath13 along the cd using the _ xmm - newton spectra fitting results in @xcite . we divided the data into 7 groups based on their @xmath7 values : @xmath19 , @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24 and @xmath25 . in each group we fixed @xmath13 to the interpolated temperatures for @xmath7 equal to 1.1 , 1.3 , 1.5 , 1.7 , 1.9 , 2.1 , and 2.35 , respectively . in this case we left @xmath12 free during the fits . in figure [ fig : sa_ga_kt ] we show the asymptotic power - law photon index , @xmath2 , as a function of @xmath7 in 4u 163653 . the value of @xmath12 indicated in each panel was fixed during the fittings . when the source evolves from the hard spectral state to the transitional state in the cd , the value of @xmath7 increases from 1.0 to 2.0 , @xmath2 increases from 1.8 to 2.2 and the source spectrum becomes soft . as the source moves through the vertex in the cd , @xmath26 , @xmath2 drops sharply to @xmath27 , and covers the range from @xmath27 to @xmath28 . when the source moves out of the vertex to the bottom right part of the cd @xmath2 increases to @xmath28 , and finally decreases back to @xmath23 as @xmath7 increases to @xmath32.5 . it is clear from all panels that @xmath2 shows a significant drop at @xmath26 , regardless of the value we choose for @xmath12 . moreover , at low @xmath7 values ( @xmath29 ) , @xmath2 shows a larger spread for low @xmath12 values . this likely indicates that the lower blackbody temperature ( e.g. @xmath30 kev ) yields comparatively less well - constrained fits of the thermal comptonisation model in 4u 1636@xmath153 . the upper left panel of figure [ fig : sa_ga_pha ] shows @xmath2 as a function of @xmath7 when we used the interpolated values of @xmath13 and left @xmath12 free . the trend in this figure is similar to those in figure [ fig : sa_ga_kt ] . in the upper right panel figure [ fig : sa_ga_pha ] we show the electron temperature , @xmath14 , as a function of @xmath7 . as the source evolves from the hard to the transitional , and then to the soft spectral state @xmath14 first decreases from @xmath31 kev to @xmath32 kev and then stays more or less constant . in order to visualise the evolution of the individual components and the total spectrum in different states , we chose four observations covering the whole range of @xmath7 values . the four observations , obs1 , obs2 , obs3 and obs4 with @xmath331.05 , 1.90 , 2.15 and 2.35 , respectively , are shown with red filled circles in the upper panels of figure [ fig : sa_ga_pha ] . in the middle and lower panels of figure [ fig : sa_ga_pha ] we show the pca / hexte model spectra of these 4 observations . the spectral components in the plots are diskbb ( red / dashed - dotted ) , bbodyrad ( pink / dotted ) , gaussian ( green / dotted ) and nthcomp ( blue dashed - three dotted ) , respectively . the best - fitting results for the four observations are given in table [ tab - nthcomp ] . + + in the middle - left panel of figure [ fig : sa_ga_pha ] ( obs1 ) the source is in the low / hard state . the broad band x - ray spectrum is dominated by the hard / comptonised component , nthcomp . in this observation the power - law index is @xmath34 and the electron temperature is @xmath35 kev when the source evolves from the low / hard state towards the vertex in the cd , the value of @xmath7 increases from 1.0 to 2.0 , @xmath2 increases from @xmath36 to @xmath37 and the spectrum becomes soft ( obs2 in the right panel of figure [ fig : sa_ga_pha ] ) . the electron temperature decreases from @xmath38 kev to @xmath39 kev . compared to obs1 , it is clear that the contribution of bbodyrad and diskbb increases , and the emission above 80 kev drops , significantly . when the source is at the transitional state , @xmath40 , the value of @xmath2 covers a large range between 1.3 and 2.5 . the emission from the diskbb and nthcomp components is more or less the same below 5 kev ( bottom left panel of figure [ fig : sa_ga_pha ] ) . the nthcomp component is relatively flat but the high - energy cut - off is at much lower energy than in obs1 . our results are consistent with those of @xcite for the combined _ xmm - newton rxte _ spectra ( see their figure 4 ) . when the source moves out from the transitional state to the high - soft state ( obs4 ) , similar to what @xcite show in their figure 4 , the normalization of diskbb decreases and the emission from diskbb is lower than that of nthcomp below 5 kev . = 2.0 mm from figure [ fig : sa_ga_pha ] and the discussion above , it is apparent that the power - law photon index , @xmath2 , in many of the observations shows a significant drop when the source is in the transitional state . due to the large spread of @xmath2 when 2 @xmath41 @xmath7 @xmath41 2.2 the trend of @xmath2 in this area of the plot is not clear . we therefore first sorted the data in figure [ fig : sa_ga_pha ] according to @xmath7 and rebinned them using a step of 0.025 , 0.005 and 0.02 in the range of 1.0@xmath421.8 , 1.8@xmath422.2 and @xmath432.2 , respectively . the weighted average @xmath2 as a function of @xmath7 is shown in figure [ fig : sa_gamma_rebin ] . from this figure it is apparent that , as @xmath7 increases from @xmath44 to @xmath45 , @xmath2 decreases abruptly from @xmath46 to @xmath47 , around @xmath7 @xmath45 @xmath2 increases again with @xmath7 from @xmath47 to @xmath48 and , as @xmath7 increases from @xmath45 to @xmath49 , @xmath2 decreases from @xmath48 to @xmath50 . in the transitional state @xmath2 shows a large spread , but the plot in figure [ fig : sa_gamma_rebin ] shows that @xmath2 changes significantly with @xmath7 and that the variation of @xmath2 is larger than the statistical fluctuations in the data . the large range of @xmath2 at the transitional state might be due to the fact that the model assumptions ( e.g. , a geometrically thin disc , or a spherical corona ) are no longer valid . figure [ fig : sa_qpo ] shows the centroid frequency of all the khz qpos detected in 4u 163653 as a function of @xmath7 . the filled circles ( black ) and the filled squares ( red ) represent the observations with upper and lower khz qpos , respectively ( see also * ? ? ? the upper khz qpo is detected when @xmath51 , and its frequency increases as @xmath7 increases . the lower khz qpo however , is only detected when @xmath52 , and its frequency covers a broad range over a narrow @xmath7 range . , in nthcomp as a function of @xmath7 in 4u 1636@xmath153 . the data are the same as in the upper left panel of figure [ fig : sa_ga_pha ] , but have been rebinned as described in the text . , width=316 ] in 4u 163653 . the filled circles ( black ) and filled squares ( red ) represent the observations with upper and lower khz qpos , respectively . , width=316 ] in the upper panel of figure [ fig : sa_qpo_sigma ] we plot the power - law photon index , @xmath2 , as a function of @xmath7 ( the same data shown in figure [ fig : sa_ga_pha ] ) . the filled circles ( black ) and the filled squares ( red ) represent the observations with upper and lower khz qpos , respectively . interestingly , observations with lower khz qpos correspond with those observations in which @xmath2 shows a large spread . and @xmath53 as a function of @xmath7 in 4u 163653 . the filled circles ( black ) and filled squares ( red ) represent the observations with upper and lower khz qpos , respectively . the gray crosses represent observations in which no khz qpo was detected . , title="fig:",width=316 ] and @xmath53 as a function of @xmath7 in 4u 163653 . the filled circles ( black ) and filled squares ( red ) represent the observations with upper and lower khz qpos , respectively . the gray crosses represent observations in which no khz qpo was detected . , title="fig:",width=316 ] in the case of comptonisation in spherically symmetric region @xmath2 can be expressed as : @xmath54^{1/2 } - \frac{1}{2 } , \label{eq_gamma}\ ] ] where @xmath53 is the optical depth of the medium , and @xmath55 is the electron temperature of the comptonising region @xcite . we calculated @xmath53 for each observation and plotted them as a function of @xmath7 in the lower panel of figure [ fig : sa_qpo_sigma ] . from this figure it is apparent that in the low / hard state the optical depth remains more or less constant at @xmath56 as @xmath7 increases from @xmath57 to @xmath58 . when the source enters the transitional state , with @xmath59 , @xmath53 increases very abruptly with @xmath7 . in the observations with a lower khz qpo @xmath53 covers the range from @xmath60 to @xmath61 . a large value of @xmath53 has also been found in the soft state of other ns - lmxbs ( e.g. 4u 1608@xmath152 * ? ? ? the large variation of @xmath2 ( and @xmath53 ) over a small region in the cd indicates that , in the transitional state , either the spectrum changes significantly or the model is no longer physically appropriate . even if the latter was the case , it is still interesting that the lower khz qpo is detected only in this part of the diagram . we fitted the @xmath0 kev x - ray spectra of all _ rxte _ observations of the neutron - star low - mass x - ray binary 4u 163653 with a model that includes a comptonised component . we subsequently studied the relation between the parameters of this component and the presence of khz qpos as a function of the source state , using the parameter @xmath7 to measure the position of the source in the cd . we found that , while the upper khz qpo is present in almost all states of the source , the lower khz qpo appears only during the transition from the hard to the soft state , in observations in which the optical depth of the comptonised component is relatively high and increases very sharply with @xmath7 , from @xmath62 to @xmath63 . in the low - hard state , the power - law index of the comptonised component , @xmath2 , increases gradually from @xmath64 to @xmath65 as the source spectrum softens and the source moves in the cd from the hard to the transitional state . in these observations @xmath2 is well correlated with the source hard colour and @xmath7 . the power density spectra of these observations show only the upper khz qpo , with the qpo frequency being also correlated both with hard colour and @xmath7 . similar correlations have been reported in other ns - lmxbs ( e.g. , 4u 0614 + 09 , 4u 160852 , 4u 172834 , and aql x1 ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? these correlations between spectral and timing properties in 4u 163653 are consistent with the scenario in which the truncation radius of the accretion disc decreases as mass accretion increases : as the x - ray luminosity of the source ( and the inferred mass accretion rate through the disc ) increases , the inner truncation radius of the accretion disc decreases ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and the disc temperature @xcite , and hence the soft flux , in the system increases . this in turn boosts the cooling of the corona through inverse compton scattering , which leads to a drop of the electron temperature and an increase of the optical depth of the corona @xcite , and the energy spectrum of the source softens . in the case of 4u 163653 this picture is further supported by fits to the _ xmm - newton _ and _ rxte _ ( pca@xmath66hexte ) spectra in the @xmath67 kev band over a range of source spectral states @xcite . finally , if the frequency of the upper khz qpo reflects the keplerian frequency at the inner edge of the accretion disc , a smaller truncation radius yields a higher qpo frequency . the above description only links the dynamics ( frequency ) of ( at least ) the upper khz qpo to the spectral properties of the source , but it does not address the mechanism that determines the emission properties of the qpos . for example , the steep increase of the rms amplitude of the khz qpos with energy ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) implies a modulation of the emitted flux of up to @xmath68% at @xmath69 kev , where the contribution of the disc is negligible ( see figure [ fig : sa_ga_pha ] ) . our findings suggest that the mechanisms that lead to the emission of the lower and the upper khz qpo are not the same . this is similar to what was proposed by @xcite based on the energy and frequency dependence of the time lags of the khz qpos in this source . we detect the upper khz qpo over all source spectral states , except the extreme soft state , whereas we only detect the lower khz qpo in the transition from the hard to the soft state , at @xmath70 , where @xmath2 drops abruptly from @xmath37 to @xmath47 . the parameter @xmath2 , together with the electron temperature , @xmath55 , in the nthcomp spectral model component is related to @xmath53 , the optical depth of the corona responsible for the comptonised emission , through eq . ( [ eq_gamma ] ) . since during the observations in the transitional state , where the lower khz qpo is present , @xmath2 drops abruptly ( see , e.g. , the upper left panel of figure [ fig : sa_ga_pha ] ) , whereas @xmath71 continues decreasing gradually ( see the upper right panel of figure [ fig : sa_ga_pha ] ) , in the context of the nthcomp model the drastic changes of @xmath2 likely reflect substantial changes of the optical depth of the corona . the lower panel of figure [ fig : sa_qpo_sigma ] shows that in the observations with the lower khz qpo @xmath53 is high and increases very sharply with @xmath7 , from @xmath62 to @xmath63 . notwithstanding that in several of these observations we also detected the upper khz qpo , this qpo is also present in observations in which the optical depth of the corona is low in comparison , @xmath72 , and does not change much with @xmath7 . notice , by the way , that the drop of @xmath2 in the transitional and soft states does not contradict the fact that at @xmath73 the spectrum of the source softens . indeed , although in the transitional and soft states @xmath2 decreases to values that are similar to those in the hard state , suggesting an overall hardening of the spectrum , @xmath55 is much higher in the hard than in the transitional and soft states , such that the @xmath0 kev spectrum of the source does in fact softens as @xmath7 increases ( cf . figure [ fig : sa_ga_pha ] ) . @xcite ( see also * ? ? ? * ) proposed a model in which a feed - back loop between the corona and the source of soft photons ( the disc and/or the neutron - star surface ) excites a global mode in the system , such that the temperature and electron density of the corona and the temperature of the source of soft photons oscillate coherently at a given frequency , giving rise to a qpo . this model can explain the rms and time - lag energy spectra of the lower khz qpo @xcite in the ns lmxb 4u 160852 . this scenario also provides a framework to interpret our findings in terms of a coupling and an efficient energy exchange between the disc and the corona . since in the observations in which the lower khz qpo is present the changes of the temperature of the corona are relatively small ( at least compared to the relative changes of the optical depth ) , the frequency range spanned by this qpo likely reflects the range of optical depths ( electron density ) in the corona at which the resonance is excited . we define the optical depth of the corona as @xmath74 @xcite , where @xmath75 is the size of the corona , @xmath76 @xmath77 is the thomson cross section of the electron , and @xmath78 is the electron density . if we take 5 and 20 as extreme values of @xmath53 in our fits ( see figure [ fig : sa_qpo_sigma ] ) , then @xmath79 @xmath80 @xmath81 and @xmath82 @xmath81 for @xmath75 in the range of 4 and 15 km @xcite , respectively . these values are consistent with the typical electron density in the corona of neutron star or black hole systems ( e.g. , * ? ? ? it is interesting that in 4u 163653 the frequency at which ( i ) the lower khz qpo is most often detected @xcite , ( ii ) the quality factor ( @xmath83 , defined as the qpo frequency divided by the full - width at half - maximum ) of the lower khz qpo is the highest @xcite , and ( iii ) the coherence between the low- and high - energy signals is the highest @xcite , is always @xmath84 hz . furthermore , @xcite recently showed that the phase lag between the low- and high - energy photons in the lower khz qpo in 4u 163653 is largest at @xmath85 , which also corresponds to a qpo frequency of @xmath84 hz . all the above suggests that this frequency corresponds to a global oscillation mode of the accretion flow in this source ( see * ? ? ? the mode could possibly be excited over a fairly large range of values of the properties of the corona ( e.g. , @xmath53 , @xmath55 and @xmath75 ) , and hence over a range of frequencies of the qpo , as observed . on the other hand , the excitation and damping mechanisms of this global mode of the corona will play a role upon the range of frequencies and the coherence of the qpo . in a follow - up paper ( ribeiro et al . 2017 , in preparation ) we discuss correlations between the rms amplitude of the qpos and the spectral properties of the source , which provide further support to this scenario . this research has made use of data obtained from the high energy astrophysics science archive research center ( heasarc ) , provided by nasa s goddard space flight center and nasa s astrophysics data system bibliographic services . we thank tomaso belloni for useful comments and discussions and wenfei yu for carefully reading and commenting the manuscript . er acknowledges the support from conselho nacional de desenvolvimento cientfico e tecnolgico ( cnpq - brazil ) m. , van der klis m. , van paradijs j. , lewin w. h. g. , lamb f. , vaughan b. , kuulkers e. , augusteijn t. , zhang w. , marshall f. e. , swank j. h. , lapidus i. , lochner j. c. , strohmayer t. e. , 1996 , apj , 469 , l13

we fitted the @xmath0-kev spectrum of all the observations of the neutron - star low - mass x - ray binary 4u 1636@xmath153 taken with the _ rossi x - ray timing explorer _ using a model that includes a thermal comptonisation component . we found that in the low - hard state the power - law index of this component , @xmath2 , gradually increases as the source moves in the colour - colour diagram . when the source undergoes a transition from the hard to the soft state @xmath2 drops abruptly ; once the source is in the soft state @xmath2 increases again and then decreases gradually as the source spectrum softens further . the changes in @xmath2 , together with changes of the electron temperature , reflect changes of the optical depth in the corona . the lower kilohertz quasi - periodic oscillation ( khz qpo ) in this source appears only in observations during the transition from the hard to the soft state , when the optical depth of the corona is high and changes depends strongly upon the position of the source in the colour - colour diagram . our results are consistent with a scenario in which the lower khz qpo reflects a global mode in the system that results from the resonance between , the disc and/or the neutron - star surface , and the comptonising corona . [ firstpage ] accretion , accretion discs stars : neutron x - rays : binaries stars : individual : 4u 163653

non - gaussianity from the simplest inflation models , that are based on a slowly rolling scalar field , is very small @xcite ; however , a very large class of more general models , e.g. , models with multiple scalar fields , features in inflation potential , non - adiabatic fluctuations , non - canonical kinetic terms , deviations from the bunch - davies vacuum , among others , predict substantially higher level of primordial non - gaussianity ( * ? ? ? * for a review and references therein ) . primordial non - gaussianity can be described in terms of the 3-point correlation function of bardeen s curvature perturbations , @xmath4 , in fourier space : @xmath5 depending on the shape of the 3-point function , i.e. , @xmath6 , non - gaussianity can be broadly classified into two classes @xcite . first , the local , `` squeezed , '' non - gaussianity where @xmath7 is large for the configurations in which @xmath8 . second , the non - local , `` equilateral , '' non - gaussianity where @xmath7 is large for the configuration when @xmath9 . the local form arises from a non - linear relation between inflaton and curvature perturbations @xcite , curvaton models @xcite , or the new ekpyrotic models @xcite . the equilateral form arises from non - canonical kinetic terms such as the dirac - born - infeld ( dbi ) action @xcite , the ghost condensation @xcite , or any other single - field models in which the scalar field acquires a low speed of sound @xcite . while we focus on the local form in this paper , it is straightforward to repeat our analysis for the equilateral form . the local form of non - gaussianity may be parametrized in real space as @xcite : @xmath10 where @xmath0 characterizes the amplitude of primordial non - gaussianity . different inflationary models predict different amounts of @xmath0 , starting from @xmath11 to @xmath12 , beyond which values have been excluded by the cosmic microwave background ( cmb ) bispectrum of wmap temperature data , @xmath13 , at the @xmath14 level @xcite . so far all the constraints on primordial non - gaussianity use only temperature information of the cmb . by also having the e - polarization information together with cmb temperature information , one can improve the sensitivity to the primordial fluctuations @xcite . although the experiments have already started characterizing e - polarization anisotropies @xcite , the errors are large in comparison to temperature anisotropy . the upcoming experiments such as planck satellite will characterize e - polarization anisotropy to a higher accuracy . it is very timely to develop the tools which can optimally utilize the combined cmb temperature and e - polarization information to constrain models of the early universe . throughout this paper we use the standard lambda cdm cosmology with the following cosmological parameters : @xmath15 , @xmath16 , @xmath17 , @xmath18 , @xmath19 , and @xmath20 . for all of our simulations we used healpix maps with @xmath21 pixels . in our recent paper @xcite we described a fast cubic ( bispectrum ) estimator of @xmath0 , using a combined analysis of the temperature and e - polarization observations . the estimator was optimal for homogeneous noise , where optimality was defined by saturation of the fisher matrix bound . in this paper we generalize our previous estimator of @xmath0 to deal more optimally with a partial sky coverage and the inhomogeneous noise . the generalization is done in an analogous way to how @xcite generalized the temperature only estimator developed by @xcite ; however , the final result of @xcite ( their eq . ( 30 ) ) is off by a factor of two , which results in the error in @xmath0 that is much larger than the fisher matrix prediction , as we shall show below . the fast bispectrum estimator of @xmath0 from the combined cmb temperature and e - polarization data can be written as @xmath22 , where @xcite @xmath23 @xmath24 @xmath25 @xmath26 @xmath27 @xmath28 and @xmath29 is a fraction of the sky observed . indices @xmath30 and @xmath31 can either be @xmath32 or @xmath33 . here , @xmath34 is 1 when @xmath35 , 6 when @xmath36 , and 2 otherwise , @xmath37 is the theoretical bispectrum for @xmath38 @xcite , @xmath39 is the power spectrum of the primordial curvature perturbations , and @xmath40 is the radiation transfer function of adiabatic perturbations . it has been shown that the above mentioned estimator is optimal for the full sky coverage and homogeneous noise @xcite . to be able to deal with the realistic data , the estimator has to be able to deal with the inhomogeneous noise and foreground masks . the estimator can be generalized to deal with a partial sky coverage and the inhomogeneous noise by adding a linear term to @xmath41 : @xmath42 . for the temperature only case , this has been done in @xcite . following the same argument , we find that the linear term for the combined analysis of cmb temperature and polarization data is given by @xmath43 where @xmath44 and @xmath45 are the @xmath46 and @xmath47 maps generated from monte carlo simulations that contain signal and noise , and @xmath48 denotes the average over the monte carlo simulations . the generalized estimator is given by @xmath49 which is the main result of this paper . note that @xmath50 , and this relation also holds for the equilateral shape . therefore , it is straightforward to find the generalized estimator for the equilateral shape : first , find the cubic estimator of the equilateral shape , @xmath51 , and take the monte carlo average , @xmath52 . let us suppose that @xmath51 contains terms in the form of @xmath53 , where @xmath46 , @xmath47 , and @xmath54 are some filtered maps . use the wick s theorem to re - write the average of a cubic product as @xmath55 . finally , remove the mc average from single maps , and replace maps in the product with the simulated maps : @xmath56 . this operation gives the correct expression for the linear term , both for the local form and the equilateral form . one can find the estimator of @xmath0 from the temperature data only by setting @xmath57 . we have compared our formula in the temperature - only limit with the original formula derived by @xcite ( their eq . ( 30 ) ) , and found a discrepancy . to see the discrepancy , let us re - write the estimator as : @xmath58 . our formula gives @xmath59 , while eq . ( 30 ) of @xcite gives @xmath60 . to make sure that our normalization gives the minimum variance estimator , we have done monte carlo simulations with varying @xmath61 . we find that @xmath59 minimizes the variance ( as shown in fig . [ normalization ] ) . we conclude that the analysis given in @xcite resulted in the larger - than - expected uncertainty in @xmath0 because of this error in their normalization of the linear term . the main contribution to the linear term comes from the inhomogeneous noise and sky cut . for the temperature only case , most of the contribution to the linear term comes from the inhomogeneous noise , and the partial sky coverage does not contribute much to the linear term . this is because the sky - cut induces a monopole contribution outside the mask . in the analysis one subtracts the monopole from outside the mask before measuring @xmath41 , which makes the linear contribution from the mask small @xcite . for a combined analysis of the temperature and polarization maps , however , the linear term does get a significant contribution from a partial sky coverage ( see the right panel of fig . [ fnl_stdev ] ) . subtraction of the monopole outside of the mask is of no help for polarization , as the monopole does not exist in the polarization maps by definition . ( the lowest relevant multipole for polarization is @xmath62 . ) the estimator is still computationally efficient , taking only @xmath63 ( times the @xmath31 sampling , which is of order 100 ) operations in comparison to the full bispectrum calculation which takes @xmath64 operations . here @xmath3 refers to the total number of pixels . for planck , @xmath65 , and so the full bispectrum analysis is not feasible while our analysis is . in the left panel of figure [ fnl_stdev ] , we show the variance of @xmath0 using the estimator ( with and without the linear term ) for the gaussian cmb simulations in the presence of inhomogeneous noise and partial sky coverage . for this analysis we use the noise properties that are expected for the planck satellite , assuming the cycloidal scanning strategy @xcite . inhomogeneous nature of the noise is depicted in the lower map of figure [ noise_sabb ] where we show the number of observations ( @xmath66 ) for the different pixels in the sky . as for the foreground masks , we use wmap kp0 intensity mask and p06 polarization mask . we find that , with the inclusion of the linear term , the variance reduces by more than a factor of 5 . the linear term greatly reduces the variance approaching the fisher matrix bound . however , the estimator is close to , but not exactly the same as the fisher variance prediction in the noise dominated regime . nevertheless , we do not observe the increase of variance at higher @xmath67 : the variance becomes smaller as we include more multipoles . this result is in contradiction with the result of @xcite and @xcite . we attribute this discrepancy to the error in the normalization of linear term in their formula . in the right panel of figure [ fnl_stdev ] , we show the variance of @xmath0 again using gaussian simulations , but now in the presence of flat sky cut and in the _ absence _ of any noise . the purpose of the plot is to demonstrate ( as pointed out in the previous section ) that for the combined cmb temperature and polarization analysis , sky - cut does contribute significantly to the linear term . we find that the generalized estimator does a very good job in reducing the variance excess , and the simulated variance of @xmath0 does accurately saturate the fisher matrix bound . can our estimator recover the correct @xmath0 , i.e. , is our estimator unbiased ? we have tested our estimator against simulated non - gaussian cmb temperature and e - polarization maps . the non - gaussian cmb temperature and e - polarization maps were generated using the method described in @xcite . we find that our estimator is unbiased , i.e. , we can recover the @xmath0 value which was used to generate the non - gaussian cmb maps . the results for the unbiasedness of the estimator are shown in table [ nong_fnl ] . the analysis also shows the unbiasedness of the estimator described in @xcite . .unbiasedness of the generalized estimator . non - gaussian cmb maps with @xmath68 are used for @xmath69 . the standard deviation of @xmath0 , @xmath70 , was obtained using gaussian simulations . [ cols="^,^,^,^,^",options="header " , ] figures [ sabb ] and [ noise_sabb ] show the maps @xmath71 and @xmath72 , which appear in the linear term ( eq . [ slin ] ) of the estimator . these maps are calculated using 100 monte carlo simulations of the data . since the linear term contributes only in the presence of inhomogeneities , we also show these maps calculated with noise - only simulations ( i.e. no signal ) . notice how these maps correlate with the inhomogeneous noise ( as shown in the lower map of figure [ noise_sabb ] ) . upcoming cmb experiments will provide a wealth of information about the cmb polarization anisotropies together with temperature anisotropies . the combined information from the cmb temperature and polarization data improves the sensitivity to primordial non - gaussianity @xcite . the promise of learning about the early universe by constraining the amplitude of primordial non - gaussianity is now well established . in this paper we have generalized the bispectrum estimator of non - gaussianity described in @xcite , to deal with the inhomogeneous nature of noise and incomplete sky coverage . the generalization from @xcite enables us to increase optimality of the estimator significantly , without compromising the computational efficiency of the estimator : the estimator is still computationally efficient , scaling as @xmath73 compared to the @xmath74 scaling of the full bispectrum @xcite calculation for sky maps with @xmath3 pixels . for the planck satellite , this translates into a speed - up by factors of millions , reducing the required computing time from thousands of years to just hours and thus making @xmath0 estimation feasible . the speed of our estimator allows us to study its statistical properties using monte carlo simulations . we have used gaussian and non - gaussian simulations to characterize the estimator . we have shown that the generalized fast estimator is able to deal with the partial sky coverage very well and in fact the variance of @xmath0 saturates the fisher matrix bound . in the presence of both the realistic noise and galactic mask , we find that the generalized estimator greatly reduces the variance in comparison to the @xcite estimator of non - gaussianity using combined cmb temperature and polarization data . since the estimator is able to deal with the partial sky coverage very effectively , the estimator can also be used to constrain primordial non - gaussianity using the data from ground and balloon based cmb experiments which observe only a small fraction of the sky . the estimator also solves the problem @xcite of non - trivial polarization mode coupling due to foreground masks . earlier this issue was dealt with by removing the most contaminated @xmath75 modes from the analysis ( usually @xmath76 ) . the naive approach of using galactic masks to deal with the polarization contamination is to be refined . both temperature and polarization foregrounds are expected to produce non - gaussian signals . some sources of non - primordial non - gaussianity are cmb lensing , point sources , and the sunyaev zeldovich effect . understanding the non - gaussianity from the polarization foreground sources and refining the estimator to be able to deal with it will be the subject of our future work . some of the results in this paper have been derived using the cmbfast package by uros seljak and matias zaldarriaga @xcite and the healpix package @xcite . this work was partially supported by the national center for supercomputing applications under tg - mca04t015 and by university of illinois . we also utilized the teragrid cluster ( www.teragrid.org ) at ncsa . bdw acknowledges the friedrich wilhelm bessel research award by the alexander von humboldt foundation . bdw and apsy also thank the max planck institute for astrophysics for hospitality . bdw and apsy are supported in part by nsf grant numbers ast 0507676 and 0708849 , nasa / jpl subcontract no . ek acknowledges support from the alfred p. sloan foundation . komatsu , e. , kogut , a. , nolta , m. n. , bennett , c. l. , halpern , m. , hinshaw , g. , jarosik , n. , limon , m. , meyer , s. s. , page , l. , spergel , d. n. , tucker , g. s. , verde , l. , wollack , e. , & wright , e. l. 2003 , apj . , 148 , 119 , t. e. , ade , p. a. r. , bock , j. j. , bond , j. r. , borrill , j. , boscaleri , a. , cabella , p. , contaldi , c. r. , crill , b. p. , de bernardis , p. , de gasperis , g. , de oliveira - costa , a. , de troia , g. , di stefano , g. , hivon , e. , jaffe , a. h. , kisner , t. s. , jones , w. c. , lange , a. e. , masi , s. , mauskopf , p. d. , mactavish , c. j. , melchiorri , a. , natoli , p. , netterfield , c. b. , pascale , e. , piacentini , f. , pogosyan , d. , polenta , g. , prunet , s. , ricciardi , s. , romeo , g. , ruhl , j. e. , santini , p. , tegmark , m. , veneziani , m. , & vittorio , n. 2006 , , 647 , 813 , l. , hinshaw , g. , komatsu , e. , nolta , m. r. , spergel , d. n. , bennett , c. l. , barnes , c. , bean , r. , dore , o. , halpern , m. , hill , r. s. , jarosik , n. , kogut , a. , limon , m. , meyer , s. s. , odegard , n. , peiris , h. v. , tucker , g. s. , verde , l. , weiland , j. l. , wollack , e. , & wright , e. l. 2007 , apjs , 170 , 335 , d. n. , bean , r. , dore , o. , nolta , m. r. , bennett , c. l. , hinshaw , g. , jarosik , n. , komatsu , e. , page , l. , peiris , h. v. , verde , l. , barnes , c. , halpern , m. , hill , r. s. , kogut , a. , limon , m. , meyer , s. s. , odegard , n. , tucker , g. s. , weiland , j. l. , wollack , e. , & wright , e. l. 2006 , apjs , in press ( astro - ph/0603449 )

in our recent paper @xcite we described a fast cubic ( bispectrum ) estimator of the amplitude of primordial non - gaussianity of local type , @xmath0 , from a combined analysis of the cosmic microwave background ( cmb ) temperature and e - polarization observations . in this paper we generalize the estimator to deal with a partial sky coverage as well as inhomogeneous noise . our generalized estimator is still computationally efficient , scaling as @xmath1 compared to the @xmath2 scaling of the brute force bispectrum calculation for sky maps with @xmath3 pixels . upcoming cmb experiments are expected to yield high - sensitivity temperature and e - polarization data . our generalized estimator will allow us to optimally utilize the combined cmb temperature and e - polarization information from these realistic experiments , and to constrain primordial non - gaussianity .

recently the manifolds of exceptional holonomy attracted much attention . these are 7dimensional manifolds of @xmath5 holonomy and 8dimensional manifolds of @xmath7 holonomy . they are considered in the context of the string theory compactifications . the supersymmetric nonlinear sigma models on the manifolds of exceptional holonomy are described by conformal field theories , their superconformal chiral algebras were constructed in @xcite . we will call them the @xmath5 and @xmath7 superconformal algebras . these are nonlinear @xmath8algebras ( @xcite , for review see @xcite ) of central charge @xmath9 and @xmath10 respectively . the conformal field theories were further studied in @xcite . the @xmath7 algebra is identified @xcite with the @xmath11superconformal algebra @xcite , existing at generic values of the central charge . it consists of the @xmath12 superconformal algebra extended by its spin2 superprimary field . the unitary representation theory of the @xmath11algebra is studied in @xcite , where complete list of unitary representations is determined ( including the @xmath13 model , corresponding to the @xmath7 manifolds ) . in this paper we identify the @xmath5 algebra with the @xmath0superconformal algebra ( in notations of @xcite ) at the central charge @xmath14 and the coupling constant ( see below ) @xmath15 . the @xmath0algebra was first constructed in @xcite ( see also @xcite ) . it is superconformal @xmath8algebra , which besides the energy momentum supercurrent ( the first `` @xmath2 '' in @xmath0 ) contains two supercurrents of spins @xmath2 and @xmath16 . the @xmath0algebra has two generic parameters . along with the central charge there is a free coupling @xmath17 ( the self coupling of the spin@xmath2 superprimary field ) , which is not fixed by jacobi identities . in @xcite the @xmath0algebra is shown to be the symmetry algebra of the quantized toda theory corresponding to the @xmath18 lie superalgebra ( the only simple lie superalgebra with free parameter ) . in the same ref.@xcite the free field representation of the @xmath0algebra is constructed . we study different aspects of the @xmath0algebra in the present paper . first we find that the @xmath0algebra is the symmetry algebra of the diagonal coset @xmath19 we define highest weight representations of the algebra and study their unitarity . the unitary minimal models are described by the coset ( [ coset ] ) . their central charge and coupling @xmath17 are given by @xmath20 we also obtain all the values of @xmath21 and @xmath17 , where the @xmath0algebra has continuous spectrum of unitary representations . one such model ( @xmath14 , @xmath15 ) , which corresponds to the @xmath5 algebra , is discussed in details , the full spectrum of unitary representations is obtained . we also present the complete list of the minimal model representations and their fusion rules . the diagonal coset constructions of type @xmath22 were found very useful in the description of minimal models of different conformal algebras . the minimal models of the virasoro algebra @xcite ( @xmath23 ) correspond to the diagonal coset construction @xcite @xmath24 the coset ( [ coset n=1 ] ) is found @xcite to form the minimal models of the @xmath12 superconformal algebra ( @xcite and appendix [ appn=1 ] ) . the minimal models of the @xmath25 algebra @xcite are the @xmath26 diagonal cosets @xmath27 we present here the first example ( to our knowledge ) of the conformal chiral algebra , corresponding to the diagonal coset of type @xmath28 . it is nontrivial fact that the coset space ( [ coset ] ) has the same symmetry algebra for different @xmath29 and @xmath30 . it can be explained , probably , by the connection of the @xmath0algebra to the lie superalgebra @xmath18 , which has a free parameter unlike the other simple lie algebras . the @xmath0algebra contains two fields of spin @xmath2 and three fields of spin @xmath16 , making enough room for embedding of different subalgebras , such as the @xmath31 ( virasoro ) and the @xmath12 conformal algebras . besides the trivial @xmath12 subalgebra ( generated by the super energy momentum tensor ) there are 3 different @xmath12 superconformal subalgebras of the @xmath0algebra . these embeddings play a crucial role in the understanding of the representation theory of the algebra . there are four types of highest weight representations of the algebra : neveu schwarz ( ns ) , ramond and two twisted sectors . ( the twisted sectors are defined only in the case of vanishing coupling @xmath17 . ) the minimal models are labeled by two natural numbers : @xmath29 and @xmath30 . the ns and ramond minimal model representations can be arranged in the form of 3dimensional table , similarly to the 2dimensional tables of representations of the @xmath31 and the @xmath12 conformal algebras . the fusion rules also satisfy the `` @xmath32 pattern '' of the @xmath31 and @xmath12 minimal model fusions . the set of the @xmath5 algebra representations consists of 4 sectors : ns , ramond and two twisted . there are continuous spectrum representations in every sector . we prove , that the @xmath5 conformal algebra is the extended version of the @xmath11algebra at @xmath14 . due to this fact we get all the @xmath5 unitary representations from the known spectrum @xcite of the @xmath11algebra . the paper is organized as follows . after reviewing the structure of @xmath0 in section [ structure of the algebra ] we prove in section [ coset constr ] that the algebra is the symmetry algebra of the coset space ( [ coset ] ) . in section [ n=1 superconformal subalgebras ] we discuss different embeddings of the @xmath12 superconformal algebra into the @xmath0algebra and obtain the unitarity restrictions on the values of @xmath21 and @xmath17 . in section [ highest weight representations ] the highest weight representations of the algebra are introduced , the zero mode algebras in different sectors are discussed . in section [ minimal models ] we concern with the minimal models of the algebra : we explain how the spectrum of unitary representations is obtained , discuss the fusion rules , and give two examples of the @xmath0minimal models in terms of the @xmath33 superconformal minimal models . section [ g2 algebra ] is devoted to the @xmath5 algebra , the superconformal algebra associated to the manifolds of @xmath5 holonomy . we find it convenient to put some useful ( but in some cases lengthy ) information in the closed form in the five appendices . we have to note that substantial part of the calculations was done with a help of _ mathematica _ package @xcite for symbolic computation of operator product expansions . [ structure of the algebra ] here we review the structure of @xmath0algebra , which was first constructed in @xcite . the @xmath0algebra is an extension of @xmath12 superconformal algebra by two superconformal multiplets of dimensions @xmath34 and @xmath16 . a superconformal multiplet @xmath35 of dimension @xmath36 consists of two virasoro primary fields of dimensions @xmath36 and @xmath37 . under the action of the supersymmetry generator @xmath38 they transform as @xmath39 the @xmath0algebra multiplets are denoted by @xmath40 , @xmath41 ( @xmath42 ) , @xmath43 ( @xmath44 ) . the structure of the algebra is schematically given by @xmath45 where @xmath46 and the @xmath21 dependence of the coefficients is omitted . the explicit operator product expansions ( opes ) are fixed by the fusions ( [ algebra_structure ] ) and by @xmath12 superconformal invariance . we reproduce the opes in appendix [ apa ] . unitarity is introduced by the standard conjugation relation @xmath47 for any generator except @xmath48 . the commutation relation @xmath49 } = u_{n+m}\ , $ ] , following from ( [ gphi ] ) , requires the conjugation @xmath50 . [ coset constr ] we start from a few supporting arguments , that the coset theory ( [ coset ] ) indeed possesses the @xmath0superconformal symmetry . first , we note that formally @xmath51 and the coset ( [ coset ] ) can be written in the form of kazama suzuki coset @xcite @xmath52 , where @xmath53 and @xmath54 is its @xmath32 diagonal subalgebra . it means , that the chiral algebra contains @xmath12 superconformal algebra , obtained from the affine currents by a superconformal generalization of sugawara construction ( see ref . @xcite for details ) . all other currents , that constitute the chiral algebra should come in pairs of superpartners with difference of scaling dimensions equal to @xmath55 . the central charge ( [ c cpl c ] ) of the coset models ( [ coset ] ) has limiting point ( when @xmath56 ) @xmath57 . all the known examples of minimal series have limiting central charge @xmath58 , where @xmath59 and @xmath60 are the number of bosonic and fermionic fields in the correspondent chiral algebra . adopting the argument to our case we get , that the chiral algebra consists of three supercurrents ( including the super virasoro operator ) . the next argument follows from the simple observation @xcite , that if there is a sequence of subalgebra inclusions @xmath61 then the coset theory can be decomposed to the direct sum @xmath62 in the case of coset ( [ coset ] ) the inclusion sequence is @xmath63 with 3 different choices of @xmath64 : @xmath65 + @xmath66 @xmath67 the correspondent decompositions are : @xmath68 all three contain coset spaces of type ( [ coset n=1 ] ) with @xmath69 , @xmath70 and @xmath71 respectively . therefore the chiral algebra contain 3 different @xmath12 superconformal subalgebras ( not including the trivial one , generated by the super energy momentum tensor ) . this is possible if there are at least 3 operators of scaling dimension 2 . they have 3 superpartners of dimensions @xmath34 or @xmath72 . one field of dimension @xmath34 can not serve as a superconformal generator for 3 different superconformal subalgebras . the case , when all three are of dimension @xmath34 is also excluded , because then the algebra is trivially a sum of 3 commuting @xmath12 superconformal algebras . collecting the arguments we get that the only possibility that the chiral algebra consists of 6 fields of dimensions @xmath34 , @xmath34 , 2 , 2 , 2 , @xmath72 , which can be combined to three supercurrents of dimensions @xmath34 , @xmath34 , 2 , giving the @xmath0algebra . in this section we prove by explicit construction that the coset ( [ coset ] ) contains the @xmath0algebra . the method we use was first proposed in @xcite for coset ( [ coset n=1 ] ) . the @xmath32 affine algebra is generated by 3 currents @xmath73 , @xmath74 : @xmath75 the @xmath76 algebra consists of 3 commuting copies of the @xmath77 algebra at levels @xmath29 , @xmath30 and @xmath16 . the @xmath32 on level @xmath16 is realized by free fermions in the adjoint representation of @xmath32 : @xmath78 then the affine currents are expressed as @xmath79 the affine algebra @xmath80 is diagonally embedded in @xmath81 : @xmath82 the coset space @xmath83 contains operators , constructed from the @xmath76 currents , which commute with the currents of @xmath84 . the energy momentum tensor of the coset @xmath83 is given by the sugawara construction : @xmath85 where @xmath86 the general dimension@xmath2 operator can be written as @xmath87 it should commute with the @xmath88 currents . this requirement leads to condition @xmath89 one has two independent dimension@xmath2 operators ( @xmath38 and @xmath90 ) in the coset theory , since there are two free parameters @xmath91 and @xmath92 . in order to close the algebra one needs 3 more operators @xmath93 , @xmath94 , @xmath48 of scaling dimensions 2 , 2 , @xmath95 . the coset construction of all the 6 operators is given in appendix [ apb ] . we have explicitly checked , that the set of 6 operators @xmath96 , @xmath38 , @xmath90 , @xmath93 , @xmath94 , @xmath48 satisfy the opes of the @xmath0algebra with central charge @xmath21 and coupling @xmath17 given by ( [ c cpl cpl ] ) . [ n=1 superconformal subalgebras ] we start this section by observation , that 3 bosonic operators @xmath97 do not generate a closed subalgebra because of @xmath98 term in the ope of @xmath93 with @xmath94 ( [ opemw ] ) . we will construct in this section different @xmath12 superconformal subalgebras of @xmath0and discuss the unitarity restrictions . the most general dimension@xmath34 operator is and @xmath90 in the case of twisted sectors . ] @xmath99 we calculate its ope with itself : @xmath100 where @xmath101 and take the dimension2 operator @xmath102 as the virasoro generator of the subalgebra . these two operators , @xmath103 and @xmath102 , generate a closed subalgebra if the following two equations are satisfied : @xmath104 then the central charge of the subalgebra is @xmath105 . formally the lefthand side of the first equation should be multiplied by @xmath106 . we removed it in order to exclude the trivial solution @xmath107 , corresponding to the obvious @xmath12 subalgebra . the operator @xmath108 generates another closed subalgebra , namely virasoro algebra with central charge @xmath109 , and it is commutative with the @xmath12 superconformal subalgebra of @xmath103 and @xmath102 . the equations ( [ eq : alpha_beta ] ) are polynomial equations in @xmath110 and @xmath106 of orders 2 and 3 . generically there are 6 solutions . one should take into account the @xmath111 symmetry @xmath112 and @xmath113 of the equations , corresponding to the @xmath111 transformation of the @xmath12 superconformal algebra @xmath114 . so at any generic @xmath21 and @xmath17 there are three @xmath12 superconformal subalgebras . however , we are interested in real subalgebras , i.e. preserving the conjugation relation : the conjugation of the subalgebra @xmath115 should be consistent with the conjugation relations of the @xmath0algebra . this is true , if @xmath110 and @xmath106 are real . generically there can be 1 or 3 real solutions of ( [ eq : alpha_beta ] ) . unitary representations of an algebra are necessarily unitary representations of all its real subalgebras . the @xmath12 superconformal algebra has unitary representations only at @xmath116 or at @xmath117 ( [ c_k ] ) . the central charges of all the real @xmath12 superconformal subalgebras should be from this set . we study the solutions of the set of equations ( [ eq : alpha_beta ] ) in the region @xmath118 . ( at @xmath119 the coupling @xmath120 becomes imaginary . ) the results are presented in figure [ figure1 ] . the region @xmath121 is divided to two parts by the curve ( thick curve in figure [ figure1 ] ) @xmath122 in the region i ( @xmath123 ) there is one real solution of ( [ eq : alpha_beta ] ) , in the region iii ( @xmath124 ) there are 3 real solutions . the curves in figure [ figure1 ] have constant subalgebra central charge along it : @xmath125 . we call them the unitary curves . taking different @xmath12 subalgebras one gets different curves : @xmath126 all the region under discussion is spanned by the curves ( [ curvea ] , [ curveb ] ) . there are no real solutions of ( [ eq : alpha_beta ] ) corresponding to @xmath127 . in the region iii the unitarity is restricted to the intersections of the unitary curves ( the dots in figure [ figure1 ] ) . there are intersections of exactly 3 curves in every intersection point : two curves of type ( [ curvea ] ) with @xmath69 and @xmath70 and the third of type ( [ curveb ] ) with @xmath71 . the intersection points are given by ( [ c cpl c ] , [ c cpl cpl ] ) . the formula ( [ c cpl c ] ) is exactly the formula for central charge of coset theories ( [ coset ] ) . the central charges of three real subalgebras ( we will call them the first , the second and the third @xmath12 subalgebras ) at @xmath21 and @xmath17 at the intersection point are @xmath128 , @xmath129 , @xmath130 , they also coincide with the central charges of the three @xmath12 subalgebras of the coset ( [ coset ] ) ( see the decompositions ( [ decom2 ] , [ decom3 ] , [ decom1 ] ) ) . we conclude that all unitary models of the @xmath0algebra in the region iii are given by coset models ( [ coset ] ) . one can solve the equations ( [ eq : alpha_beta ] ) for the values of @xmath21 and @xmath17 from ( [ c cpl c ] , [ c cpl cpl ] ) to get the linear connection between the generators @xmath96 , @xmath93 , @xmath94 and the virasoro generators of the three subalgebras : @xmath131 in the region i the unitarity is restricted to the unitary curves ( [ curvea ] ) . the region i models with @xmath21 and @xmath17 satisfying ( [ curvea ] ) are expected to have continuous spectrum of unitary representations . on the separating curve ( [ divide curve ] ) the unitarity is restricted to the limiting points of ( [ c cpl c ] , [ c cpl cpl ] ) at one @xmath132 fixed and another @xmath132 taken to infinity . ( and @xmath133 , when both @xmath134 . ) the virasoro subalgebras , generated by @xmath108 , give no new restrictions on unitarity of the @xmath0algebra . the point @xmath14 , @xmath15 , which corresponds to the conformal algebra on @xmath5 manifold , is in the region i and lies on the ( [ curvea ] ) curve with @xmath135 . it means that the algebra have one real @xmath12 subalgebra and its central charge is @xmath136 in agreement with results of ref.@xcite . the generators of this real subalgebra are @xmath137 ( this is true for any @xmath15 model . ) the important question for understanding the structure of unitary representations of the @xmath0algebra is how the algebra is decomposed to the representations of its real @xmath12 subalgebras ? the decomposition is @xmath138 under subalgebras corresponding to the ( [ curvea ] ) curve and @xmath139 under subalgebras corresponding to the ( [ curveb ] ) curve , where @xmath140 is the vacuum representation of the @xmath12 superconformal algebra , @xmath141 and @xmath142 are its degenerate representations , having null vector on level @xmath2 . the @xmath102 , @xmath103 and @xmath96 fields are in the @xmath140 representation . three other fields of @xmath0form @xmath142 ( or @xmath141 ) representation , they can be understood in this context as @xmath142 ( spin@xmath2 field ) , @xmath143 ( spin@xmath16 field ) and @xmath144 ( spin@xmath95 field ) . @xmath145 is proportional to @xmath146 ( @xmath147 ) , since there is a null state on level @xmath2 . [ highest weight representations ] the @xmath0commutation relations admit two consistent choices of generator modes in the general case : ns and ramond sectors ; and two more then the coupling @xmath15 : first twisted ( tw1 ) and second twisted sectors ( tw2 ) . ns sector . : : the modes of the bosonic operators ( @xmath148 , @xmath149 , @xmath150 ) are integer ( @xmath151 ) and the modes of the fermionic operators ( @xmath152 , @xmath153 , @xmath154 ) are half integer ( @xmath155 ) . ramond sector . : : the modes of all the operators are integer . first twisted sector . : : the modes of @xmath148 , @xmath150 , @xmath156 operators are integer ( @xmath151 ) and the modes of @xmath152 , @xmath157 , @xmath154 operators are half integer ( @xmath155 ) . second twisted sector . : : the modes of @xmath148 , @xmath158 , @xmath150 , @xmath159 operators are integer ( @xmath151 ) and the modes of @xmath153 , @xmath157 operators are half integer ( @xmath155 ) . how can one understand the existence of four different sectors in terms of the coset construction ( appendix [ apb ] ) ? in order to get ns or ramond sectors of the algebra one should take all the three fermions of @xmath160 in ns or ramond sectors respectively . the modes of @xmath161 and @xmath162 currents are integer . the twisted sectors ( @xmath163 , since @xmath15 ) are obtained in less obvious way . first twisted sector : one takes one ramond fermion ( say , @xmath164 ) and two ns fermions ( @xmath165 and @xmath166 ) , the modes of @xmath167 , @xmath168 , @xmath169 are integer and the modes of @xmath170 , @xmath171 , @xmath172 are half integer . second twisted sector : one ns fermion ( @xmath166 ) and two ramond fermions ( @xmath164 and @xmath165 ) , the modes of @xmath170 , @xmath168 , @xmath172 are integer and the modes of @xmath167 , @xmath171 , @xmath169 are half integer . one can not define the modes of separate bosonic currents ( e.g. @xmath173 ) in the twisted sectors . the commutation relations of the @xmath0algebra include products of the generators . the formula ( [ noepsilon ] ) for the mode expansion of composite operators in various sectors is derived in appendix [ noexp ] . now are ready to define the highest weight representations in all sectors . the highest weight state is annihilated by positive modes of all generators : @xmath174 to deal with the zero modes one should discuss the sectors separately . there are 3 zero modes : @xmath175 , @xmath176 , @xmath177 . it is convenient to choose the highest weight state to be the eigenstate of these 3 operators and to label the highest weight representation by the correspondent eigenvalues . this is possible if the set of zero modes is commutative . one finds from ( [ opemw ] ) that the commutator @xmath178 } = \frac{9 \ , { \mu}}{2\ , c } ( m_0 + { { : } g h { :} } _ 0)\ ] ] is not zero . we rewrite the commutator by expanding @xmath179 in the modes of @xmath38 and @xmath90 : @xmath180 } = \frac{9 \ , { \mu}}{2\ , c } \sum_{r=1/2}^\infty \left ( g_{-r } \ , h_r - h_{-r } \ , g_r \right)\ ] ] the action of righthand side of ( [ commw ] ) on highest weight statevanishes . this is what one effectively needs in order to choose the highest weight state to be the eigenstate of both @xmath176 and @xmath177 . we define the notion of `` effective '' commutator : the commutation relation , which is true modulo terms , their action on highest weight state is zero . concluding : the `` effective '' commutators of all three zero modes vanish , and one can label the highest weight representation by three weights , the eigenvalues of the zero modes : @xmath181 one gets all states in the representation acting by negative modes on the highest weight state . there are 6 zero modes : @xmath175 , @xmath176 , @xmath177 , @xmath182 , @xmath183 , @xmath184 . since @xmath175 is commutative with all other zero modes , it can be represented by a number @xmath54 . the ( anti)commutation relations of other zero modes are @xmath185 } & = { \{g_0,u_0\}}=0\ , , \\ { \left[g_0,w_0\right ] } & = u_0 \ , , \\ { \{h_0,h_0\ } } \label{zmahh } & = 2\ , ( h - c/24)+ { \lambda}\ , m_0 + 4/3 \ , { \mu}\ , w_0 \ , , \\ { \left[h_0,m_0\right ] } \label{zmahm } & = 2/3 \ , { \mu}\ , u_0 \ , , \\ { \left[h_0,w_0\right ] } \label{zmahw } & = { \lambda}/2 \ , u_0 \ , , \\ { \left[m_0,w_0\right ] } \label{zmamw } & = { \{h_0,u_0\}}= \frac{9 \ , { \mu}}{4\ , c } \ , ( 2\ , g_0 \ , h_0 - m_0 ) \ , , \\ { \left[m_0,u_0\right ] } & = \frac{9 \ , { \mu}}{2\ , c } \ , \left ( 2\,(h - c/24)\ , h_0 -g_0 \ , m_0 \right ) \ , , \\ { \left[w_0,u_0\right ] } & = \frac{1}{8\,c}\ , \big ( 54 \ , { \mu}\ , u_0 - 54 \ , ( h - c/24 ) \ , \left ( 2\ , g_0 + { \lambda}\ , h_0 \right)\big . \nonumber \\ & \quad \big . -27 \ , { \lambda}\ , g_0 \ , m_0 - 72 \ , { \mu}\ , g_0 \ , w_0 + 54 \ , h_0 \ , m_0 \big ) \ , , \\ { \{u_0,u_0\ } } \label{zmauu } & = \frac{9}{4\,c } \ , \big ( -12\ , ( h - c/24)^2- ( h - c/24 ) \left(6\ , { \lambda}\,m_0 + 8\,{\mu}\ , w_0 \right ) \big . \nonumber \\ & \quad \big . + 4 \ , { \mu}\ , g_0 \ , u_0 + 3\ , m_0 \ , m_0 \big)\ , .\end{aligned}\ ] ] they define finite dimensional @xmath1 superalgebra . the commutation relations ( [ zmagg][zmahw ] ) are exact , and the commutators ( [ zmamw][zmauu ] ) are `` effective '' , i.e. modulo terms which annihilate highest weight states . the irreducible representations of the zero mode algebra are one dimensional or two dimensional and labeled by three weights : @xmath54 , @xmath186 and @xmath187 . in one dimensional representation the zero modes are given by @xmath188 such a representation exists only if the following condition is satisfied : @xmath189 taking @xmath190 one gets @xmath191 . such representation is called ramond ground state . in the limit @xmath192 the weight @xmath187 approaches @xmath193 like @xmath194 and then @xmath195 and @xmath196 . the definition of the two dimensional representations of the zero mode algebra ( [ zmagg][zmauu ] ) is more complicated . the bosonic zero modes @xmath175 , @xmath176 and @xmath177 can not be diagonalized simultaneously , since @xmath176 and @xmath177 do not commute , even `` effectively '' . one can label the highest weight representations by @xmath54 , @xmath197 and @xmath198 . in the following sections we will use another labels , but they will be always linearly dependent on the @xmath54 , @xmath186 , @xmath187 . the maximal set of commuting operators contains 3 operators , which can be chosen as following : @xmath175 , some fermionic operator @xmath199 and its square @xmath200 . in section [ g2 algebra ] it will be convenient to choose the @xmath199 operator as the zero mode of the @xmath12 subalgebra supersymmetry generator . there are 3 zero modes : @xmath175 , @xmath177 and @xmath183 . the zero mode algebra is obtained by setting @xmath15 in ( [ zmahh ] ) and ( [ zmahw ] ) : @xmath201 } & = 0 \ , .\end{aligned}\ ] ] its irreducible representations are one dimensional . the highest weight state is labeled by two weights : @xmath202 there are four zero modes : @xmath175 , @xmath177 , @xmath182 and @xmath184 . again @xmath175 is commutative with other zero modes and represented by its eigenvalue @xmath54 . the commutation relations are @xmath203 } & = u_0 \ , , \\ { \{g_0,u_0\ } } & = 0\ , , \\ { \left[w_0,u_0\right ] } & = \frac{1}{2\,c } \big ( 27\ , ( 9/48-(h - c/24))\ , g_0 + 9\,{\mu}\ , ( u_0 - 2\ , g_0 \ , w_0 ) \big ) \ , , \\ { \{u_0,u_0\ } } & = \frac{1}{c } \big ( 27\ , ( h - c/24)\,(9/48-(h - c/24 ) ) \big . \nonumber \\ & \quad \big . -18\ , { \mu}\,(h - c/24)\ , w_0 + 9\,{\mu}\,g_0\,u_0 \big ) \ , .\end{aligned}\ ] ] the last two commutators are `` effective '' . similarly to the ramond sector the irreducible representations of the zero mode algebra are one or two dimensional . they are labeled by two weights . in the one dimensional representation the zero modes are @xmath204 the representation exists only if @xmath205 the two dimension representation is constructed in the following way . one can choose to diagonalize @xmath175 and @xmath177 , and then the zero mode algebra is satisfied by @xmath206 g=(h - c/24)^{1/2 } , \qquad u = g\,(w_2-w_1 ) . { \end{array}}\ ] ] there is a connection between @xmath207 and @xmath208 : @xmath209 the two dimensional highest weight state is labeled by two weights : @xmath54 and @xmath207 . [ minimal models ] as we have shown in section [ n=1 superconformal subalgebras ] the existence of @xmath12 superconformal subalgebras restricts the values of @xmath21 and @xmath17 corresponding to unitary models of the @xmath0algebra . the unitary highest weight representations of an algebra are unitary with respect to all its real subalgebras . so there are also restrictions on the weights of unitary highest weight representations , coming from the non - unitarity theorem ( appendix [ appn=1 ] ) of the n=1 superconformal algebra . in the region iii of @xmath21 , @xmath17 values ( [ c cpl c ] , [ c cpl cpl ] ) there are 3 different @xmath12 superconformal subalgebras . the ns representation is labeled by three weights : @xmath54 , @xmath186 , @xmath187 , which are linear functions of three weights @xmath210 , @xmath211 , @xmath212 of the three @xmath12 subalgebras of central charge @xmath128 , @xmath129 , @xmath213 respectively . the connection between the two sets of weights is taken from ( [ h_to_ddd ] , [ m_to_ddd ] , [ w_to_ddd ] ) . the necessary condition for ns representation to be unitary is that the weights @xmath210 , @xmath211 , @xmath212 are included in the correspondent kac tables ( [ eq : n=1_dimensions ] ) of conformal dimensions of the @xmath12 superconformal algebra . therefore there is a finite number of unitary highest weight representations in the region iii models , and we can call them the minimal models of the @xmath0algebra . the highest weight representation of the @xmath0algebra can be decomposed to the sum of representations of the @xmath12 subalgebra . let s take for example the third subalgebra ( @xmath214 ) . as we have shown in section [ n=1 superconformal subalgebras ] the generators of the @xmath0algebra fall to the @xmath215 and @xmath216 representations of the third subalgebra . the fusion rule for @xmath216 is @xmath217 therefore the highest weight state of the @xmath0algebra with @xmath218 has descendants lying in the @xmath219 , @xmath220 , representation of the third subalgebra , and the @xmath0highest weight representation is decomposed to the sum of the third subalgebra representations with dimensions from the same row of the kac table . applying these conclusions to the ramond sector one gets that @xmath221 in the two dimensional representation looks like @xmath222 in the basis , where it is diagonal ; @xmath223 is the dimension of ramond representation of the @xmath12 superconformal algebra . although the zero modes of the three @xmath12 subalgebras @xmath224 , @xmath225 , @xmath221 , can not be diagonalized simultaneously , one still can label the ramond representation by 3 pairs of ramond dimensions ( nearest ramond neighbors in the correspondent kac tables ) . taking trace of zero modes in ( [ h_to_ddd ] ) one obtains the conformal dimension @xmath226 for the twisted sectors the situation is different . the twisted representations exist only in the minimal models with @xmath227 . since one can not mix @xmath38 and @xmath90 generators in the twisted sector , the only @xmath12 subalgebra is the third subalgebra , its generators are given by ( [ gsub_cpl_zero ] ) and ( [ tsub_cpl_zero ] ) . the two weights of the highest weight representation in the tw1 sector can be chosen to be the @xmath221 eigenvalue @xmath212 and the conformal dimension @xmath54 . the tw1 representation is of ramond type with respect to the third @xmath12 subalgebra . the conditions ( [ cond_1dim_tw2 ] ) for existence of the one dimensional representation in the tw2 sector of the @xmath227 minimal model are rewritten as @xmath228 the two dimensional representation of the tw2 type is labeled by two weights : @xmath54 and a pair @xmath229 of the nearest ns dimensions as the eigenvalues of the @xmath221 operator . the connection ( [ connec w1_w2 ] ) , being rewritten in terms of @xmath230 and @xmath231 , states exactly that @xmath230 and @xmath231 are nearest ns neighbors in the row of the correspondent kac table . all the conditions of unitarity so discussed are not sufficient . in the case @xmath232 one gets additional restrictions on unitary representations by noting that the energy momentum tensor can be decomposed to two commuting parts @xmath233 where @xmath234 is the generator of the virasoro algebra of central charge @xmath235 corresponding to the unitary minimal models of the virasoro algebra @xcite . ( one could see the decomposition from ( [ decom3 ] ) and ( [ decom1 ] ) . ) this fact restricts the values of @xmath236 and @xmath237 . for the @xmath232 models all the discussed restrictions on the weights are in fact sufficient conditions of unitarity of ns and ramond representations . ( we do not prove it here . ) the discussion for @xmath232 applies to the @xmath238 case as well . by taking formally @xmath239 one should obtain the minimal models of @xmath12 superconformal algebra ( appendix [ appn=1 ] ) . in addition we know the examples ( see section [ examples ] ) of explicit construction of @xmath240 and @xmath241 minimal models of the @xmath0algebra in terms of @xmath33 models . based on these facts and with the help of the coset construction ( [ coset ] ) we guess the unitary spectrum of the general ( arbitrary @xmath29 and @xmath30 ) @xmath0minimal model . the full list of minimal model unitary representations is presented in appendix [ list rep ] . the unitarity was also checked by computer calculations of the @xmath0algebra kac determinant on the few first levels . the list of ns and ramond representations forms a three dimensional table with indices @xmath242 , running in the range ( [ s range ] ) . the twisted sector representations form a two dimensional table with indices @xmath243 and @xmath244 ( [ t range ] ) . there is a same number of ns and ramond representations and the same number of tw1 and tw2 representations . substantial part of the spectrum could be predicted using the magic relation between the dimensions of any @xmath12 minimal model ( [ eq : n=1_dimensions ] ) : @xmath245 the relation is to be understood in the context of the fusion rule @xmath246 taking @xmath247 we get , that @xmath248 fields are local or semilocal with respect to @xmath249 . the @xmath0algebra is decomposed to @xmath250 representation of the first @xmath12 subalgebra . therefore the field @xmath248 ( of the first @xmath12 subalgebra ) is a valid representation of the whole @xmath0algebra , since it is local ( or semilocal ) with respect to all the generators of @xmath0 . this representation is of ramond or ns type , depending on @xmath251 is even or odd respectively . the conformal dimension @xmath54 of such a field coincides with the weight of the first @xmath12 subalgebra . for @xmath252 the @xmath248 field lies in the highest weight representation of the @xmath0algebra , for @xmath253 it is descendantof some highest weight representation . we call such a representation the purely internal representation with respect to the first @xmath12 subalgebra . obviously the set of purely internal representations is closed under fusion rules . ( the similar situation is encountered in the case of the @xmath11algebra , which have purely internal representations with respect to its virasoro subalgebra ( section 4 of @xcite ) . ) of course , there are also purely internal representations with respect to the second and to the third @xmath12 subalgebras . the representations @xmath254 , @xmath255 , @xmath256 are purely internal of the first , second and third subalgebras respectively . such representations have simple fusion rules . it is interesting to note , that @xmath257 fields of the @xmath0algebra are local or semilocal with respect to the @xmath258 field . it is well known that the fusion rules of ns and ramond sector representations have @xmath111 grading : ( ns @xmath259 , ramond @xmath260 under addition modulo 2 ) . the @xmath0algebra has two additional twisted sectors : tw1 and tw2 . the full set of fusion rules has @xmath261 grading : ns @xmath262 , ramond @xmath263 , tw1 @xmath264 , tw2 @xmath265 . the fusion rules of different sectors are summarized in the table : [ cols="^,^,^,^,^",options="header " , ] the fusions of the @xmath0representations have to be consistent with fusion rules of its subalgebras . in the case of minimal models there are three @xmath12 superconformal subalgebras and the fusions of the @xmath0ns and ramond representations are completely fixed by the fusions of the correspondent @xmath12 minimal models ( [ eq : n=1_fusions ] ) . the fusion of @xmath266 and @xmath267 representation ( see appendix [ list rep ] ) of the @xmath0algebra is @xmath268 where @xmath269 , @xmath270 and @xmath271 are raised by steps of 2 . the selection of one index is independent on two others and satisfies the `` @xmath32 pattern '' . the fusion rules of the @xmath31 and the @xmath12 minimal models satisfy the same pattern , the only difference is that in our case the table of representations is three dimensional . the proof is based on the first column of formula ( [ m n min_mod ] ) . the @xmath272 representation is decomposed to the sum of representations of the first @xmath12 subalgebra from the column number @xmath269 of the correspondent @xmath12 kac table . thus the column selection rule of the first @xmath12 subalgebra is preserved and coincides with the @xmath269 selection in ( [ eq : sww_fusions ] ) . similarly the @xmath270 and @xmath271 selection rules are adopted from the column and row selection rules of the second and the third @xmath12 subalgebras respectively . the twisted representations of the minimal models are labeled by two numbers : @xmath243 and @xmath244 ( appendix [ list rep ] ) . but there is only one @xmath12 subalgebra ( the third one ) in the twisted sector . one can reed from ( [ list tw reps ] ) that @xmath243 is the row number in the correspondent kac table , meaning that @xmath243 has common selection rules with @xmath271 . the selection of @xmath244 in the fusion rules can not be fixed by the described methods . the `` corner '' entries of the three dimensional table of ns and ramond representations are the @xmath273 , @xmath274 , @xmath275 , @xmath276 representations . the first one is the vacuum representation . the three others have the following fusion `` square '' : @xmath277 where @xmath278 denotes the identity ( vacuum ) representation . if such a field @xmath279 is of the ramond type , then the fusion of it with the other fields defines one - to - one transformation , mapping ns fields to ramond ones . ( this is an analogy of the @xmath280 flow of the @xmath33 superconformal algebra @xcite . ) if the field @xmath279 is of the ns type , then its conformal dimension @xmath54 is integer or half integer and the @xmath0algebra can be extended to include this field . in the case , then both @xmath29 and @xmath30 are even , all three `` corner '' fields are of the ns type ; in other cases ( at least one @xmath132 is odd ) one `` corner '' fields is of the ns type and two are of the ramond type ( and then there are two different ns r isomorphisms ) . the following two examples are the @xmath281 , @xmath15 ( @xmath240 ) and @xmath282 , @xmath15 ( @xmath241 ) minimal models of the @xmath0algebra . the former model is realized as @xmath283 orbifold of a tensor product of free fermion and free boson on radius @xmath284 , the latter is realized as @xmath283 orbifold of the sixth ( @xmath282 ) minimal model of the @xmath285 superconformal algebra . the boson on radius @xmath284 is equivalent to the first minimal model of the @xmath285 superconformal algebra . the generators of @xmath0are constructed in the following way : @xmath286 where @xmath287 , @xmath288 , @xmath5 and @xmath289 are the ( real ) generators of the @xmath285 superconformal algebra at @xmath290 , @xmath291 is the free fermion field and @xmath292 is its energy momentum operator . the expressions in ( [ c32 ] ) are invariant under @xmath111 transformation . one can build all the highest weight representations in the model ( 5 in ns , 5 in ramond , 3 in every twisted sector ) by appropriate combinations of representations of the @xmath285 minimal model and the ising model . the list of representations is presented in table [ table_k_1=k_2=1 ] in appendix [ list rep ] . the ns sector of the sixth ( @xmath282 ) @xmath285 minimal model contains highest weight state @xmath293 of conformal dimension @xmath2 and zero @xmath280 charge . the @xmath285 superconformal algebra can be extended by @xmath285 superprimary field corresponding to this state @xcite . then the fields , invariant under @xmath111 transformation form the @xmath0algebra : @xmath294 where again @xmath287 , @xmath288 , @xmath5 and @xmath289 are the ( real ) generators of the @xmath285 superconformal algebra and @xmath295 is a superpartner of @xmath296 : @xmath297 . the highest weight representations of the @xmath285 minimal model can be easily transformed to representations of the @xmath0algebra . the list of highest weight representations consists of 16 ns , 16 ramond , 6 tw1 and 6 tw2 representations ( too long to be reproduced here explicitly ) . in this section we discuss the @xmath0algebra at @xmath298 . as we have shown in section [ n=1 superconformal subalgebras ] there is one real @xmath12 subalgebra . it has central charge @xmath299 and thus coincides with the tricritical ising model . the subalgebra is generated by operators @xmath103 ( [ gsub_cpl_zero ] ) and @xmath102 ( [ tsub_cpl_zero ] ) . the @xmath0algebra is decomposed to the @xmath250 representation of the @xmath12 subalgebra . since at @xmath299 the fields @xmath300 are identical there is a new null state on level @xmath16 : @xmath301 ( the null state on level @xmath2 is already encoded in the structure of the @xmath0algebra , but the existence of the null state ( [ null state ] ) is a special feature of the tricritical ising model . ) translating ( [ null state ] ) to the language of generators of the @xmath0algebra one gets that the null field is @xmath302 this is an ideal of the @xmath0algebra at @xmath298 . ( the existence of the ideal is known since @xcite . ) in ref . @xcite the conformal algebra associated to the manifolds of @xmath5 holonomy is derived . up to the ideal ( [ eq : ideal ] ) the @xmath5 conformal algebra coincides with the @xmath0algebra at @xmath298 . in the free field representation , used by the authors of @xcite to obtain the @xmath5 algebra , the ideal ( [ eq : ideal ] ) vanishes identically . the authors of ref . @xcite used different basis of generators of the algebra . their basis is connected to ours by @xmath303 the @xmath96 and @xmath38 generators are the same . we have explicitly checked , that the opes in the first appendix of @xcite coincide ( up to the ideal ( [ eq : ideal ] ) ) with the opes of the @xmath0algebra . some unitary highest weight representations of the @xmath5 algebra are found in @xcite . in this section we complete the list of unitary representations . our calculation is based on the fact , that the @xmath304 fields of the @xmath5 algebra generate closed subalgebra modulo the same ideal ( [ eq : ideal ] ) and its descendants . this subalgebra is the @xmath11superconformal algebra @xcite of central charge @xmath9 . the @xmath5 algebra can be seen as an extended version of the @xmath11algebra . it is interesting to note that the @xmath11algebra at another value of central charge ( @xmath13 ) is the superconformal algebra associated to manifolds of @xmath7 holonomy @xcite . the complete spectrum of unitary representations of the @xmath11algebra is found in @xcite . the @xmath14 unitary model spectrum is presented in table [ sw spectrum ] , where @xmath305 stands for real positive number in the continuous spectrum representations . @xmath306 there are ns and ramond representations , they are labeled by two numbers : the conformal dimension @xmath54 and the internal dimension @xmath307 ( the weight of the @xmath308 virasoro subalgebra of the @xmath11algebra ) . the @xmath90 field of the @xmath0algebra is identified with the @xmath309 ns representation of the @xmath11algebra , @xmath93 is its superpartner . the @xmath310 supermultiplet is purely internal with respect to the @xmath308 virasoro subalgebra of the @xmath11algebra . due to this fact we know its fusion rules with all other representations of the @xmath11algebra . in order to get the representations of @xmath0one have to combine the table [ sw spectrum ] representations in multiplets under the action of the @xmath311 field . it can be easily done , however this way one gets only two weights . but we know that the @xmath0ns and ramond representations are labeled by three weights . it is convenient to choose them : the conformal dimension @xmath54 ; the weight of the @xmath12 subalgebra , which coincides with the @xmath11algebra internal dimension @xmath307 ; and the eigenvalue @xmath187 of the @xmath93 zero mode . ( in the case of two dimensional ramond representation @xmath187 stands for @xmath312 . ) the twisted representations are labeled by two weights : @xmath54 and @xmath307 . one obtains the third weight @xmath187 by using again the null field ( [ eq : ideal ] ) . acting by it on the highest weight state one should get zero . consider , for example , the ns sector . take the @xmath313 descendant of the ideal : @xmath314 the eigenvalue of its zero mode should be set to zero for a consistent representation . this leads to a connection between @xmath187 and two other weights @xmath54 and @xmath307 : @xmath315 the similar connection can be found in the ramond case . concluding , the unitary representations of the @xmath5 algebra are ( again @xmath316 ) : + ns sector : @xmath317 ramond sector : @xmath318 tw1 sector : @xmath319 tw2 sector : @xmath320 the ns and ramond discrete spectrum states ( the first two in ( [ eq : g2 spectrum ns ] ) and ( [ eq : g2 spectrum r ] ) ) were found in @xcite . the first ramond representation ( @xmath321 ) is purely internal with respect to the tricritical ising model . due to this fact we know its fusions . since its square is identity , the field serves as an isomorphism mapping , connecting ns and ramond sectors , and connecting the tw1 and tw2 sectors . the representations of the same line number in ( [ eq : g2 spectrum ns ] ) and ( [ eq : g2 spectrum r ] ) and in ( [ eq : g2 spectrum tw1 ] ) and ( [ eq : g2 spectrum tw2 ] ) are isomorphic . in this paper we study the @xmath0superconformal algebra . we show by explicit construction that the coset ( [ coset ] ) contains the @xmath0algebra . the space of parameters ( @xmath21 and @xmath17 ) is divided to two regions ( figure [ figure1 ] ) . in the first region there are unitary models at discrete points in the @xmath322 space . these are the minimal models of the algebra , they are described by the coset ( [ coset ] ) . in this region the @xmath0algebra has three different nontrivial @xmath12 subalgebras . the conformal dimensions with respect to these subalgebras serve as the weights of the @xmath0highest weight representations . the fusion rules are also dictated by the fusions of the @xmath12 subalgebras . the characters of highest weight representations are not discussed in this paper . we suppose that the characters can be easily obtained from the coset construction . in the second region of parameters the @xmath0algebra has one @xmath12 superconformal subalgebra . the unitary models `` lie '' on unitary curves and have continuous spectrum of unitary representations . one of the continuous unitary models is the @xmath14 , @xmath15 model . the @xmath0algebra at these values of parameters coincides ( up to a null field ) with the superconformal algebra , associated to the manifolds of @xmath5 holonomy . from the other point of view it is an extended version of the @xmath11algebra , which at another value of the central charge ( @xmath13 ) corresponds to the manifolds of @xmath7 holonomy . we find the unitary spectrum of the @xmath5 holonomy algebra . the connection of various realizations of the @xmath5 superconformal algebra with the geometric properties of the @xmath5 manifolds is the open problem for study . @xmath96 and @xmath38 generate @xmath12 superconformal algebra of central charge @xmath21 . @xmath310 and @xmath323 are its superprimary fields . the nontrivial opes are : @xmath324 where @xmath325 and the fields in the right hand sides of the opes are taken in the point @xmath186 . at @xmath326 all unitary representations of the @xmath12 superconformal algebra are described by its minimal models . their central charge is @xmath327 the conformal dimensions of the unitary highest weight representations @xmath328 of the @xmath329 minimal model are given in the kac table @xmath330 the fusion rules are given by @xmath32 like selection rules for every index ( @xmath187 and @xmath251 ) : @xmath331 where the indices @xmath187 and @xmath251 in the sums are raised by steps of 2 . here we present the complete list of unitary highest weight representations of the @xmath0minimal models . the central charge @xmath21 and the coupling @xmath17 of the @xmath334 minimal model are given in ( [ c cpl c ] , [ c cpl cpl ] ) . the list of ns and ramond sector representations can be presented in the form of three dimensional table with indices @xmath269 , @xmath270 , @xmath271 : @xmath335 the representation @xmath336 is of ns or ramond type depending on @xmath337 is odd or even respectively . the highest weight representation is labeled by 3 weights @xmath210 , @xmath211 , @xmath212 , the conformal dimensions with respect to the three @xmath12 superconformal subalgebras . their values are taken from the correspondent @xmath12 kac tables ( [ eq : n=1_dimensions ] ) : @xmath338 where the indices are connected to @xmath269 , @xmath270 , @xmath271 by @xmath339 @xmath340 , @xmath341 , @xmath342 are values of the @xmath343 function : @xmath344 we define the function @xmath343 by its graph : the number @xmath345 in ( [ m n min_mod ] ) can be 0 or 1 . it is 0 in the ns sector ( @xmath337 odd ) . in the ramond sector it is given by @xmath346 where @xmath347 is a derivative of @xmath343 with respect to @xmath305 . the @xmath348 function is not continuous , but the values in the points of discontinuity are not important . @xmath345 distinguishes between one ( @xmath349 ) and two dimensional ( @xmath350 ) ramond representations . the conformal dimension @xmath54 ( the eigenvalue of @xmath175 operator ) is calculated from @xmath210 , @xmath211 , @xmath212 weights by @xmath351 in the case of one dimensional ( ns or ramond ) representation . in the case of two dimensional ramond representation the @xmath210 , @xmath211 , @xmath212 in ( [ h ddd ] ) should be substituted by half of the sum of the correspondent @xmath12 ramond dimensions ( [ h ddd r ] ) . the @xmath227 minimal models contain two additional twisted sectors : tw1 and tw2 . the list of tw1 and tw2 representations can be arranged in the two dimensional table with indices @xmath243 and @xmath244 : @xmath354 the @xmath355 even entries are of tw2 type and the @xmath355 odd entries are of tw1 type . the representations in the twisted sectors are labeled by the conformal dimension @xmath54 and the weight of the third @xmath12 subalgebra @xmath212 : @xmath356 { d^{(3)}}&= & { \displaystyle}{d^{(2\,k)}}_{m , n}\ , , \quad \begin{aligned } m&=t_1 , \\ n&=t_1+\text{sgn}(t_2-t_1 ) \pm r , \end{aligned } & \quad & r&=&\left\ { \begin{array}{ll } 0 , & \quad\text{tw1},\\ \text{sgn}(t_2-t_1 ) , & \quad\text{tw2}. \end{array } \right . { \end{array}}\ ] ] again @xmath345 distinguishes between one ( @xmath349 ) and two dimensional ( @xmath350 ) tw2 representations . there are @xmath357 tw1 representations and the same number of tw2 representations . we want to illustrate the formulas of the present appendix by some explicit examples . the simplest model is the @xmath240 ( @xmath358 ) model , discussed in section [ k1_k2_1 ] . its highest weight representations are presented in table [ table_k_1=k_2=1 ] . we use the @xmath359 { d^{(1)}},{d^{(2)}},{d^{(3 ) } } \end{array } \right)}$ ] notation for the ns and ramond representations and the @xmath359 { d^{(3 ) } } \end{array } \right)}$ ] notation for the twisted sectors . one of the four ns / ramond weights is dependent on other three and is presented for convenience only . the ramond and tw1 sectors are _ slanted_. the @xmath360 ramond representation is two - dimensional . @xmath361 { \begin{array}{c } s_3\\ \downarrow \vspace{6 cm } \end{array } } { \setlength\arraycolsep{2pt } \begin{array}{|cc| } \hline { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } 0 \\[4pt ] 0,0,0 \end{array } \right ) } & { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } { \it{\frac{7}{16 } } } \\[4pt ] { \it{\frac{3}{80},\frac{7}{16 } , \frac{3}{8 } } } \end{array } \right ) } \\[7pt ] { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } { \it{\frac{1}{16 } } } \\[4pt ] { \it{\frac{3}{80},\frac{3}{80},\frac{1}{16 } } } \end{array } \right ) } & { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } \frac{1}{8 } \\[4pt ] 0,\frac{1}{10},\frac{1}{16 } \end{array } \right ) } \\[7pt ] { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } \frac{1}{6 } \\[4pt ] \frac{1}{10},\frac{1}{10},\frac{1}{6 } \end{array } \right ) } & { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } { \it{\frac{5}{48 } } } \\[4pt ] { \it{\frac{3}{80},\frac{3}{80},\frac{1}{24 } } } \end{array } \right ) } \\[7pt ] { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } { \it{\frac{9}{16 } } } \\[4pt ] { \it { { \frac{3}{80 } \atop \frac{7}{16 } } , { \frac{3}{80 } \atop \frac{7}{16 } } , { \frac{1}{16 } \atop \frac{9}{16 } } } } \end{array } \right ) } & { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } \frac{1}{8 } \\[4pt ] \frac{1}{10},0 , \frac{1}{16 } \end{array } \right ) } \\[7pt ] { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } \frac{1}{2 } \\[4pt ] \frac{1}{10},\frac{1}{10},0 \end{array } \right ) } & { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } { \it{\frac{7}{16 } } } \\[4pt ] { \it{\frac{7}{16},\frac{3}{80},\frac{3}{8 } } } \end{array } \right)}\\[17pt ] \hline \end{array } } \\ \\[-8pt ] \phantom{s_3 s_3 } s_1=1 \end{array } \,\begin{array}{c } \phantom{s_3 s_3 s_3 } s_2 \rightarrow \hfill\\[2pt ] { \begin{array}{c } s_3\\ \downarrow \vspace{6 cm } \end{array } } { \setlength\arraycolsep{2pt } \begin{array}{|cc| } \hline { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } { \it{\frac{7}{16 } } } \\[4pt ] { \it{\frac{7}{16},\frac{3}{80},\frac{3}{8 } } } \end{array } \right ) } & { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } \frac{1}{2 } \\[4pt ] \frac{1}{10},\frac{1}{10},0 \end{array } \right ) } \\[7pt ] { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } \frac{1}{8 } \\[4pt ] \frac{1}{10},0,\frac{1}{16 } \end{array } \right ) } & { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } { \it{\frac{9}{16 } } } \\[4pt ] { \it { { \frac{3}{80 } \atop \frac{7}{16 } } , { \frac{3}{80 } \atop \frac{7}{16 } } , { \frac{1}{16 } \atop \frac{9}{16 } } } } \end{array } \right ) } \\[7pt ] { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } { \it{\frac{5}{48 } } } \\[4pt ] { \it{\frac{3}{80},\frac{3}{80},\frac{1}{24 } } } \end{array } \right ) } & { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } \frac{1}{6 } \\[4pt ] \frac{1}{10},\frac{1}{10 } , \frac{1}{6 } \end{array } \right ) } \\[7pt ] { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } \frac{1}{8 } \\[4pt ] 0,\frac{1}{10},\frac{1}{16 } \end{array } \right ) } & { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } { \it{\frac{1}{16 } } } \\[4pt ] { \it{\frac{3}{80},\frac{3}{80},\frac{1}{16 } } } \end{array } \right ) } \\[7pt ] { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } { \it{\frac{7}{16 } } } \\[4pt ] { \it{\frac{3}{80},\frac{7}{16},\frac{3}{8 } } } \end{array } \right ) } & { \rule{0cm}{28pt } \left ( \setlength\arraycolsep{0pt } \begin{array}{c } 0 \\[4pt ] 0,0,0 \end{array } \right ) } \\[17pt ] \hline \end{array } } \\ \\[-8pt ] \phantom{s_3 s_3 } s_1=2 \end{array } \quad \ ] ] the second example is the @xmath363 ( @xmath364 ) minimal model . since the list of representations is too long , we reproduce only the conformal dimensions @xmath54 of the highest weight representations ( table [ k_1=2_k_2=3 ] ) . @xmath365 { \begin{array}{c } s_3\\ \downarrow \vspace{4.2 cm } \end{array } } \setlength\arraycolsep{2pt } \begin{array}{|cccc| } \hline \rule{0cm}{15pt } { 0 } & { \it{\frac{27}{80 } } } & { \frac{9}{10 } } & { \it{\frac{31}{16 } } } \\[7pt ] { \it { \frac{5}{48 } } } & { \frac{1}{15 } } & { \it{\frac{121}{240 } } } & { \frac{7}{6 } } \\[7pt ] { \frac { 5}{18 } } & { \it{\frac{83}{720 } } } & { \frac{8}{45 } } & { \it{\frac{103}{144 } } } \\[7pt ] { \it{\frac { 37}{48 } } } & { \frac{7}{30 } } & { \it{\frac{41}{240 } } } & { \frac{1}{3 } } \\[7pt ] { \frac{5}{6 } } & { \it{\frac{161}{240 } } } & { \frac{7}{30 } } & { \it{\frac{13}{48 } } } \\[7pt ] { \it{\frac{175}{14 4 } } } & { \frac{61}{90 } } & { \it{\frac{443}{720 } } } & { \frac{5}{18 } } \\[7pt ] { \frac{7}{6 } } & { \it{\frac{241}{240 } } } & { \frac{17}{30 } } & { \it{\frac{29}{48 } } } \\[7pt ] { \it{\frac{23}{16 } } } & { \frac{9}{10 } } & { \it{\frac{67}{80 } } } & { 1 } \\[7pt ] \hline \end{array } \\ \\[-8pt ] \phantom{s_3 s_3 } s_1=1 \end{array } \quad \begin{array}{c } \phantom{s_3 s_3 s_3 s_3 } s_2 \rightarrow \hfill\\[2pt ] { \begin{array}{c } s_3\\ \downarrow \vspace{4.2 cm } \end{array } } \setlength\arraycolsep{2pt } \begin{array}{|cccc| } \hline \rule{0cm}{15pt } { \it{\frac{3}{8 } } } & { \frac{27}{80 } } & { \it{\frac{31}{40 } } } & { \frac{23}{16 } } \\[7pt ] { \frac{5}{48 } } & { \it{\frac{53}{120 } } } & { \frac{121}{240 } } & { \it{\frac{25}{24 } } } \\[7pt ] { \it{\frac{11}{72 } } } & { \frac{83}{720 } } & { \it{\frac{199}{360 } } } & { \frac{103}{1 44 } } \\[7pt ] { \frac{13}{48 } } & { \it{\frac{13}{120 } } } & { \frac{41}{240 } } & { \it{\frac{17 } { 24 } } } \\[7pt ] { \it{\frac{17}{24 } } } & { \frac{41}{240 } } & { \it{\frac{13}{120 } } } & { \frac{13 } { 48 } } \\[7pt ] { \frac{103}{144 } } & { \it{\frac{199}{360 } } } & { \frac{83}{720 } } & { \it{\frac{1 1}{72 } } } \\[7pt ] { \it{\frac{25}{24 } } } & { \frac{121}{240 } } & { \it{\frac{53}{120 } } } & { \frac { 5}{48 } } \\[7pt ] { \frac{23}{16 } } & { \it{\frac{31}{40 } } } & { \frac{27}{80 } } & { \it{\frac{3 } { 8 } } } \\[7pt ] \hline \end{array } \\ \\[-8pt ] \phantom{s_3 s_3 } s_1=2 \end{array } \quad \begin{array}{c } \phantom{s_3 s_3 s_3 s_3 } s_2 \rightarrow \hfill\\[2pt ] { \begin{array}{c } s_3\\ \downarrow \vspace{4.2 cm } \end{array } } \setlength\arraycolsep{2pt } \begin{array}{|cccc| } \hline \rule{0cm}{15pt } { 1 } & { \it{\frac{67}{80 } } } & { \frac{9}{10 } } & { \it{\frac{23}{16 } } } \\[7pt ] { \it{\frac{29}{48 } } } & { \frac{17}{30 } } & { \it{\frac{241}{240 } } } & { \frac{7}{6 } } \\[7pt ] { \frac{5}{18 } } & { \it{\frac{443}{720 } } } & { \frac{61}{90 } } & { \it{\frac{175}{144 } } } \\[7pt ] { \it{\frac{13}{48 } } } & { \frac{7}{30 } } & { \it{\frac{161}{240 } } } & { \frac{5}{6 } } \\[7pt ] { \frac{1}{3 } } & { \it{\frac{41}{240 } } } & { \frac{7}{30 } } & { \it{\frac{37}{48 } } } \\[7pt ] { \it{\frac{103}{144 } } } & { \frac{8}{45 } } & { \it{\frac{83}{720 } } } & { \frac{5}{18 } } \\[7pt ] { \frac{7}{6 } } & { \it{\frac{121}{240 } } } & { \frac{1}{15 } } & { \it{\frac{5}{48 } } } \\[7pt ] { \it{\frac{31}{16 } } } & { \frac{9}{10 } } & { \it{\frac{27}{80 } } } & { 0 } \\[7pt ] \hline \end{array } \\ \\[-8pt ] \phantom{s_3 s_3 } s_1=3 \end{array}\ ] ] here we derive the formula for the mode expansion of normal ordered product of operators in various sectors . the normal ordered product @xmath366 is defined as the zero order term in ope : @xmath367 the well known formula for the mode expansion of @xmath366 @xmath368 is valid only if @xmath187 has the same modding as @xmath369 , i.e. @xmath187 runs on integer or half integer numbers depending on the spin of @xmath370 is integer or half integer respectively . so in the ns sector the expansion ( [ noexpansion ] ) works . in the case of ramond or twisted sectors the formula should be modified . the idea of the following calculation is taken from ref.@xcite ( section 3 ) , where the mode expansion of @xmath371 was obtained using the same method . ( @xmath372 and @xmath373 are the supersymmetry generators of the @xmath285 superconformal algebra . ) let s calculate the integral : @xmath374 where the first integration is around @xmath186 and the second is around @xmath193 . the integral ( [ nointegral ] ) is equal to the @xmath251mode @xmath375 expanding @xmath380 and integrating one gets @xmath381 where @xmath187 runs on @xmath382 . @xmath383 is usually chosen to be @xmath193 or @xmath55 to produce the correct modding for operator @xmath370 . in the case @xmath384 we get back the formula ( [ noexpansion ] ) as expected . note that the formula ( [ noepsilon ] ) is valid for any @xmath383 ( not only 0 or @xmath55 ) consistent with the chosen modding . s. l. shatashvili and c. vafa , `` superstrings and manifold of exceptional holonomy , '' selecta math . * 1 * ( 1995 ) 347381 [ arxiv : hep - th/9407025 ] . + s. l. shatashvili and c. vafa , `` exceptional magic , '' nucl . * 41 * ( 1995 ) 345 . a. b. zamolodchikov , `` infinite additional symmetries in two - dimensional conformal quantum field theory , '' theor . * 65 * ( 1985 ) 1205 . p. bouwknegt and k. schoutens , `` w symmetry in conformal field theory , '' phys . * 223 * ( 1993 ) 183 [ arxiv : hep - th/9210010 ] . j. m. figueroa - ofarrill , `` a note on the extended superconformal algebras associated with manifolds of exceptional holonomy , '' phys . b * 392 * ( 1997 ) 77 [ arxiv : hep - th/9609113 ] . d. gepner and b. noyvert , `` unitary representations of sw(3/2,2 ) superconformal algebra , '' nucl . b * 610 * ( 2001 ) 545 [ arxiv : hep - th/0101116 ] . t. eguchi and y. sugawara , `` cft description of string theory compactified on non - compact manifolds with g(2 ) holonomy , '' phys . b * 519 * ( 2001 ) 149 [ arxiv : hep - th/0108091 ] . k. sugiyama and s. yamaguchi , `` cascade of special holonomy manifolds and heterotic string theory , '' nucl . b * 622 * ( 2002 ) 3 [ arxiv : hep - th/0108219 ] . r. blumenhagen and v. braun , `` superconformal field theories for compact g(2 ) manifolds , '' jhep * 0112 * ( 2001 ) 006 [ arxiv : hep - th/0110232 ] . r. roiban and j. walcher , `` rational conformal field theories with g(2 ) holonomy , '' jhep * 0112 * ( 2001 ) 008 [ arxiv : hep - th/0110302 ] . t. eguchi and y. sugawara , `` string theory on g(2 ) manifolds based on gepner construction , '' arxiv : hep - th/0111012 . r. blumenhagen and v. braun , `` superconformal field theories for compact manifolds with spin(7 ) holonomy , '' jhep * 0112 * ( 2001 ) 013 [ arxiv : hep - th/0111048 ] . j. m. figueroa - ofarrill and s. schrans , `` the conformal bootstrap and super w algebras , '' int . j. mod . a * 7 * ( 1992 ) 591 . + j. m. figueroa - ofarrill and s. schrans , `` extended superconformal algebras , '' phys . lett . * b257 * ( 1991 ) 69 . s. komata , k. mohri and h. nohara , `` classical and quantum extended superconformal algebra , '' nucl . * b359 * ( 1991 ) 168 . r. blumenhagen , `` covariant construction of n=1 super - w algebras , '' nucl . * b381 * ( 1992 ) 641 . r. blumenhagen , w. eholzer , a. honecker and r. hubel , `` new n=1 extended superconformal algebras with two and three generators , '' int . j. mod . phys . a * 7 * ( 1992 ) 7841 [ arxiv : hep - th/9207072 ] . w. eholzer , a. honecker and r. hubel , `` representations of n=1 extended superconformal algebras with two generators , '' mod . lett . a * 8 * ( 1993 ) 725 [ arxiv : hep - th/9209030 ] . s. mallwitz , `` on sw minimal models and n=1 supersymmetric quantum toda field theories , '' int . j. mod . a * 10 * ( 1995 ) 977 [ arxiv : hep - th/9405025 ] . a. a. belavin , a. m. polyakov and a. b. zamolodchikov , `` infinite conformal symmetry in two - dimensional quantum field theory , '' nucl . phys . * b241 * ( 1984 ) 333 . p. goddard , a. kent and d. olive , `` virasoro algebras and coset space models , '' phys . b * 152 * , 88 ( 1985 ) . p. goddard , a. kent and d. olive , `` unitary representations of the virasoro and supervirasoro algebras , '' commun . * 103 * , 105 ( 1986 ) . h. eichenherr , `` minimal operator algebras in superconformal quantum field theory , '' phys . b * 151 * ( 1985 ) 26 . m. a. bershadsky , v. g. knizhnik and m. g. teitelman , `` superconformal symmetry in two - dimensions , '' phys . b * 151 * ( 1985 ) 31 . d. friedan , z. a. qiu and s. h. shenker , `` superconformal invariance in two - dimensions and the tricritical ising model , '' phys . b * 151 * ( 1985 ) 37 . s. l. lukyanov and v. a. fateev , `` physics reviews : additional symmetries and exactly soluble models in two - dimensional conformal field theory , '' _ chur , switzerland : harwood ( 1990 ) 117 p. ( soviet scientific reviews a , physics : 15.2)_. k. thielemans , `` a mathematica package for computing operator product expansions , '' int . j. mod . c * 2 * ( 1991 ) 787 . y. kazama and h. suzuki , `` new n=2 superconformal field theories and superstring compactification , '' nucl . * b321 * ( 1989 ) 232 . y. kazama and h. suzuki , `` characterization of n=2 superconformal models generated by coset space method , '' phys . * b216 * ( 1989 ) 112 . a. schwimmer and n. seiberg , `` comments on the n=2 , n=3 , n=4 superconformal algebras in two - dimensions , '' phys . b * 184 * ( 1987 ) 191 . t. inami , y. matsuo and i. yamanaka , `` extended conformal algebra with n=2 supersymmetry , '' int . j. mod . phys . a * 5 * ( 1990 ) 4441 . s. odake , `` extension of n=2 superconformal algebra and calabi - yau compactification , '' mod . a * 4 * ( 1989 ) 557 .

the @xmath0superconformal algebra is a @xmath1 algebra with two free parameters . it consists of 3 superconformal currents of spins @xmath2 , @xmath2 and 2 . the algebra is proved to be the symmetry algebra of the coset @xmath3 . at the central charge @xmath4 the algebra coincides with the superconformal algebra associated to manifolds of @xmath5 holonomy . the unitary minimal models of the @xmath0algebra and their fusion structure are found . the spectrum of unitary representations of the @xmath5 holonomy algebra is obtained . * unitary minimal models of @xmath6 + superconformal algebra and + manifolds of @xmath5 holonomy + *

cosmological inflation @xcite @xcite @xcite @xcite was proposed to address the horizon problem , flatness problem and monopole problem in the context of big bang cosmology . by postulating that in the early universe there was a brief period of rapid , exponential expansion one can explain , without fine - tuning , the observed facts that the universe is the same in different regions which are causally disconnected ( the horizon problem ) , the universe appears to be spatially flat ( the flatness problem ) and that there appears to be a much lower density of grand unified monopoles than one would naively expect . however , the inflation hypothesis itself has several unanswered questions : ( i ) what is the detailed mechanism for inflation ? ( ii ) what precedes the inflationary phase or how does inflation turn on " ? ( iii ) how does the universe make a graceful exit from this early , inflationary phase to standard friedman - robertson - walker ( frw ) radiation dominated expansion i.e. how does inflation turn off " . in many of the original models @xcite @xcite @xcite inflationary expansion was driven by a phase transition at the grand unified scale . the mechanism for inflation we propose here is based on particle creation from the gravitational field and it need not occur at the same time / energy scale compared to the canonical examples of inflationary mechanisms . specifically , we focus on particle creation connected with the hawking - like radiation that occurs in frw space - time . this is similar to black hole evaporation , but time reversed . for an astrophysical size black hole hawking radiation is at first a very weak channel for mass / energy loss for the black hole . as the black hole decreases in mass due to loss from hawking radiation it gets hotter and evaporates at a faster rate . beyond some size hawking radiation becomes very strong so that near the end stages of evaporation the black hole will radiate explosively . however , near the end stages of evaporation one can no longer trust the semi - classical calculation @xcite leading to hawking radiation . one common speculation is that near the end stages of evaporation where quantum gravity should become important , that hawking radiation will turn off " . one concrete proposal along these lines is the suggestion that in the quantum gravity regime space - time becomes non - commutative which leads naturally to a turning off of hawking radiation in the late stages of black hole evaporation @xcite . applying these ideas to frw space - time leads to a time reversed version of black hole evaporation . during the very earliest stages of the universe when the energy density is large , so that one is in the quantum gravity regime , the hawking radiation from the frw would be turned off until the universe expanded to the point when quantum gravity started to give way to semi - classical gravity . at this point the hawking radiation of frw space - time would turn on " and as we show below , would drive a period of exponential expansion . as the universe expanded the hawking temperature of the frw universe would decrease until the universe becomes dominated by ordinary radiation rather than hawking radiation . at this point the universe would make a gracefully transition from inflationary expansion to the power law expansion associated with a universe dominated by ordinary radiation . already in the 1930s schrdinger @xcite put forward the idea that particle creation can influence cosmological evolution . more recently parker @xcite and others @xcite-@xcite have followed this early work of schrdinger with studies of how particle creation can affect the structure of cosmological space - times . as pointed out in @xcite there are two points about cosmological history which are well addressed by these particle creation models . first , one can explain very well the enormous entropy production in the early universe via the _ irreversible energy flow from the gravitational field to the created particles_. second , since the matter creation is an _ irreversible _ process one avoids the initial singularity in cosmological space - times @xcite . in this model the universe begins from an instability of the vacuum instead of a singularity . the universe then rapidly moves through an inflationary phase followed by a radiation dominated era and finally followed by a matter / dust dominated era . our particle creation / hawking radiation model for inflation is closely tied to thermodynamics in a given space - time so we begin by collecting together some thermodynamic results . the first law of thermodynamics reads @xmath0 , where @xmath1 is the heat flow into / out of the system during some interval of cosmic time from @xmath2 to @xmath3 , @xmath4 is the energy density , @xmath5 is the volume and @xmath6 is the thermodynamic pressure . dividing this equation by @xmath7 , gives the following differential form for the first law of thermodynamics , @xmath8 for most cosmological models the assumption is made that the universe is a _ closed , adiabatic _ system which means @xmath9 . with this assumption the second law of thermodynamics , @xmath10 , leads to a non - change in the entropy , i.e. @xmath11 , during the cosmic time interval @xmath7 . this line of reasoning contradicts the observed fact that the universe has an enormous entropy . this contradiction can be addressed by having irreversible particle creation from the gravitational field i.e. hawking radiation from an frw space - time . this irreversible particle production leads to entropy production . the change in heat , @xmath1 , is now completely due to the change of the number of particles coming from particle creation . therefore there is a transfer of energy from the gravitational field to the created matter and the universe is treated like an _ open , adiabatic _ thermodynamic system @xcite . we review the relevant parts of the frw space - time . the standard frw metric is @xmath12 , \label{frw}\end{aligned}\ ] ] where @xmath13 is the scale factor and @xmath14 is the spatial curvature of the universe @xmath15 is flat , @xmath16 is open and @xmath17 is closed ) . the einstein field equations ( @xmath18 ) for this metric have a time - time ( @xmath19 ) component and space - space @xmath20 component given respectively by @xmath21 in the above equations @xmath4 is the energy density and @xmath6 is pressure of the matter source fluid / field . combining these two equations gives the standard conservation relationship @xmath22 , which clearly , describes the universe as a closed , adiabatic system with @xmath9 . as mentioned above this leads to @xmath11 which then seems to contradict the very large observed entropy of the universe . allowing for matter creation alters things . first in the presence of matter creation the equations are altered . the first equation on the left of remains the same but the second equation is altered and one has an additional equation for the time rate of change of particle number density . these modified and additional equations are @xcite , @xmath23 the overdot implies a time derivative , @xmath24 is particle number density , @xmath25 is the matter creation rate and @xmath26 is the pressure due to matter creation . the matter creation rate and the matter creation pressure are connected by the following relationship @xcite , @xmath27 if one assumes that @xmath4 and @xmath6 describe a normal fluid so that one has the energy condition @xmath28 ( assuming that @xmath29 this condition is known as the weak energy condition @xcite ) and in addition that the matter creation rate is positive @xmath30 one can see that @xmath26 of is positive and thus contributes a negative pressure to . such negative pressures can drive accelerated expansion such as during the early inflationary phase of the universe or during the current dark energy " dominated era of the universe . it would be economical if this negative pressure that occurs due to hawking radiation in frw space - time could drive both the inflationary era _ and _ the present accelerated phase of the universe which is normally attributed to dark energy . we will show that while this particle creation pressure can drive inflation it can not drive the present accelerated expansion . in the form one could easily explain both inflation and the current accelerated expansion by simply choosing a matter creation rate @xmath31 to produce whatever acceleration ( if @xmath30 ) or deceleration ( if @xmath32 ) one wants . for example , if one wants exponential expansion , @xmath33 one should choose @xmath34 @xcite . however this choice has very little physical motivation beyond giving one the result one wanted in advance . the strength of our proposal is that the particle creation comes from a specific mechanism hawking radiation in frw space - time and as such leads to definite predictions which allow the model to be verified or ruled out . we will see that our mechanism does in fact lead to a particle production rate @xmath35 . we now move on to a discussion of hawking radiation and associated temperature in frw space - time . since the frw space - time is dynamical , the definition of the cosmological event horizon is subtle . however one can define the apparent horizon knowing the local properties of the space - time . in order to do this one can rewrite in the following form @xcite @xmath36 where , @xmath37 and @xmath38 diag@xmath39 and @xmath40 . the position of the apparent horizon is given by the root ( @xmath41 ) of the equation @xmath42 . expanding this equation over @xmath43 sector and simplifying we get the position of the apparent horizon ( @xmath44 ) @xcite @xmath45_{\tilde{r}=\tilde{r}_a } = 0 \implies { \tilde r}_a = \frac{c}{\sqrt{h^2+\frac{kc^2}{a^2}}}. ~,\label{rah}\ ] ] using the above one can find the hawking temperature of the apparent horizon @xcite @xmath46 the first equality above is the standard relationship between the hawking temperature and surface gravity @xmath47 at the horizon of a given space - time . for frw space - time the surface gravity is @xmath48 . thus in general the temperature , @xmath49 , depends on both @xmath41 and its time derivative , @xmath50 . however , during an inflationary phase the universe s scale factor takes the form @xmath51 so that @xmath52 . if @xmath53 satisfies @xmath54 ( later we show this is the case for our model of inflation ) we have from @xmath55 and @xmath56 . thus the temperature in simplifies to@xcite @xmath57 in the final approximation we are again assuming @xmath54 which as mentioned above we will justify later . before moving to a detailed calculation of how the particle creation pressure affects the evolution of the early universe in the case when this pressure comes from the particle creation from hawking radiation , we give some numerical comparisons which show that this mechanism is of the correct order of magnitude to explain inflation . considering @xmath58 to be inverse of the planck time ( @xmath59 ) gives from @xmath60 . on the other hand at planck energy ( @xmath61 ) , gives a planck temperature of @xmath62 . thus the hawking temperature of frw space - time at very early is around the planck temperature . this large temperature associated with hawking radiation of frw space - time in the early universe is a good indication that our proposed mechanism has the proper order of magnitude to be a major factor in the early evolution of the universe . our proposed hawking radiation mechanism for inflation is the inverse of black hole evaporate . for astrophysical black holes the evaporation process begins very weakly for a black hole having the mass of our sun the temperature of the black body radiation emitted is @xmath63 . however at the end stages of evaporation when the black holes has a small mass the evaporation will proceed explosively . at this point one is not justified in using the approximations that led to hawking radiation as a thermal spectrum and it is said that one must have in hand a quantum theory of gravity to understand these end stages of black hole evaporation . for frw space - time one is not justified in using hawking radiation results at the very early stage of the universe ; one should have a theory of quantum gravity to understand this regime . as the universe expands there will be a point at which the approximations leading to hawking radiation from frw space - time become valid . it is at this point that our hawking radiation from frw space - time mechanism for inflation turns on " and inflation begins . as the universe inflates further the hawking temperature naturally decreases and our inflation mechanism will automatically turn off " . one can see that our proposed process is the inverse of black hole evaporation since the direction of radiation flux of the hawking radiation for the apparent horizon in a frw space - time is the opposite from that of a schwarzschild black hole event horizon . for black holes , the created particles escape outside the event horizon towards asymptotic infinity , while for the apparent horizon of frw space - time the created particles come inward from the horizon . due to the isotropy of frw space - time , the radiation is isotropic from all directions . the net result is an effective power gain in the universe , given by the stephan - boltzmann ( s - b ) radiation law . in summary the difference in the radiation direction from a schwarzschild black hole and from the frw space - time is as follows : for black holes the time rate of energy change @xmath64 is negative ( i.e. they lose power during hawking evaporation ) while for the frw space - time the time rate of energy change , @xmath64 , is positive ( i.e. the universe gains energy ) . according to the stephan - boltzmann radiation law the time rate of energy gain due to hawking radiation is @xmath65 where @xmath66 is the s - b constant and @xmath67 is the area of apparent horizon . now one can substitute into but in that case the right hand side of ( which is the rate of change in energy flux through the apparent horizon ) has to be evaluated at @xmath68 , so that @xmath69_{\tilde{r}=\tilde{r}_a}=\sigma a_h \left(\frac{\hbar h}{2\pi k_b}\right)^{4 } , \label{genrl}\ ] ] where we have used . to calculate the left hand side we first consider the volume of a sphere of arbitrary radius by ignoring the curvature term i.e. we take @xmath15 and the volume is given by @xmath70 . note that here we take the radius at arbitrary @xmath71 . only after performing the @xmath2 derivative in we set @xmath72 . on the other hand for the right hand side which represents the flow of energy across the apparent horizon we take @xmath73 where @xmath74 comes from . using these expressions for the area and volume in yields a modified continuity equation @xmath75 if one ignored the effect of the hawking radiation / particle creation term on the right hand side of , by setting @xmath76 in , then becomes @xmath77 which is the usual continuity equation in the absence of particle creation . using @xmath15 and @xmath78 we now rewrite using the first equation in as @xmath79 where we have taken the equation of state for ordinary matter as @xmath80 and the time dependent equation of state due to particle creation is @xmath81 the equation of state parameter due to particle creation is @xmath82 the constant @xmath83 above is essentially the inverse of the planck energy density @xmath84 . as we will show later it is this constant @xmath83 sets the time and length scale for our inflation mechanism . this may also be different from the usual scale of inflation which is set by the grand unified scale . moving the @xmath85 term in from the right hand side to the left hand side one can see that this particle creation term acts like a negative pressure . for the present universe this term is negligible . the present value of the energy density of the universe is @xmath86 so that @xmath87 term on the right hand side of is effectively zero . thus this effective negative pressure can not explain the current accelerated expansion of the universe one still needs dark energy . however in the early universe @xmath4 can be large enough so that the particle creation pressure on right hand side of dominates , and as we will see this can drive inflation and also give a natural turn off " for inflation . at this point it should be mentioned that does not violate wald s first axiom @xcite on the energy - momentum tensor which is nothing but the usual conservation equation @xmath88 @xcite . to see this , we note that in the absence of particle creation , the right hand side of vanishes and the energy - momentum tensor has the form @xmath89 which satisfies the conservation equation . however in the presence of particle creation the above definition of @xmath90 fails to simultaneously describe the conservation law and particle creation . in order to take both features into account one needs to consider a modification @xmath91 which can deal with particle creation . such a scenario is normally discussed in relationship to particle creation from black holes . since , under the appropriate choice of vacuum state ( i.e. the unruh vacuum ) black holes emit real particles in the form of thermal radiation , so that there is a power loss associated with hawking radiation , it may appear that wald s first axiom is violated . however , as demonstrated in @xcite , for such cases it is the regularized energy - momentum tensor @xmath92 which satisfies the conservation equation @xmath93 . for the unruh vacuum the regularized energy - momentum tensor is @xmath94 thus it is clear that when one is dealing with particle creation it is the regularized ( modified ) energy - momentum tensor that satisfies wald s axioms . this is exactly the picture in our case . looking into the relation one can see that the conservation equation in our case is given by @xmath95 , where the modified energy - momentum tensor has the form @xmath96 in the above relation @xmath97 , @xmath98 is independent of time and given by whereas the remaining part , @xmath99 , only contains the contribution from @xmath100 the particle creation pressure due to hawking radiation . in addition to the negative pressure associated with particle creation due to hawking radiation one can also calculate the effective particle creation rate , @xmath101 , and compare with general result given in . using the equation of state @xmath102 one can re - write as @xmath103 equating with gives the time dependent matter creation rate associated with particle creation due to hawking radiation in frw space - time @xmath104 recall that in order to have exponential expansion @xmath105 one needs the creation rate from to be @xmath106 @xcite . thus from , in order to have exponential expansion ( i.e. inflation ) one needs @xmath107 to be approximately the same size as @xmath108 i.e. one needs @xmath109 if one assumes the equation of state for ordinary radiation i.e. @xmath110 . since @xmath111 where @xmath112 , this equality ( i.e. @xmath109 ) will occur when the density @xmath113 which is approximately the planck density . this density corresponds to the density in the early universe . thus the rough calculations again point toward there being a large enough matter creation rate , @xmath114 , in the early universe to drive inflationary expansion . however as the universe expands and @xmath115 drops the creation rate @xmath114 will decrease and this hawking radiation driven mechanism for inflation will turn off . we now give a detailed calculation of inflation driven by hawking radiation . inserting @xmath116 into one can integrate the resulting equation to find the energy density @xmath4 as a function of scale factor @xmath117 @xmath118 @xmath119 is a constant and in the last equality we have taken the equation of state of the ordinary matter to be that of radiation ( i.e. @xmath120 ) since we want the early hawking radiation inflation phase to be followed by a universe dominated by ordinary radiation . the dimensions of @xmath119 depend on the value of the equation of state parameter @xmath121 . note , in the classical limit ( @xmath122 ) , @xmath123 , the frw hawking radiation effect turns off , and gives @xmath124 which is the well known result for a universe dominated by ordinary matter with an equation of state @xmath80 . there are two limits of this @xmath4 from : ( i ) @xmath125 so that @xmath126 and the hawking radiation effect dominates ; ( ii ) @xmath127 so that @xmath128 which is the energy density of an ordinary radiation dominated universe . in case ( i ) the energy density is constant so that one has an effective cosmological constant which , as shown below , leads to exponential , inflationary expansion . in both cases ( i ) and ( ii ) the universe is radiation dominated but for case ( i ) this means hawking radiation of an frw space - time and in case ( ii ) this means ordinary radiation . as one can see from the two limiting case behaviors of @xmath4 these two types of radiation result in very different evolution . we now want to find the time - dependence of the scale factor @xmath13 . we begin by substituting @xmath4 ( from ) into the first equation in to get a differential equation for @xmath117 as a function of @xmath2 .(recall we are assuming that @xmath129 in is zero or negligible compared to the other terms ) . it is possible to integrate the resulting equation for @xmath117 to obtain @xmath130 = \frac{8}{3}\sqrt{\frac{2\pi g d}{c^2}}t -(k-1 ) \sqrt{\alpha d } ~. \label{soln}\ ] ] we have written the integration constant as @xmath131 where @xmath132 is some positive number greater than @xmath133 . this will make it easier to write out some of the later formulas . one important point to make about the scale factor , @xmath13 , in is that it has an early exponential expansion phase ( the second , logarithm term on the left hand side ) which naturally transitions to a power law expansion ( the first , power law term on the left hand side ) . we will discuss these two regimes in more detail in the following subsections . that these two phases come out naturally from the proposed inflation mechanism , without need for fine - tuning some inflaton potential , is a very attractive feature . in the next following three subsections we will analyze the the early time , exponential behavior of , the later time , power - law behavior of , and then we will discuss the possible values of @xmath119 and @xmath132 . we first examine the limit of in the very early universe where @xmath13 is of a size such that one has the limit @xmath125 . in this limit becomes @xmath134 ~. \label{inflation - era}\ ] ] thus in this limit we find exponential expansion ( inflation ) with a hubble constant given by @xmath135 at this point we can return and justify some of our earlier assumption and approximations . first , after we assumed that @xmath136 is valid for @xmath137 near the planck size or larger ( e.g. for @xmath138 ) . for @xmath13 of the planck scale one has @xmath139 as compared to @xmath140 from eq . . second , we assumed that @xmath56 . this is also justified since from @xmath141 and during the inflationary phase @xmath58 is approximately constant with its value given by . thus @xmath142 . during inflation the standard lore is that the radius of the universe should increase by a factor of @xmath143 . thus we need @xmath144 where @xmath145 , with @xmath146 being the end and beginning time for this hawking radiation driven inflation . from we have @xmath147 sec@xmath148 so we find that gives @xmath149 sec . note that if one took the ratio in to be 10 orders larger ( i.e. @xmath150 ) this would yield @xmath151 sec . in other words the time scale for the length of this inflation is set by @xmath58 in and independent of @xmath119 and @xmath132 in . because @xmath58 in is so large one does not need a very long time , @xmath152 , in order to inflate the universe by many order of magnitude . in contrast to the above mechanism of inflation , which is driven by near - planck scale physics , the standard picture of inflation is that it is driven by physics at the grand unified scale i.e. by a grand unified phase transition . in this standard scenario inflation is thought to go from @xmath153 sec . until @xmath154 sec . or @xmath155 sec . thus for inflation driven by a phase transition at the grand unified scale one has @xmath156 sec . is plotted with respect to @xmath2 ( in units of planck time @xmath157 ) using equation . in ( a ) we fix @xmath158 and in ( b ) @xmath159 . in both cases we take @xmath160 @xmath161 . in this range of time @xmath13 increases exponentially from planck size to about @xmath162 following the equation . because of an extremely large value of the hubble constant the lifetime of this inflation is very small . this is the reason why in(b ) apparently time is not changing along @xmath163 axis . in fact the change in @xmath2 takes place after the eight decimal places and thus does not appear in the plot.__,width=566 ] in we show two plots of @xmath164 from for the early , inflationary part of . in this figure we have set @xmath165 and two different values of @xmath132 are shown . this value of @xmath119 is justified in a subsequent section . from the two different values of @xmath132 we see that this parameter controls when inflation starts but it does not influence how long inflation last , which in this model is @xmath166 sec . the scale function @xmath13 given in will leave the regime where the very early universe approximation in is valid , and then at some time will reach the point where @xmath167 . after this intermediate stage @xmath13 from will continue to increase until the regime is reached where @xmath127 . in this limit gives @xmath168 furthermore if the above condition is satisfied in a manner that @xmath169 , one finally finds @xmath170 this is the usual @xmath171 power law expansion for a radiation dominated universe . thus after the inflationary stage given by the solution given in transitions into radiation dominated expansion given by . is plotted with respect to @xmath2 ( in units of planck time @xmath157 ) again by using . in ( a ) we consider @xmath158 and in ( b ) @xmath159 . as before we take @xmath172 and time changes after eight decimal places in ( b ) . both figures show that in these intermediate values of @xmath13 the inflationary behavior naturally makes a transition to an ordinary radiation dominated era for @xmath173 . these figures nicely capture the end of inflation and the beginning of ordinary radiation domination.__,width=566 ] in we show two plots of @xmath117 vs. @xmath2 from which shows the beginning of the transition from exponential inflation to @xmath171 power law inflation . again in this figure we have set @xmath165 and have the same values of @xmath132 as in . again the two different values of @xmath132 control when inflation starts but they do not influence its duration . in this subsection we want to investigate possible values of the integration constants @xmath119 and @xmath132 . @xmath119 can be set from the late time energy density of radiation . from @xcite one finds that @xmath174 which is the ratio of the radiation energy density to the critical energy density . using the value of the critical energy density ( @xmath175 ) we get @xmath176 for the present radiation energy density . equating this with @xmath177 and taking @xmath178 as the present scale factor of the universe yields @xmath165 . thus the amplitude of is @xmath179 . since for our inflationary phase scale factor @xmath13 given by we require @xmath125 ( which because of the @xmath180 power can translate to @xmath181 ) we see that in this picture inflation stops at a scale of @xmath182 rather than @xmath183 . however given the uncertainty in when exactly inflation ends this is not a fatal problem . the scale of the universe is still inflated by the same orders of magnitude it just starts inflating a smaller scale and ends at a smaller scale . moving on to the constant @xmath132 one can see from figs . 1 and 2 that this constant sets the time scale for when inflation starts . in plots 1(a ) and 2(a ) where @xmath132 is chosen to be @xmath184 we find that inflation starts at @xmath2 a few times larger than the planck time @xmath157 . from the figures we see that for @xmath158 inflation starts at @xmath185 to @xmath186 . on the other hand from plots 1(b ) and 2(b ) we see that for @xmath159 inflation starts at about @xmath187 sec . this start time corresponds to the standard picture where inflation is driven by a grand unified phase transition . note that even though @xmath132 can shift the starting time of inflation , it can not control the duration which is fixed at @xmath188sec in the previous section we sketched a model for inflation driven by hawking radiation of frw space - time which has a natural turn off " or graceful exit from inflation . we now offer speculation that this model of inflation driven by hawking radiation may also have a natural turn on " or entrance to inflation . as already noted the process for inflation suggested here is the reverse of black hole evaporation . during post - inflation ( i.e. late stage ) the hawking radiation of frw space - time will be a weak / minor effect just as hawking radiation is weak / minor effect at the beginning ( i.e. early stage ) of black hole evaporation . during inflation ( i.e. the very early stage ) described in the section above the frw hawking radiation effect is dominate , just as during the end ( i.e. very late stage ) of black hole evaporation the hawking radiation is dominate . during the very late stages of evaporation of a black hole there are speculations that quantum gravity effects will turn off hawking radiation . one particularly concrete example of this is in the non - commutative geometry scenario @xcite where , as the planck scale is approached , space - time become non - commutative @xmath189=i \theta ^{\mu \nu } ~,\ ] ] where @xmath190 is an anti - symmetric rank 2 tensor which has the dimensions of distance squared . as a result of this non - commutativity black holes can not evaporate to arbitrarily small size , but due to the implied uncertainty relationship between spatial coordinates @xmath191 for example @xmath192 a black hole can not shrink to zero size since then one would have @xmath193 in violation of this uncertainty relationship . detailed analysis @xcite shows that as a black hole evaporates in the non - commutative space - time characterized by it reaches some maximum temperature after which the black hole temperature will decrease as the black hole continues to evaporate . at some point the hawking temperature of the black hole goes to zero , the evaporation process stops and one is left with a non - radiating remnant @xcite . applying this picture to the frw hawking radiation model of inflation one would find that in the very early universe , as during the late stages of black hole evaporation , the size of the universe would be small and the frw hawking temperature would be zero . thus at this early stage there would be no inflation since the frw hawking radiation would be turned off " . the universe would expand normally " according to a power law like . at some point the universe would reach a size large enough not to be dominated by the uncertainty relationship coming from the noncommutative space - time relationship of . at this point the frw hawking radiation would turn on " and drive inflation until the universe transitioned from the regime @xmath125 to the regime @xmath127 . when the universe entered this regime ( i.e. @xmath127 ) it would undergo power law type of expansion given in rather than the inflationary expansion of . in this paper we have proposed a mechanism for inflation based on the particle creation due to hawing radiation in an frw space - time . this mechanism differs from the model of inflation driven by some phase transition at the grand unified scale . this can be seen in the different time scales inflation driven by a grand unified phase transition is thought to start at @xmath153 sec . and last until @xmath194 sec . , thus having @xmath195 sec . because of the large value of @xmath58 in ( or alternatively the small value of @xmath83 in ) the time scale of our proposed mechanism for inflation is @xmath196 sec . which is different than the standard time for inflation . there are two constant , @xmath119 and @xmath132 , which arise in the solution of the scale factor . the constant @xmath119 is determined by matching the theoretical late - time energy density ( @xmath128 ) with the observed value of the present day radiation energy density ( @xmath197 ) and the present day value of @xmath178 . in this way we obtain @xmath165 . we also get a the amplitude of the inflationary period expression for @xmath13 as given in namely @xmath198 . this means that this model of inflation ends when @xmath199 . this is six orders of magnitude smaller than the standard picture of inflation which ends at @xmath200 . however the scale factor in our model still inflates in size by a factor of @xmath143 . in this picture inflation exits at a smaller scale factor than in the canonical picture . the other constant @xmath132 simply shifts when inflation starts , but does not control the duration . from figs . 1 and 2 one can see for @xmath201 inflation starts near the planck time while for @xmath202 inflations starts near @xmath203 sec the standard starting time in inflation driven by a grand unified phase transition . because for some values of @xmath132 the starting time of inflation can be near the planck time one should worry , for these values of @xmath132 , about the validity of the calculation of the hawking radiation . for one , near the planck scale the constants @xmath204 , @xmath205 and @xmath206 could be different from the present day values . in particular since @xmath83 in and therefore @xmath58 in depend on @xmath204 to the seventh power , having a different value of @xmath204 at these early , near - planck times by even one order of magnitude would greatly change the scale of hawking radiation driven inflation mechanism proposed here . if @xmath204 were one order of magnitude smaller in these very early times the energy scale of the hawking radiation driven inflation would shift to be more in line with that of the grand unified phase transition mechanism for inflation . in this paper we simply stick to the simplest assumption that @xmath204 , @xmath205 and @xmath206 have the present , constant values even at these early , near - planck times . we hope later to investigate the possibility that @xmath204 , @xmath205 and/or @xmath206 have different value at these early times . in this picture of hawking radiation driven inflation the time scale is set by @xmath83 in . setting aside this definite scale prediction for a moment - allowing for an arbitrary scale @xmath83 - we note that one might regard , and the resulting scale factor @xmath13 in , as a good phenomenological model for the time development of the size of the universe which naturally includes exponential expansion with power law expansion in a single expression . the inflation mechanism presented here is the time reversal of black hole evaporation . for a black hole in the _ early stages _ of evaporation via hawking radiation , the radiation is a weak effect , barely changing the mass and space - time of the black hole ; for an frw universe in its _ late stages _ the hawking radiation is a weak effect having effectively no effect on the expansion rate of the universe . for a black hole in the _ late stages _ of evaporation via hawking radiation , the radiation is a dominant effect , which plays a significant role in the change of the black hole s mass and the structure of the space - time ; for an frw universe in its _ early stages _ the hawking radiation is a huge effect and leads to an enormous expansion rate for the universe . in the _ very late stages _ of black hole evaporation it is postulated that quantum gravity effects will shut off hawking radiation ; for an frw space - time we postulate that in the _ very early stages _ quantum gravity effects will shut off hawking radiation and the associated exponential expansion . there have been other works that have studied the role of particle creation in the evolution of the universe @xcite @xcite . the present proposal is similar to the work of @xcite which views particle creation as an irreversible process from energy transfer and entropy production from the gravitational field to the particles . the difference in the present work is that we have proposed a very specific particle creation mechanism namely the hawking radiation associated with frw space - time . the frw hawking radiation gives rise to an effective negative pressure evolution equation for the energy density , @xmath4 , . the resulting @xmath4 given in leads to a time dependent scale factor @xmath13 given in which has two regimes one where @xmath125 with the resulting @xmath13 being exponential / inflationary expansion as given in and one where @xmath127 with the resulting @xmath13 being power law expansion as given in . there is a natural transition from inflationary expansion to power law expansion so that this model for inflation has a graceful exit from inflationary behavior . finally based on the inverse similarity between black hole evaporation and this frw hawking radiation model of the evolution of the scale factor @xmath13 , where the period of late time black hole evaporation corresponds to early period of the universe ( and visa versa ) , we have given some speculation as to how frw hawking radiation mechanism for inflation may turn on " due to non - commutative space - time effects . thus the frw hawking radiation picture for the evolution of @xmath13 provides not only a graceful exit to inflation as well as a possible graceful entrance . one final comment - this inflation mechanism has a feedback mechanism which forces the scale factor , @xmath13 , to be uniform . for example , if one assumed that the scale factor also had a dependence on @xmath207 ( i.e. @xmath208 ) the hawking radiation inflation mechanism would tend to erase this @xmath207 dependence . if , @xmath208 were smaller for some @xmath207 this would imply a higher hawking temperature and more rapid expansion . this would push those regions of @xmath207 with smaller scale factor , @xmath117 , to expand more rapidly until they were the same as the scale factor in other regions . if , @xmath208 were larger for some @xmath207 this would imply a lower hawking temperature and less rapid expansion . this would push those regions of @xmath207 with larger scale factor , @xmath117 , to expand less rapidly until they were the same as the scale factor in other regions .

we present a model for cosmological inflation which has a natural turn on " and a natural turn off " mechanism . in our model inflation is driven by the hawking - like radiation that occurs in friedman - robertson - walker ( frw ) space - time . this hawking - like radiation results in an effective negative pressure fluid " which leads to a rapid period of expansion in the very early universe . as the universe expands the frw hawking temperature decreases and the inflationary expansion turns off and makes a natural transition to the power law expansion of a radiation dominated universe . the turn on " mechanism is more speculative , but is based on the common hypothesis that in a quantum theory of gravity at very high temperatures / high densities hawking radiation will stop . applying this speculation to the very early universe implies that the hawking - like radiation of the frw space - time will be turned off and therefore the inflation driven by this radiation will turn off .

alice is a general - purpose heavy ion experiment designed to study the physics of strongly interacting matter and the quark gluon plasma ( qgp ) in nucleus - nucleus collisions at the lhc @xcite . resonances , in particular the @xmath0 meson , are useful probes to study the high density medium created in ultrarelativistic heavy ion collisions . the hadronic cross section associated with the @xmath0 meson is small , which makes this particle rather insensitive to the presence of other hadrons in the late stage of the collision . therefore , the production of @xmath0 mesons has been suggested as a signature for strangeness production mechanisms owing to an early partonic phase @xcite . preliminary results from the star collaboration suggest that such mechanisms might have been observed at rhic energies when comparing the @xmath0 yields obtained in au - au collisions , and more recently in cu - cu collisions , with those obtained in pp collisions @xcite . although the lifetime of the @xmath0 meson in vacuum is larger than that expected for the qgp state , significant medium modifications of its spectral properties have been predicted @xcite . as a consequence , the branching ratio for its decay into kaon and lepton pairs may change . the observation of such modifications might also provide information on the mechanism relevant for @xmath0 production in high energy collisions , which at present remains an open question . in addition , the @xmath0 meson is an interesting particle in itself as a hadronic measurement that could be done at several energies and for various collision systems at the lhc . analysing resonances in pp collisions is important as a benchmark for the heavy ion run . moreover , it is also a significant analysis as @xmath0 production has not been measured to very high precision at the tevatron @xcite , so even low energy " points from the lhc startup would become the best data at that energy . furthermore , @xmath0 mesons could be used as indicators of strangeness production along with the particles with open strangeness ( @xmath1 ) @xcite . it indicates the level of strangeness production as there are predictions that the strange sea could be large @xcite . at the lhc it would be possible to access the very low - x region , about which not much is known additionally , hera measurements will need verification and improvements @xcite . a key feature of the alice experiment is its very good capability to identify charged and some neutral particles using a variety of detector techniques , which can resolve the different particle ambiguities at different momentum ranges . alice can combine the pid information from single detectors to cover the different kinematic limits . table [ pidkaonstable ] shows the momentum range over which kaons can be identified using the specified detector and technique . in particular , the tpc and its , which give de / dx measurements , covering the full central region and can be used to identify charged particles from below 1 gev/_c_. the tof is used for hadron identification by time of flight . the efficiency and contamination for kaons is shown in figure [ fig : pidkaons ] . the high momentum particle identification ( hmpid ) system was not included in this figure but can be used for the identifications of kaons in the medium range of momentum at a limited solid angle coverage . in addition , de / dx measurements can be used again to identify kaons at high momenta using the relativistic rise phenomenon @xcite . @l*15@l range ( gev/_c _ ) & pid technique / sub - detector + 0.1 - 0.5 & de / dx ( its+tpc ) + 0.35 - 2.5 & time of flight + 1 - 3 & hmpid + 5 - 50 & relativistic rise ( its+tpc ) + 0.3 - 13 & secondary vertex reconstruction + [ pidkaonstable ] apart from the identification of charged particles , resonance studies require a good understanding of the combinatorial background . figure [ fig : phiinvmass ] ( left ) shows the invariant mass distribution of kaon pairs without using the pid system . a sample of 7x@xmath2 minimum bias pythia pp events at @xmath3= 14 tev was analysed ( same statistics for the rest of plots ) . the sample was divided into three multiplicity ranges . the figures shown in this paper correspond to the ( intermediate ) range between 5 and 25 in the negative charged track multiplicity at mid - rapidity . mixing particles from different events has been a technique used in the past to estimate the combinatorial background . the limitation of this method is that some event similarity " condition needs to be imposed , which strongly depends on the multiplicities presented in the event . combining particles from the same event is also another option , using like - sign pairs . both methods were implemented in this analysis , providing similar results as was previously reported in @xcite . these techniques are currently being optimised to make use of distributed events by grid services where the final analysis is foreseen . the selections for kaon tracks have yet to be finalised for the first analysis with real data , but some preliminary results will be presented @xcite . the simulation of pp interactions is generated by pythia 6.214 monte carlo event generator . the output is then passed to the alice simulation and reconstruction software using the standard computing framework . results presented here are based on a fast - simulation method described in detail in @xcite . because the short lifetime of resonances , they all decay at the primary vertex , so the track selection exclude particles from secondary interactions . all tracks are required to come from the primary vertex by using impact parameter cuts . figure [ fig : phiinvmass ] ( right ) shows the invariant mass distribution and the estimation of the background using the like - sign method and assuming perfect pid efficiency for kaons . the reconstructed mass and width , as well as the number of @xmath0 mesons after the background subtraction , were found to be consistent with the generated values . a similar study , using the full set of alice detectors to simulate realistic conditions for pid , leads to mass spectra ( not shown here ) very similar to those in figure [ fig : phiinvmass ] ( right ) . more details are given in reference @xcite . invariant mass spectrum for pp collisions : ( left ) without making use of the alice pid capabilities ; ( right ) assuming perfect particle identification efficiency for charged kaons . the background was estimated by like - sign method ( open points).,title="fig:",scaledwidth=45.0% ] invariant mass spectrum for pp collisions : ( left ) without making use of the alice pid capabilities ; ( right ) assuming perfect particle identification efficiency for charged kaons . the background was estimated by like - sign method ( open points).,title="fig:",scaledwidth=45.0% ] because of the narrow width and good signal - to - background ratio for the @xmath0 , alice should be able to detect this resonance early in the running of lhc . however , during that period the particle identification system might not be fully available . a study was carried out to investigate the prospects for measuring @xmath0 mesons without accessing any of the pid information provided by the detectors . it was found that by applying a relatively high momentum cut on the candidate pair , the @xmath0 meson peak can only be resolved to an acceptable level . figure [ fig : phinopid ] shows the predictions for the @xmath4 invariant mass spectrum for pp collisions for pairs with @xmath5 gev/_c _ ( left hand side ) and after the subtraction of the background ( right ) , showing that a resonance signal above the combinatorial background can be obtained from 7x@xmath2 events . invariant mass spectrum for pp collisions without accessing the pid information from the detectors ( left ) and after the subtraction of the background ( right ) . the background was estimated by like - sign method ( open points ) . a cut on the transverse momentum of the pair of 2.4@xmath62.8 gev/_c _ was used.,title="fig:",scaledwidth=45.0% ] invariant mass spectrum for pp collisions without accessing the pid information from the detectors ( left ) and after the subtraction of the background ( right ) . the background was estimated by like - sign method ( open points ) . a cut on the transverse momentum of the pair of 2.4@xmath62.8 gev/_c _ was used.,title="fig:",scaledwidth=45.0% ] the alice experiment will allow the study of resonance production in great detail due to its very good particle identification and tracking capabilities . a study of the @xmath0 meson in hadronic ( and leptonic ) decay channels will be possible . like - sign and mixing - event methods were tested for background subtraction . they give consistent results . the @xmath0 meson analysis might be one of the first resonance measurements in alice because of the good signal - to - background ratio . alice should be able to study the @xmath0 meson in pp collisions before the pid system becomes available . results presented here indicate that it will be possible to measure the mass distribution and momentum spectra for the @xmath0 meson through its hadronic decay @xmath7 during the first physics " days at the lhc . this work was supported by the uk overseas research students awards scheme ( orsas ) and the university of birmingham , and by the eu al@xmath8an programme . 10 b. alessandro _ et al . _ [ alice collaboration ] , j. phys . g * 32 * ( 2006 ) 1295 . j. rafelski and b. muller , phys . * 48 * ( 1982 ) 1066 . p. koch , b. muller and j. rafelski , phys . * 142 * ( 1986 ) 167 . s. a. bass _ et al . _ , nucl . a * 661 * ( 1999 ) 205 . et al . _ [ star collaboration ] _ j. phys . g : nucl . part . _ ( these proceedings ) f. klingl , t. waas and w. weise , phys . b * 431 * ( 1998 ) 254 . r. rapp and j. wambach , adv . * 25 * ( 2000 ) 1 . t. alexopoulos _ et al . _ [ e735 collaboration ] , z. phys . c * 67 * ( 1995 ) 411 . o. villalobos baillie [ alice collaboration ] , j. phys . g * 31 * ( 2005 ) s701 . j. r. ellis , nucl . a * 684 * ( 2001 ) 53 . s. chekanov _ et al . _ [ zeus collaboration ] , nucl . b * 718 * ( 2005 ) 3 . s. s. adler _ et al . _ [ phenix collaboration ] , phys . c * 72 * ( 2005 ) 014903 .

the alice experiment at the cern large hadron collider ( lhc ) will allow the study of resonance production in nucleus - nucleus and proton - proton collisions . this paper presents results based on physics performance studies to discuss prospects in alice for @xmath0(1020 ) meson production in pp interactions during the lhc startup .

we study the statistics of the exchange of energy between two subvolumes @xmath1 and @xmath2 of a quantum system . we assume that no degrees of freedom leave or enter @xmath1 and that @xmath1 is coupled to @xmath2 only through one degree of freedom @xmath3 . @xmath3 shall be coupled to the variable @xmath4 in @xmath2 , see fig . [ vbarv ] . a generalization of our approach to more than one coupling variable is straightforward . our model hamiltonian reads @xmath5 with @xmath6=[j_v , h_{\bar{v}}]=[h_v , x_{\bar{v}}]=[h_v , h_{\bar{v}}]=0 $ ] . and @xmath2 that are coupled via the variables @xmath3 and @xmath4 . , width=188 ] to calculate the statistics of the flow of energy into @xmath1 in a period of time @xmath7 $ ] we assume it to be in an eigenstate @xmath8 of @xmath9 at time @xmath10 and calculate moments of @xmath9 at time @xmath11 , defining the generating function @xmath12 ( we have set @xmath13 . ) the average is taken over the initial state @xmath8 . the exponential @xmath14 generates moments of the energy in @xmath1 at time @xmath15 . through the exponentials @xmath16 the initial energy @xmath17 in @xmath1 is subtracted from that such that @xmath18 generates moments of @xmath19 , the flux of energy into @xmath1 during @xmath7 $ ] , @xmath20 a generalization of @xmath18 that generates correlators of the energy flux at finite frequency is the functional @xmath21&= & \left\langle \overleftarrow{t } e^{i\int{dt\ , [ h- \dot{\xi}^-(t ) h_v ] } } \right.\nonumber \\ & & \left . \;\ ; \overrightarrow{t } e^{-i\int{dt\,[h - \dot{\xi}^+(t ) h_v ] } } \right\rangle . \end{aligned}\ ] ] the symbols @xmath22 denote ( inverse ) time ordering and we have collected the two source functions into a vector in a `` keldysh space '' , @xmath23 . @xmath24 correlates the energy flux at different times , @xmath25 { { \cal z } } \big| _ { { { \xi } } = 0}.\ ] ] @xmath24 arises naturally as one models a linear detector that measures the energy flux @xmath26 into @xmath1 @xcite . the detector read - off @xmath27 is then a linear functional of @xmath26 , @xmath28 where the response function @xmath29 is causal , @xmath30 for @xmath31 , and it depends on the internal dynamics of the detector . moments of the detector read - off are then generated by @xmath24 according to @xmath32\big|_{\xi=0}.\ ] ] evidently , eq . ( [ eq : heatfreqorig ] ) reduces to ( [ eq : heat ] ) for the measurement of the time integrated energy flow during @xmath7 $ ] , if @xmath33 [ with the step function @xmath34 for @xmath35 and @xmath36 for @xmath37 . in the interaction picture with respect to the uncoupled problem @xmath38 , @xmath39 and @xmath40 , @xmath24 takes the form @xmath41&= & \big\langle \overleftarrow{t } e^{i\int{dt\ , [ x_{\bar{v}}(t ) j_v(t ) - \dot{\xi}^-(t ) h_v ] } } \nonumber \\ & & \overrightarrow{t } e^{-i\int{dt\,[x_{\bar{v}}(t ) j_v(t ) - \dot{\xi}^+(t ) h_v ] } } \big\rangle . \end{aligned}\ ] ] we rewrite eq . ( [ eq : heatfreq ] ) by breaking the time - ordered product up into a large number of time - development exponentials for infinitesimal time steps @xmath42 . applying to each one of them the identity @xmath43 x_{\bar{v}}(t)}\ ] ] we find that @xmath44= \left\langle \overleftarrow{t } e^{i\int{dt\ , j_v[t-\xi^-(t ) ] x_{\bar{v}}(t ) } } \overrightarrow{t } e^{-i\int{dt\ , j_v[t-\xi^+(t ) ] x_{\bar{v}}(t ) } } \right\rangle.\ ] ] the generating functional of the statistics of energy flow into a volume @xmath1 takes the form of a partition functional of the entire system with shifted time arguments of all variables in @xmath1 . an analogous structure has been found before for the statistics of transfer of other globally conserved quantities , like charge @xcite and momentum @xcite . also there the generating functional has the structure of eq . ( [ eq : shiftham ] ) . the source term locally shifts the variable conjugated to the measured quantity ( the phase for a measurement of charge , position for a momentum measurement , and time for the energy measurement considered here ) . in a path integral formulation of eq . ( [ eq : heatfreqorig ] ) we mark fields that evolve the system forward and backward in time with superscripts @xmath45 and @xmath46 respectively and we collect them in vectors like the source @xmath47 . we may change integration variables as @xmath48 in the action corresponding to @xmath9 if @xmath49 at all times ( only then this map is bijective ) . the path integral takes then a form very similar to eq . ( [ eq : shiftham ] ) , @xmath50&= & \int { { \cal d } \vec{j}_v { \cal d } \vec{x}_{\bar{v } } \,e^{-i\left ( { \cal s}_v[\vec{j}_v ] + { \cal s}_{\bar{v}}[\vec{x}_{\bar{v}}]\right ) } } \nonumber \\ & & \;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\ ; \times e^ { -i\int{dt\ , \vec{j}_v(t-\vec{\xi}(t ) ) \tau_3 \vec{x}_{\bar{v}}(t)}}. \end{aligned}\ ] ] we have introduced actions @xmath51 and @xmath52 for the uncoupled volumes @xmath1 and @xmath2 that are obtained by integrating the actions corresponding to @xmath9 and @xmath53 over all variables except @xmath3 and @xmath4 respectively . @xmath54 is the third pauli matrix and we use the notation @xmath55^{\alpha}={j}^{\alpha}_v(t-{\xi}^{\alpha}(t))$ ] ( @xmath56 ) . the assumption @xmath57 has the following physical interpretation . from eq . ( [ eq : genhv ] ) and its finite frequency generalization it is seen that one can infer a probability distribution of energy flow by fourier transforming the generating function @xmath58 . to resolve energy differences of the order of @xmath59 in this distribution one needs to know @xmath24 for values the condition @xmath49 means then that the energy detector smears out the measurement over a time @xmath61 , that is , it does not attempt to measure the energy faster than it is allowed to by the uncertainty relation . we first illustrate our method with a particularly simple example : we calculate the statistics of heat dissipated from a voltage biased mesoscopic conductor into a bolometer attached to its right contact ( see fig . [ geometry of contact ] ) . to this end , we express the action of a simple connector in circuit theory by the keldysh green s functions @xmath62 of the adjacent reservoirs @xcite @xmath63 \label{connector}\ ] ] where the @xmath64 denote energy independent transmission probabilities and the brackets @xmath65 anticommutation . the trace includes integration over time . in the following , we address the zero frequency limit only . in this case , the time integration in eq . ( [ connector ] ) is conveniently rewritten as energy integration . as formulated in eq . ( [ eq : shiftham ] ) , we obtain the statistics of heat dissipation into the right contact by shifting all observables of the right contact in time by @xmath66 on the upper keldysh contour and by @xmath67 on the lower contour . the diagonal elements of @xmath68 in the keldysh space are left unchanged by this transformation which can be cast in the following form ( we introduce the difference @xmath69 ) : @xmath70 the time shift introduces a rotation of the green s function @xmath71 analogous to charge counting statistics @xcite . @xmath72 denotes the energy dependent fermi function of contact @xmath73 that depends on temperature @xmath74 and electrochemical potential @xmath75 . evaluating eq . ( [ connector ] ) in the zero frequency limit we find the cumulant generating function @xcite @xmath76 = \\ \quad\\ \frac{\tau}{2\pi } \int d\epsilon \sum_n \ln\left\{1+\gamma_nf_l(1-f_r)(e^{-i\xi(\epsilon-\mu_r)}-1)\right.\\ \quad\\ \left.+\gamma_nf_r(1-f_l)(e^{i\xi(\epsilon-\mu_r)}-1 ) \right\}. \end{array}\ ] ] cumulants of dissipated energy can now be calculated easily by taking derivatives with respect to @xmath77 . the mean dissipation without applied voltage ( @xmath78 ) is for instance given by @xmath79 for almost equal temperatures @xmath80 we recover the mesoscopic version of the wiedemann - franz law for thermal conductance @xcite . ( upper panel ) and charge @xmath81 statistics ( lower panel ) for a tunnel contact in equilibrium and non - equilibrium due to an applied bias or temperature difference . the energy is normalized to temperature @xmath82 or bias @xmath83 respectively . the product of @xmath84 and dimensionless conductance @xmath85 is assumed to be large . , width=302 ] we note that eq . ( [ heat statistics ] ) is similar to the statistics of charge transfer@xcite : charge statistics is recovered by substituting @xmath86 where @xmath87 is the elementary charge and the field @xmath88 generates cumulants of charge transfer . as for charge statistics , there is a classical probabilistic interpretation of eq.([heat statistics ] ) : an electron of total energy @xmath42 in channel @xmath89 passes the mesoscopic conductor with probability @xmath64 . it then dissipates the energy @xmath90 in the right contact . it is the energy integration in eq . ( [ heat statistics ] ) which makes the statistics of heat transfer interesting . unlike in the binomial charge statistics the exponent @xmath91 assigns energy weights to each electron . therefore , energy transfer is not quantized . we now turn to a specific example : the tunnel junction with @xmath92 . in this case , the logarithm of eq . ( [ heat statistics ] ) can be expanded and the energy integration involves only elementary integrals ( for equal temperatures @xmath93 on both sides of the junction ) . defining @xmath83 we find the following result for the characteristic function : @xmath94 = g\tau \coth\frac{\mu}{2kt } \left(kt\frac{\sin(\xi\mu)}{\sinh(\pi\xi kt)/\pi}-\mu\right)\\ \\ -ig\tau \frac{1-\cos(\xi\mu)}{\sinh(\pi\xi kt)/\pi}. \end{array}\ ] ] we introduced the conductance of the junction @xmath95 . for comparison , we also give the statistics of charge transfer @xcite @xmath96 = g\tau\mu\coth\frac{\mu}{2kt } \left(\cos(\chi e)-1\right ) - ig\mu\sin(\chi e).\ ] ] we observe several differences : whereas @xmath97 $ ] is strictly periodic , the characteristic function of heat dissipation @xmath98 $ ] shows damped oscillations . the lacking periodicity is due to the unquantized transfer of energy . for charge transfer , the real part of @xmath97 $ ] is even in the applied voltage @xmath99 in contrast to an odd imaginary part . odd cumulants therefore change sign under voltage inversion . this is not the case for energy , @xmath98 $ ] does not depend on the sign of the applied voltage . heat dissipation takes place in both reservoirs symmetrically regardless of the current direction ! odd cumulants of charge transfer ( the imaginary part of eq . ( [ characteristic charge tunnel ] ) ) do not depend on temperature@xcite . in contrast , we find that the asymmetry of the distribution of dissipated heat does depend on temperature ( see the imaginary part of eq . ( [ characteristic tunnel ] ) ) . another interesting case is the statistics in the presence of a temperature gradient only . analytical results are available for @xmath100 . we find @xmath101 = \\ \quad\\ 2g\tau k t_l \left\{\frac{i\pi}{\sinh(\pi k t_l \xi ) } + \ln(2 ) - \beta(-ikt_l \xi ) \right\}=\\ \quad\\ g\tau kt_l \left\{-i\frac{\pi^2}{6}kt_l\xi - \frac{3\zeta(3)}{2 } ( kt_l\xi)^2 + i\frac{7\pi^4}{360}(kt_l\xi)^3 % + \frac{15\zeta(5)}{8}\xi^4 + .. \right\ } \end{array}\ ] ] for energy transport and @xmath102 = 2g\tau k t_l \ln(2 ) \left(\cos(\chi e)-1\right)\ ] ] for charge transport ( we introduced the @xmath103-function @xmath104 ) . fig . [ different statistics ] illustrates several limiting cases : on a log scale , it compares the charge and energy transfer statistics in equilibrium , and in non - equilibrium due to an applied voltage ( see eqs . ( [ characteristic tunnel ] ) and ( [ characteristic charge tunnel ] ) ) and due to an applied temperature gradient ( see eqs . ( [ temperature gradient ] ) and ( [ temperature gradient charge ] ) ) . ) . the distribution is broader for tunneling junctions ( @xmath105 ) than for open point contacts ( @xmath106 ) . it is clearly seen that the third cumulant of the distribution changes sign . the inset shows the third cumulant @xmath107 as a function of the transparency @xmath108 . ( @xmath85 is the dimensionless conductance and @xmath109 , we assume @xmath110 to be large),width=302 ] in general , eq . ( [ heat statistics ] ) has to be evaluated numerically . [ different transparencies ] shows the probability distribution of dissipated heat for barriers with various transparencies @xmath111 in the high - voltage regime @xmath112 . noise disappears in the ballistic limit @xmath113 . it is clearly visible that the third cumulant of the distribution changes its sign as a function of @xmath108 . we find that @xmath114 ( see also the inset of fig . [ different transparencies ] ) . in this entire section , we assumed that the energy dissipation in the right reservoir is measured by an ideal bolometer which does not act back on the mesoscopic contact . a realistic bolometer is characterized by a finite thermal conductance . energy fluctuations in the right reservoir are then converted into temperature fluctuations which modulate the noise intensity of the mesoscopic contact . this backaction is similar to the backaction from a non - ideal current meter@xcite . we use now eq . ( [ eq : genpath ] ) to calculate the statistics of energy flow from a system @xmath115 into a linear medium . an example is the energy that @xmath115 emits as electromagnetic radiation . the electromagnetic field is then the linear medium into which energy flows . this energy flow may for example be measured with a photo - detector . the distribution of the number of photons emitted by a source current @xmath116 can be expressed in terms of correlators of @xmath116 @xcite . in ref . @xcite this relation has been established perturbatively in a weak coupling of the photo - detector to the electromagnetic field . in ref . @xcite it has been used to calculate the fluctuations in the number of photons emitted by a mesoscopic conductor . eq.([eq : genpath ] ) allows us to obtain these results non - perturbatively . as a second application of our method we calculate the fluctuations in the amount of heat that is dissipated in a macroscopic electrical circuit around a mesoscopic conductor . such fluctuations could be measured with a bolometer . in our analysis we divide space into three regions . the system @xmath115 , a volume @xmath1 into which the energy flow is measured and a region @xmath117 . we call @xmath118 the environment of @xmath115 . the subspaces are connected via two variables @xmath119 and @xmath4 in @xmath117 , that couple to @xmath116 in @xmath115 and @xmath3 in @xmath1 respectively ( see fig . [ couplefig ] ) . with the hamiltonians @xmath120 , @xmath121 , and @xmath9 of @xmath115 , @xmath117 , and @xmath1 we then have the total hamiltonian @xmath122 where @xmath123 and @xmath124 are sources that generate moments of @xmath4 and @xmath119 . is an arbitrary quantum system . the energy flow into the part @xmath1 of its linear environment is considered in this section . the variables @xmath116 , @xmath119 and @xmath3 , @xmath4 couple the three constituent volumes . , width=302 ] the generating functional for correlators of the energy flow into @xmath1 takes the form of eq . ( [ eq : genpath ] ) with @xmath125 } & = & \int { { \cal d } \vec{x}_s { \cal d } \vec{j}_{s } \ , e^{-i \left({\cal s}_s[\vec{j}_s ] + { \cal s}_{\tilde{e}}[\vec{x}_{s},\vec{x}_{\bar{v}}]\right ) } } \nonumber \\ & & \;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\ ; \times e^ { -i \int{dt\ , \vec{j}_s(t ) \tau_3 \vec{x}_{s}(t ) } } \end{aligned}\ ] ] with the actions @xmath126 and @xmath127 corresponding to @xmath120 and @xmath121 . @xmath128 , being the action of a linear system , is quadratic and depends only on the response functions of @xmath117 and its temperature @xmath129 . we characterize the response of @xmath130 when disconnected from @xmath115 ( corresponding to the hamiltonian @xmath131 ) by four functions , @xmath132 the volume @xmath1 is described by the response to a source @xmath133 coupling to @xmath3 in the absence of @xmath117 , corresponding to the hamiltonian @xmath134 , @xmath135 the environment s action is determined by the fluctuation dissipation theorem @xcite . the action for @xmath3 when @xmath1 is isolated reads @xmath136= \vec{j}_v\otimes m_{iso } \otimes \vec{j}_v.\ ] ] the matrix multiplication @xmath137 extends over keldysh as well as time indices , @xmath138(t , t '' ) \equiv \int{dt'\,a(t , t ' ) b(t',t'')}$ ] and correspondingly for vectors like @xmath139 . in a steady state @xmath140 depends only on the time difference @xmath141 and in fourier representation @xmath142 it reads @xmath143 with the bose - einstein distribution @xmath144 . ( we define the response functions such that they have negative imaginary part . ) we analyze first the particularly simple case that the energy flow into the entire environment to @xmath115 is measured , @xmath145 . then we have @xmath146 , @xmath147 and @xmath148 . we rewrite eq . ( [ eq : genpath ] ) by introducing a coupling matrix @xmath149 @xmath150= \int { { \cal d } \vec{j}_s { \cal d } \vec{j}_{v } e^{-i\left ( { \cal s}_s[\vec{j}_s ] + { \cal s}_{iso}[\vec{j}_v]+ \vec{j}_v \otimes \tau_3 \sigma \otimes \vec{j}_{s}\right)}}.\ ] ] the gaussian integrals over @xmath151 in ( [ eq : vise ] ) are easily done , resulting in @xmath152= \int { { \cal d } \vec{j}_s\ , e^{-i\left ( { \cal s}_s[\vec{j}_s ] + { \cal s}_{e}[\vec{j}_{{s } } ] + { \cal a}_{\vec{\xi}}[\vec{j}_s]\right ) } } \ ] ] with an action @xmath153 that describes the influence of the environment on @xmath115 , @xmath154= -{\textstyle \frac{1}{4 } } \vec{j}_s\tau_3\otimes m_{iso}^{-1 } \otimes \tau_3 \vec{j}_s,\ ] ] and a source term @xmath155 that vanishes at @xmath156 , @xmath157&= & \vec{j}_s\otimes a_{\vec{\xi } } \otimes \vec{j}_s , \ ; a_{\vec{\xi}}= -{\textstyle \frac{1}{4 } } \tau_3 a_{\vec{\xi } } \tau_3 , \label{eq : adirect } \\ a_{\vec{\xi}}&=&\sigma^{\dagger } \otimes m_{iso}^{-1 } \otimes \sigma - m_{iso}^{-1}. \end{aligned}\ ] ] in the general case @xmath158 , the environment s action is determined by @xmath159 , @xmath160&= & \vec{j}_s \otimes g_{ss } \otimes \vec{j}_s,\\ g_{\alpha\gamma}&= & { \textstyle \frac{1}{2 } } [ r_{\alpha\gamma } \tau_3 + ( r_{\alpha\gamma}-r_{\gamma\alpha}^ * ) g_0],\;\ ; \ ; \alpha,\gamma \in \{s , v\}. \nonumber \\\end{aligned}\ ] ] the source term then takes the form @xmath161^{-1}-g_{vv}^{-1 } \right\ } \otimes g_{vs}. \nonumber \\ \end{aligned}\ ] ] eq . ( [ eq : adirect ] ) is recovered for @xmath162 and @xmath163 , such that @xmath164 , @xmath165 , and @xmath166 . for concrete results we focus on zero frequency correlators of the energy flow , choosing @xmath167 [ c.f . eq.([eq : rn ] ) ] . we neglect transient effects , that is , we calculate only terms of leading order in the time @xmath15 over which energy is accumulated . we may then work in the discrete fourier space of @xmath15-periodic functions @xmath168 , with fourier coefficients @xmath169 and frequencies @xmath170 . in this representation the source term in eq . ( [ eq : zxi ] ) reads @xmath171= \tau \sum_l \vec{j}_{s , l } a_{\xi , l } \vec{j}_{s , l } .\ ] ] @xmath172 is now diagonal in frequency indices , @xmath173 and so is @xmath174 . at zero temperature it takes a simple form even in the general case @xmath175 , @xmath176 applied to a linear photo - detector this reproduces the distribution of photo - counts in response to a source @xmath116 obtained perturbatively in @xcite . to see this we substitute @xmath177 in eqs.([eq : zxi ] ) with ( [ eq : diagonal ] ) and ( [ eq : zerotemp ] ) to obtain the statistics of the number of absorbed quanta @xmath89 rather than that of the absorbed energy . the further substitution @xmath178 yields the generating function for factorial moments of @xmath89 . to compare the resulting factorial moments of the photocount to the formulas obtained in @xcite , we write them in the time domain , @xmath179^p \big\rangle_{{\cal s}_s+{\cal s}_e}. \nonumber \\ \end{aligned}\ ] ] this is equivalent to eq . ( 30 ) in @xcite . the detector sensitivity @xmath168 there corresponds to our @xmath180 , the density of absorbing detector modes per unit of frequency . the retarded photon propagator @xmath181 corresponds to our cross - response function @xmath182 . the time - ordering of @xmath183 and @xmath184 along the keldysh contour corresponds to the `` apex '' time order of the sources in @xcite . the expectation value in eq.([eq : factorial ] ) is taken with respect to the action @xmath185 of the source system and the piece @xmath186 that describes the back action of the environment on @xmath115 . electrical conductors couple to their electromagnetic environment - the circuit that they are embedded in - with the current @xmath187 that flows through them . to calculate the flow of energy into this environment we may therefore apply the formulas obtained in the previous section with @xmath188 , @xmath120 being the hamiltonian of the conductor . also in this case energy is transferred by means of photons . these photons may be either detected by a photocounter @xcite or with a bolometer , that measures the thermal energy exchanged by means of photons . according to eq . ( [ eq : factorial ] ) the @xmath89-th factorial moment of energy transfer is proportional to the @xmath189-th moment of fluctuations of the source @xmath116 . this suggests that a measurement of the energy flow and its fluctuations may be a useful tool for measuring higher order correlators of electrical currents whose measurement poses an experimental challenge . while experimental techniques for the measurement of the variance of electrical current fluctuations are by now well developed , so far only one experiment has been successful in measuring higher order current correlators . in @xcite the measurement of the third moment of current fluctuations through a tunnel barrier has been reported . eq.([eq : factorial ] ) suggests that the fourth moment of current fluctuations can be inferred from the variance of the heat produced by them . is the conductor that generates current fluctuations @xmath190 . these fluctuations can be characterized by a measurement of fluctuations of the heat that is dissipated in the external resistor @xmath191 . , width=321 ] to quantify this we apply now the relations obtained in the previous section to the electrical circuit depicted in fig . we assume all resistors in the circuit to be macroscopic and linear , except @xmath192 , that plays now the role of the system @xmath115 . fluctuations of the amount of heat that is dissipated in @xmath191 as measured with a bolometer are by virtue of eq.([eq : factorial ] ) related to fluctuations in the electrical current through @xmath192 . for that the volume @xmath1 should be chosen the resistor @xmath191 . the current @xmath193 flowing through the volume @xmath1 is coupled to the variable @xmath194 in the rest of the environment @xmath117 . this variable @xmath194 is the time integral of the voltage @xmath195 over @xmath191 , @xmath196 . the current @xmath187 through @xmath192 similarly couples to the environment variable @xmath197 . again we introduce external sources @xmath198 and @xmath199 . the circuit fig . [ fig2 ] is then mapped onto the general model of fig . [ couplefig ] with the choice @xmath200 , @xmath201 , @xmath202 , @xmath203 , @xmath204 , @xmath205 and @xmath188 . we need the response functions of the circuit fig . [ fig2 ] without the conductor @xmath192 , @xmath206 , \nonumber\end{aligned}\ ] ] @xmath207 @xmath208 @xmath209 ( all quantities here are frequency - dependent ) . note , that any four - terminal circuit connecting the conductor @xmath192 with @xmath191 can be modeled by the three resistors @xmath210 , @xmath211 , and @xmath212 . the resulting zero - frequency energy flow statistics takes a particularly simple form in the limit of an infinite @xmath213 . then the current flowing through @xmath192 is directly fed into @xmath191 and the source term in the generating functional ( [ eq : zxi ] ) is given by eq.([eq : diagonal ] ) with @xmath214 \left ( e^{-i \xi \omega_l}-1\right ) & 0 \end{array } \right ) .\end{aligned}\ ] ] this is a generalization of eq . ( [ eq : zerotemp ] ) to finite temperature , allowing for emission of energy quanta from @xmath1 as well as absorption . the transferred quanta have predominantly energies that are smaller than the inverse of the rc - time @xmath215 of @xmath191 . for finite @xmath210 thermal current fluctuations @xmath216 created in @xmath210 mix into the fluctuations of @xmath187 . as a consequence eq . ( [ eq : zinf ] ) is then only valid in the regime @xmath217 . the formulas for the statistics when the mixing in of thermal fluctuations occurs are more complicated . to make further progress we assume that the frequency dispersion of the measured current correlators is negligible on the scale @xmath218 , @xmath219 for @xmath220 ( here @xmath221 denotes irreducible , or cumulant correlators ) . for a mesoscopic conductor this is satisfied if @xmath218 is smaller than the voltage applied to the conductor . in this limit we find that @xmath222 ^ 2 c_4 + \int_0^{\infty}{\frac{d\omega}{2\pi } } \ , \left [ r^2_{eff}(\omega ) c^2_2 - \frac{\omega}{2 } \biglb(2n(\omega)+1\bigrb ) r_{eff}(\omega ) c_2 \right ] \right\ } \label{eq : hv2 } \end{aligned}\ ] ] with @xmath223 . we have assumed that @xmath224 and @xmath225 for @xmath226 such that the dc - component ( that will be avoided in experiments ) and the contributions from frequencies @xmath226 in eqs . ( [ eq : hv1 ] ) and ( [ eq : hv2 ] ) are negligible . we conclude that the variance of the heat dissipated in @xmath191 depends on the fourth cumulant of current fluctuations in the mesoscopic conductor , as one would expect . there are , however , also contributions from lower order current correlators . the environment circuit has a finite response time @xmath215 and therefore effectively averages fluctuations over that time . as a result , the energy fluctuations become dominated by the lowest order current correlator in the limit of a long @xmath215 , when higher order current correlators become negligible and the statistics becomes gaussian . in this limit of `` narrow - band detection '' the statistics of energy transfer is negative binomial @xcite . only deviations from this encode non - gaussian current correlations . in order to see them the statistics of charge flow through the conductor during @xmath215 has to be strongly non - gaussian . this is the case if @xmath227 , the mean number of transmitted electrons in that period , is small . more concretely , for a measurement of the fourth cumulant @xmath228 of current fluctuations in @xmath192 one would want the first term in eq . ( [ eq : hv2 ] ) to be dominant . an estimate of the three summands in eq . ( [ eq : hv2 ] ) for a tunnel contact with mean current @xmath229 ( assuming that @xmath230 for @xmath231 ) , @xmath232 ( with the conductance quantum @xmath233 ) confirms this . the variance of fluctuations of the heat flux into @xmath191 is a direct measure for the fourth cumulant of current fluctuations in a tunnel contact if @xmath234 and @xmath235 . the origin of the first condition has been explained qualitatively above . the second condition ensures that the fourth cumulant is visible on the thermal background . back action effects of the measuring resistor on the measured tunnel contact ( conductance @xmath85 ) are avoided if additionally @xmath236 @xcite . these requirements are rather restrictive , but can in principle be met in experiments . for a practical implementation , the first condition @xmath237 may be relaxed to @xmath238 by measuring the voltage dependence of @xmath239 and extracting the term that is linear in @xmath229 . we have presented a theory for the statistics of energy exchange between coupled quantum systems . as an application we have calculated the statistics of energy dissipation into the leads connected to a mesoscopic conductor . general and exact expressions can be obtained for the energy flow from a quantum system into a linear environment . we have used these results to calculate the moments of the heat dissipated in a linear circuit around a mesoscopic conductor . they have been expressed in terms of moments of the current fluctuations produced by the conductor . as one expects , the variance of this dissipated heat depends on the fourth cumulative moment of current fluctuations produced by the conductor . heat fluctuations may therefore serve as a tool to detect the fourth cumulant , whose direct measurement is difficult . we have analyzed the conditions under which a measurement of heat fluctuations can reveal higher order cumulants of the charge counting statistics . we thank carlo beenakker , markus bttiker , and yuli nazarov for valuable discussions . this research was supported by the `` nederlandse organisatie voor wetenschappelijk onderzoek '' ( nwo ) and by the `` stichting voor fundamenteel onderzoek der materie '' ( fom ) , by the swiss national science foundation , and by the european community s human potential programme for nanoscale dynamics , coherence and computation under contract hprn - ct-2000 - 00144 .

a method to calculate the statistics of energy exchange between quantum systems is presented . the generating function of this statistics is expressed through a keldysh path integral . the method is first applied to the problem of heat dissipation from a biased mesoscopic conductor into the adjacent reservoirs . we then consider energy dissipation in an electrical circuit around a mesoscopic conductor . we derive the conditions under which measurements of the fluctuations of heat dissipation can be used to investigate higher order cumulants of the charge counting statistics of a mesoscopic conductor . during the second decade of mesoscopic physics there has been an increasing interest in thermal phenomena . several experiments succeeded in measuring properties of heat transport in mesoscopic samples : experimental tools based on the coulomb blockade to measure accurately local temperatures in mesoscopic samples have been established @xcite . it has been demonstrated that the heat conductance through one - dimensional phonon modes of a micro bridge is quantized at low temperatures @xcite . the universal heat conductance per mode is given by @xmath0 . the same universal heat quantum is also found for carriers other than bosons ( see @xcite and refs . therein ) . thermopower was experimentally investigated for a quantum dot in the coulomb blockade regime @xcite and for multiwalled carbon nanotubes @xcite . the andreev reflection process is ineffective for heat transport and has been employed to measure thermopower in andreev interferometers . the thermopower depends on the magnetic flux and shows geometry dependent time inversion properties @xcite . all these experiments have in common that they investigate mean properties averaged over time . in electrical transport , however , noise measurements have become a very useful tool to study properties of non - equilibrium systems which are hidden in measurements of the mean current ( for a review see @xcite ) . a recent milestone was the first successful experimental investigation of higher order current correlators by reulet et al . @xcite . it revealed an unexpected temperature dependence that was explained by accounting for an external measurement circuit @xcite . the calculation of average currents and current fluctuations can be unified within the concept of full counting statistics that has been introduced into mesoscopic physics a decade ago . it was shown by levitov and lesovik that coherent charge transfer through a two - terminal conductor can be seen as probabilistic process defined through a set of transmission probabilities @xcite . thereafter different works investigated the statistics of other measurable quantities in mesoscopic structures such as voltage @xcite , momentum @xcite , and charge inside a mesoscopic volume @xcite . in this publication we address the statistics of fluctuations of energy exchange . we investigate the statistics of energy flow into a subvolume of a quantum system . in the first part of the paper , we write the generating function of this statistics in a keldysh - representation @xcite . the `` counting field '' that generates moments of transferred energy enters as a gauge field that shifts the time evolution of the studied subvolume relative to that of the rest of the system . in section [ sc : tunnel ] we apply our approach to a simple example which nicely demonstrates the basic features of energy transfer : we consider the energy dissipation from a biased mesoscopic conductor into one of its reservoirs . the problem has been addressed before for a conductor at zero temperature @xcite . in @xcite it was shown that dissipation happens symmetrically , that is each reservoir dissipates half of the energy . we find that this is true only on average and we determine fluctuations around this rule . in contrast to the statistics of charge transfer through such a conductor , the transfer of heat is not quantized . in section [ sc : linear ] we derive general relations for the statistics of energy exchange between a quantum system and a linear environment . moments of the exchanged energy are expressed in terms of correlators of the variable that couples the quantum system to its environment . we apply these results to a mesoscopic conductor which is embedded in a macroscopic measurement circuit in section [ sc : circuit ] . the energy transfer statistics is related to correlators of the current through the mesoscopic conductor . one expects that the second moment of energy dissipated in a series resistor depends on the fourth moment of current fluctuations in the mesoscopic conductor . this suggests to use a measurement of heat fluctuations as a probe to study higher order current correlations in mesoscopic conductors . we quantify this expectation and derive the conditions under which higher order current correlators can be extracted from an energy measurement .

in standard cosmology , the acoustic oscillations imprinted in the matter power spectrum at recombination have a length scale that can be accurately calculated based on measurements of the cmb anisotropy power spectrum @xcite . it should then be possible to measure this `` standard ruler '' scale at low redshifts , for example in large galaxy redshift surveys , and thereby constrain the matter and energy content of the universe @xcite . however , if the cmb measurements were misled by some new physics , e.g. a new undetected relativistic particle , then the misinterpretation could potentially spread to the low - redshift application and bias the inferences . here , we show that the interpretation of the low - redshift acoustic oscillations are robust if the cmb correctly tells us the baryon - to - photon ratio and the epoch of matter - radiation equality . these quantities are robustly measured in the cmb . the actual densities of matter and radiation drop out of the calculation ; only their ratio matters . the result is that even if the physical matter density @xmath0 is misinterpreted from the cmb due to undetected relativistic components , the inferences for dark energy from the combined cmb and low - redshift survey data sets are unchanged . knowledge of actual densities , e.g. @xmath1 , translates into improved constraints on the hubble constant , @xmath2 . the acoustic peak method depends upon measuring the sound horizon , which is the comoving distance that a sound wave can travel between the end of inflation and the epoch of recombination @xcite . nearly all of this distance is accumulated just prior to the epoch of recombination at @xmath3 . the sound horizon integral depends only on the hubble parameter @xmath4 and the sound speed @xmath5 in the baryon - photon plasma . if we assume dark energy is sub - dominant at @xmath6 , then @xmath7 where @xmath8 is epoch of matter - radiation equality . the sound speed depends only on the baryon - to - photon ratio and is @xmath9 with @xmath10 . these two produce the sound horizon @xmath11 where ` rec ' and ` eq ' refer to recombination and equality respectively . one sees that the aside from a prefactor of @xmath12 , the sound horizon depends only on the baryon - to - photon ratio and the redshift of equality . the epoch of recombination , being controlled by atomic physics , is very insensitive to the cosmology . for reasonable changes in the early universe and our current uncertainties of the theory of recombination @xcite , any shift in @xmath13 is negligible . the baryon - to - photon ratio is also exquisitely well measured in the cmb power spectrum by both the ratios of the odd and even acoustic peaks and by the silk damping tail @xcite . the former effect depends only on the gravitational inertia of the baryons driven by the potentials near the epoch of recombination . thus the modulation gives us a precise measurement of the baryon - to - photon ratio @xmath14 , which with our precise knowledge of @xmath15 fixes @xmath16 . moreover , for the established value of @xmath17 near @xmath18 , the effect on the sound horizon is already small . it seems very likely that the cmb will determine the baryon - to - photon ratio to sufficient accuracy for this portion of the sound horizon computation @xcite . information about matter - radiation equality is encoded in the amplitudes of the peaks through the manner in which the potentials evolve as they cross the horizon : the potential envelope @xcite . measurements of the potential envelope thus robustly constrain equality . normally , one interprets this constraint as the matter density @xmath1 , on the assumption that the photons and standard neutrino background are the full radiation density . however , one could imagine other relativistic components , and in this case , measuring the redshift of equality does not imply the matter density @xmath1 ( we continue to assume that the extra components are `` undetected '' in the cmb and return to this point in the next section ) . as we can see from eq . ( [ eq : rsound ] ) , the dependence of @xmath19 on @xmath20 is relatively small since @xmath21 , thus even a crude determination suffices to fix @xmath19 up to an overall factor of @xmath12 , i.e. , @xmath22 is very well measured . the sound horizon decreases by only 5% if @xmath20 is lowered by @xmath23 ! understanding the acoustic oscillations at @xmath3 allows us to translate knowledge of the sound horizon into knowledge of wavelength of the baryonic features in the mass power spectrum up to the same normalization uncertainty . we then wish to consider the measurement of this scale at lower redshift , such as could be accomplished in large galaxy surveys . measuring the scale along and across the line of sight , as a redshift or angular scale , constrains @xmath24 and @xmath25 , respectively . for dark energy with ( constant ) equation of state @xmath26 , the low - redshift quantities can be written as @xmath27^{-1/2},\ ] ] and ( for zero curvature ) @xmath28^{-1/2}.\ ] ] because @xmath29 is well constrained , we find that the observations actually constrain @xmath30 and @xmath31 @xcite , which contain only the terms that depend on the bare @xmath32 values , , where @xmath33 is equation [ eq : da ] . but we can write @xmath34 , where @xmath35 is the usual curvature term . this substitution shows that @xmath36 depends only on the bare @xmath32 s . ] i.e. , @xmath37 , @xmath38 , etc . in other words , the prefactors of @xmath12 have canceled out between @xmath19 and the low - redshift distances . we can thus reliably predict the distance _ ratios _ between low @xmath39 and @xmath3 as a function of redshift and hence constrain cosmology . what has happened is simply that the overall scale of the universe does nt affect any of the distance ratios . usually this scale is labeled as the hubble constant , such that @xmath40 drops out . we have rephrased the scale as @xmath1 , despite the fact that this would appear to include @xmath37 in addition to @xmath40 . another way of saying this is that the standard ruler defined by the cmb at constant redshift of equality actually scales as @xmath41 . hence , even @xmath42 redshift - space measurements of this ruler do not measure @xmath40 , but instead measure @xmath43 . parameter estimation for acoustic oscillations based on standard models @xcite will be unchanged by the presence of undetectable new radiation provided that the redshift of equality is measured correctly . only the hubble constant would be incorrect . in the context of cmb parameter estimation , it would be more useful to report @xmath44 rather than @xmath19 itself , as this is the quantity that isolates the cosmological densities at low redshift . massive neutrinos should be counted as radiation at @xmath45 but would be counted as matter at low redshift . that does create a small error in the dark energy inferences . one would be computing @xmath46 from the cmb , but the low - redshift distances in equations ( [ eq : h ] ) and ( [ eq : da ] ) would still have prefactors of @xmath47 . as the observations constrain @xmath25 and @xmath24 , the evaluation of these quantities in terms of bare @xmath32 s will be altered by @xmath48 . for the massive neutrino fraction inferred for non - degenerate neutrino species and the atmospheric neutrino mass splitting @xcite this is a 0.2% correction ; however , it could be a few percent correction at the upper limit of the allowed masses for degenerate neutrino species @xcite . fortunately , such neutrino masses should be detectable in upcoming data due to their suppression of the late - time matter power spectrum @xcite . because the integral for the sound horizon extends to early times ( essentially to the time when the cosmic perturbations were established ) , alterations to the hubble parameter ( eq . [ [ eq : h ] ] ) could alter the sound horizon even at fixed redshift of equality . for example , there might exist a non - relativistic particle that decays into relativistic unseen particles sometime prior to recombination @xcite . however , such decays create alterations to the gravitational potentials that would be detectable in the cmb if the horizon scale at that epoch is visible in the primary anisotropies ( i.e. , wavenumbers @xmath49 ) . this makes decays at @xmath50 difficult to hide . furthermore , such decays alter the transfer function and would affect the amplitude of the late - time matter power spectrum . precision measurement of the matter power spectrum out to @xmath51 from the lyman - alpha forest could push the limits on particle decays to earlier times . in the standard theory , the first 1% of the sound horizon integral is contributed by @xmath52 . it therefore seems very unlikely that a decaying particle could significantly affect the sound horizon ( relative to the standard theory at fixed redshift of equality ) and escape detection in the cmb or large - scale structure . as a corollary , it is interesting to note that when one measures the acoustic oscillation scale at @xmath13 with the cmb and at some low redshift @xmath53 in a galaxy survey , one can construct an observable ( the difference of the suitably scaled angular wavenumbers of the acoustic peaks ) that isolates the cosmology between @xmath53 and @xmath13 , i.e. , @xmath54 this integral is trivial when the universe is completely matter - dominated . hence , if we can perform a galaxy survey at some suitably high redshift , e.g. , @xmath55 , where the dark energy is supposed to be negligible , we could search for dynamical shenanigans at @xmath56 . a photometric - redshift survey over large amount of sky would be an economical route to this goal , as one can acquire very large survey volumes @xcite . we have argued that only the ratios of the radiation and matter densities enter the acoustic oscillation standard ruler method , but of course it is interesting to measure the actual values of the densities as they might reveal new relativistic components or unexpected evolution . the cmb is sensitive to the non - photon radiation density primarily through its effect on equality and the evolution of the potentials . this encodes a sensitivity to both the amount of radiation and its quadrupole moment . in general any particle that free streams rather than behaves as a fluid will have a local quadrupole that will affect the evolution of the potentials and thus the anisotropy in the cmb this breaks the degeneracy in the cmb between changes in @xmath1 and @xmath57 @xcite and allows us to constrain a `` conventional '' neutrino at the @xmath58 level @xcite . at the other extreme , extra radiation which is of the perfect fluid form leads to an increase in the small - scale power which is easily discerned from a change in @xmath1 ( see fig . [ fig : cl ] ) . only a radiation component whose higher moments track closely that of the traditional mix of photons and neutrinos could be confused with a change in @xmath1 . altering the radiation and matter density while holding the baryon ( and photon ) densities fixed would alter the baryon fraction @xmath59 and would produce significant offsets in the late - time matter power spectrum ( e.g. @xcite ) . higher baryon fractions suppress power on small scales compared to large , with the transition occurring near the sound horizon at @xmath60 . the acoustic oscillations shortward of this scale would also be increased . the small - scale amplitude ( e.g. , @xmath61 ) would be reduced . these changes are observable in galaxy surveys , weak lensing surveys , and cluster abundance measurements . of course one can also directly measure the hubble constant @xcite . in short , although the acoustic peak method depends only on the redshift of equality , other cosmological measurements both at @xmath62 and at @xmath63 are well poised to measure the actual densities and thereby constrain the presence of unknown relativistic species . we have shown that the standard ruler defined by the acoustic oscillations prior to recombination can deal gracefully with uncertainties in the matter density @xmath64 provided that the redshift of matter - radiation equality is well measured . the anisotropies of the cmb have very good leverage on this quantity , and so the acoustic peak method of probing dark energy is robust . in addition , it is likely that the cmb , perhaps in combination with other probes , will be able to constrain the actual matter and radiation densities . acknowledgements : we thank eric linder and nikhil padmanabhan for useful discussions and the lawrence berkeley laboratory for the hospitality of its summer workshop program where this work was begun . dje is supported by national science foundation grant ast-0098577 and by an alfred p. sloan research fellowship . mw is supported by the nsf and nasa . peebles , j.t . yu , astrophys . j. , 162 , 815 ( 1970 ) j.r . bond , g. efstathiou , astrophys . j. , 285 , l45 ( 1984 ) j.a . holtzman , astrophys . , 71 , 1 ( 1989 ) d.j . eisenstein , w. hu , m. tegmark , astrophys . j. , 504 , l57 ( 1998 ) a. meiksin , m. white , j.a . peacock , mnras , 304 , 851 ( 1999 ) d.j . eisenstein , in `` wide field multi - object spectroscopy '' , asp conference series , 2003 , ed . a. dey , p. 35 c. blake , k. glazebrook , astrophys . j. , 594 , 665 ( 2003 ) e.v . linder , phys . d , 68 , 083504 ( 2003 ) w. hu , z. haiman , phys . d. , 68 , 063004 ( 2003 ) h .- j . seo , d.j . eisenstein , astrophys . j. , 598 , 720 ( 2003 ) w. hu , s. dodelson , ann . astrophys . , 40 , 171 ( 2002 ) w. hu , d.j . eisenstein , m. tegmark , m. white , phys . d59 , 023512 ( 1999 ) s. seager , d.d . sasselov , d. scott , astrophys . j. lett . , 523 , 1 ( 1999 ) ; s. seager , d.d . sasselov , d. scott , astrophys . j. supp . , 128 , 407 ( 2000 ) w. hu , m. white , astrophys . j. , 471 , 30 ( 1996 ) w. hu , m. white , astrophys . j. , 479 , 568 ( 1997 ) w. hu , d. scott , n. sugiyama , m. white , phys . rev . , d52 , 5498 ( 1995 ) y. fukuda et al . , 81 , 1562 ( 1998 ) ; m. sanchez et al . d68 , 113004 ( 2003 ) ; y. ashie et al . , lett . , submitted ( 2004 ) , hep - ex/0404034 d.n . spergel et al . , astrophys . , 148 , 175 ( 2003 ) ; o. elgaroy & o. lahav , j. cosm . phys . , 04 , 004 ( 2003 ) ; s. hannestad , j. cosm . , 05 , 004 ( 2003 ) ; m. tegmark et al . , d69 , 103501 ( 2004 ) w. hu , d.j . eisenstein , & m. tegmark , phys . lett . , 80 , 5255 ( 1998 ) ; w. hu & m. tegmark , astrophys . j. , 514 , l65 ( 1999 ) ; r.a.c . croft , w. hu , & r. dav , phys . lett . , 83 , 1092 ( 1999 ) j. bardeen , j.r . bond , g. efstathiou , astrophys . j. , 321 , 28 ( 1987 ) ; j.r . bond , g. efstathiou , phys . , b265 , 245 ( 1991 ) e.j . chun , h.b . kim , j.e . kim , phys . , 72 , 1956 ( 1994 ) m. white , g. gelmini , j. silk , phys . d51 , 2669 ( 1995 ) r. bowen , s.h . hansen , a. melchiorri , j. silk , mon . not . soc . , 334 , 760 ( 2002 ) ; e. pierpaoli , mon . not . , 342 , l63 ( 2003 ) ; p. crotty , j. lesgourgues , pastro , s. , phys . , d69 , 123007 ( 2004 ) k. nishikawa , presented at the xxi int . on lepton and photon interactions at high energies ( lepton photon 2003 ) , fermilab , august , 2003 . eisenstein & w. hu , astrophys . j. 496 , 605 ( 1998 ) w.l . freedman , et al . , astrophys . j. , 553 , 47 ( 2001 )

we discuss the systematic uncertainties in the recovery of dark energy properties from the use of baryon acoustic oscillations as a standard ruler . we demonstrate that while unknown relativistic components in the universe prior to recombination would alter the sound speed , the inferences for dark energy from low - redshift surveys are unchanged so long as the microwave background anisotropies can measure the redshift of matter - radiation equality , which they can do to sufficient accuracy . the mismeasurement of the radiation and matter densities themselves ( as opposed to their ratio ) would manifest as an incorrect prediction for the hubble constant at low redshift . in addition , these anomalies do produce subtle but detectable features in the microwave anisotropies .

let @xmath0 and @xmath1 be the caldern zygmund singular integral operator . the commutator @xmath2 $ ] generated by @xmath3 and @xmath1 is defined as @xmath2(f)(x)=b(x)t(f)(x)-t(bf)(x)$ ] . by using a classical result of coifman _ et al _ @xcite , we know that the commutator @xmath4 $ ] is bounded on @xmath5 for @xmath6 . chanillo @xcite proves a similar result when @xmath1 is replaced by the fractional integral operator . however , it was observed that the commutator is not bounded , in general , from @xmath7 to @xmath5 for @xmath8 @xcite . in @xcite , the boundedness properties of the commutator for the extreme values of @xmath9 are obtained . also , in @xcite , chanillo studies some commutators generated by a very general class of pseudo - differential operators and proves the boundedness on @xmath10 for the commutators , and note that the conditions on the kernel of the singular integral operator arise from a pseudo - differential operator . as the development of singular integral operators and their commutators , multilinear singular integral operators have been well - studied . it is known that multilinear operator , as a non - trivial extension of the commutator , is of great interest in harmonic analysis and has been widely studied by many authors @xcite . in @xcite , the weighted @xmath11-boundedness of the multilinear operator related to some singular integral operators is obtained and in @xcite , the weak ( @xmath12 , @xmath13)-boundedness of the multilinear operator related to some singular integral operators is obtained . the main purpose of this paper is to establish the bmo end - point estimates for some vector - valued multilinear operators related to certain singular integral operators . first , let us introduce some notations @xcite . throughout this paper , @xmath14 will denote a cube of @xmath15 with sides parallel to the axes and centered at @xmath16 and having side length . for a locally integrable function @xmath17 and non - negative weight function @xmath18 , let , @xmath19 and @xmath20 @xmath21 . @xmath17 is said to belong to @xmath22 if @xmath23 and define @xmath24 . we denote @xmath25 and @xmath26 if @xmath27 . it is well - known that @xcite @xmath28 we also define the weighted central bmo space by @xmath29 , which is the space of those functions @xmath30 such that @xmath31 for @xmath32 @xcite , that is , @xmath33 & \quad\ , \left . \times \left(\frac{1}{|q|}\int_q w(x)^{-1/(p-1)}{\rm d}x\right)^{p-1}<\infty\right\},\quad 1<p<\infty,\\[.3pc ] a_1 & = \left\{0<w\in l^1_{\rm loc}(r^n)\hbox{:}\ \sup_{x\in q}\frac{w(q)}{|q|}\leq cw(x ) , \hbox{a.e . } \right\}\end{aligned}\ ] ] and @xmath34 @xmath35 1 . let @xmath36 and @xmath37 . we shall call @xmath38 the space of those functions @xmath17 on @xmath15 such that @xmath39 2 . let @xmath6 and @xmath18 be a non - negative weight function on @xmath15 . we shall call @xmath40 the space of that function @xmath17 on @xmath15 such that @xmath41^{-1/p}\|f\chi_{q(0 , r)}\|_{l^p(w)}<\infty.\ ] ] in this paper , we will study a class of vector - valued multilinear operators related to some singular integral operators , whose definitions are the following . fix @xmath42 and @xmath43 . let @xmath44 be a linear operator and there exists a locally integrable function @xmath45 on @xmath46 such that @xmath47 for every bounded and compactly supported function @xmath48 , where @xmath49 satisfies : @xmath50 and @xmath51 if @xmath52 . let @xmath53 be the positive integers ( @xmath54 ) , @xmath55 and @xmath56 be the functions on @xmath57 . for @xmath58 , the vector - valued multilinear operator associated with @xmath1 is defined as @xmath59 and @xmath60 set @xmath61 we write @xmath62 , @xmath63 and @xmath64 if @xmath65 . note that when @xmath66 , @xmath67 is just the multilinear commutators of @xmath68 and @xmath69 @xcite . in this paper , we will prove the bmo estimates for the vector - valued multilinear operators @xmath70 and @xmath71 . now we state our results as follows . let @xmath72 and @xmath73 for all @xmath74 with @xmath75 and @xmath54 . suppose that @xmath76 maps @xmath77 continuously into @xmath78 for any @xmath79 $ ] with @xmath80 and @xmath81 . then 1 . @xmath70 maps @xmath82 continuously into @xmath83 , that is @xmath84 2 . @xmath70 maps @xmath38 continuously into @xmath85 , that is @xmath86 let @xmath87 and @xmath88 for all @xmath74 with @xmath75 and @xmath54 . 1 . if @xmath89 and that @xmath90 is bounded on @xmath91 for any @xmath92 and @xmath93 , then @xmath71 maps @xmath94 continuously into @xmath95 , that is , @xmath96 2 . if @xmath97 and that @xmath90 is bounded on @xmath91 for any @xmath92 and @xmath98 , then @xmath71 maps @xmath40 continuously into @xmath99 , that is , @xmath100 to prove the theorems , we need the following lemmas . -.3pc @xcite . let @xmath69 be a function on @xmath15 and @xmath101 for all @xmath74 with @xmath102 and some @xmath103 . then @xmath104 @xmath35 where @xmath105 is the cube centered at @xmath16 and having side length @xmath106 . let @xmath89 , then @xmath107 . the proof of the lemma follows from @xcite and the john nirenberg lemma for bmo @xcite . it is only to prove that there exists a constant @xmath108 such that @xmath109 holds for any cube @xmath110 . without loss of generality , we may assume @xmath111 . fix a cube @xmath112 . let @xmath113 and @xmath114 , then @xmath115 and @xmath116 for @xmath75 . we split @xmath117 for @xmath118 and @xmath119 . write @xmath120 \hskip -4pc & = \int_{r^n } \frac{\prod_{j=1}^2r_{m_j+1}(\tilde a_j ; x , y)}{|x - y|^m}k(x , y)h_i(y){\rm d}y \\[.5pc ] \hskip -4pc & \quad\ , + \int_{r^n}\frac{\prod_{j=1}^2r_{m_j}(\tilde a_j ; x , y)}{|x - y|^m}k(x , y)g_i(y){\rm d}y\\[.5pc ] \hskip -4pc & \quad\ , -\sum_{|\alpha_1|=m_1}\frac{1}{\alpha_1!}\int_{r^n}\frac{r_{m_2}(\tilde a_2 ; x , y)(x - y)^{\alpha_1}}{|x - y|^m}d^{\alpha_1}\tilde a_1(y)k(x , y)g_i(y){\rm d}y\ ] ] @xmath121 \hskip -4pc \phantom{t_\delta^a(f_i)(x ) } & \quad\ , + \sum_{|\alpha_1|=m_1 , |\alpha_2|=m_2}\frac{1}{\alpha_1!\alpha_2!}\\[.5pc ] \hskip -4pc \phantom{t_\delta^a(f_i)(x ) } & \quad\ , \times \int_{r^n}\frac{(x - y)^{\alpha_1+\alpha_2 } d^{\alpha_1}\tilde a_1(y)d^{\alpha_2}\tilde a_2(y)}{|x - y|^m}k(x , y)g_i(y){\rm d}y,\end{aligned}\ ] ] then , by the minkowski s inequality , @xmath122 & \quad\,\leq\frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty |t_a(f_i)(x)-t_{\tilde a}(h_i)(x_0)|^r\right)^{1/r}{\rm d}x \\[.5pc ] & \quad\,\leq \frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty\left|\int_{r^n}\frac{\prod_{j=1}^2r_{m_j}(\tilde a_j ; x , y)}{|x - y|^m}k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}{\rm d}x \\[.5pc ] & \qquad\,+\frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty\left|\sum_{|\alpha_1|=m_1}\frac{1}{\alpha_1 ! } \right.\right.\\[.5pc ] & \qquad\ , \left.\left . \times \int_{r^n}\frac{r_{m_2}(\tilde a_2 ; x , y)(x - y)^{\alpha_1}}{|x - y|^m}d^{\alpha_1}\tilde a_1(y)k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}{\rm d}x \\[.5pc ] & \qquad\,+\frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty\left|\sum_{|\alpha_2|=m_2}\frac{1}{\alpha_2!}\right.\right.\\[.5pc ] & \qquad\ , \left.\left . \times \int_{r^n}\frac{r_{m_1}(\tilde a_1 ; x , y)(x - y)^{\alpha_2}}{|x - y|^m}d^{\alpha_2}\tilde a_2(y)k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}{\rm d}x \\[.5pc ] & \qquad\,+\frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty\left|\sum_{|\alpha_1|=m_1,\ |\alpha_2|=m_2}\frac{1}{\alpha_1!\alpha_2 ! } \right.\right.\\[.5pc ] & \qquad\ , \left.\left . \times \int_{r^n}\frac{(x - y)^{\alpha_1+\alpha_2}d^{\alpha_1}\tilde a_1(y)d^{\alpha_2}\tilde a_2(y)}{|x - y|^m}k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}{\rm d}x \\[.5pc ] & \qquad\,+\frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty \left|t_\delta^{\tilde a}(h_i)(x)-t_\delta^{\tilde a}(h_i)(x_0)\right|^r\right)^{1/r}{\rm d}x \\[.5pc ] & \quad\,:= { \rm i}_1 + { \rm i}_2 + { \rm i}_3 + { \rm i}_4 + { \rm i}_5.\end{aligned}\ ] ] now , let us estimate @xmath123 and @xmath124 , respectively . first , for @xmath125 and @xmath126 , by lemma 1 , we get @xmath127 thus , by the @xmath128-boundedness of @xmath76 , we get @xmath129 & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||t_\delta(g)|_r\|_{l^\infty } \\[.4pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{l^{n/\delta}}.\end{aligned}\ ] ] for @xmath130 , by the @xmath131-boundedness of @xmath68 for @xmath132 , @xmath133 and the hlder s inequality , we get @xmath134 & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\sum_{|\alpha_1|=m_1}\left(\frac{1}{|q|}\int_{r^n}|t_\delta(d^{\alpha_1}\tilde a_1g)(x)|_r^q{\rm d}x\right)^{1/q}\\[.4pc ] & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\sum_{|\alpha_1|=m_1}|q|^{-1/q}\left(\int_{r^n}|d^{\alpha_1}\tilde a_1(x)g(x)|_r^p{\rm d}x\right)^{1/p } \\[.4pc ] & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\\[.4pc ] & \quad\ , \times \sum_{|\alpha_1|=m_1}\left(\frac{1}{|q|}\int_{\tilde q}|d^{\alpha_1}a_1(x)-(d^{\alpha_1}a_1)_{\tilde q}|^q{\rm d}x\right)^{1/q}\||f|_r\|_{l^{n/\delta } } \\[.4pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{l^{n/\delta}}.\end{aligned}\ ] ] for @xmath135 , similar to the proof of @xmath130 , we get @xmath136 similarly , for @xmath137 , choose @xmath37 and @xmath138 such that @xmath132 and @xmath139 . we obtain , by the hlder s inequality , @xmath140 & \leq c\sum_{|\alpha_1|=m_1 , |\alpha_2|=m_2}\left(\frac{1}{|q|}\int_{r^n}|t_\delta(d^{\alpha_1}\tilde a_1d^{\alpha_2}\tilde a_2g)(x)|_r^q{\rm d}x\right)^{1/q } \\[.4pc ] & \leq c\sum_{|\alpha_1|=m_1 , |\alpha_2|=m_2}|q|^{-1/q}\left(\int_{r^n}|d^{\alpha_1}\tilde a_1(x)d^{\alpha_2}\tilde a_2(x)g(x)|_r^p{\rm d}x\right)^{1/p } \\[.4pc ] & \leq c\sum_{|\alpha_1|=m_1 , |\alpha_2|=m_2}\left(\frac{1}{|q|}\int_{\tilde q}|d^{\alpha_1}\tilde a_1(x)|^{pt_1}{\rm d}x\right)^{1/pt_1}\\[.4pc ] & \quad\ , \times \left(\frac{1}{|q|}\int_{\tilde q}|d^{\alpha_2}\tilde a_2(x)|^{pt_2}{\rm d}x\right)^{1/pt_2}\||f|_r\|_{l^{n/\delta } } \\[.4pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{l^{n/\delta}}.\end{aligned}\ ] ] for @xmath141 , we write @xmath142 \hskip -4pc & \quad\ , = \int_{r^n}\left(\frac{k(x , y)}{|x - y|^m}-\frac{k(x_0,y)}{|x_0-y|^m}\right)\prod_{j=1}^2r_{m_j}(\tilde a_j ; x , y)h_i(y){\rm d}y \\[.4pc ] \hskip -4pc & \qquad\ , + \int_{r^n}\left(r_{m_1}(\tilde a_1 ; x , y)-r_{m_1}(\tilde a_1 ; x_0 , y)\right)\frac{r_{m_2}(\tilde a_2 ; x , y)}{|x_0-y|^m}k(x_0 , y)h_i(y){\rm d}y \\[.4pc ] \hskip -4pc & \qquad\ , + \int_{r^n}\left(r_{m_2}(\tilde a_2 ; x , y)-r_{m_2}(\tilde a_2 ; x_0 , y)\right)\frac{r_{m_1}(\tilde a_1 ; x_0 , y)}{|x_0-y|^m}k(x_0 , y)h_i(y){\rm d}y\\[.4pc ] \hskip -4pc & \qquad\ , -\sum_{|\alpha_1|=m_1}\frac{1}{\alpha_1!}\int_{r^n}\left[\frac{r_{m_2}(\tilde a_2 ; x , y)(x - y)^{\alpha_1}}{|x - y|^m}k(x , y ) \right.\\[.4pc ] \hskip -4pc & \qquad\ \!\left . -\frac{r_{m_2}(\tilde a_2 ; x_0 , y)(x_0-y)^{\alpha_1}}{|x_0-y|^m}k(x_0,y)\right]d^{\alpha_1}\tilde a_1(y)h_i(y){\rm d}y\\[.4pc ] \hskip -4pc & \qquad\ , -\sum_{|\alpha_2|=m_2}\frac{1}{\alpha_2!}\int_{r^n}\left[\frac{r_{m_1}(\tilde a_1 ; x , y)(x - y)^{\alpha_2}}{|x - y|^m}k(x , y ) \right.\\[.4pc ] \hskip -4pc & \qquad\ \!\left . -\frac{r_{m_1}(\tilde a_1 ; x_0 , y)(x_0-y)^{\alpha_2}}{|x_0-y|^m}k(x_0,y)\right]d^{\alpha_2}\tilde a_2(y)h_i(y){\rm d}y\ ] ] @xmath143 \hskip -4pc & \qquad\ , \left . -\frac{(x_0-y)^{\alpha_1+\alpha_2}}{|x_0-y|^m}k(x_0,y)\right]d^{\alpha_1}\tilde a_1(y)d^{\alpha_2}\tilde a_2(y)h_i(y){\rm d}y \\[.4pc ] \hskip -4pc & \quad\ , = { \rm i}_5^{(1 ) } + { \rm i}_5^{(2)}+{\rm i}_5^{(3)}+{\rm i}_5^{(4)}+{\rm i}_5^{(5)}+{\rm i}_5^{(6)}.\end{aligned}\ ] ] by lemma 1 and the following inequality @xcite @xmath144 we know that , for @xmath125 and @xmath145 , @xmath146 & \leq c k|x - y|^{m_j}\sum_{|\alpha|=m_j}\|d^\alpha a_j\|_{\rm bmo}.\end{aligned}\ ] ] note that @xmath147 for @xmath125 and @xmath148 , and we obtain , by the conditions on @xmath49 , @xmath149 \hskip -4pc & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\\[.3pc ] \hskip -4pc & \quad\ , \times \sum_{k=0}^\infty\int_{2^{k+1}\tilde q\setminus2^k\tilde q}k^2\left(\frac{|x - x_0| } { |x_0-y|^{n+1-\delta}}+\frac{|x - x_0|^\varepsilon}{|x_0-y|^{n+\varepsilon- \delta}}\right)|f_i(y)|{\rm d}y.\end{aligned}\ ] ] thus , by the minkowski s inequality , @xmath150 \hskip -4pc & \quad\ , \times \sum_{k=0}^\infty\int_{2^{k+1}\tilde q\setminus2^k\tilde q}k^2\left(\frac{|x - x_0|}{|x_0-y|^{n+1-\delta } } + \frac{|x - x_0|^\varepsilon}{|x_0-y|^{n+\varepsilon-\delta}}\right)|f(y)|_r{\rm d}y \\[.3pc ] \hskip -4pc & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right ) \sum_{k=1}^\infty k^2(2^{-k}+2^{-\varepsilon k})\||f|_r\|_{l^{n/\delta } } \\[.3pc ] \hskip -4pc & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{l^{n/\delta}}.\end{aligned}\ ] ] for @xmath151 , by the formula @xcite : @xmath152 @xmath35 and lemma 1 , we have @xmath153 & \quad\ , \leq c\sum_{|\beta|<m_j}\sum_{|\alpha|=m_j}|x - x_0|^{m_j-|\beta|}|x- y|^{|\beta|}\|d^\alpha a_j\|_{\rm bmo}.\end{aligned}\ ] ] thus @xmath154 & \quad\ , \times \sum_{k=0}^\infty\int_{2^{k+1}\tilde q\setminus2^k\tilde q}k\frac{|x - x_0|}{|x_0-y|^{n+1-\delta}}|f(y)|_r{\rm d}y \\[.3pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{l^{n/\delta}}.\end{aligned}\ ] ] similarly , @xmath155 for @xmath156 , taking @xmath157 such that @xmath158 , then @xmath159 \hskip -4pc & \quad\,\times |r_{m_2}(\tilde a_2 ; x , y)||d^{\alpha_1}\tilde a_1(y)||h(y)|_r{\rm d}y \\[.3pc ] \hskip -4pc & \quad\,+c\sum_{|\alpha_1|=m_1}\int_{r^n}|r_{m_2}(\tilde a_2 ; x , y)-r_{m_2}(\tilde a_2 ; x_0 , y)|\\[.2pc ] \hskip -4pc & \quad\ , \times\frac{|(x_0-y)^{\alpha_1}k(x_0,y)| } { |x_0-y|^m}|d^{\alpha_1}\tilde a_1(y)||h(y)|_r{\rm d}y\\[.3pc ] \hskip -4pc & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\sum_{|\alpha_1|=m_1}\sum_{k=1}^\infty k(2^{-k}+2^{-\varepsilon k})\\[.3pc ] \hskip -4pc & \quad\,\times \left(\frac{1}{|2^k\tilde q|}\int_{2^k\tilde q}|d^{\alpha_1}\tilde a_1(y)|^t{\rm d}y\right)^{1/t}\||f|_r\|_{l^{n/\delta } } \\[.3pc ] \hskip -4pc & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{l^{n/\delta}}.\end{aligned}\ ] ] similarly , @xmath160 for @xmath161 , taking @xmath162 such that @xmath163 , then @xmath164 & \quad\ , \left . -\frac{(x_0-y)^{\alpha_1+\alpha_2}k(x_0,y ) } { |x_0-y|^m}\right| |d^{\alpha_1}\tilde a_1(y)||d^{\alpha_2}\tilde a_2(y)||h(y)|_r{\rm d}y \\[.4pc ] \hskip -4pc & \leq c\sum_{|\alpha_1|=m_1,|\alpha_2|=m_2}\sum_{k=1}^\infty(2^{-k}+2^{-\varepsilon k})\||f|_r\|_{l^{n/\delta } } \\[.4pc ] \hskip -4pc & \quad\ , \times\left(\frac{1}{|2^k\tilde q|}\int_{2^k\tilde q}|d^{\alpha_1}\tilde a_1(y)|^{t_1}{\rm d}y\right)^{1/t_1}\\[.4pc ] \hskip -4pc & \quad\ , \times \left(\frac{1}{|2^k\tilde q|}\int_{2^k\tilde q}|d^{\alpha_2}\tilde a_2(y)|^{t_2}{\rm d}y\right)^{1/t_2 } \\[.4pc ] \hskip -4pc & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{l^{n/\delta}}.\end{aligned}\ ] ] thus @xmath165 it suffices to prove that there exists a constant @xmath108 such that @xmath166 holds for any cube @xmath167 with @xmath168 . without loss of generality , we may assume @xmath111 . fix a cube @xmath167 with @xmath168 . let @xmath169 and @xmath170 be the same as the proof of ( a ) . write , for @xmath117 with @xmath118 and @xmath119 , @xmath171 & \quad\,\leq\frac{1}{|q| } \int_q\left(\sum_{i=1}^\infty |t_a(f_i)(x)-t_{\tilde a}(h_i)(0)|^r\right)^{1/r}{\rm d}x \\[.4pc ] & \quad\,\leq \frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty\left|\int_{r^n}\frac{\prod_{j=1}^2r_{m_j}(\tilde a_j ; x , y)}{|x - y|^m}k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}{\rm d}x\ ] ] @xmath172 & \qquad\ \!\!\left \times \int_{r^n}\frac{r_{m_2}(\tilde a_2 ; x , y ) ( x - y)^{\alpha_1}}{|x - y|^m}d^{\alpha_1}\tilde a_1(y)k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}{\rm d}x \\[.3pc ] & \qquad\ , + \frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty\left|\sum_{|\alpha_2|=m_2}\frac{1}{\alpha_2!}\right.\right.\\[.3pc ] & \qquad\ \!\!\left . \times\int_{r^n}\frac{r_{m_1}(\tilde a_1 ; x , y ) ( x - y)^{\alpha_2}}{|x - y|^m}d^{\alpha_2}\tilde a_2(y)k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}{\rm d}x \\[.3pc ] & \qquad\ , + \frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty\left|\sum_{|\alpha_1|=m_1,\ |\alpha_2|=m_2}\frac{1}{\alpha_1!\alpha_2!}\right.\right.\\[.3pc ] & \qquad\ \!\!\left . \times \int_{r^n}\frac{(x - y)^{\alpha_1+\alpha_2}d^{\alpha_1}\tilde a_1(y)d^{\alpha_2}\tilde a_2(y)}{|x - y|^m}k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}{\rm d}x \\[.3pc ] & \qquad\,+\frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty \left|t_\delta^{\tilde a}(h_i)(x)-t_\delta^{\tilde a}(h_i)(0)\right|^r\right)^{1/r}{\rm d}x \\[.3pc ] & \quad\,:= { \rm j}_1+{\rm j}_2+{\rm j}_3+{\rm j}_4+{\rm j}_5.\end{aligned}\ ] ] similar to the proof of ( a ) , we get , for @xmath81 , @xmath173 , @xmath174 and @xmath175 , @xmath176 & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\left(\frac{1}{|q|}\int_q|t_\delta(g)(x)|_r^q{\rm d}x\right)^{1/q } \\[.3pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)d^{-n(1/p-\delta / n)}\||f|_r\chi_{\tilde q}\|_{l^p } \\[.3pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p^{\delta}},\\[.3pc ] { \rm j}_2 & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\sum_{|\alpha_1|=m_1}\frac{1}{|q|}\int_q |t_\delta(d^{\alpha_1}\tilde a_1g)(x)|_r{\rm d}x \\[.3pc ] & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\sum_{|\alpha_1|=m_1}\left(\frac{1}{|q|}\int_{r^n}|t_\delta(d^{\alpha_1}\tilde a_1g)(x)|_r^t{\rm d}x\right)^{1/t}\ ] ] @xmath177 & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\\[.4pc ] & \quad\ , \times \sum_{|\alpha_1|=m_1 } \left(\frac{1}{|q| } \int_{\tilde q } & \quad\ , \times |q|^{\delta / n-1/p}\||f|_r\chi_{\tilde q}\|_{l^p } \\[.4pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p^{\delta } } , \\[.4pc ] { \rm j}_3&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p^{\delta}},\\[.4pc ] { \rm j}_4&\leq c\sum_{|\alpha_1|=m_1 , |\alpha_2|=m_2}\frac{1}{|q|}\int_q |t_\delta(d^{\alpha_1}\tilde a_1d^{\alpha_2}\tilde a_2g)(x)|_r{\rm d}x \\[.4pc ] & \leq c\sum_{|\alpha_1|=m_1 , |\alpha_2|=m_2}\left(\frac{1}{|q|}\int_{r^n}|t_\delta(d^{\alpha_1}\tilde a_1d^{\alpha_2}\tilde a_2f_1)(x)|_r^q{\rm d}x\right)^{1/q } \\[.4pc ] & \leq c\sum_{|\alpha_1|=m_1 , |\alpha_2|=m_2}|q|^{-1/s}\left(\int_{r^n}|d^{\alpha_1}\tilde a_1(x)d^{\alpha_2}\tilde a_2(x)g(x)|_r^s{\rm d}x\right)^{1/s } \\[.4pc ] & \leq c\sum_{|\alpha_1|=m_1}\left(\frac{1}{|q|}\int_{\tilde q}|d^{\alpha_1}\tilde a_1(x)|^{st_1}{\rm d}x\right)^{1/st_1}\\[.4pc ] & \quad\ , \times \sum_{|\alpha_2|=m_2}\left(\frac{1}{|q|}\int_{\tilde q}|d^{\alpha_2}\tilde a_2(x)|^{st_2}{\rm d}x\right)^{1/st_2}|q|^{\delta / n-1/p}\||f|_r\chi_{\tilde q}\|_{l^p } \\[.4pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p^{\delta}}.\end{aligned}\ ] ] for @xmath178 , we write , for @xmath125 , @xmath179 & \quad\,=\int_{r^n}\left(\frac{k(x , y)}{|x - y|^m}-\frac{k(0,y)}{|y|^m}\right)\prod_{j=1}^2r_{m_j}(\tilde a_j ; x , y)h_i(y){\rm d}y \\[.3pc ] & \qquad\,+\int_{r^n}(r_{m_1}(\tilde a_1 ; x , y)-r_{m_1}(\tilde a_1 ; 0 , y))\frac{r_{m_2}(\tilde a_2 ; x , y)}{|y|^m}k(0 , y)h_i(y){\rm d}y\ ] ] @xmath180 & \qquad\,-\sum_{|\alpha_1|=m_1}\frac{1}{\alpha_1!}\int_{r^n}\left[\frac{r_{m_2}(\tilde a_2 ; x , y)(x - y)^{\alpha_1}}{|x - y|^m}k(x , y ) \right.\\[.5pc ] & \qquad\quad\ , \left . -\frac{r_{m_2}(\tilde a_2 ; 0 , y)(-y)^{\alpha_1}}{|y|^m}k(0,y)\right]d^{\alpha_1}\tilde a_1(y)h_i(y){\rm d}y \\[.5pc ] & \qquad\,-\sum_{|\alpha_2|=m_2}\frac{1}{\alpha_2!}\int_{r^n}\left[\frac{r_{m_1}(\tilde a_1 ; x , y)(x - y)^{\alpha_2}}{|x - y|^m}k(x , y)\right.\\[.5pc ] & \qquad\quad\ , \left . -\frac{r_{m_1}(\tilde a_1 ; 0 , y)(-y)^{\alpha_2}}{|y|^m}k(0,y)\right]d^{\alpha_2}\tilde a_2(y)h_i(y){\rm d}y \\[.5pc ] & \qquad\,+\sum_{|\alpha_1|=m_1,\ |\alpha_2|=m_2}\frac{1}{\alpha_1!\alpha_2!}\int_{r^n}\left[\frac{(x - y)^{\alpha_1+\alpha_2}}{|x - y|^m}k(x , y)\right.\\[.5pc ] & \qquad\quad\ , \left . -\frac{(-y)^{\alpha_1+\alpha_2}}{|y|^m}k(0,y)\right]d^{\alpha_1}\tilde a_1(y)d^{\alpha_2}\tilde a_2(y)h_i(y){\rm d}y \\[.6pc ] & \quad\,= { \rm j}_5^{(1)}+{\rm j}_5^{(2)}+{\rm j}_5^{(3)}+{\rm j}_5^{(4)}+{\rm j}_5^{(5)}+{\rm j}_5^{(6)}.\end{aligned}\ ] ] similar to the proof of ( a ) , we get , for @xmath174 and @xmath181 , @xmath182 \hskip -4pc & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\\[.4pc ] \hskip -4pc & \quad\ , \times \sum_{k=0}^\infty\int_{2^{k+1}\tilde q\setminus2^k\tilde q}k^2\left(\frac{|x| } { |y|^{n+1-\delta}}+\frac{|x|^\varepsilon}{|y|^{n+\varepsilon-\delta}}\right)|f(y)|_r{\rm d}y \\[.4pc ] \hskip -4pc & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\\[.4pc ] \hskip -4pc & \quad\ , \times \sum_{k=1}^\infty k^2(2^{-k}+2^{-\varepsilon k})(2^k d)^{-n(1/p-\delta / n)}\||f|_r\chi_{2^k\tilde q}\|_{l^p } \\[.4pc ] \hskip -4pc & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p^{\delta}},\ ] ] @xmath183 \hskip -4pc & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p^{\delta}},\\[.5pc ] \hskip -4pc \left(\sum_{i=1}^\infty|{\rm j}_5^{(3)}|^r\right)^{1/r}&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p^{\delta}},\\[.5pc ] \hskip -4pc \left(\sum_{i=1}^\infty|{\rm j}_5^{(4)}|^r\right)^{1/r}&\leq c\sum_{|\alpha_1|=m_1}\int_{r^n}\left|\frac{(x - y)^{\alpha_1}k(x , y)}{|x - y|^m}-\frac{(-y)^{\alpha_1}k(0 , y ) } { |y|^m}\right|\\[.5pc ] \hskip -4pc & \quad\ , \times |r_{m_2}(\tilde a_2 ; x , y)||d^{\alpha_1}\tilde a_1(y)||h(y)|_r{\rm d}y \\[.5pc ] \hskip -4pc & + c\sum_{|\alpha_1|=m_1}\int_{r^n}|r_{m_2}(\tilde a_2 ; x , y)-r_{m_2}(\tilde a_2 ; 0 , y)|\\[.5pc ] \hskip -4pc & \quad\ , \times \frac{|(-y)^{\alpha_1}k(0 , y)| } { |y|^m}|d^{\alpha_1}\tilde a_1(y)||h(y)|_r{\rm d}y \\[.5pc ] \hskip -4pc & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\\[.6pc ] \hskip -4pc & \quad\ , \times \sum_{k=1}^\infty k(2^{-k}+2^{-\varepsilon k})(2^k d)^{-n(1/p-\delta / n)}\||f|_r\chi_{2^k\tilde q}\|_{l^p } \\[.5pc ] \hskip -4pc & \times\sum_{|\alpha_1|=m_1}\left(\frac{1}{|2^k\tilde q|}\int_{2^k\tilde q}|d^{\alpha_1}\tilde a_1(y)|^{p'}{\rm d}y\right)^{1/p ' } \\[.5pc ] \hskip -4pc & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p^{\delta}},\\[.5pc ] \hskip -4pc \left(\sum_{i=1}^\infty|{\rm j}_5^{(5)}|^r\right)^{1/r}&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p^{\delta}},\\[.5pc ] \hskip -4pc \left(\sum_{i=1}^\infty|{\rm j}_5^{(6)}|^r\right)^{1/r}&\leq c\sum_{k=1}^\infty(2^{-k}+2^{-\varepsilon k } ) ( 2^k d)^{-n(1/p-\delta / n)}\||f|_r\chi_{2^k\tilde q}\|_{l^p } \\[.5pc ] \hskip -4pc & \quad\,\times \sum_{|\alpha_1|=m_1}\left(\frac{1}{|2^k\tilde q|}\int_{2^k\tilde q}|d^{\alpha_1}\tilde a_1(y)|^{t_1}{\rm d}y\right)^{1/t_1}\ ] ] @xmath184 \hskip -4pc & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p^{\delta}}.\end{aligned}\ ] ] thus @xmath185 this completes the proof of theorem 1 . it suffices to show the conclusion for the case @xmath25 by lemma 2 , that is , it is only to prove that there exists a constant @xmath108 such that @xmath186 holds for any cube @xmath110 . without loss of generality , we may assume @xmath111 . fix a cube @xmath112 . let @xmath169 and @xmath170 be the same as the proof of theorem 1 . write , for @xmath117 with @xmath118 and @xmath119 , @xmath187 & \quad\ , \leq\frac{1}{|q| } \int_q\left(\sum_{i=1}^\infty|t_a(f_i)(x)-t_{\tilde a}(h_i)(x_0)|^r\right)^{1/r}{\rm d}x \\[.3pc ] & \quad\ , \leq \frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty\left|\int_{r^n}\frac{\prod_{j=1}^2r_{m_j}(\tilde a_j ; x , y)}{|x - y|^m}k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}{\rm d}x \\[.3pc ] & \qquad\ , + \frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty\left|\sum_{|\alpha_1|=m_1}\frac{1}{\alpha_1 ! } \right.\right.\\[.3pc ] & \qquad\quad\ , \left.\left . \times \int_{r^n}\frac{r_{m_2}(\tilde a_2 ; x , y ) ( x - y)^{\alpha_1}}{|x - y|^m}d^{\alpha_1}\tilde a_1(y)k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}{\rm d}x\\[.3pc ] & \qquad\ , + \frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty\left|\sum_{|\alpha_2|=m_2}\frac{1}{\alpha_2!}\right.\right.\\[.3pc ] & \qquad\quad\ , \left.\left . \times \int_{r^n}\frac{r_{m_1}(\tilde a_1 ; x , y ) ( x - y)^{\alpha_2}}{|x - y|^m}d^{\alpha_2}\tilde a_2(y)k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}{\rm d}x \ ] ] @xmath188 & \qquad\quad\ , \left.\left . \times \int_{r^n}\frac{(x - y)^{\alpha_1+\alpha_2}d^{\alpha_1}\tilde a_1(y)d^{\alpha_2}\tilde a_2(y)}{|x - y|^m}k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}{\rm d}x \\[.3pc ] & \qquad\ , + \frac{1}{|q|}\int_q\left(\sum_{i=1}^\infty |t_\delta^{\tilde a}(h_i)(x)-t_\delta^{\tilde a}(h_i)(x_0)|^r\right)^{1/r}{\rm d}x \\[.3pc ] & : = { \rm l}_1+{\rm l}_2+{\rm l}_3+{\rm l}_4+{\rm l}_5.\end{aligned}\ ] ] by the @xmath189-boundedness of @xmath90 for @xmath190 and using the same argument as in the proof of theorem 1 , we get , for @xmath191 , @xmath192 & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||t(g)|_r\|_{l^\infty } \\[.4pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{l^\infty(w ) } , \\[.4pc ] { \rm l}_2 & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\sum_{|\alpha_1|=m_1}\frac{1}{|q|}\int_q |t(d^{\alpha_1}\tilde a_1g)(x)|_r{\rm d}x \\[.4pc ] & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\sum_{|\alpha_1|=m_1}\left(\frac{1}{|q|}\int_{r^n}|t(d^{\alpha_1}\tilde a_1g)(x)|_r^s{\rm d}x\right)^{1/s } \\[.4pc ] & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\sum_{|\alpha_1|=m_1}\left(\frac{1}{|q|}\int_{r^n}|d^{\alpha_1}\tilde a_1(x)g(x)|_r^s{\rm d}x\right)^{1/s } \\[.4pc ] & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\\[.4pc ] & \quad\ , \times \sum_{|\alpha_1|=m_1}\left(\frac{1}{|q|}\int_{\tilde q}|d^{\alpha_1}a_1(x ) -(d^{\alpha_1}a_1)_{\tilde q}|^s{\rm d}x\right)^{1/s}\||f|_r\|_{l^\infty } \\[.4pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{l^\infty(w)},\\[.4pc ] { \rm l}_3 & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{l^\infty(w)},\ ] ] @xmath193 & \leq c\sum_{|\alpha_1|=m_1 , |\alpha_2|=m_2}\left(\frac{1}{|q|}\int_{r^n}|t(d^{\alpha_1}\tilde a_1d^{\alpha_2}\tilde a_2g)(x)|_r^s{\rm d}x\right)^{1/s } \\[.4pc ] & \leq c\sum_{|\alpha_1|=m_1 , |\alpha_2|=m_2}w(q)^{-1/s}\left(\int_{r^n}|d^{\alpha_1}\tilde a_1(x)d^{\alpha_2}\tilde a_2(x)g(x)|_r^s{\rm d}x\right)^{1/s } \\[.4pc ] & \leq c\sum_{|\alpha_1|=m_1 , |\alpha_2|=m_2}\left(\frac{1}{|q|}\int_{\tilde q}|d^{\alpha_1}\tilde a_1(x)|^{st_1}{\rm d}x\right)^{1/st_1}\\[.4pc ] & \quad\ , \times \left(\frac{1}{|q|}\int_{\tilde q}|d^{\alpha_2}\tilde a_2(x)|^{st_2}{\rm d}x\right)^{1/st_2}\||f|_r\|_{l^\infty } \\[.4pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{l^\infty(w)},\\[.4pc ] { \rm l}_5&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{l^\infty(w)}.\end{aligned}\ ] ] it suffices to prove that there exists a constant @xmath108 such that @xmath194 holds for any cube @xmath167 with @xmath168 . without loss of generality , we may assume @xmath111 . fix a cube @xmath167 with @xmath168 . let @xmath169 and @xmath170 be the same as the proof of theorem 1 . write , for @xmath117 with @xmath195 and @xmath119 , @xmath196 \hskip -4pc & \quad\,\leq\frac{1}{w(q)}\int_q\left(\sum_{i=1}^\infty \hskip -4pc & \quad\,\leq \frac{1}{w(q)}\int_q\left(\sum_{i=1}^\infty\left|\int_{r^n}\frac{\prod_{j=1}^2r_{m_j}(\tilde a_j ; x , y)}{|x - y|^m}k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}w(x){\rm d}x \\[.4pc ] \hskip -4pc & \qquad\,+\frac{1}{w(q)}\int_q\left(\sum_{i=1}^\infty\left|\sum_{|\alpha_1|=m_1}\frac{1}{\alpha_1 ! } \right.\right.\\[.4pc ] \hskip -4pc & \qquad\quad\ , \times \left.\left . \int_{r^n}\frac{r_{m_2}(\tilde a_2 ; x , y ) ( x - y)^{\alpha_1}}{|x - y|^m}d^{\alpha_1}\tilde a_1(y)k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}w(x){\rm d}x \ ] ] @xmath197 \hskip -4pc & \qquad\quad\ , \times \left.\left . \int_{r^n}\frac{r_{m_1}(\tilde a_1 ; x , y ) ( x - y)^{\alpha_2}}{|x - y|^m}d^{\alpha_2}\tilde a_2(y)k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}w(x){\rm d}x \\[.4pc ] \hskip -4pc & \qquad\,+\frac{1}{w(q)}\int_q\left(\sum_{i=1}^\infty\left|\sum_{|\alpha_1|=m_1,\ |\alpha_2|=m_2}\frac{1}{\alpha_1!\alpha_2!}\right.\right.\\[.4pc ] \hskip -4pc & \qquad\quad\ , \times \left.\left . \int_{r^n}\frac{(x - y)^{\alpha_1+\alpha_2}d^{\alpha_1}\tilde a_1(y)d^{\alpha_2}\tilde a_2(y)}{|x - y|^m}k(x , y)g_i(y){\rm d}y\right|^r\right)^{1/r}w(x){\rm d}x \\[.4pc ] \hskip -4pc & \qquad\,+\frac{1}{w(q)}\int_q\left(\sum_{i=1}^\infty |t_\delta^{\tilde a}(h_i)(x)-t_\delta^{\tilde a}(h_i)(0)|^r\right)^{1/r}w(x){\rm d}x \\[.4pc ] \hskip -4pc & \quad\,:= { \rm m}_1+{\rm m}_2+{\rm m}_3+{\rm m}_4+{\rm m}_5.\end{aligned}\ ] ] similar to the proof of theorem 1 , we get @xmath198 & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\left(\frac{1}{w(q)}\int_q|t(g)(x)|_r^pw(x){\rm d}x\right)^{1/p } \\[.3pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)w(\tilde q)^{-1/p}\||f|_r\chi_{\tilde q}\|_{l^p(w ) } \\[.3pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p(w)}.\end{aligned}\ ] ] for @xmath199 , since @xmath98 , @xmath18 satisfies the reverse of hlder s inequality : @xmath200 for all cube @xmath110 and some @xmath201 @xcite . thus , taking @xmath202 such that @xmath203 and @xmath204 , then @xmath205 & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\ ] ] @xmath206 & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\sum_{|\alpha_1|=m_1}w(q)^{-1/s}\|d^{\alpha_1}\tilde a_1|g|_r\|_{l^s(w ) } \\[.3pc ] & \leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}w(q)^{-1/s}\sum_{|\alpha_1|=m_1}\left(\int_{\tilde q}|d^\alpha\tilde a_1(y)|^{st'}{\rm d}y\right)^{1/st'}\\[.3pc ] & \quad\ , \times \left(\int_{\tilde q}|f(x)|_r^{st}w(x)^t{\rm d}x\right)^{1/st } \\[.3pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)|q|^{1/st'}w(q)^{-1/s}\\[.3pc ] & \quad\ , \times \left(\int_{\tilde q}|f(x)|_r^pw(x){\rm d}x\right)^{1/p}\left(\int_{\tilde q}w(x)^q{\rm d}x\right)^{(p - s)/pqs } \\[.3pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)w(\tilde q)^{-1/p}\||f|_r\chi_{\tilde q}\|_{l^p(w ) } \\[.3pc ] & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p(w ) } , \\[.3pc ] { \rm m}_3 & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p(w)}.\end{aligned}\ ] ] for @xmath207 , taking @xmath208 such that @xmath209 , @xmath210 and @xmath211 , then , by the reverse of hlder s inequality , @xmath212 \hskip -4pc & \leq c\sum_{|\alpha_1|=m_1 , |\alpha_2|=m_2}\left(\frac{1}{w(q)}\int_{r^n}|t(d^{\alpha_1}\tilde a_1d^{\alpha_2}\tilde a_2g)(x)|_r^sw(x){\rm d}x\right)^{1/s } \\[.3pc ] \hskip -4pc & \leq c\sum_{|\alpha_1|=m_1 , |\alpha_2|=m_2}w(q)^{-1/s}\left(\int_{r^n}|d^{\alpha_1}\tilde a_1(x)d^{\alpha_2}\tilde a_2(x)g(x)|_r^sw(x){\rm d}x\right)^{1/s } \\[.3pc ] \hskip -4pc & \leq c\sum_{|\alpha_1|=m_1}\left(\int_{\tilde q}|d^{\alpha_1}\tilde a_1(x)|^{st_1}{\rm d}x\right)^{1/st_1}\\[.3pc ] \hskip -4pc & \quad\times \sum_{|\alpha_2|=m_2}\left(\int_{\tilde q}|d^{\alpha_2}\tilde a_2(x)|^{st_2}{\rm d}x\right)^{1/st_2 } w(q)^{-1/s } \left(\int_{\tilde q}|f(x)|_r^{st_3}w(x)^{t_3}{\rm d}x\right)^{1/st_3 } \ ] ] @xmath213 \hskip -4pc & \leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p(w)}.\end{aligned}\ ] ] for @xmath214 , we write , for @xmath125 , @xmath215 & \quad\,=\int_{r^n}\left(\frac{k(x , y)}{|x - y|^m}-\frac{k(0,y)}{|y|^m}\right)\prod_{j=1}^2r_{m_j}(\tilde a_j ; x , y)h_i(y){\rm d}y \\[.5pc ] & \qquad\,+\int_{r^n}(r_{m_1}(\tilde a_1 ; x , y)-r_{m_1}(\tilde a_1 ; 0 , y))\frac{r_{m_2}(\tilde a_2 ; x , y)}{|y|^m}k(0 , y)h_i(y){\rm d}y \\[.5pc ] & \qquad\,+\int_{r^n}(r_{m_2}(\tilde a_2 ; x , y)-r_{m_2}(\tilde a_2 ; 0 , y))\frac{r_{m_1}(\tilde a_1 ; 0 , y)}{|y|^m}k(0 , y)h_i(y){\rm d}y \\[.5pc ] & \qquad\,-\sum_{|\alpha_1|=m_1}\frac{1}{\alpha_1!}\int_{r^n}\left[\frac{r_{m_2}(\tilde a_2 ; x , y)(x - y)^{\alpha_1}}{|x - y|^m}k(x , y ) \right.\\[.5pc ] & \qquad\quad\,\left . -\frac{r_{m_2}(\tilde a_2 ; 0 , y)(-y)^{\alpha_1}}{|y|^m}k(0,y)\right]d^{\alpha_1}\tilde a_1(y)h_i(y){\rm d}y \\[.5pc ] & \qquad\,-\sum_{|\alpha_2|=m_2}\frac{1}{\alpha_2!}\int_{r^n}\left[\frac{r_{m_1}(\tilde a_1 ; x , y)(x - y)^{\alpha_2}}{|x - y|^m}k(x , y)\right.\\[.5pc ] & \qquad\quad\,\left . -\frac{r_{m_1}(\tilde a_1 ; 0 , y)(-y)^{\alpha_2}}{|y|^m}k(0,y)\right]d^{\alpha_2}\tilde a_2(y)h_i(y){\rm d}y \\[.5pc ] & \qquad\,+\sum_{|\alpha_1|=m_1,\ |\alpha_2|=m_2}\frac{1}{\alpha_1!\alpha_2!}\int_{r^n}\left[\frac{(x - y)^{\alpha_1+\alpha_2}}{|x - y|^m}k(x , y)\right.\\[.5pc ] & \qquad\quad\,\left . -\frac{(-y)^{\alpha_1+\alpha_2}}{|y|^m}k(0,y)\right]d^{\alpha_1}\tilde a_1(y)d^{\alpha_2}\tilde a_2(y)h_i(y){\rm d}y \\[.5pc ] & \quad\,= { \rm m}_5^{(1)}+{\rm m}_5^{(2)}+{\rm m}_5^{(3)}+{\rm m}_5^{(4)}+{\rm m}_5^{(5)}+{\rm m}_5^{(6)}.\end{aligned}\ ] ] similar to the proof of theorem 1 and notice that @xmath216 , we get @xmath217 @xmath218 \hskip -4pc&\quad\ , \times \sum_{k=0}^\infty\int_{2^{k+1}\tilde q\setminus2^k\tilde q } k^2\left(\frac{|x|}{|y|^{n+1}}+\frac{|x|^\varepsilon}{|y|^{n+\varepsilon}}\right)|f(y)|_r{\rm d}y\\[.4pc ] \hskip -4pc&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\\[.4pc ] \hskip -4pc & \quad\ , \times \sum_{k=1}^\infty k^2(2^{-k}+2^{-\varepsilon k } ) w(2^k\tilde q)^{-1/p}\left(\int_{2^k\tilde q}|f(y)|_r^pw(y){\rm d}y\right)^{1/p } \\[.4pc ] \hskip -4pc&\times\left(\frac{1}{|2^k\tilde q|}\int_{2^k\tilde q}w(y){\rm d}y\right)^{1/p}\left(\frac{1}{|2^k\tilde q|}\int_{2^k\tilde q}w(y)^{-1/(p-1)}{\rm d}y\right)^{(p-1)/p } \\[.4pc ] \hskip -4pc&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p(w)},\\[.4pc ] \hskip -4pc\left(\sum_{i=1}^\infty|{\rm m}_5^{(2)}|^r\right)^{1/r}&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right ) \sum_{k=0}^\infty\int_{2^{k+1}\tilde q\setminus2^k\tilde q}k\frac{|x|}{|y|^{n+1}}|f(y)|_r{\rm d}y \\[.4pc ] \hskip -4pc&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p(w ) } , \\[.4pc ] \hskip -4pc\left(\sum_{i=1}^\infty|{\rm m}_5^{(3)}|^r\right)^{1/r}&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p(w)}.\end{aligned}\ ] ] for @xmath219 , choose @xmath173 , notice that @xmath220 , we get @xmath221 \hskip -4pc & \quad\ , \times \sum_{k=0}^\infty \int_{2^{k+1}\tilde q\setminus2^k\tilde q}k\left(\frac{|x|}{|y|^{n+1}}+\frac{|x|^\varepsilon}{|y|^{n+\varepsilon}}\right)|d^{\alpha_1}\tilde a_1(y)||f(y)|_r{\rm d}y \\[.4pc ] \hskip -4pc&\leq c\sum_{|\alpha_2|=m_2}\|d^{\alpha_2}a_2\|_{\rm bmo}\sum_{k=0}^\infty\left(\frac{d}{(2^kd)^{n+1}}+\frac{d^\varepsilon}{(2^kd)^{n+\varepsilon}}\right)\\[.4pc ] \hskip -4pc&\quad\ , \times \left(\int_{2^{k+1}\tilde q}|f(y)|_r^s{\rm d}y\right)^{1/s}{\rm d}y \left(\int_{2^{k+1}\tilde q}|d^{\alpha_1}\tilde a_1(y)|^{s'}{\rm d}y\right)^{1/s'}\ ] ] @xmath222 \hskip -4pc&\quad\ , \times\left(\int_{2^k\tilde q}|f(y)|_r^pw(y){\rm d}y\right)^{1/p } \left(\frac{1}{|2^k\tilde q|}\int_{2^k\tilde q}w(y){\rm d}y\right)^{1/p}\\[.4pc ] \hskip -4pc&\quad\ , \times\left(\frac{1}{|2^k\tilde q|}\int_{2^k\tilde q}w(y)^{-s/(p - s)}{\rm d}y\right)^{(p - s)/ps } \\[.4pc ] \hskip -4pc&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p(w ) } , \\[.4pc ] \hskip -4pc\left(\sum_{i=1}^\infty|{\rm m}_5^{(5)}|^r\right)^{1/r}&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p(w)}.\end{aligned}\ ] ] for @xmath223 , choose @xmath224 such that @xmath225 and @xmath209 . notice that @xmath226 , we get @xmath227 & \quad\ , \times |d^{\alpha_1}\tilde a_1(y)||d^{\alpha_2}\tilde a_2(y)||h(y)|_r{\rm d}y \\[.4pc ] \hskip -4pc&\leq c\sum_{k=0}^\infty\left(\frac{d}{(2^kd)^{n+1}}+\frac{d^\varepsilon}{(2^kd)^{n+\varepsilon}}\right ) \left(\int_{2^{k+1}\tilde q}|f(y)|_r^{t_3}{\rm d}y\right)^{1/t_3}{\rm d}y \\[.4pc ] \hskip -4pc&\times\sum_{|\alpha_1|=m_1}\left(\int_{2^{k+1}\tilde q}|d^{\alpha_1}\tilde a_1(y)|^{t_1}{\rm d}y\right)^{1/t_1}\\[.4pc ] & \quad\times \sum_{|\alpha_2|=m_2}\left(\int_{2^{k+1}\tilde q}|d^{\alpha_2}\tilde a_2(y)|^{t_2}{\rm d}y\right)^{1/t_2 } \\[.4pc ] \hskip -4pc&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\sum_{k=1}^\infty(2^{-k}+2^{-\varepsilon k } ) w(2^k\tilde q)^{-1/p}\\[.4pc ] & \quad\ , \times \left(\int_{2^k\tilde q}|f(y)|_r^pw(y){\rm d}y\right)^{1/p } \left(\frac{1}{|2^k\tilde q|}\int_{2^k\tilde q}w(y){\rm d}y\right)^{1/p}\\[.4pc ] \hskip -4pc&\times\left(\frac{1}{|2^k\tilde q|}\int_{2^k\tilde q}w(y)^{-t_3/(p - t_3)}{\rm d}y\right)^{(p - t_3)/pt_3 } \\[.4pc ] \hskip -4pc&\leq c\prod_{j=1}^2\left(\sum_{|\alpha_j|=m_j}\|d^{\alpha_j}a_j\|_{\rm bmo}\right)\||f|_r\|_{b_p(w)}.\end{aligned}\ ] ] thus @xmath228 this completes the proof of theorem 2 . now we shall apply the theorems of the paper to some particular operators such as the caldern zygmund singular integral operator and fractional integral operator . caldern zygmund singular integral operator . let @xmath1 be the caldern zygmund operator @xcite . then it is easy to see that @xmath1 satisfies the conditions in theorem 2 . thus the conclusions of theorem 2 hold for @xmath229 . fractional integral operator with rough kernel . for @xmath36 , let @xmath68 be the fractional integral operator with rough kernel defined by @xcite @xmath230 the vector - valued multilinear operator related to @xmath68 is defined by @xmath231 and @xmath232 is homogeneous of degree zero on @xmath15 , @xmath233 and @xmath234 for some @xmath235 , that is , there exists a constant @xmath236 such that for any @xmath237 , @xmath238 . then @xmath68 satisfies the conditions in theorem 1 . thus the conclusions of theorem 1 hold for @xmath67 . the author would like to express his deep gratitude to the referee for his valuable comments and suggestions . this work is supported by the nnsf ( grant : 10271071 ) .

in this paper , we prove some bmo end - point estimates for some vector - valued multilinear operators related to certain singular integral operators . [ theore]_proof of theorem _ [ theoree]_proof of theorem _

we perform numerical simulations of the deformation and failure of defected monolayer graphene using the lammps molecular dynamics simulator package @xcite . the carbon - carbon atom interaction is modeled with the `` adaptive intermolecular reactive bond order '' ( airebo ) potential @xcite . in order to simulate a realistic bond failure behavior , the shortest - scale adaptive cutoff of the airebo potential has to be fine - tuned @xcite , as detailed in appendix [ sec : potential ] . the simulated system consists of single layer , monocrystalline graphene sheets , composed of a variable number @xmath0 of atoms : @xmath0 varies from approximately @xmath1 to @xmath2 atoms . the sheets are prepared by placing the atoms on a hexagonal lattice ; the characteristic lattice length scale @xmath3 is chosen so that the system is initially in an equilibrium configuration . the sheets have an almost square shape lying on the xy coordinate plane ; their lateral size depends on @xmath0 and varies between 50 and @xmath4 ( 5 and 36 nm ) . when placing defects on the sheets , a fixed fraction of atoms is randomly removed ; this corresponds to vacancy concentrations @xmath5 , 0.2 and 0.5% . while the graphene layer is essentially 2d , the atom positions are integrated in all the three spatial directions ; also , the layers have no periodic boundary conditions . the simulations are performed by stretching the samples along the x coordinate axis , corresponding to the `` armchair '' direction of the graphene hexagonal structure . we select two boundary strips of atoms at the opposite x - ends of the sheet . these strips are @xmath6 wide , corresponding to 4 atom layers . hence , the atoms are free to move in the y and z directions , but follow an imposed motion along the stretching direction ( x ) . this constraint induces an initial pre - stress on the sheet that is visible in the stress - strain curve ( see fig.[fig : examplefailure]b ) . the y - end boundaries are left free . the system is thermostated by means of a berendsen @xcite thermostat with a temperature ranging from 1k to 800k , and a characteristic relaxation time equal to 0.1 ps ; the simulation timestep is set to 0.5 fs to insure a correct time integration of the atoms dynamics . these parameters lead to a slightly underdamped atom dynamics . before the stretching protocol is started , the system is allowed to relax to thermal equilibrium from the initial constrained state . afterwards , one of the lateral strips is set in motion , so that the sample is subject to a constant engineering strain rate @xmath7 independent of the system size . the strain rates lie between @xmath8 and @xmath9 . as for other molecular dynamics simulations , the simulated strain rates are much higher than those applied experimentally , but the deformation speed is still much lower than the sound speed in graphene . the chosen strain rate is reached by adiabatically ramping up @xmath7 , in order to minimize the creation of shock waves in the material . as a matter of fact , visual inspection of velocity fields shows that shock waves are rapidly damped and do not significantly influence the system dynamics . simulations are carried on until the graphene sheet fractures . failure statistics are sampled over 100 realizations for each condition in which we vary vacancy concentration @xmath10 , temperature @xmath11 , strain rate @xmath7 and system size @xmath0 . the only the exception is provided by systems characterized by @xmath12k , @xmath13 , @xmath14 and @xmath15 atoms , where 50 samples were simulated . atoms , with a vacancy concentration ( porosity ) @xmath16 . the color bar indicates the @xmath17 component of stress tensor per - atom . a ) graphical view of the failure process ( from left to right ) . the crack nucleates from one of the defects already present in the material ( not necessarily the most stressed ) and rapidly grows untill the graphene sheet complete failure is achieved . b ) the stress strain curve displays temperature dependent fracture strength . the pre - stressed initial condition ( @xmath18 ) is due to the constraint applied to the atoms belonging to the 4 outmost layers of the sheet , which are subject to the stretching along x. ] size and on the vacancy concentration @xmath10 . simulations are carried out with @xmath12k and @xmath13 . the lines are the theoretical prediction as discussed in the supporting information . they do not arise as direct fit of the numerical curves , but result from the analytical evaluation of the integral expression of @xmath19 . b ) the failure stress survival distribution at @xmath12k , and @xmath20 for different system sizes with vacancy concentration equal to @xmath21% ( blue ) , @xmath22% ( green ) and @xmath23% ( red ) . when the survival probability distributions are rescaled by @xmath0 according to the predictions of the extreme value theory , the data collapse into a single curve that only depends on the vacancy concentration @xmath10 . ] an example of the fracture process is shown in fig . [ fig : examplefailure]a , where the graphene structure is seen from above at four different times during the nucleation and ensuing growth of the crack ( see also video 1 ) . the color code represents the xx component of the symmetric per - atom stress tensor @xmath17 , including both potential and kinetic terms . typical stress strain curves are reported in fig . [ fig : examplefailure]b , showing that the tensile strength depends on temperature @xmath11 . our results provide a clear indication that it also depends on system size @xmath0 , vacancy concentration @xmath10 and strain rate @xmath7 , as we discuss below . [ fig : distr]a reports the average failure stress @xmath24 as a function of system size for different values of the porosity @xmath10 , showing that the larger and more defective a sample is , the weaker it is . a more complete description of the failure statistics is obtained by the survival distribution @xmath25 , defined as the probability that a sample has not yet failed at a stress @xmath26 . the numerical results for @xmath25 are reported in fig . [ fig : distr]b . if a system of volume @xmath27 fails according to extreme value statistics , the survival distribution should depend on the volume as @xmath28 , where @xmath29 is the survival distribution of a representative element of volume @xmath30 , the smallest independent unit in the sample @xcite . if we express the volume in terms of the number of atoms @xmath0 and their atomic volume @xmath31 , the survival probability can be written as @xmath32 $ ] , where @xmath33 is a suitable function which is a power law @xmath34 in case of weibull distribution @xcite , and exponential @xmath35 for gumbel distribution @xcite . [ fig : distr ] shows that the @xmath0-dependence of the survival distribution follow the prescriptions of extreme value theory , but @xmath33 is not a power law , indicating that the weibull distribution does not represent the data . this is confirmed by the size scaling of the average failure stress that does not follow a power law , as would be expected from the weibull distribution . the survival distribution depends also on temperature and strain rate , as shown in fig . [ fig : distrt - rate ] , which is hard to reconcile with the weakest link hypothesis underlying the weibull distribution . indeed , by monitoring the local stress field @xmath17 before failure , we estimate that only in less than 20% of the samples ( for @xmath15 ) the final crack nucleates in the most stressed region . in 50 - 60% of the cases , the final crack is nucleated in regions that ranked fourth of more in terms of stress . this is a clear indication that failure is not dictated by the weakest link . , @xmath36 and @xmath37 ) depends on temperature ( a ) and strain rate ( b : @xmath22 , @xmath12k and @xmath37 ) . the dashed lines represent the best least square fit according to the theory of breaking kinetics discussed in the text . ] to understand our simulation results , we generalize extreme value theory taking into account thermal fluctuations . we describe the system as a set of @xmath38 representative elements of volume @xmath30 ( slabs ) such that the thermally activated failure of a single element induces global failure . each representative @xmath39 element obeys linear elasticity up to a critical strain @xmath40 , so that the elastic energy of the sample under an external stress @xmath26 is given by @xmath41 , where @xmath42 where @xmath43 is the young modulus . the sample is loaded at constant strain rate @xmath7 so that @xmath44 and critical strains are distributed according to a probability density function @xmath45 . assuming the slabs noninteracting and identicals , the survival probability for the entire sample is given by the product of the survival probabilities of each representative element @xmath46^n$ ] , according to the theory of breaking kinetics @xcite . the representative volume survival probability is defined as @xmath47 where @xmath48 represents the survival probability of a single slab characterized by a failure strain @xmath49 . [ survival_cumulative ] reduces to the standard extreme value theory when @xmath50 , but otherwise depends on temperature and strain rate . in general , however , the theory predicts that @xmath51 should not depend on the system size , as verified by our simulations ( see fig . [ fig : distr]b ) . to estimate the survival distribution of the single slab @xmath48 we make the phenomenological hypothesis that the material failure arises as a thermally activated process . historically , the idea that the solid failure can be described by means of the kramer s theory , where the intrinsic energy barrier is reduced proportionally to the applied field , has firstly appeared in material science to treat the kinetic fracture of solids under applied stresses , and dates back to the works of tobolsky and eyring @xcite and , later , of zhurkov @xcite . more recently it has been successfully applied to the study the failure of fibers @xcite , gels @xcite , wood and fiber - glasses @xcite where the potential energy barrier is given by the griffith crack nucleation energy @xcite . most of previous work focused on the thermal dependence of the average strength or the failure time in creep experiments and did not address the survival distribution and its size dependence . to this end , we start from recent theories developed for single - molecule pulling , where the molecule rate coefficient for rupture ( or unbinding ) is modified by the presence of an external time - dependent force @xcite . in our case , the stress - dependent failure rate of a single element characterized by a failure strain @xmath49 , is given by an arrhenius like form @xcite @xmath52 } \label{rate}\ ] ] where @xmath53 is the kramer s escape rate from the potential well described in eq . [ potential_0 ] @xcite , @xmath54 with a characteristic frequency @xmath55 . in our numerical simulations one end of the graphene sheet is held fixed , while the other is pulled at constant strain rate @xmath7 : this can be interpreted as the action of a stiff device @xcite for which eq.[rate ] has been derived . @xmath48 obeys to the following first - order rate equation @xcite @xmath56 where @xmath44 . the survival probability is then readily obtained as @xmath57}. \label{survival_adim}\ ] ] notice that eq . [ survival_adim ] , only holds for @xmath58 since otherwise the element fails with probability one ( when @xmath59 the kramer s theory incorrectly predicts @xmath60 , since it only holds for energy barriers @xmath61 @xcite ) . finally , inserting eq . [ survival_adim ] in eq . [ survival_cumulative ] and , in turns , into the constitutive equation for the theory of breaking kinetics , we obtain @xmath62}\right)^n.\end{aligned}\ ] ] the survival distribution reported in eq . [ survival_n ] is written as a convolution of the disorder distribution @xmath45 with a temperature and rate dependent kernel . it is instructive to study its limiting behaviors since this allows to assess the relevance of thermal and rate dependent effects for fracture statistics . our starting point is the expression for the conditional survival probability @xmath48 reported in eq . [ survival_adim ] . it is convenient to study its behavior in term of the dimensionless parameter @xmath63 , the ratio between the elastic energy of a representative volume element and the thermal energy . in terms of @xmath64 we can write @xmath65 , where @xmath66 . \label{eq : lambda}\ ] ] thermal fluctuations can be neglected when @xmath67 , yielding the usual disorder - induced survival probability distribution @xmath68 it is interesting to consider first the limit of @xmath69 , corresponding to very low temperature and large representative volume elements . in this limit , the exponential factors in @xmath70 dominates and the function goes to zero even for small strain rates . in more generality , thermal fluctuations become negligible when @xmath71 . \label{eq : thermal}\ ] ] therefore there is a temperature and stress dependent critical strain rate above which we can neglect thermal fluctuations . another interesting limit is the low stress regime ( i.e. @xmath72 ) where @xmath73 hence , thanks to eq . [ survival_cumulative ] , the survival distribution for a representative element is given by @xmath74 where @xmath75 . therefore , the survival probability distribution function for the entire system can be recast as @xmath76 displaying a linear dependence on the applied stress , irrespective of the failure strain distribution function @xmath45 . [ survival_n ] provides an excellent fit to the results obtained from numerical simulations of defected graphene at different defect concentrations @xmath10 , temperature @xmath11 and loading rate @xmath7 . to fit the numerical simulations with eq . [ suppl : survival_n_highstress ] , we first need to establish the form of @xmath45 . this is a phenomenological function describing the distribution of failure strains of representative volume elements at zero temperature . a reasonable estimate of its functional form can be obtained from simulations at low temperature ( i.e. @xmath77k ) , where thermal fluctuations are negligible , as discussed in details in appendix [ sec : fit ] . the numerical outcomes indicate that @xmath45 follow the gumbel distribution @xcite ( see fig . [ fig : lowt ] ) . we then insert the resulting form of @xmath45 in eq . [ survival_n ] which we adopt as a fitting function for the numerical survival probability @xmath25 , with @xmath55 and @xmath30 as fitting parameters . the representative volume @xmath30 ranges between @xmath78 and @xmath79 , while the characteristic frequency is found in the range @xmath80 and @xmath81 ( see fig . [ fig : fitting parameters ] ) moreover , from the survival distribution we also calculate , without additional fitting , the system size dependence of the average tensile strength @xmath19 , which displays an excellent agreement with simulations results as shown in fig . [ fig : distr]a . further details on the fitting methodology and the analytical expressions used in our model are reported in appendix [ sec : fit ] . k , @xmath82 and @xmath37 for different values of the vacancy concentration @xmath10 . the numerical data are fitted with the exponential function @xmath83 ( solid lines ) , leading to a gumbel distribution for the failure strains @xmath84 ( see eq.[suppl : rho ] ) . for @xmath16 we obtain @xmath85 , @xmath86 . for @xmath87 : @xmath88 , @xmath89 . for @xmath90 @xmath91 , @xmath92 . ] and b ) the activation frequency as a function of temperature ( @xmath37 , @xmath36 ) . these values are obtained by the least square fit of the numerical survival probability distribution with the expression [ suppl : survival_n ] ( dashed lines in fig.3a ) . the same values , c ) and d ) , as a function of the strain rate ( @xmath37 , @xmath12k ) , obtained from the best fits shown in fig.3b ( @xmath87 ) and fig.[fig : p0.5 ] ( @xmath90 ) ( dashed lines ) . ] in conclusions , we have performed extensive numerical simulations for the tensile failure of defected graphene focusing on the size effects of the strength distribution for different temperatures and loading rate . the results of numerical simulations show deviations from the weakest - link hypothesis but can be explained by taking explicitly into account the effect of thermally activated crack nucleation . the resulting theory describes well our results and could prove useful to understand the tensile strength distribution of other nanomaterials such as carbon nanotubes or other nanowires . at present it is not possible to compare our numerical and theoretical predictions directly to experiments . experimental measurements of the strength of graphene sheets are mostly based on indentation tests @xcite , while tensile tests only recently appeared in the literature @xcite but thermal , rate and size effects have not been studied . furthermore , most experimental studies are focusing on the strength of graphene in pristine conditions @xcite without defects or pre - existing cracks . our theory is , however , very general yielding predictions that should be applicable also to other carbon nanomaterials and allows to formulate general considerations on the relevance of thermal effects for fracture . [ survival_n ] suggests that thermal fluctuations can be neglected for large enough strain rate , since in this limit @xmath93 and the sample fails according to the weakest link statistics . in our simulations , we have @xmath94pa , @xmath95 so that at room temperature we estimate @xmath96 . if we use this value in eq . [ eq : lambda ] , we find that for @xmath97 the exponential terms do not vanish close to failure ( i. e. for @xmath98 ) and thermal effects should therefore be relevant . indeed using @xmath99 in eq . [ eq : thermal ] , one can readily show that thermal effects start to become relevant for @xmath100k in agreement with our simulations . the same argument suggests that in macroscopic samples , with larger representative volume elements , thermally activated failure can often be ignored , even at room temperature . consider for instance a ceramic material , like sintered @xmath101-alumina @xcite , with @xmath102 and a typical tensile strength of @xmath103 . assuming that the representative volume element corresponds to a grain size of @xmath104 , we can estimate @xmath105 . now the exponential factors impose that @xmath67 even at low strain rates , implying that the strength distribution should be described by conventional extreme value theory . indeed , experiments show that the strength distribution is described by weibull statistics with parameters that are largely temperature independent @xcite . our theory thus provides a simple way to estimate the relevance of thermal and rate dependent effects for fracture . this result could have important implications for applications to micro- and nano - mechanical devices whose reliability may crucially depend on the control of thermally activated failure . the carbon - carbon atom interactions were modeled using the `` adaptive intermolecular reactive bond order '' ( airebo ) potential @xcite , which was originally developed as an extension of the `` reactive bond order '' potential ( rebo ) @xcite . in turn , the rebo potential was developed to describe covalent bond breaking and forming with associated changes in atomic hybridization within a classical potential ; it has proven an useful tool for modelling complex chemistry in large many - atom systems . the airebo potential improves the rebo potential with an adaptive treatment of non - bonded and dihedral angle interactions that is employed to capture the bond breaking and bond reformation between carbon atom chains . the analytical form of the airebo potential ( as discussed in the documentation @xcite ) is written as : @xmath106\ ] ] the @xmath107 term has the same functional form as the hydrocarbon rebo potential developed in @xcite . we will not cover here the details of the energetic terms which are thoroughly discussed in the mentioned reference . in short , the rebo term gives the model its _ short to medium _ range reactive capabilities , describing short - ranged c - c , c - h and h - h interactions ( @xmath108 ) . these interactions have strong coordination - dependence through a bond order parameter , which adjusts the attraction between the @xmath109 atoms based on the position of other nearby atoms and thus has 3- and 4-body dependencies . a more detailed discussion of formulas for this part of the potential are given in @xcite . the @xmath110 term adds longer - ranged interactions ( @xmath111 ) using a form similar to the standard lennard - jones potential . it contains a series of switching functions so that the short - ranged lj repulsion ( @xmath112 ) does not interfere with the energetics captured by the @xmath113 term . the extent of the @xmath110 interactions is determined by a cutoff argument ; in general the resulting @xmath110 cutoff is approximately 10 , in this work we consider a cutoff of approximately 14 . finally , the @xmath114 term is an explicit 4-body potential that describes various dihedral angle preferences in hydrocarbon configurations . the airebo potential has been extensively used to simulate and predict mechanical properties of carbon - based materials , i.e. fullerene , carbon nanotube and graphene @xcite . furthermore , it offers a valid tradeoff between accuracy and computational efficiency ; a realistic fracture of large system sizes can be simulated in reasonably short time scales ( a few hours on recent computers ) . other interaction models can offer little improvement to the actual realism of the simulation , at the cost of much larger computational costs : for example , the reaxff potential , or dft semiclassical approaches could describe more accurately the fast time scales of chemical reactions , but this would not change the ultimate failure length of the c - c bond : the expected maximum elongation for a c - c bond in graphene is around 0.178 nm . on the other hand , the use of faster but too simplistic models ( e.g. lennard - jones potentials , mass and spring systems , or other elastic models ) fail to significantly reproduce a realistic behavior . however , in order to simulate a realistic bond failure behavior , the short - scale c - c adaptive cutoff ( @xmath115 ) of the airebo potential has to be tuned . in fact , it has been observed @xcite that , during simulations of fracture of covalent bonds and without cutoff tuning , the shortest - scale potential introduces a sharp increase of bond forces near the cutoff distances , which in turn causes spurious increase in fracture stress and strain @xcite . it should also be noted here that this phenomenon is specifically relevant for perfect graphene and cnt lattices , while it is much less pronounced in defected samples , due to the disorder induced in the lattice by the atom vacancies . this issue has been solved in the past by incrementing the short - scale cutoff lenght of the potential ; the cited papers increase this parameter to 2.0 . this , however , has the side effect of leading to a singular behavior in the atomic pair potential when the atom atom distance is exactly 2 . . the simulated graphene sheet is composed by @xmath37 carbon atoms at @xmath12k , without in - built defects ( @xmath116 ) and pulled at constant strain rate @xmath13 . for @xmath117 nm , the stress displays a spurious increase while for @xmath118 nm the pair potential shows an unphysical singularity ( not shown ) . the chosen value of @xmath115 is set to @xmath119 nm . ] we performed stretching simulations varying the cutoff parameter from @xmath120 nm ( default value ) to @xmath121 nm in both armchair ( x ) and zigzag ( y ) directions of the graphene sheet with no vacancies ( @xmath116 ) . the stress - strain curve obtained from the numerical simulations are shown in fig . [ fig : cutoff ] . for @xmath122 nm , a sharp increase on tensile stress for large strains is observed , leading to an unphysical ultra - high failure stress and corresponding failure strain . increasing the @xmath115 in the range @xmath123 nm strongly suppresses this phenomenon . moreover , the stress - strain data reported in fig . [ fig : cutoff ] , clearly display that the failure strain varies from 0.13 to 0.25 when @xmath115 is in the range @xmath124 , whereas the failure stress exhibites a much weaker fluctuation ( from @xmath125 pa to @xmath126 pa ) . finally we notice that for defected samples like those investigated in the present article , i.e. @xmath127 , the values of the failure stresses and strains do show a much less marked dependence on the choice of @xmath115 , whenever @xmath124 . to fit the numerical simulations with eq . [ survival_n ] , we first obtain @xmath45 from simulations at low temperature ( i.e. @xmath77k ) . as shown in fig.[fig : lowt ] , the numerical survival distribution function @xmath128 , obtained at @xmath77k and @xmath13 , can be nicely fitted with the following exponential form @xmath129 . the theoretical prediction for the survival probability distribution furnished by eq.[suppl : survival_n_highstress ] requires @xmath130 , once we assume that @xmath131 when @xmath132 . hence , we obtain @xmath133 which corresponds to a gumbel distribution of failure strains @xcite . the numerical values of the fitting parameters @xmath134 are reported in the caption of fig.[fig : lowt ] for three vacancy concentrations @xmath10 . we notice that the simulated samples for @xmath77k are 250 in the case of @xmath16 and 850 for @xmath87 and 800 @xmath90 . , @xmath37 and three strain rates @xmath7 . the vacancy concentration is set to @xmath90 . the fitting function is provided by eq.[suppl : survival_n ] : the values of the fitting parameters @xmath30 and @xmath55 are reported in fig.[fig : fitting parameters ] . ] we then perform the least square fit of the numerical survival probabilities @xmath128 , obtained for different values of @xmath11 , @xmath7 and porosities @xmath10 ( see fig.s 3a , b , [ fig : p0.5 ] , [ fig : various_p ] ) , with the following function where the fitting parameters are the representative volume element @xmath30 and the characteristic frequency @xmath55 . the atomic volume @xmath31 has been evaluated by considering a density of 38.18 atoms per @xmath136 and a sheet thickness equal to @xmath137 nm , yielding @xmath138 . for any value of @xmath10 , the corresponding values of @xmath139 and @xmath140 obtained from the best fit of the data in fig.[fig : lowt ] are plugged into eq.[suppl : survival_n ] . the fitted @xmath30 and @xmath55 corresponding to fig.s 3a , b , [ fig : p0.5 ] , [ fig : various_p ] are reported in fig.[fig : fitting parameters ] . @xmath142}+\right.\\ + \left.\frac{\omega_0}{\sqrt{\pi}\dot{\varepsilon}}\left(\frac{v_0e } { k_bt}\right)^{3/2}\int_{\sigma / e}^\infty d\varepsilon_c\,ae^{\frac{\varepsilon_c}{\varepsilon_0}-ae^{\frac{\varepsilon_c}{\varepsilon_0}}}\left(\varepsilon_c-\frac{\sigma}{e}\right)e^{-(\varepsilon_c-\sigma / e)^2\frac{v_0e}{k_bt}}e^{-\frac{\omega_0}{\dot{\varepsilon}}\sqrt{\frac{v_0e}{\pi k_bt}}\left[e^{-(\varepsilon_c-\sigma / e)^2\frac{v_0e}{k_bt}}-e^{-\varepsilon_c^2\frac{v_0e}{k_bt}}\right]}\right\}. \end{array } \label{suppl : probability_n_final}\ ] ] @xmath143}+\right.\\ + \left.\frac{\omega_0}{\sqrt{\pi}\dot{\varepsilon}}\left(\frac{v_0e } { k_bt}\right)^{3/2}\int_{\sigma / e}^\infty d\varepsilon_c\,ae^{\frac{\varepsilon_c}{\varepsilon_0}-ae^{\frac{\varepsilon_c}{\varepsilon_0}}}\left(\varepsilon_c-\frac{\sigma}{e}\right)e^{-(\varepsilon_c-\sigma / e)^2\frac{v_0e}{k_bt}}e^{-\frac{\omega_0}{\dot{\varepsilon}}\sqrt{\frac{v_0e}{\pi k_bt}}\left[e^{-(\varepsilon_c-\sigma / e)^2\frac{v_0e}{k_bt}}-e^{-\varepsilon_c^2\frac{v_0e}{k_bt}}\right]}\right\}. \end{array}. \label{suppl : mean_stress}\ ] ] this quantity can be analytically calculated and plotted as a function of @xmath0 , setting @xmath144 , as shown in fig.2a for @xmath12k , @xmath13 and three values of the vacancy concentration @xmath10 . we emphasize that in this case no fit , but just the numerical evaluation of the integral expression of @xmath19 [ suppl : mean_stress ] is provided , making use of the proper values of @xmath145 , obtained by fitting the survival probabilities displayed in fig.[fig : lowt ] and fig.[fig : various_p ] . , @xmath13 and @xmath37 for three vacancy concentrations @xmath10 . the numerical data are fitted with the expression [ suppl : survival_n ] ( dashed lines ) using the proper values of @xmath139 and @xmath146 reported in fig.[fig : lowt ] . the fitted values of @xmath30 and @xmath55 are shown in fig.[fig : fitting parameters ] for @xmath87 and @xmath90 . for @xmath16 the least square fit gives @xmath147 and @xmath148 . the set of parameters @xmath149 and @xmath55 which characterize uniquely the theoretical expression [ suppl : survival_n ] ( dashed lines ) are inserted into eq.[suppl : mean_stress ] to calculate the mean average rupture stress @xmath19 as a function of @xmath0 , shown in fig.2a . ] 48ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1146/annurev.physchem.58.032806.104502 [ * * , ( ) ] , , \doibase doi : 10.1016/j.pmatsci.2011.01.005 [ * * , ( ) ] , @noop _ _ ( , ) @noop _ _ , edited by ( , ) \doibase http://dx.doi.org/10.1016/0038-1098(85)91061-0 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.108.065504 [ * * , ( ) ] link:\doibase 10.1103/physrevapplied.2.034008 [ * * , ( ) ] http://stacks.iop.org/0022-3727/42/i=21/a=214007 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2130713 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4754709 [ * * , ( ) ] link:\doibase 10.1126/science.1157996 [ * * , ( ) ] link:\doibase 10.1021/nl203547z [ * * , ( ) ] link:\doibase 10.1038/ncomms4782 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) http://stacks.iop.org/0295-5075/103/i=6/a=68004 [ * * , ( ) ] @noop * * , ( ) http://stacks.iop.org/0957-4484/24/i=50/a=505703 [ * * , ( ) ] http://stacks.iop.org/0957-4484/18/i=15/a=155708 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.3488620 [ * * , ( ) ] link:\doibase 10.1021/nl901448z [ * * , ( ) ] , link:\doibase 10.1016/j.spmi.2011.11.019 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.185505 [ * * , ( ) ] link:\doibase 10.1006/jcph.1995.1039 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.481208 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.448118 [ * * , ( ) ] http://stacks.iop.org/0022-3727/42/i=21/a=214012 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1016/s0006 - 3495(03)74449-x [ * * , ( ) ] link:\doibase 10.1073/pnas.1534554100 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1073/pnas.0903003106 [ * * , ( ) ] , @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) in @noop _ _ ( )

nanoscale materials display enhanced strength and toughness but also larger fluctuations and more pronounced size effects with respect to their macroscopic counterparts . here we study the system size - dependence of the failure strength distribution of a monolayer graphene sheet with a small concentration of vacancies by molecular dynamics simulations . we simulate sheets of varying size encompassing more than three decades and systematically study their deformation as a function of disorder , temperature and loading rate . we generalize the weakest - link theory of fracture size effects to rate and temperature dependent failure and find quantitative agreement with the simulations . our numerical and theoretical results explain the crossover of the fracture strength distribution between a thermal and rate - dependent regime and a disorder - dominated regime described by extreme value theory . nanomaterials have remarkable mechanical properties , such as enhanced strength and toughness @xcite , but display considerable size effects and sample - to - sample fluctuations , which represent an issue for engineering applications . our current understanding of fracture size - effects in macroscopic disordered media relies on extreme value theory which relates the strength to the statistics of the weakest region in the sample @xcite . while the theory does not consider the effect of stress concentrations and crack interactions , numerical models for the failure of elastic networks with disorder show that an extreme value distribution describes failure at large enough scales , although the form usually deviates from the standard weibull distribution @xcite . understanding size effects in nanomaterials is still an intriguing open issue also because of the presence of rate - dependent thermal effects that would invalidate the weakest - link hypothesis @xcite . yet , the weibull distribution is commonly used to fit experimental data in carbon based nanomaterials @xcite , although the tensile strength is observed to depend on the strain rate @xcite . testing fracture properties of graphene is quite challenging due to the difficulty in applying high tensile stresses in a controlled fashion on nanoscale objects @xcite . therefore numerical simulations represents a viable alternative to understand the size dependence of its mechanical behavior @xcite . numerical simulations of defected carbon nanutubes suggest that failure is described by the weibull distribution in quasistatic , zero - temperature conditions @xcite . finite temperature molecular dynamics simulations reveal , however , that the average tensile strength of nanotubes @xcite and graphene @xcite depends on temperature and loading rate . despite these insightful results , a comprehensive theory describing the size dependent fracture strength distribution of carbon nanomaterials , elucidating the role of thermal fluctuations and strain rate , is still lacking . here we perform large scale molecular dynamics simulations of the deformation and failure of defected monolayer graphene sheets for a wide range of sample sizes , vacancy concentration , temperature and strain rate . to explain the observed temperature and rate dependence of the tensile strength distribution , we generalize extreme value theory to the case of thermally activated rate dependent fracture . the resulting theory is shown to be in excellent agreement with our simulations and provides a general framework to explain rate - dependent thermal effects in the failure of disordered nanomaterials . based on our theory , we derive a simple criterion that allows to assess the relative importance of structural disorder and thermal fluctuations in determining failure . using this rule , one can readily show that the failure of nanoscale samples is more prone to thermal induced failure , while the fracture macroscopic samples are more likely to be ruled by quenched disorder . this confirms previous results showing that in the limit of very large samples failure is ruled by extreme value statistics ( although not necessarily by the weibull law ) @xcite . the paper is organized as follows . in section [ sec : model ] we describe the molecular dynamics simulation model and in section [ sec : simulations ] discuss the numerical results . the theory is described in details in section [ sec : theory ] where we also compare its prediction with experiments . section [ sec : discussion ] discusses the general implications of our work to understand size effects in materials at different scales . appendix [ sec : potential ] provides details on the choice of interatomic potential and appendix [ sec : fit ] discusses the fitting method .

the total energy formula obtained by sun is @xmath1 \label{uv}\ ] ] where @xmath2 is the total energy per atom , @xmath3 and @xmath4 are the equilibrium bulk modulus and equilibrium volume respectively . @xmath5 and @xmath6 are parameters and are related by the following relations : @xmath7 , @xmath8 , which are obtained by imposing the volume analyticity condition . since in this case the energy of the free atoms is zero , cohesive energy of the solid at @xmath9 is the energy at which the @xmath2 is minimum which happens to be at @xmath4 . thus the formula for cohesive energy @xmath10 turns out to be @xmath11 also , it turns out that @xmath12 . the values of @xmath3 , @xmath4 and @xmath13 are listed for various materials in the paper@xcite . cohesive energies calculated from the above formula are quite erroneous . calculated values for some materials using eq.([ecoh ] ) are compared with experimental values@xcite in table._1_. also we compare the energy per particle vs volume curve of aluminum with the data obtained from ab - initio calculations@xcite in fig.([1 ] ) . it can be seen that there is a serious mismatch between the two . however from fig.([1 ] ) , we can notice that the slopes of the mglj eos and that of the ab - initio curve are similar which is the reason for pressure calculated from mglj eos being accurate . .[11 ] cohesive energy [ cols="^,^,^ " , ] energy vs volume curve for aluminum at temperature @xmath9 . crosses are ab - initio data@xcite . solid line is obtained using eq.([uv ] ) , title="fig : " ] the mglj potential is given by @xmath14 \label{glj}\ ] ] the parameters @xmath15 , @xmath16 and @xmath17 are related to @xmath3 , @xmath4 and @xmath18 denoted as @xmath19 through the following relations . @xmath20 where @xmath21 is the structural constant which is @xmath22 for @xmath23 solids ans @xmath24 for @xmath25 solids and @xmath15 is the depth of the potential and @xmath26 is the number of first nearest neighbors . it can be seen that thermodynamic properties calculated using mglj potential with parameters of sun diverge for materials with @xmath19 is less than @xmath0 . for example , consider the excess internal energy per particle ( @xmath27 ) obtained through the energy equation@xcite . @xmath28 where @xmath29 is the density of the system and @xmath30 is the radial distribution function . since @xmath30 becomes @xmath31 asymptotically , the integral requires that each term of @xmath32 decays faster than @xmath33 . however , if @xmath19 is less than @xmath0 , the attractive component of @xmath32 decays slower than @xmath33 allowing @xmath27 in eq.([ee ] ) to diverge and for most of the materials @xmath19 is less than @xmath0 . this renders the potential , as parameterized by sun , to be inapplicable to calculate thermodynamic properties as they involve evaluation of integrals similar to eq.([ee ] ) . also the potential can not be used in molecular simulations as the tail correction for internal energy is similar to eq.([ee ] ) with lower limit being replaced by the cutoff radius of the potential . we noted that the mglj eos predicts cohesive energies erroneously . also we showed that the mglj potential can not be used in liquid state theories and molecular simulations for materials with @xmath34 less than @xmath0 as the thermodynamic quantities calculated using it diverge . this may be remedied by adjusting parameter @xmath16 so that @xmath10 is properly reproduced . also , including sufficient number of neighbors so that the total energy per particle converges would improve the results . lincoln et . al.@xcite obtained parameters of morse potentials for various fcc and bcc materials by including up to @xmath35 neighbor shell . in a separate work , we have done the improvements mentioned above and obtained the parameters by fitting the mglj eos to ab - initio data . same method is followed for eos obtained from other pair potentials and the results are analyzed@xcite . i am thankful to dr . chandrani bhattacharya , discussions with whom led to this paper . i thank dr . n.k . gupta for his encouragement . 18 g. kresse ; j. hafner , phys . rev . b 47 , 558 ( 1993 ) , g. kresse , furthmuller , j. computat mater sci 6 , 15(1996),g . kresse , furthmuller , phys . rev . b 54 , 11 169(1996),g . kresse , d . joubert , phys rev b 59 , 1758(1999 )

the cohesive energies of solids calculated using mglj eos proposed by sun jiuxun ( sun jiuxun , j. phys . : condens . matter 17 , l103 ( 2005 ) ) are seen to be erroneous . also we observed that the thermodynamic properties calculated using the mglj potential diverge for materials whose pressure derivative of bulk modulus at equilibrium is less than @xmath0 . thus the mglj potential can not be used in liquid state theories and molecular simulations to obtain thermodynamic properties . sun jiuxun @xcite recently suggested an equation of state(eos ) based on modified generalized lennard - jones ( mglj ) potential . the mglj eos is obtained by modifying the generalized lennard - jones potential in such a way that the eos obtained is volume analytic and satisfies spinodal condition . the mglj eos has three parameters and are related to lattice parameter , bulk modulus and derivative of the bulk modulus at equilibrium . in the paper@xcite it was shown that the pressure ( p ) vs compression ratio curve obtained using the mglj eos is quite accurate . the idea of generating an eos starting from a potential is interesting and has the advantage that the potential of the material also can be known in addition to the eos . however , we found some problems with the mglj eos and the potential so obtained from it . they are , ( a ) the cohesive energy calculated from the mglj eos is quite erroneous and ( b ) the thermodynamic properties calculated using the mglj potential with parameters of sun diverge for most of the materials . details about each problem are as follows :

for self - contain the bwma relations have been given in the appendix a through the graphs for three states . denoting the eigenvalues of a braiding matrix @xmath15 with three eigenvalues by @xmath16 , @xmath17 , and @xmath18 , where @xmath15 satisfies braid relation @xmath19 and without loss of generality by setting the eigenvalues of @xmath15 to be @xmath16 , @xmath17 and @xmath18 with @xmath20 we have for @xmath15 with three distinct eigenvalues . @xmath21 defining @xmath22 ( [ eq1.3 ] ) becomes @xmath23 where @xmath15 and @xmath24 occupy @xmath25 and @xmath26 sites and satisfy the relations shown in appendix a , i. e. they form bwm algebra . noting that a loop takes the value @xmath27 following the philosophy for t - l algebra to set up the topological basis @xmath28 and @xmath29 @xcite , we shall find the uni - orthogonal basis @xmath28 , @xmath29 and @xmath30 for @xmath15 and @xmath24 such that @xmath31 with @xmath32{cross}}}\quad\mathrm{and}\quad e_{k , k+1}={\raisebox{-0.4\height}{\includegraphics[height=1cm]{uniil}}},\quad d={\raisebox{-0.4\height}{\includegraphics[height=0.6cm]{nu}}}={\raisebox{-0.33\height}{\includegraphics[height=0.5cm]{loop}}}\quad \label{eq1.8}\ ] ] where the eigenvalues @xmath33 may be complex . the graphic expressions @xcite of bwma are shown in appendix a. to satisfy all the relations for bwma the base takes the forms : @xmath34{tsep}}}\label{eq1.9}\\ |e_i\rangle&=&f_i\left\{{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{tcross}}}+\alpha_i{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{uu}}}+\beta_i{\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{tsep}}}\right\}\quad ( i=1,2)\label{eq1.10 } .\end{aligned}\ ] ] in terms of the graphic calculations @xcite it can be proved that the ( [ eq1.7 ] ) together with @xmath35 lead to the constraints to the parameters @xmath36 , @xmath37 and normalization constant @xmath38 : @xmath39 and @xmath40\}^{1/2 } \label{eq1.13}\ ] ] for @xmath41 , i.e. @xmath42 ( hermitian ) , whereas @xmath43 for @xmath44 , i.e. @xmath45 ( unitary ) . the ( [ eq1.12 ] ) takes the same form for @xmath15 being hermitian or unitary . the only difference between hermitian and unitary consists in the different normalized constants @xmath38 and the parameters being complex for @xmath1 . the proof can be seen in the appendix b. in terms of the uni - orthogonally topological basis the direct calculation gives the @xmath12-d matrix forms of @xmath24 and @xmath15 acting on 1-st and 2-nd sites , 2-nd and 3-rd sites , respectively : @xmath50 where a represents the braiding between the 1-st and 2-nd sites , whereas b for the braiding between the 2-nd and 3-rd sties . the explicit 3-d matrix forms are shown to be@xmath51 : @xmath52 , \qquad a=\left[\begin{array}{ccc } \lambda_1&&\\ & \lambda_2&\\ & & \lambda_3\\ \end{array}\right ] = \left[\begin{array}{ccc } \lambda_1&&\\ & -\lambda_1^{-1}&\\ & & \sigma\\ \end{array}\right]\ ] ] @xmath53\ ] ] @xmath54\ ] ] when @xmath55,@xmath56,@xmath57 , i.e. the standard braiding matrix given in @xcite , we have @xmath58\ ] ] @xmath59\label{eq2.6}\ ] ] the other relations can be sees in appendix a. when @xmath15 is unitary with @xmath60 at root of unity , namely @xmath61 the basis reads ( @xmath62 ) @xmath63{tcross}}}+\alpha_i{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{uu}}}+\beta_i{\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{tsep}}}\right),\ ] ] @xmath64{ntcros}}}+\alpha_i^{*}{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{nn}}}+\beta_i^{*}{\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{tncup}}}\right).\ ] ] the matrix forms of @xmath24 and @xmath15 read @xmath65,\ ] ] @xmath66,\ ] ] @xmath67|e_1\rangle-(1+\lambda_1\lambda_3+\lambda_1^{-1}\lambda_3d)f_1f_2^{-1}|e_2\rangle + \lambda_1^{-1}f_1^{*-1}|e_3\rangle\right\},\\ s_{23}|e_2\rangle=&d^{-1}(\lambda_1+\lambda_1^{-1})\cdot\\ & \left\{(1-\lambda_1^{-1}\lambda_3-\lambda_1\lambda_3d)f_1^{-1}f_2|e_1\rangle + \lambda_1[\lambda_3(d-1)+\lambda_1]|e_2\rangle -\lambda_1f_2^{*-1}|e_3\rangle\right\},\\ s_{23}|e_3\rangle=&d^{-1}(\lambda_1+\lambda_1^{-1})^{-1}\cdot\\ & \left\{\lambda_1^{-1}f_1^{-1}|e_1\rangle + \lambda_1f_2^{-1}|e_2\rangle + ( \lambda_1+\lambda_1^{-1})\lambda_3^{-1}|e_3\rangle\right\}. } \ ] ] the 3-d matrix form is found : @xmath68,\ ] ] @xmath69&(1-\lambda_1^{-1}\lambda_3-\lambda_1\lambda_3d)f_1^{-1}f_2&\lambda_1^{-1}f_1^{-1}\\ -(1+\lambda_1\lambda_3+\lambda_1^{-1}\lambda_3d)f_1f_2^{-1}&\lambda_1[\lambda_3(d-1)+\lambda_1]&\lambda_1f_2^{-1}\\ \lambda_1^{-1}f_1^{*-1}&-\lambda_1f_2^{*-1}&(\lambda_1+\lambda_1^{-1})\lambda_3^{-1}\\ \end{array}\right],\ ] ] with @xmath70 i.e. @xmath24 must be real . @xmath71 following ref . @xcite for @xmath60 at root of unity it allows non - standard braiding matrices , say , for @xmath72 some of @xmath73 , and for the others @xmath74 , then the general form of @xmath75 is @xmath76 it leads to @xmath77_q\\ \end{array}\ ] ] where @xmath78 is the difference between the positive power number and negative ones in the power of @xmath60 in the third eigenvalues of @xmath15 for the fundamental representations of @xmath72 , @xmath79 and @xmath80 . as was pointed out in @xcite that for @xmath72 algebra the corresponding braiding matrix @xmath15 has three distinct eigenvalues @xmath55,@xmath81 whereas as the third one is given by @xmath82[eq5.1 ] where @xmath83 can be either @xmath60 ( standard solution ) or @xmath84 for nonstandard solution , so in general , we are able to take @xmath85 where @xmath86 can be arbitrary integers . to satisfy the spectral parameter dependent yang - baxter equation , the corresponding @xmath87-matrix takes the form @xcite @xmath88 where @xmath89 , @xmath90 , @xmath91 . + in order to obtain the rational limit of the @xmath87 the type - i solution is given by @xmath92 under the rational limit and @xmath93 we set @xmath94 for @xmath95 it leads to @xmath96 that under the rescaling @xmath97 becomes @xmath98 where @xmath99 , @xmath100 , and @xmath101 . + following the standard way of baxter @xcite the hamiltonian can be given by @xmath102 here the @xmath25 site has been indicated explicitly because @xmath103 and @xmath104 occupy the @xmath105-th and @xmath106-th sites . the @xmath107 is a new term added to the permutation - like operator @xmath108 due to bwma . + in particular when @xmath109 , i.e. @xmath110 we have @xmath111 it is worthy noting that @xmath109 means the solution of ybe being `` nonstandard''@xcite the m works in the block for @xmath112 , where @xmath113 means the third component of spin-@xmath2 at @xmath105-th site , or in terms of the basis for spin-@xmath2 : @xmath114 where the @xmath115 represents the state which occupies @xmath105-th and @xmath116-th sites . in general , @xmath116 is not necessary to be @xmath117 . @xmath118 where @xmath119 is real and @xmath120 , @xmath121 ( i.e. @xmath122 , here the braiding matrix is not unitary ) , then @xmath108 and @xmath107 satisfy bwm algebra . it is easy to check that for spin-@xmath2 and @xmath123 , in terms of + @xmath124 $ ] , @xmath125 $ ] and @xmath126 $ ] , + we have folowing baxter @xcite , @xmath127 for @xmath128 . therefore , for spin-@xmath2 @xmath129 it is interesting to note that the hamiltonian ( [ eq5.14 ] ) is known well for long time . especially , it is not permutation operator , but plays role , say , in the haldane conjuncture @xcite . here we have obtained @xmath130 for spin-1 whose hamiltonian is associated with bwm algebra . furthermore , as a demonstration example we show how to solve the model with @xmath131 in terms of the topological basis given by ( [ eq1.9 ] ) , ( [ eq1.10 ] ) . graphically the hamiltonian can be expressed by the operators for @xmath131 @xmath132{crosss}}}-{\raisebox{-0.4\height}{\includegraphics[height=1cm]{uniil}}}\right)=j\sum_{k=1}^{4}\left(\hat{t}_{k , k+1}-\hat{m}_{k , k+1}\right)\ ] ] its 9-d representation is given by acting the operator @xmath133 on 9-d basis . whereas acting ( [ eq5.15 ] ) on the 3-d basis ( [ eq1.9]),([eq1.10 ] ) for @xmath109 , we find @xmath134 the @xmath133 can be diagonalized in terms of the eigenstates @xmath135 : ( @xmath136 ) @xmath137 where @xmath138 and @xmath139 how to extend the approach to any @xmath140 by using the topological basis more than four sites to solve the eigenvalues problem with the help of topological basis is far beyond the current discussion . here we only discuss a four spin@xmath141 model which may be a hint to look for how to solve the @xmath140-site chain problem based on the topological basis . the relations ( [ eq5.11 ] ) and ( [ eq5.12 ] ) are defined for any @xmath105 and @xmath116 . to obtain the hamiltonian ( [ eq5.13 ] ) , the nearest neighborhood has been imposed through putting @xmath123 . however for any @xmath142 and @xmath143 , the operator @xmath144 can be recast to @xmath145 it can be checked in terms of ( [ eq5.11 ] ) , ( [ eq5.12])and ( [ eq6.1 ] ) , it holds for any @xmath142 and @xmath105 : @xmath146 where at any @xmath142-th site , @xmath147[eq6.3 ] noting that ( [ eq6.2 ] ) is valid for any i and k. because @xmath148 and total spin @xmath149 for @xmath131 we have @xmath150[eq6.4 ] there appear the additional terms other than @xmath151 , i.e. the term @xmath152 and @xmath153 . taking into account of @xmath154 we find @xmath155 where @xmath156,@xmath157 , @xmath158 , @xmath159 and @xmath160 substituting ( [ eq5.17 ] ) and ( [ eq6.5 ] ) into ( [ eq6.6]),we obtain for @xmath131 @xmath161[eq6.7 ] it is interesting to note that the topological eigenstates @xmath162 are spin singlet . from the point of view of lie algebra , the direct product of four spin 1 can be decomposed to 5 subspaces ; however , only the singlet with multiplicities three is the eigenstates of @xmath133 for @xmath131 . the yang - baxterization ( parametrization ) of @xmath164 braiding matrices can be made in the standard way , say , following jimbo , jones , and others @xmath165 @xcite . for @xmath166 ybe there is another way to introduce spectral parameter to a given braiding matrix . the basic idea comes from the wigner d - function @xcite . @xmath167 @xmath168 it can be proved that @xmath169 provided it holds @xmath170 \label{eq:6.4}\ ] ] when @xmath171 , ( [ eq:6.4 ] ) reduces to @xcite @xmath172 which is the condition that ( [ eq:6.4 ] ) reduces to braid relation . denoting by @xmath173 the ( @xmath166 ) ybe reads @xcite @xmath174 or @xmath175 is satisfied under ( [ eq:6.4 ] ) . the examples for @xmath176 had been given in the ref.[1 ] . here it holds for any ( @xmath177)-dimensional representations . let us condiser spin-1 example . the @xmath178 takes the form @xcite @xmath179 the 3-d braiding relation is given by @xmath180 where @xmath181 @xmath182 for the type - i , i.e. for @xmath183 , @xmath184 in ( [ eq:6.7 ] ) , after the unitary transformation , ( [ eq:6.7 ] ) becomes @xmath185 on the other hand , for @xmath0 on substituting @xmath186 into ( [ eq2.2 ] ) and ( [ eq2.4 ] ) the ( @xmath187 ) matrices @xmath188 and @xmath189 given by the topological basis and on account of @xmath190^{\frac{1}{2}},\quad f_2=[2s(s-1)]^{\frac{1}{2}}\ ] ] we obtain for @xmath191 @xmath192 that under similar transformation becomes @xmath193 ( [ eq:6.15 ] ) is identified with ( [ eq:6.11 ] ) . namely , as was pointed out for @xmath42 that the type - i ( @xmath187 ) braiding matrices based on the topological basis are the same as those given by @xmath178 with @xmath194 . for the type - ii , i.e. @xmath195 , @xmath196 with the same transformation @xmath197 as given by ( [ eq:6.11 ] ) the @xmath198-function gives @xmath199 on the other hand , for @xmath45 the ( @xmath187 ) braiding matrices ( [ eq3.7 ] ) and ( [ eq3.8 ] ) based on the topological basis and on account of @xmath200 are equal to @xmath201 @xmath202 that after the transformation becomes @xmath203 ( [ eq6.18 ] ) are the same as those given by @xmath198-function , subjecting to the unitary transformation v for the type - ii as shown in ( [ eq6.16 ] ) . in short conclusion it turns out that the yang - baxterization for ( @xmath166 ) ybe takes the different way from the ( @xmath164 ) representations . the resultant solutions of ybe are simply the wigner s @xmath204-functions . when @xmath171 it reduces to the braiding matrix . two examples for spin-@xmath205 and spin-@xmath2 have been checked in ref.[1 ] and in this section , respectively . however , the explicit correspondences between @xmath206 and ( @xmath164 ) braiding matrices for any @xmath207 for type - ii are still a challenge problem . it is emphasized that the yang - baxterization for ( @xmath166 ) solutions of ybe through the d - function is a quite new parametrization way based on the topological basis in different from refs.@xcite . as was pointed out in ref[1 ] that the extreme of @xmath208-functions may take multiple values for the different @xmath143 with @xmath209 components . however , for any half - integer @xmath143 ( also for @xmath143 integer , but @xmath210 should be excluded , obviously ) there exist the common maximum @xmath196 and minimum @xmath184 . the bell basis can be regarded as a linear combination of the natural basis @xmath211 , i.e. subject to the rotational transformation with @xmath13 @xcite . in general @xmath212 where @xmath213 is an @xmath214 vector serving as natural basis . for @xmath215 , @xmath216 and @xmath217 stands for the bell basis @xcite which possesses the maximum of entanglement . for any @xmath218 other than @xmath219 ( the maximum of @xmath220-norm of @xmath221 ) , it decreases the entanglement . we naturally think the extreme of d - function should indicate the entangling degree . one of the descriptions of entangling degree is von neumann entropy . we should show that the extreme points of the entropy and @xmath220-norm of d - function shares the common @xmath218-values . the von neumann entropy is defined by @xmath222 where @xmath223 is the reduced density matrix of a quantum state . following the ybe in @xmath224 dimensions the schmidt decomposition of a entangled state in @xmath224 dimensions , ( i.e. acting on topological basis ) @xmath225 can be written in the form for a fixed m : @xmath226 where m and m take over @xmath227 , and @xmath228 , @xmath229 are `` natural states '' . since the reduced density operator of the subsystem a is @xmath230 and for a fixed m @xmath231 the extremes of @xmath232 are shown in the following examples in comparison to the @xmath220-norm , i.e. @xmath233 . example : when @xmath184 , the bipartite state is a direct product sate which is separable , therefore @xmath232 and @xmath234 arrive at their minimum value simultaneously and the equality of ( [ eq6.19 ] ) holds . when @xmath196 , @xmath235 , the bipartite state reaches the maximum of entanglement . we have @xmath236 shown in fig.1 and fig.2 . example 2 : @xmath237 when @xmath184 , the state is a separable , @xmath232 and @xmath234 arrive at their minimum value simultaneously and the equality of ( [ eq6.19 ] ) holds . when @xmath196 , @xmath232 and @xmath234 reach their maximum value simultaneously . of course , the maximum value of @xmath238 is not the same as of @xmath232 , since @xmath239 , but both of them occur at @xmath196 , see fig.3 and fig.4 . the explicit forms of the common minimum @xmath184 and maximum @xmath196 for both @xmath220-norm of @xmath240 and @xmath241 for @xmath242 can be seen in appendix e. next , the fig.@xmath243 shows the derivatives of @xmath232 and @xmath234 with respect to @xmath218 . the zero points correspond to the extreme points of @xmath232 and @xmath234 . except the two common zero points @xmath13 and @xmath14 , other four zero points of @xmath232 do not coincide with the zero points of @xmath234 . in general , we can prove that @xmath244 and @xmath234 always share the same common extreme points @xmath13 and @xmath14 in the period @xmath245 $ ] for arbitrary @xmath143 and @xmath86 ( m=0 is excluded ) . firstly , let s consider the case of @xmath246 . because @xmath247 , we have @xmath248 for arbitrary @xmath86 , and it exactly denotes a separable state . therefore @xmath232 and @xmath234 both take the minimum value at @xmath246 , and @xmath249 when @xmath13 , we have @xcite @xmath250 and @xmath251 in ref.@xcite , it had proved that the @xmath220-norm of d - functions reaches the extreme value at @xmath13 . we just need to prove @xmath232 also to have extreme at @xmath13 . considering @xmath252 hence when @xmath253 is odd , @xmath254 , thus @xmath255 . when @xmath253 is even , according to eq . ( [ eq2 ] ) , we have @xmath256 i.e. @xmath257 , therefore @xmath255 . to sum up , @xmath13 is the common extreme points of @xmath232 and @xmath234 . similar to the standard strategy of the construction of the 2-d topological basis for tqft associated with temperley - lieb algebra @xcite , the extension has been made to construct the 3-d basis for bwm algebra . the point is to introduce the basis ( [ eq1.9 ] ) and ( [ eq1.10 ] ) , then all of the 3-d representations of @xmath48 , @xmath49 , @xmath46 and @xmath47 can be shown in terms of the graphic technique @xcite that yield the @xmath12-d representations of the corresponding braiding @xmath258 and @xmath24-operations . for both @xmath42 and @xmath45 we have given the explicit matrix forms of them in the sec . [ sec2 ] and sec . [ sec3 ] . in comparison with the case of tla now the @xmath24-involved relations appear . matching for the @xmath48 , @xmath49 , @xmath46 and @xmath47 in @xmath259 matrix form the corresponding @xmath188 , @xmath189 , @xmath260 and @xmath261 have been found in @xmath262 matrix forms that are nothing but the natural extension of @xmath263 matrix forms for tla . the physical meaning of the tqft associated with tla has been well established @xcite . however , the physical meaning of the counterpart for bwma deserves to be explored in the future . in ref @xcite a physical model was proposed , but here we present a different approach . as for the interaction model arisen from bwma it deserves more discussions . the connection of ( [ eq5.14 ] ) with bwma has been shown for spin-@xmath2 . how to solve the model for any n based on the 3-d representation in terms of the topological basis is an open problem . more interesting point should be emphasized . we have shown that the @xmath220-norm of @xmath264-function and von neumann entropy share the common maximum and minimum points of @xmath218 , i.e @xmath219 and @xmath265 , respectively , as least for any half - integer @xmath143 . as was proved that @xmath266 are @xmath267 solutions of ybe . the connection of the extreme points for both the @xmath220-norm of @xmath264 and von neumann entropy explores the deep meaning of @xmath220-norm in quantum mechanics . it also happened in other model @xcite . it describes quantum entanglement in the quantum information . the reason is that braiding operation is very natural to describe entanglement for two particles . because the entanglement only exists in the interaction between `` particles '' a and b themselves . an intuitive explanation of @xmath206-dimensional braiding can easily be made in terms of @xmath268-function . suppose two parallel lines along @xmath269 axis ( time pointed ) represent a and b at the site @xmath2 and @xmath7 , and form @xmath270 plane . in the spherical coordinates an entanglement between a and b occurs by over crossing the two lines with angle @xmath218 . the ( [ eq63 ] ) is to fix @xmath119 for a given @xmath218 . different @xmath218 also gives rise the rotation of @xmath270 plane about @xmath269 axis by @xmath119 . the @xmath13 means that the two lines are perpendicular to each other locally at different @xmath119 . whereas @xmath246 means parallel to each other . the former corresponds with the maximum entangled state and the later with decomposable one , i.e. disentangled . since ybe is the factorization condition of multi - body s - matrix to 2-body scatterings . we look over the facts such as factorization of s - matrix , topological basis , d - function , @xmath220-norm and von neumann entropy . now all of them have been connected within the frame work of ybe for quantum information , especially for the entanglement . we appreciate the interesting discussion with prof . z. h. wang and prof . j. birman , and x. b. peng . this work is in part supported by nsf of china with the grant no.11075077 and 11275024 . additional support was provided by the ministry of science and technology of china , no.2011aa120101 . 99 k. niu , k. xue and m. l. ge , j. phys . a , 44 , 265304(2011 ) h.n.v.temperley and e.h.lieb , proc.r.soc.london . sera.322,251(1971 ) the early review see , l. takhtadzhan and l. faddeev , russian mathematical surveys 34 , 11 ( 1979 ) : l. faddeev , soviet sci c : math . phys . rev 1 , 107 ( 1980 ) . r.j.baxter , exactly solved models in statistical mechanics , academic press , london , 1982 p.p . kulish , e.k . sklyamin , lecture notes in phys . , 151 , pp 61 - 119 . v. drinfeld , procedings of icm , california ( berkeley ) , academic press , 798 ( 1986 ) . for the collection of articles , see m.jimbo(ed ) , yang - baxter eq . in integrable systems , world scientific , singapore , 1990 . also , c.n.yang and m.l.ge(eds ) , braid group , knot theory and statistical mechanics , world scientific pub . singapore , 1990 . j. baxter , exactly solved models in statistical mechanics , dover pub . , inc mineola , new york , 1982 . z.wang , topologization of electron liquids with chern - simons theory and quantum computation . differential geometry an physics , 106 - 120 , nankai tracts . math.10 , world sci . publ , hackensack , nj , 2006 . z. wang , nankai lectures on tqft , june 5 - 7 , 2006 . m.h.freedman , m. larsen , z. wang , commu . phys , 227 , 605(2002 ) . s. das sarma , m. freedman , c. nayak , phys.rev.lett.94,166802(2005 ) y. zhang , l. kauffman , and m. ge , quantum information processing 4 , 159 ( 2005 ) . j. l. chen , k. xue , andm . l. ge , phys . a 76 , 042324 ( oct 2007 ) . hu , k. xue , andm . l. ge , phys . rev . a 78 , 022319 ( 2008 ) . j. l. chen , k. xue , and m. l. ge , annals of physics 323 , 2614 ( 2008 ) . j.preskill lecture notes for physics 219 : quantum computation , 2004 http://www.theory.caltech.edu / preskill / ph229/. c.nayak , f. wilczek , nucl.phys.b 479 , 529(1996 ) ; c. nayak et al , rev . 80 , 1083(2008 ) ; j. k. slingerland , f. a. bais , nucl . b612[fs]229 - 290 , 2001 ; e. h.rezayi , n.read , nucl.phys.b56,16864(1996 ) j. birman and h. wenzl , trans . 313 , 249 ( 1989 ) . j. murakami , osaka j. math . 24 , 745 ( 1987 ) . h. wenzl , ann . 128 , 179(1988 ) . m. jimbo , commun . 102 , 537(1986 ) ; m. jimbo , lett . math . phys . 10 , 63 - 69 ( 1985 ) . y. cheng , m. l. ge , k. xue , commun . 136 , 195(1991 ) . m. l. ge and a.c.t . wu , j. phys . a24,l725(1991 ) p. fendley and e. fradkin , phys . b 72 , 024412 ( 2005 ) . l. kauffman , knots in physics , world scientific pub . singapore , 1991 . f. d. m. haldane , phys . 50 , 1153(1983 ) . i. j. affleck , phys . matter , 1 , 3049(1989 ) . jones , commun . math . 125 , 459 , 1991 m. l. ge , y. s. wu , k. xue , inter . j. mod . 6a , 3735(1991 ) . d. a. varshalovich , a. n. moskalev , v. k. khersonskii , quantum theory of angular momentum , world scientific pub . singapore , 1988 , ( ref . p 53 ) ; a.m. pelelomov , sov . 20,703 ( 1997 ) . a. benvegnu , m. spera , rev . math . phys . 18 , 1075 ( 2006 ) . hu , k. xue and a76 , 042324 , ( 2007 ) . l. kauffman and s. lomonaco , jr , new journal of physics 6 , 134 ( 2004 ) . chen , k. xue , m .- phys . 323 , 2614 ( 2008 ) ; b .- x xie , k. xue and m .- l ge , phys . rev a 77 , 064101 ( 2008 ) . e. baake , m. baak and h. wager , phys . e57 , 1191 - 1192 ( 1998 ) . for the self - contain we list the graphic expressions for the later use . the bwa reads @xmath271{cross}}},\quad s^{-1}={\raisebox{-0.4\height}{\includegraphics[height=0.8cm , angle=90]{cross}}},\quad e={\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{un}}}\quad ( \mathrm{operators})\ ] ] @xmath272 @xmath273 @xmath274 @xmath275 @xmath276 the ( [ eq.a2 ] ) reads graphically as @xmath277{ssssijk}}}={\raisebox{-0.4\height}{\includegraphics[height=2cm]{sssijk}}}\ ] ] and the @xmath24 takes the simple graph and satisfies t - l algebra @xmath278{unii}}},\quad { \raisebox{-0.4\height}{\includegraphics[height=1.6cm]{complexijk}}}=d\,{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{unll}}},\quad d={\raisebox{-0.33\height}{\includegraphics[height=0.5cm]{loop}}},\quad ( \mathrm{t - l})\ ] ] all the other relations can then be expressed in terms of the graphs . say , @xmath279 @xmath280 @xmath281{sicrossei}}}={\raisebox{-0.4\height}{\includegraphics[height=1.5cm]{eisicross}}}=\sigma{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{un}}}\ ] ] @xmath282{ncross}}}=\sigma{\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{ncup}}},\qquad { \raisebox{-0.4\height}{\includegraphics[height=0.8cm]{ucros}}}=\sigma{\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{ucup}}},\qquad { \raisebox{-0.4\height}{\includegraphics[height=0.8cm]{ncros}}}=\sigma^{-1}{\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{ncup}}}\ ] ] @xmath283 @xmath281{eisisj}}}={\raisebox{-0.4\height}{\includegraphics[height=1.5cm]{sisjej}}}={\raisebox{-0.4\height}{\includegraphics[height=1.2cm]{eiej}}}\ ] ] @xmath284 @xmath281{eisjej}}}={\raisebox{-0.4\height}{\includegraphics[height=1.5cm]{sihsjhej}}}\rightarrow{\raisebox{-0.4\height}{\includegraphics[height=1.5cm]{ucrossnn}}}={\raisebox{-0.4\height}{\includegraphics[height=1.5cm]{tcrosh}}}\ ] ] @xmath285 @xmath281{eisjejb}}}={\raisebox{-0.4\height}{\includegraphics[height=1.5cm]{sihejb}}}\rightarrow{\raisebox{-0.4\height}{\includegraphics[height=1.5cm]{ucrossnn}}}={\raisebox{-0.4\height}{\includegraphics[height=1.5cm]{lun}}}\ ] ] @xmath281{eiejsj}}}={\raisebox{-0.4\height}{\includegraphics[height=1.2cm]{sihej}}}\ ] ] @xmath286 the dependent relations @xmath287 can also easily be expressed in terms of the similar graphs . it can be checked that all of @xmath288,@xmath289,@xmath188 and @xmath189 obey bwm algebra in 3-d representation : @xmath290 @xmath291 @xmath292 @xmath293 @xmath294 and other dependent relations : @xmath295 @xmath296 the basis @xmath297 is defined by ( [ eq1.7 ] ) , so @xmath298 i. e. @xmath299 for @xmath62 ( no sum over repeating indices ) @xmath300{uu}}}+\alpha_i [ { \raisebox{-0.4\height}{\includegraphics[height=0.8cm]{crosuu}}}+w{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{uu}}}-w{\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{tsep}}}+\sigma\beta_i{\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{tsep}}}]\right\ } \nonumber\\ & = \alpha_{i}f_{i}\left\{{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{tcross}}}+\alpha_i^{-1}(1+\alpha_{i}w){\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{uu}}}+\alpha_i^{-1}(\sigma\beta_i-\alpha_{i}w){\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{tsep}}}\right\ } } \ ] ] in order that @xmath301 are eigenstates for @xmath48 it should hold @xmath302 we then have @xmath303 i. e. @xmath304 because of @xmath305{ncross}}}=\sigma{\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{ncup}}}$ ] ( @xmath306 ) , @xmath305{ncros}}}=\sigma^{-1}{\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{ncup}}}$ ] , it holds ( @xmath62 ) @xmath307{ccrossc}}}+\alpha_i { \raisebox{-0.33\height}{\includegraphics[height=0.5cm]{loop}}}+\beta_i{\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{tsquare}}}\right ) \qquad ( { \raisebox{-0.33\height}{\includegraphics[height=0.5cm]{loop}}}=d ) \nonumber \\ & = & d^{-1}f_i\left(\sigma^{-1}d+\alpha_i d+\beta_i d^2\right)=0 \nonumber\end{aligned}\ ] ] where @xmath308 , @xmath309 and ( [ eq.b2 ] ) have been used . substituting ( [ eq.b2 ] ) into ( [ eq1.10 ] ) the basis @xmath297 take the forms @xmath310{tsep}}}\\ |e_i\rangle&=&f_i\left({\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{utcross}}}+\lambda_i { \raisebox{-0.4\height}{\includegraphics[height=0.8cm]{uu}}}+d^{-1}(\sigma^{-1}+\lambda_i){\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{tsep}}}\right)\end{aligned}\ ] ] the relations for bwa in the appendix a have been used . to verify @xmath311 we have to distinguish two types of the braiding matrices from each other for @xmath42 ( hermitian ) and @xmath45 ( unitary ) , respectively . \(a ) for @xmath42 there are @xmath312{tcross}}}|{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{utcross}}}\rangle = d[(\sigma^{-1}-\sigma)w+d ] \end{aligned}\ ] ] @xmath312{tcross}}}|{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{uu}}}\rangle=\sigma d \qquad\\end{aligned}\ ] ] after calculation we find @xmath313+w\sigma d+(\beta_1 + \beta_2)\sigma^{-1}d\\ & & + ( \lambda_1 \beta_2-\beta_1 \lambda_2)d+(\lambda_1 \lambda_2+\beta_1 \beta_2)d^2\\ & = & d\{\sigma^{-1}(w+\beta_1)+\lambda_2 \beta_1\}=0\end{aligned}\ ] ] the @xmath314 leads to ( [ eq1.13 ] ) . \(b ) for @xmath45 we have ( @xmath62 ) @xmath315{ntcros}}}+\alpha_{i}^{*}{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{nn}}}+\beta_{i}^{*}{\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{tncup}}}\right\},\\ { \raisebox{-0.4\height}{\includegraphics[height=0.8cm]{crossuu}}}=s_{12}{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{uu } } } , \quad \mathrm{and}\quad \langle e_1|e_2\rangle={\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{tloop}}}=d^2 } \ ] ] now @xmath309 , @xmath37 are complex and @xmath316 that gives @xmath317 @xmath318^{-1/2 } \end{aligned}\ ] ] for @xmath42 ( @xmath62 ) @xmath319{unot}}}\left\{{\raisebox{-0.4\height}{\includegraphics[height=0.8cm]{utcross}}}+\lambda_i { \raisebox{-0.4\height}{\includegraphics[height=0.8cm]{uu}}}+\beta_i { \raisebox{-0.4\height}{\includegraphics[height=0.5cm]{tsep}}}\right\}\\ & = & d f_i ( \sigma^{-1}+\beta_i d+\lambda_i ) |e_3\rangle\end{aligned}\ ] ] @xmath320 @xmath321 @xmath322 @xmath323|e_1\rangle\right.\\ & + ( \lambda_1-\lambda_2)^{-1}f_2^{-1}[(w+\beta_i)\lambda_1-(\lambda_i - w)\lambda_3]|e_2\rangle\\ & \left.+[(w+\beta_i)\lambda_3^{-1}+(\lambda_i - w)\lambda_3+d]|e_3\rangle\right\ } } \ ] ] @xmath324+\lambda_3^{-1}|e_3\rangle\right\}\ ] ] obviously the base @xmath28 , @xmath29 and @xmath30 form a closed set for the operations @xmath15 and @xmath24 . for @xmath45 ( @xmath62 ) @xmath325 @xmath326 from ( [ eq1.6])-([eq1.13 ] ) it follows that @xmath122 and for @xmath109 as well as @xmath327 ( for @xmath328 ) we get : @xmath329 and @xmath330 that leads to : @xmath331 @xmath332 @xmath333 @xmath334 @xmath335 @xmath336 @xmath337 @xmath338 example 3 : @xmath339 when @xmath184 , the state is separable , and @xmath340 arrive at their minimum value simultaneously . the point @xmath196 is both the local minimum point of @xmath232 and @xmath234 , however , @xmath341 . in addition , @xmath232 and @xmath234 shares another two common local maximum points in the period @xmath342 $ ] , and the two common maximum points both correspond to the maximally entangled state , therefore @xmath343 at the two points . see fig.e1 and fig.e2 . when @xmath184 , the state is a separable sate and @xmath340 reach their minimum value simultaneously . when @xmath196 , @xmath232 and @xmath234 reach their maximum values simultaneously , however both the values are not the same because @xmath345 . see fig.e3 and fig.e4 . example 5 : @xmath346 . when @xmath184 , the state is separable , and @xmath340 arrives at the minimum value . the point @xmath196 is both the local minimum point of @xmath232 and @xmath234 , with @xmath341 . it is worth noting that @xmath232 and @xmath347 norm @xmath234 both have other four local extreme points , in the period @xmath348 . it is shown in fig . @xmath349 and fig.@xmath350 .

the topological basis associated with birman - wenzl - murakami algebra ( bwma ) is constructed and the three dimensional forms of braiding matrices s have been found for both @xmath0 and @xmath1 . a familiar spin-@xmath2 model related to braiding matrix associated with bwma is discussed . the extreme points @xmath3 and @xmath4 of @xmath5-norm and von neumann entropy are shown to be connected to each other . through the general discussion and examples we then point out that the @xmath5-norm describes quantum entanglement . in the ref.@xcite two types of braiding matrices with two distinct eigenvalues , i , e . those associated with temperely - lieb algebra ( tla)@xcite and their corresponding solutions of yang - baxter equation ( ybe)@xcite had been discussed . based on the topological basis @xcite the @xmath6-dimensional braiding operators were mapped onto 2-dimensional ones @xcite . we had shown that the two types of braiding matrices are related to the extremes of @xmath5-norm of wigner s d - function@xcite . especially the @xmath7-d braiding matrix corresponding the bell basis ( the type - ii ) is connected to the maximum of the @xmath5-norm , whereas the permutation and extensions ( the type - i ) to the minimum . it hints that the @xmath5-norm should relate to the quantum information . it is natural to extend the discussions in the ref.@xcite to the solution of ybe with three distinct eigenvalues . among them , the most important ones belong to birman - wenzl - murakami algebra @xcite(bwma ) . as is well - known that the forms of braiding matrices with three distinct eigenvalues were given by references , say in @xcite , and the connection with bwma was shown in ref.@xcite for both standard and non - standard solutions . in parallel to the ref.@xcite , in this paper we shall firstly set up the topological base @xmath8 , @xmath9 , and @xmath10 associated with bwma , then map the @xmath11-d braiding matrices to @xmath12-d forms . the physical application of bwma is raised through a familiar model which is different from the model discussed in ref . @xcite . we shall point out that in general the extremes of @xmath5-norm of the d - functions ( @xmath13 and @xmath14 ) are related to those of von neumann entropy . we also take the spin-@xmath2 models as examples to favor the statement .

phenomenology of models based on extra spatial dimensions @xcite is quite popular now . besides offering a solution to the hierarchy problem of the standard model @xcite these models allow for the low - energy unification of the gauge couplings @xcite , provide a rich tev scale new physics phenomenology , existence of gravity at the tev scale and may even offer candidate(s ) for the cold dark matter of the universe @xcite . in the simplest string theory inspired extension of the standard model ( sm ) based on one extra spatial dimension , originally proposed by randell - sundram ( rs ) @xcite , gravitons are the only propagating particles in the bulk . such gravitons will therefore have kaluza - klein ( kk ) excitations which will appear in experiments as a widely separated resonances . this contrasts with the kk spectrum of models with compact extra dimensions such as add @xcite where there are a very large number of closely spaced graviton modes . the coupling of these graviton excitations to the sm is through the 4-d reduced planck mass @xmath0 , which may be on the tev scale , rather than the inaccessibly large planck mass at @xmath1 gev . the couplings to standard model particles are therefore be proportional to @xmath2 , thereby allowing graviton excitations to decay into all the sm particles including a fermion pair or a pair of gauge bosons . at tev scale energies , when such graviton excitations are produced , this variety of possible decay modes will give rise to vast phenomenology at the tev scale ; @xmath3 , with @xmath4 as a warped factor which arise due to the compactification of the extra dimension on a circle with radius @xmath5 . the factor @xmath6 is due to the fact that sm in located on the circle at @xmath7 and @xmath8 is the curvature parameter . although lots of variations of the rs model have been proposed over the years @xcite and their phenomenology @xcite has been studied in detail , in this paper we will consider the original scenario . in particular , we assume that the whole the sm is localized on the tev brane , so that the mass of gravitons is given by @xmath9 , where @xmath10 are the roots of the first - order bessel function . in order to be useful in the resolution of the hierarchy problem and keep gravity weak enough to be treated perturbatively , @xmath11 should lie in the range @xmath12 . the focus of this paper will be the distinctive triphoton signature at the large hadron collider ( lhc ) produced by the rs model and other models like it . the importance of such a signature lies in the fact that this signature is experimentally clean and a distinctive signature for models of this type . we will also discuss graviton mass reconstruction and the angular distribution of the graviton decay which which are important tools for characterizing the physics which produces the triphoton signal . the paper is organized as follow : in section 2 , we discuss the graviton production in association with a photon and its decays into the sm particles . section 3 focuses on the numerical analysis of signal and background as well as graviton reconstruction in detail . finally , in section 4 , we summarize our findings . triphoton signal in the rs model will arise due to the associated production of a on shell graviton with a photon while the graviton subsiquently decays into an additional photon pair . in this section we discuss the production process as well as the various other dominant two body decays of the graviton . the parton - level matrix - element for the production process @xmath13 as calculated in @xcite is , @xmath14 with , + @xmath15 @xmath16 + where , @xmath17 and @xmath18 . @xmath19 is the number of colors , @xmath20 and @xmath21 is the quark charge , @xmath22 for up and down type quarks respectively . the kinematics of this process implies that @xmath23 . note that this cross - section is symmetric with respect to the interchange of @xmath24 and @xmath25 . the lhc production cross - section for this process is presented in figure [ fig : gprd ] . we use a wide range of graviton mass well above the tevatron bounds @xcite for three different sets of center - of - mass energy , @xmath26 as 7 tev , 10 tev , and , 14 tev respectively . we use cteq6l-1@xcite parton densities at @xmath27 , and the renormalization and factorization scales are set as , @xmath28 . from the figure [ fig : gprd ] , it is quite clear that due to the low cross - section , it is hard to observe such productions with the early lhc data with an integrated luminosity of 100 pb@xmath29 and @xmath30 tev . for instance only 2 events would be produced if @xmath31 tev ; at the higher energy @xmath32 tev , the number of events is roughly doubled . tev has been assumed in this chart.,width=336,height=288 ] in figure [ fig : gdec ] , we present the branching fractions of graviton to various two - body sm mode . the dominant mode is the dijet channel with a branching ratio @xmath33 ( where , @xmath34 ) . the fraction for the diphoton mode is smaller , @xmath35 , but it is important as this will lead to a clean signature at the lhc . the production of a graviton associated with a photon occurs in the high x - region , so generally the associated photon as well as the photons produced in the graviton decay will carry high transverse momentum . large transverse momentum cuts on the triphoton signal will therefore be helpful in selecting graviton events and rejecting sm backgrounds where photons tend to have low transverse momentum . we use madgraph @xcite to produce signal events with a photon and a graviton . later , these events are interfaced to pythia @xcite for the analysis purpose . decay of gravtiton is done using the decay table in pythia . braching fractions for different decay modes for @xmath36 tev are shown in fig . [ fig : gdec ] . before selecting our event samples , we order the photons on the basis of their transverse momentum i.e. , @xmath37 . in order to analyze the actual event rate expected to observe at the lhc , we first employ the following three basic cuts : * @xmath38 gev and @xmath39 , * photon - photon separation , @xmath40 ; @xmath41 , @xmath42 , @xmath43 being psuedo - rapidity of photon @xmath44 and is defined as @xmath45 . * missing transverse energy , @xmath46 . [ fig : spt1 ] the sm background is mostly composed of ( a ) the direct triphoton production i.e. where three photons are produced by sm processes , and ( b ) processes involving fake photons , in other words , where the detector incorrectly identifies particles which are not photons as photons . in order to estimate the sm background we use madgraph with the same parton densities as for the signal case . at 14 tev , the total cross - section for the direct triphoton production in the sm is @xmath47 pb . the most likely fake photon backgrounds are those due to the misidentification of jet or an electron as a photon we can thus subdivide the processes in case ( b ) as follows : \(i ) @xmath48 , @xmath49 @xmath50 , and , \(ii ) @xmath51 , @xmath52 and @xmath53 . \(iii ) @xmath54 , @xmath55 , @xmath56 . we found that the bare cross - sections for these processes are @xmath57 pb , @xmath58 pb and , @xmath59 pb . with the fake probability for a jet to be identified as a photon , @xmath60 , as given in @xcite , the contributions due to these channels to the triphoton background will be @xmath61 pb , @xmath62 pb and @xmath47 pb respectively . cross - sections for the processes involving electrons ( and positrons ) are @xmath63 pb , @xmath64 pb , and , @xmath65 pb . by assuming a fake rate similar to that of a jet the net contribution to the background from these processes will be @xmath66 pb which is mostly due to the process involving one electron or a positron arising due to a @xmath67 production , where @xmath67 decays into an electron or a positron and the corresponding neutrino or anti - neutrino . contribution due to processes involving one or more jets and electrons ( or positrons ) is @xmath68 pb which is also insignificant even though the bare rates are @xmath69 pb , @xmath70 pb and @xmath71 pb for @xmath72 , @xmath55 , @xmath56 respectively . the background form fake photons due to jets is the highest with a combined cross - section of @xmath73 pb which is about two times larger than the direct sm triphoton background . in figures [ fig : ptg ] , [ fig : pts ] and [ fig : invm ] , we present @xmath74 , @xmath75 and invariant mass @xmath76 distributions with basic cuts for three different values of graviton mass in case of signal , and , the net background due to direct tri - photons and faked backgrounds due to electorns ( or positrons ) and photons as discussed above . from figures [ fig : ptg ] , it is clear that a demand of @xmath77 gev and @xmath78 gev will ensure that the triphotons are indeed due to the a heavy graviton production . we employ further cuts on the scalar sum of photons transverse momenta , @xmath75 and invariant mass of harder photons , @xmath76 . effects of these individual and combined cuts on the signal and background event rates for an lhc luminosity @xmath79 fb@xmath29 is presented in table [ t : lhc ] . as it clear from the table [ t : lhc ] that , though a large fraction of the background is already eliminated at the basic level itself due to the demand that the missing transverse energy per event should not exceed @xmath80 gev limit . yet a cut on scalar sum of the transverse momenta of all the photons , @xmath75 of about 550 gev or alternatively , @xmath76 of about 800 gev in addition to the basic and photon selection cuts reduces the background significantly . a combination of all these cuts reduces to background events to 14 while keeping signals events to 293 , 213 and 105 respectively for the graviton mass 1 , 1.5 and 2 tev respectively for an integrated luminosity of 300 fb@xmath29 . in order to reconstruct the graviton mass we use the following techniques : we first reconstruct all possible diphoton pairs with the @xmath74 ordered photons . later we drop those pairs in each event which are nearer to each other , i.e. , those with @xmath81 gev where , @xmath82 is the difference of reconstructed invariant masses . clearly with this technique we are able to help a graviton mass with an uncertainty within @xmath83 per cent even in the presence of background as shown in figure [ fig : invm ] . [ fig : ct1 ] .efficiency of cuts on signal and background triphoton events for an integrated lhc luminosity of @xmath84 fb@xmath29 [ cols="^,^,^,^,^,^",options="header " , ] once we have reconstructed the graviton mass , the next task is to measure its spin in order to ensure that the photons are indeed due to decay of a rs graviton . in order to do so , we first identify the correct photon pair which yield the right graviton mass peak in a clean signature case . next , we boost the photon pair into the rest frame of the graviton . in this boosted frame we produce angular distribution of either of the photons as the two have nearly identical shape . we plot angular distribution for the first photon in figure [ fig : angle ] . we have studied triphoton signals in details . with a detailed study on background arising due to various sm processes including the processes where a photon can be faked by a jet or an electron or positron . we found that the processes is relatively clean and the background may be greatly reduced ( see the table [ t : lhc ] ) . we have also shown that the graviton mass reconstruction and its spin measurement is possible in this signals though we need to wait for 300 fb@xmath29 data in order to have sufficient statistics at the lhc . see for example , r. rattazzi , _ proc . of cargese school of particle physics and cosmology _ , corsica , france , 4 - 16 aug 2003 ; j. hewett and m. spiropulu , ann . nucl . and part . sci . * 52 * , 397 ( 2002 ) ; c. csaki , arxiv : hep - ph/0404096 ; r. sundrum , arxiv : hep - th/0508134 . s. weinberg , phys . rev . d * 13 * , 974 ( 1976 ) , e. gildener , phys . d * 14 * , 1667 ( 1976 ) ; phys . rev . d * 19 * , 1277 ( 1979 ) ; l. susskind , phys . rev . d * 20 * , 2619 ( 1979 ) ; g. t hooft , in _ recent developments in gauge theories _ , proceedings of the nato advanced summer institute , cargese 1979 , ( plenum , 1980 ) . l. randall and m. d. schwartz , jhep * 0111 * , 003 ( 2001 ) [ arxiv : hep - th/0108114 ] ; l. randall and m. d. schwartz , phys . lett . * 88 * , 081801 ( 2002 ) [ arxiv : hep - th/0108115 ] ; w. d. goldberger and i. z. rothstein , phys . d * 68 * , 125011 ( 2003 ) [ arxiv : hep - th/0208060 ] ; k. w. choi and i. w. kim , phys . d * 67 * , 045005 ( 2003 ) [ arxiv : hep - th/0208071 ] ; k. agashe , a. delgado and r. sundrum , annals phys . * 304 * , 145 ( 2003 ) [ arxiv : hep - ph/0212028 ] . k. agashe and g. servant , phys . lett . * 93 * , 231805 ( 2004 ) [ arxiv : hep - ph/0403143 ] ; k. agashe and g. servant , jcap * 0502 * , 002 ( 2005 ) [ arxiv : hep - ph/0411254 ] ; a. pomarol , phys . lett . * 85 * , 4004 ( 2000 ) [ arxiv : hep - ph/0005293 ] . l. randall and r. sundrum , phys . lett . * 83 * , 4690 ( 1999 ) [ arxiv : hep - th/9906064 ] . l. randall and r. sundrum , phys . lett . * 83 * , 3370 ( 1999 ) [ arxiv : hep - ph/9905221 ] . n. arkani - hamed , s. dimopoulos and g. r. dvali , phys . b * 429 * , 263 ( 1998 ) [ arxiv : hep - ph/9803315 ] ; i. antoniadis , n. arkani - hamed , s. dimopoulos and g. r. dvali , phys . b * 436 * , 257 ( 1998 ) [ arxiv : hep - ph/9804398 ] ; n. arkani - hamed , s. dimopoulos and g. r. dvali , phys . d * 59 * , 086004 ( 1999 ) [ arxiv : hep - ph/9807344 ] . n. arkani - hamed , s. dimopoulos , g. r. dvali and n. kaloper , phys . rev . lett . * 84 * , 586 ( 2000 ) [ arxiv : hep - th/9907209 ] ; j. d. lykken and l. randall , jhep * 0006 * , 014 ( 2000 ) [ arxiv : hep - th/9908076 ] ; i. oda , phys . b * 480 * , 305 ( 2000 ) [ arxiv : hep - th/9908104 ] ; c. csaki and y. shirman , phys . d * 61 * , 024008 ( 2000 ) [ arxiv : hep - th/9908186 ] ; i. oda , phys . b * 472 * , 59 ( 2000 ) [ arxiv : hep - th/9909048 ] ; t. j. li , arxiv : hep - th/9911234 ; h. davoudiasl , j. l. hewett and t. g. rizzo , phys . * 84 * , 2080 ( 2000 ) [ arxiv : hep - ph/9909255 ] ; i. i. kogan , s. mouslopoulos , a. papazoglou , g. g. ross and j. santiago , nucl . b * 584 * , 313 ( 2000 ) [ arxiv : hep - ph/9912552 ] . h. davoudiasl , j. l. hewett and t. g. rizzo , phys . b * 473 * , 43 ( 2000 ) [ arxiv : hep - ph/9911262 ] ; ibid . phys . rev . lett . * 84 * , 2080 ( 2000 ) [ arxiv : hep - ph/9909255 ] . a. pomarol , phys . b * 486 * , 153 ( 2000 ) [ arxiv : hep - ph/9911294 ] ; y. grossman and m. neubert , phys . b * 474 * , 361 ( 2000 ) [ arxiv : hep - ph/9912408 ] ; k. agashe , a. delgado , m. j. may and r. sundrum , jhep * 0308 * , 050 ( 2003 ) [ arxiv : hep - ph/0308036 ] . v. m. abazov _ et al . _ [ d0 collaboration ] , phys . rev . * 100 * , 091802 ( 2008 ) [ arxiv:0710.3338 [ hep - ex ] ] . h. l. lai _ et al . _ , phys . d * 51 * , 4763 ( 1995 ) [ arxiv : hep - ph/9410404 ] . f. maltoni and t. stelzer , jhep * 0302 * , 027 ( 2003 ) [ arxiv : hep - ph/0208156 ] . k. hagiwara , j. kanzaki , q. li and k. mawatari , eur . j. c * 56 * , 435 ( 2008 ) [ arxiv:0805.2554 [ hep - ph ] ] . t. sjostrand , s. mrenna and p. skands , jhep * 0605 * , 026 ( 2006 ) [ arxiv : hep - ph/0603175 ] . g. aad _ et al . _ [ the atlas collaboration ] , arxiv:0901.0512 [ hep - ex ] .

we investigate triphoton signals of the randall - sundram model at the large hadron collider . such a signal can be an important probe to the rs model as these are relatively cleaner from the hadronic activity and bear significant rate . the corresponding standard model background has also been studied in detail . we also show that a clear graviton reconstruction is possible in such signal . + +

as we all know , chiral symmetry breaking was originally explained quite well by nambu - jona - lasinio ( njl ) model as early as 1961 @xcite . like in superconductivity , the strong attractive force between quark and antiquark in the @xmath2 channel can create non - perturbative ground state with @xmath3 condensation . due to the pair condensation , the original symmetry between massless left and right - handed quarks is broken down to @xmath4 , and then the quarks obtain constituent mass . the remaining residual interactions between the constituent quarks bind them into collective excitations , i.e. , hadrons in the chiral symmetry breaking vacuum . especially in the pseudoscalar channel the residual strong interaction creates massless pions as goldstone bosons in the chiral limit . when a small current quark mass @xmath5 is introduced in the theory , chiral symmetry is explicitly broken , and pion obtains its small physical mass @xmath6 . although the njl model has two serious drawbacks , i.e. , lacks of confinement and renormalizability , it is still regarded as an applicable model at low momentum , especially for dealing with processes of pion , such as pion - pion scattering near threshold . traditionally , the scheme of the njl model is represented by two schwinger - dyson ( sd ) equations , one is for the constituent quark propagator , and the other is for the composite meson propagator . at the lowest level , the applications of the njl model are based upon mean - field approximation @xcite - @xcite , i.e. , hartree approximation to the gap equation for quark mass and the random - phase approximation ( rpa ) to the bethe - salpeter equation for meson mass . it is clear , that at this level the solution of the gap equation determines the meson propagators , but the solution of meson sd equation has no feedback to the quark propagator . since the constituent quark propagator is the fundamental element , from which all the quantities , including quark mass , meson masses and quark - antiquark condensate , are calculated , it is necessary to consider the back contribution of meson modes to the quark propagator . among efforts @xcite - @xcite to go beyond the mean - field approximation , refs . @xcite and @xcite are in a chirally symmetric self - consistent approximation , namely the chiral properties such as the goldstone s theorem , the goldberger - treiman relation and the conservation of the quark axial current are exactly preserved in the chiral limit of the njl model . by using effective action method in a semi - bosonized way , and expanding the action to one quark - loop and one - meson - loop in @xcite , or directly evaluating the feynman diagrams under the constraint to keep the chiral relations at quark level in @xcite . in this paper , we extend the method of @xcite to a general scheme with explicit chiral symmetry breaking in the su(2 ) njl model . different from the case in the chiral limit , we must be careful to deal with the form of internal meson propagators . in a way different from @xcite , we regard the constituent quark as the fundamental element and only expand quark s propagator in the power of small external momentum in the calculation of feynman diagrams . in the process to go beyond the mean - field approximation , we have to deal with the divergent integrals of quark loops and meson loops . we adopt pauli - villars regulation @xcite to treat divergent integrals resulted from quark loops , and choose a covariant cutoff @xmath7 for the meson momentum . there are four parameters in our treatment , namely the current quark mass @xmath5 , quark coupling constant @xmath8 , fermionic cut - off @xmath9 and bosonic cut - off @xmath7 , to be fixed . in the mean - field approximation , the three parameters @xmath5 , @xmath8 , @xmath9 are usually fixed by comparing with the pion mass @xmath10 mev , pion decay constant @xmath11 mev and the quark condensate @xmath12 mev . in the near future , the dirac experiment will measure the @xmath13 scattering lengths in good precision , which will shed some light on the quark condensate @xcite . to see clearly the quark condensate dependence of the four parameters , we give only the quark condensate a reasonable constraint : -300 mev @xmath14 -200 mev . the outline of this paper is as follows : in section 2 , we briefly review the general scheme represented by two schwinger - dyson equations in the su(2 ) njl model . in section 3 , we introduce the method of external momentum expansion , and prove a general relation between the pion polarization function and the axial - vector matrix element . we also deduce the internal meson propagator to @xmath1 order in the @xmath0 expansion . our numerical results with mesonic contributions and the effect of explicit chiral symmetry breaking will be shown in section 4 . the conclusions are given at the end . in this section , we briefly review the traditional scheme of su(2 ) njl model with large @xmath0 expansion . the two - flavor njl model is defined through the lagrangian density , @xmath15,\end{aligned}\ ] ] here @xmath8 is the effective coupling constant of dimension @xmath16 , and @xmath5 is the current quark mass , assuming isospin degeneracy of the @xmath17 and @xmath18 quarks , and @xmath19 are quark fields with flavor , colour and spinor indices suppressed . the traditional non - perturbative method of njl model is inspired from many - body theory . the complete description is represented by two schwinger - dyson ( sd ) integral equations , i.e. , the constituent quark propagator , see fig . 1a , and the composite meson propagator , see fig . ( 350,100)(0,0 ) ( 60,80)[]a ( 50,60)(90,60 ) ( 200,60)(220,60 ) ( 200,85)(10,8)(0)0.8 ( 102,60)[]= ( 110,60)(150,60 ) ( 165,60)[]+ ( 180,60)(200,60 ) ( 200,60)(200,75)4 ( 200,60)2.5 ( 200,75)2.5 ( 60,40)[]b ( 55,20)(90,20)4 ( 55,20)2.5 ( 90,20)2.5 ( 250,20)(270,20)4.5 ( 240,20)(10,8)(90)0.8 ( 110,20)[]= ( 130,20)2.5 ( 165,20)2.5 ( 130,20)(165,20)4 ( 180,20)[]+ ( 205,20)(230,20)4 ( 205,20)2.5 ( 230,20)2.5 ( 250,20)2.5 ( 270,20)2.5 the two sd equations must couple to each other self - consistently and keep chiral symmetry in the chiral limit . the one - vertex grey bubble @xmath20 in fig . 1@xmath21 represents the quark self - energy , and the two - vertex grey bubble @xmath22 in fig . 1@xmath23 indicates the meson polarization function . in the case of su(2 ) , meson modes refer to only pseudoscalar mesons @xmath24 and scalar mesons @xmath25 . it is difficult to give the full expressions of @xmath20 and @xmath22 in fig . usually an approximation scheme called large @xmath0 expansion is adopted in njl model , i.e. , the both kernels expanded in @xmath26 , where @xmath0 is the number of color . in this scheme , @xmath27 is a constant when @xmath28 , so coupling constant scales like @xmath29 . one can find the detailed description of large @xmath0 expansion in @xcite . v.dmitrainovi ' c et.al . proved in their paper @xcite that the @xmath20 and @xmath22 shown in fig . 2 are self - consistent leading and subleading order in @xmath0 expansion and can keep all the chiral relations in the chiral limit . the leading @xmath30 order of quark self - energy @xmath20 in fig . 2 is named @xmath31 after hartree approximation , and the subleading @xmath1 order is called @xmath32 . including current quark mass @xmath5 , the gap equation for quarks can be expressed as @xmath33 where @xmath34 , and @xmath35 is given in appendix in order to compactly discuss physics in the text . the leading @xmath36 and subleading @xmath30 order of meson polarization function @xmath22 are expressed as @xmath37 and @xmath38 respectively , here @xmath39 represents @xmath24 or @xmath25 . it is clear to see that the back interaction which conserves all the chiral properties is reflected in the contribution from the meson propagator to the quark mass . the total meson propagator can be expressed in terms of the total polarization function @xmath40 where @xmath41 , and the rpa part is @xmath42 for @xmath24 and @xmath43 for @xmath25 , and the function @xmath44 is defined in appendix . ( 450,100)(0,0 ) ( 30,50)(15,10)(0)0.8 ( 30,35)2.5 ( 120,50)(15,0,360 ) ( 120,35)2.5 ( 30,19)[]@xmath45 ( 120,19)[]@xmath31 ( 200,50)(15,0,360 ) ( 200,70)(15,320,220)4 ( 75,50)[]= ( 160,50)[]+ ( 200,35)2.5 ( 189,61)2.5 ( 210,61)2.5 ( 200,19)[]@xmath32 ( 200,75)[]@xmath24,@xmath25 ( 450,100)(0,0 ) ( 30,50)(15,10)(90)0.8 ( 30,15)[]@xmath22 ( 15,50)2.5 ( 45,50)2.5 ( 67.5,50)[]= ( 105,50)(15,0,360 ) ( 90,50)2.5 ( 120,50)2.5 ( 110,15)[]@xmath37 ( 135.5,50)[]+ ( 180,50)(15,0,360 ) ( 180,65)(180,35)4 ( 165,50)2.5 ( 180,65)2.5 ( 180,35)2.5 ( 195,50)2.5 ( 180,15)[]@xmath46 ( 180,50)[]@xmath24,@xmath25 ( 215.5,50)[]+ ( 260,50)(15,0,360 ) ( 260,70)(15,320,220)4 ( 249,60)2.5 ( 248,41)2.5 ( 271,60)2.5 ( 272,41)2.5 ( 260,75)[]@xmath24,@xmath25 ( 260,15)[]@xmath47 ( 295.5,50)[]+ ( 340,50)(10,0,360 ) ( 400,50)(10,0,360 ) ( 330,50)2.5 ( 340,60)2.5 ( 340,40)2.5 ( 410,50)2.5 ( 400,60)2.5 ( 400,40)2.5 ( 370,65)[]@xmath24 ( 370,35)[]@xmath25 ( 370,15)[]@xmath48 in the njl model , mesons are bound states of constituent quark and anti - quark , the meson mass @xmath49 satisfies the meson propagator s pole condition @xmath50 and the coupling constant @xmath51 is determined by the residue at the pole @xmath52 performing direct calculation of the feynman diagrams in fig . 2 leads to the explicit expressions for the quark mass @xmath53 \int\frac{d^4p}{(2\pi)^4 } { \rm tr}[\gamma_5 s(p+q ) \gamma_5 s(p ) 1 s(p ) ] \nonumber \\ & - & 4 gn_c \int\frac{d^4q}{(2\pi)^4}[-{\rm i}d^{(rpa)}_{\sigma}(q ) ] \int\frac{d^4p}{(2\pi)^4}{\rm tr}[1 s(p+q ) 1 s(p ) 1 s(p)],\end{aligned}\ ] ] and for pion polarization fluctuations @xmath54 \int\frac{d^4p}{(2\pi)^4 } { \rm tr}[\gamma_5 s(p+q ) \gamma_5 s(p)\gamma_5 s(p - k ) \gamma_5 s(p - k+q ) ] \nonumber \\ & & + \int\frac{d^4q}{(2\pi)^4}[-{\rm i}d^{(rpa)}_{\sigma}(q ) ] \int\frac{d^4p}{(2\pi)^4 } { \rm tr}[\gamma_5 s(p+q ) 1 s(p ) \gamma_5 s(p - k ) 1 s(p - k+q ) ] ] , \\ & \delta \pi^{(c)}_{\pi}(k ) & = -12in_c[\int\frac{d^4q}{(2\pi)^4}[-{\rm i}d^{(rpa)}_{\pi}(q ) ] \int\frac{d^4p}{(2\pi)^4 } { \rm tr}[\gamma_5 s(p ) \gamma_5 s(p+q ) \gamma_5 s(p ) \gamma_5 s(p - k ) ] \nonumber \\ & & - { 1 \over 3 } \int\frac{d^4q}{(2\pi)^4}[-{\rm i}d^{(rpa)}_{\sigma}(q ) ] \int\frac{d^4p}{(2\pi)^4 } { \rm tr}[\gamma_5 s(p ) 1 s(p+q ) 1 s(p ) \gamma_5 s(p - k ) ] ] , \\ & \delta \pi^{(d)}_{\pi}(k ) & = i } \gamma^{\sigma \pi \pi}(k , q ) ] [ -{\rm i}d^{(rpa)}_{\pi}(q ) ] [ -{\rm i}d^{(rpa)}_{\sigma}(q - k)][-{\rm i } \gamma^{\sigma \pi \pi}(q , k)].\end{aligned}\ ] ] here , the expression @xmath55 is the ( @xmath56)-vertex @xmath57,\end{aligned}\ ] ] @xmath58 the quark propagator and @xmath59 the internal meson propagator . in all the formulae of this paper , @xmath60 means the @xmath24 external momentum , @xmath61 the internal quark momentum and @xmath62 the internal meson momentum ; another important quantity is the pion decay constant @xmath63 which is calculated from the vacuum to one - pion axial - vector matrix element , replacing one vertex @xmath64 of @xmath22 in fig . 2 by @xmath65 . similar to the expression of @xmath66 , we express @xmath67 as @xmath68 since the feynman diagrams for @xmath66 and @xmath63 have the same structure , we expect to find a general relation between them . starting from the axial - vector matrix element @xmath69 , and using the single - quark axial ward identity involving the quark propagator @xmath70 , @xmath71 one can deduce the general relations functions @xmath72 the detailed proof can be found in ref.@xcite . using pion s pole condition eq . ( [ pole ] ) and gap equation eq . ( [ gap ] ) , we get finally the relation @xmath73 where we have defined the on - shell quantities @xmath74 . while in the chiral limit , @xmath63 satisfies the goldberger - treiman relation @xmath75 . we are familiar with one - loop feynman diagrams , but it is difficult to analytically calculate the two - loop feynman diagrams such as those in @xmath22 . since pion s external momentum @xmath76 is small , we will calculate the subleading corrections in @xmath22 to @xmath77 term only , but keep the complete contributions from the leading order . in the njl model , the constituent quarks are the fundamental elements , and the mesons are their bound . in fact , all the quantities in the njl model are calculated from the quark propagator @xmath58 . therefor , if we want to build up a self - consistent expansion in the caculation of the the 2-loop diagrams , we should only expand the quark propagator in the external momentum @xmath60 @xmath78 with this expansion , all the two - loop diagrams can be expanded naturally in @xmath60 in a unified way . the lorentz covariance gives the expansion form of @xmath79 as @xmath80 in the following , we consider @xmath81 to the @xmath77 term only . in this approximation , the pole condition for the pion propagator is simplified as @xmath82 with the expression of @xmath83 and the gap equation eq . ( [ gap ] ) we obtain the gap equation to determine the pion mass , @xmath84 similarly , the expansion of the residue of the pole to the order @xmath77 gives the pion - quark coupling constant @xmath85 where @xmath86 , and the quark - loop integral @xmath87 is defined in appendix . it is tedious to calculate @xmath88 directly . here we will use another equivalent way by calculating the first term of the axial - vector matrix element in the external momentum expansion . we have shown the general relations between pion polarization function and axial - vector matrix element in eq . ( [ fgpi ] ) . if @xmath89 , i.e. , in the chiral limit , eq . ( [ fgpi ] ) becomes @xmath90 thus the eq . ( [ fgpi ] ) in general case can be rewritten as @xmath91 with the expansion form of @xmath66 , the second equation of ( [ fgpi1 ] ) becomes @xmath92 by replacing in the feynman diagrams in @xmath93 one vertex @xmath94 by @xmath95 , the obtained subleading axial - vector matrix element has the following form in external momentum expansion : @xmath96 comparing the two equations ( [ expfgpi1 ] ) and ( [ expfey ] ) , we derive the pion polarization fluctuation in terms of the first coefficient of the expansion of the axial - vector matrix element . @xmath97 with @xmath98(-3q^2l(q ) ) \nonumber \\ & & + \int\frac{d^4q}{(2\pi)^4}[-{\rm i}d^{(rpa)}_{\sigma}(q ) ] ( 4k(q)+3(4m^2-q^2)l(q ) ) \ } , \\ & m^{(c)}(0 ) & = i~n_c \ { 6 \int\frac{d^4q } { ( 2\pi)^4}[-{\rm i}d^{(rpa)}_{\pi}(q)](k(q)+3k(0)-3q^2m(q ) ) \nonumber \\ & & + 2 \int\frac{d^4q}{(2\pi)^4}[-{\rm i}d^{(rpa)}_{\sigma}(q ) ] ( 5k(q)+3k(0)-3(q^2 - 4m^2)m(q ) ) \ } , \\ & m^{(d)}(0 ) & = -32in_c~\int\frac{d^4q}{(2\pi)^4 } n_c [ -{\rm i}d^{(rpa)}_{\pi}(q)][-{\rm i}d^{(rpa)}_{\sigma}(q ) ] \{-(i(q)+2m^2k(0))(i(q)-i(0 ) ) \nonumber \\ & & + ( i(q)+i(0)-(q^2 + 2m^2)k(q))i(q ) -q^2i(q)(i(q)+2m^2k(0))[-{\rm i}d^{(1)}_{\sigma}(q ) ] \},\end{aligned}\ ] ] where @xmath99 , @xmath100 are defined in appendix . in calculating @xmath101 , we have expanded the internal @xmath25 propagator as @xmath102 = [ -{\rm i}d^{(rpa)}_{\sigma}(q)](1+q_{\nu}k^{\nu}[-{\rm i}d^{(1)}_{\sigma}(q ) ] [ -{\rm i}d^{(rpa)}_{\sigma}(q ) ] ) , \end{aligned}\ ] ] with @xmath103 = 8n_c(i(q)+((4m^2-q^2)/2q^2)(i(q)-i(0)+q^2k(q ) ) [ -{\rm i}d^{(rpa)}_{\sigma}(q)])$ ] . till now , we have assumed that @xmath20 and @xmath22 shown in fig . 2 are expanded properly to leading and subleading order in @xmath0 . in the chiral limit beyond mean - field approximation , the leading meson polarization function @xmath104 has the order of @xmath36 , which gives the rpa meson propagator @xmath105 in the order of @xmath1 , thus @xmath32 is the subleading @xmath1 order of quark self - energy . however , it is not the case when current quark mass is introduced beyond mean - field approximation . now we first analyze the one - quark - loop pion propagator @xmath106 beyond mean - field approximation and including @xmath5 explicitly . the definition of rpa pion propagator is @xmath107 substituting the expression of @xmath108 , and using the gap equation eq . ( [ gap ] ) , the denominator is @xmath109 with the pion pole condition eq . ( [ pole ] ) , we can deduce @xmath110 so the rpa pion propagator ( [ rpapro1 ] ) can be expressed as @xmath111 here @xmath112 is the subleading diagrams contribution , which is in the order of @xmath30 . this term induces the rpa propagator expanded to @xmath113 order , where @xmath114 could go to @xmath115 . if we substitute this complete form of internal pion propagator into @xmath45 , it is clear that the @xmath1 part of rpa propagator corresponds to the @xmath1 part of @xmath32 , and other suppressed parts of rpa propagator correspond to the suppressed parts of @xmath32 . in order to keep the proper subleading order @xmath1 of @xmath32 , the internal meson propagator must be in its leading order @xmath1 . we neglect the contributions resulted from @xmath112 , and get the leading order of pion propagator in @xmath1 , @xmath116 similarly , the @xmath1 sigma propagator is @xmath117 it can be seen that @xmath8 is not included explicitly in the two rpa meson propagators , and this simplifies our numerical calculations . now we turn to the numerical evaluation . as evaluated in the last two chapters , we have three equations , the gap equation , eq . ( [ gap ] ) , pion pole condition , eq . ( [ pole ] ) , and pion decay constant , eq . ( [ onshell ] ) . i the chiral limit , pion pole condition eq . ( [ pole ] ) is trivial and the pion decay constant satisfies the goldberger - treiman relation . in order to deal with the divergence resulted from quark - loop integral and meson - loop integral , like in @xcite , we introduce two cut - off , the quark momentum cut - off @xmath9 in pauli - villars regularization and the meson momentum cut - off @xmath7 in covariant regularization . as pointed out in introduction , there are four parameters , the current quark mass @xmath5 , coupling constant @xmath8 , quark momentum cut - off @xmath9 and meson momentum cut - off @xmath7 to be fixed . by comparing with two observables @xmath118 , @xmath119 and one reasonable empirical range of @xmath120 , we can not give fixed values of these four parameters . here we regard the ratio @xmath121 as one free parameter . for each @xmath122 , we can get a series of solutions from the above conditions . the meson cloud effect is now characterized by @xmath122 , the larger @xmath122 means more meson contributions . specially , when @xmath123 , i.e. , @xmath124 , it returns to the mean - field approximation . in the case of explicit chiral symmetry breaking @xmath125 , giving a @xmath122 , for different @xmath126 , we find a series of @xmath9 and @xmath5 from the above three equations , and then @xmath8 , @xmath127 and other quantities could be calculated . in the case of chiral limit @xmath128 , giving a @xmath122 , for different @xmath126 , we find @xmath9 and @xmath8 from goldberger - treiman equation and the gap equation , then @xmath127 and other quantities are calculated . in both cases , the mesonic contributions to the quark self - energy @xmath126 and to the pion polarization @xmath66 are negative , and all the ratios of the mesonic corrections to the one - quark - loop contributions to @xmath126 , @xmath66 and physical quantities @xmath63 and quark condensate increases with increasing @xmath122 , and can reach 30 - 40@xmath129 for values of parameters in the range @xmath130 and @xmath131 . these results are qualitatively the same as those in @xcite and @xcite . our numerical results are shown in figs . 3 - 5 , which concentrate on investigating the effect of explicit chiral symmetry breaking . in all figures , the solid lines correspond to the case of explicit chiral symmetry breaking @xmath125 , and the dashed lines correspond to the case of chiral limit @xmath128 , and @xmath132 correspond to @xmath133 respectively . in fig . [ mcond_fig ] , we plot the quark scalar condensate @xmath134 as a function of constituent quark mass @xmath126 for different values of @xmath122 . this figure is different from the fig . 6 and fig . 7 in @xcite . our numerical results show that : + 1 ) . for each curve , the quark condensate has a plateau , in which the quark condensate changes slowly with @xmath126 , and the width of the plateau becomes more and more narrow with increasing @xmath122 . for each curve , we define the plateau in the range of @xmath135 gev . + 2 ) . for each @xmath122 , the plateau of quark condensate is in the experimental range 0.2 gev @xmath136 0.3 gev , and the plateau becomes higher and higher with increasing @xmath122 , which shows that the meson contributions affect the quark condensate explicitly . + 3 ) . the solid and dashed lines of @xmath21 are almost coincide , which means that the effect of current quark mass is invisible in the mean - field approximation . while beyong mean - field approximation , i.e. , considering the meson contributions , the two curves in the case of @xmath23 or @xmath137 are separate in the region of the plateau . in table 1 , corresponding to the plateau region @xmath138 , ( 460,240)(0,0 ) ( 0,0)(460,240 ) ( 0,60)(460,60 ) ( 0,120)(460,120 ) ( 0,180)(460,180 ) ( 40,0)(40,180 ) ( 40,30)(460,30 ) ( 40,90)(460,90 ) ( 40,150)(80,150 ) ( 240,150)(320,150 ) ( 80,0)(80,240 ) ( 20,30)[]z=1.5 ( 60,15)[]@xmath125 ( 60,45)[]@xmath128 ( 20,90)[]z=1 ( 60,75)[]@xmath125 ( 60,105)[]@xmath128 ( 20,150)[]z=0 ( 60,135)[]@xmath125 ( 60,165)[]@xmath128 ( 160,0)(160,240 ) ( 240,0)(240,240 ) ( 320,0)(320,240 ) ( 400,0)(400,240 ) ( 120,220)[]-@xmath139 ( 120,150)[]0.2096 + 0.0015 ( 120,105)[]0.2397 + 0.0015 ( 120,75)[]0.2360 + 0.0015 ( 120,45)[]0.2599 + 0.0015 ( 120,15)[]0.2566 + 0.0015 ( 120,200)[](gev ) ( 200,210)[]m(gev ) ( 200,150)[]0.47@xmath1400.1 ( 200,105)[]0.39@xmath1400.05 ( 200,75)[]0.40@xmath1400.05 ( 200,45)[]0.38@xmath1400.04 ( 200,15)[]0.39@xmath1400.04 ( 280,210)[]@xmath5(mev ) ( 280,135)[]8.9@xmath1400.30 ( 280,75)[]7.78@xmath1400.20 ( 280,15)[]7.29@xmath1400.20 ( 360,210)[]@xmath9(gev ) ( 360,150)[]0.615@xmath1410.002 ( 360,75)[]0.71@xmath1410.01 ( 360,105)[]0.725@xmath1410.01 ( 360,15)[]0.785@xmath1410.008 ( 360,45)[]0.801@xmath1410.008 ( 430,210)[]g@xmath142 ( 430,150)[]4.8@xmath1400.9 ( 430,75)[]4.7@xmath1400.5 ( 430,105)[]4.8@xmath1400.5 ( 430,15)[]5.2@xmath1400.5 ( 430,45)[]5.3@xmath1400.5 table . quantities in the region of defined plateaus . we list the range of consituent quark mass @xmath126 , current quark mass @xmath5 , quark momentum cut - off @xmath9 and a dimensionless quantity @xmath143.this table shows that , in the mean - field approximation , the values of constituent quark mass @xmath126 and the current mass @xmath5 in the plateau is a little higher than the empirical values @xmath144 and @xmath145mev , and the quark condensate in the plateau is much lower than 0.25 gev ; however , these quantities in the plateaus at @xmath146 are more reasonable comparing with the empirical values . in fig . [ mlf_fig ] , we show the quark momentum cut - off @xmath9 as a function of the constituent quark mass @xmath126 , and in fig . [ mm0_fig ] we show the current quark mass @xmath5 as a function of the constituent quark mass @xmath126 for different values of @xmath122 . there is also a plateau for each curve . although the ranges of constituent quark mass @xmath126 corresponding to the @xmath9 plateau , @xmath5 plateau and quark condensate plateau are not exactly the same , they are mostly around @xmath147 . this range @xmath148 is determined dominantly by the gap equation in the mean - field approximation , the effects of meson cloud contributions are also shown clearly . comparing the plateaus at @xmath123 and @xmath149 , we see that @xmath9 and @xmath134 are all modified by the order of about @xmath150 . but the corresponding plateaus at @xmath151 modified too much . in fact , all these figures show that the plateaus of @xmath151 are abnormal comparing that of @xmath152 . based on the above analysis , it seems that @xmath151 is not a reasonable choice , and @xmath153 is more reasonable . in the region @xmath154 gev , all these three figures figs . [ mcond_fig]-[mm0_fig ] show that : + 1 ) in the mean - field approximation , the solid line and dashed line of @xmath123 coincide , which means that the effects of the current quark mass are invisible in this case . + 2 ) when beyond mean - field approximation @xmath155 , i.e , considering meson contributions , the effects of current quark mass are explicitly reflected by the separation of the solid and dashed lines . + as we pointed out in introduction , in njl model all information is calculated from constituent quark propagator , and the current quark mass @xmath5 induces the massive pion @xmath6 . in the mean - field approximation , there is no feedback of meson modes to quark propagator , so the effects of @xmath5 or @xmath6 can not be reflected . only beyond mean - field approximation , when the back interaction of meson mode to quark propagator are considered , can the effects of @xmath5 or @xmath6 be reflected on the quantities calculated from the fundamental element , quark propagator . we establish a general scheme of su(2 ) njl model including the current quark mass explicitly and considering the meson cloud contributions . to ensure the quark self - energy expanded in the proper order of @xmath26 , we discuss in details that the internal meson propagator must be in @xmath1 order . and we also develop a unified way to calculate all feynman diagrams in the njl model by only expanding the constituent quark propagator in external momentum @xmath60 . our numerical results show that : a ) when @xmath153 , the njl parameters are more reasonable than that in the mean - field approximation ; b ) the effect of current quark mass can be reflected only beyond mean - field approximation , and it is characterized explicitly by the separation of the solid and dashed lines in the range of the parameters corresponding to @xmath126 between @xmath156 . the fundamental functions are defined as following : @xmath157 @xmath158 @xmath159 @xmath160 @xmath161 and their corresponding pauli - villars expressions are : @xmath162 @xmath163 @xmath164 @xmath165 @xmath166 where @xmath167 $ ] , @xmath168 $ ] , @xmath169 for a=0,1,2 , and @xmath170 .

based on the self - consistent scheme beyond mean - field approximation in the large @xmath0 expansion , including current quark mass explicitly , a general scheme of su(2 ) njl model is developed . to ensure the quark self - energy expanded in the proper order of @xmath0 , an approximate internal meson propagator is deduced , which is in order of @xmath1 . in our scheme , adopting the method of external momentum expansion , all the feynman diagrams are calculated in a unified way by only expanding the quark propagator . our numerical results show , that different from the mean field approximation in which the explicitly chiral symmetry breaking is invisible , the effect of finite pion mass can be seen clearly when beyond mean - field approximation .

the process of the energy interchange between radiation and matter provided by compton scattering is relevant in many areas of physics . for example , in cosmology it keeps the matter at the same temperature as radiation @xcite . compton scattering is also a unique spectroscopy for condensed matter physics , which has acquired greater importance with the advent of modern synchrotron sources @xcite . for instance , it has been used to extract information about wave functions of valence electrons in a variety of systems ranging from ice @xcite and water @xcite to alloys @xcite and correlated electron systems @xcite . moreover , compton scattering can potentially help delineate confinements @xcite and spin polarization effects @xcite in nanoparticles . the compton scattering cross section strength is determined by the classical electron radius , also known as the thomson scattering length , @xmath2 where @xmath3 is the electron charge , @xmath4 is the electron mass , @xmath5 is the speed of light and @xmath6 is the dielectric constant . unfortunately , the small size of @xmath7 makes compton experiments in condensed matter systems difficult . this is why only a few experiments have been done , even with the best synchrotron sources . the classical proton radius is even smaller by a factor of @xmath8 , where @xmath9 is the proton mass . therefore , nuclei are practically invisible in x - ray compton scattering experiments . in 1952 , max born suggested that the electronic radius @xmath7 is connected to an absolute length scale @xmath1 @xcite . thus , if the electromagnetic interaction strength is modified , @xmath1 must change as well . understanding this variation could enable us to enhance the compton scattering cross sections by _ engineering _ an effective quantum electro - dynamics ( qed ) interaction . the effective coupling constant @xmath10 can be modified through the dielectric response @xmath6 , for instance , if the incident photon energy is tuned near to the binding energy of a deep core electron level in certain materials . this work shows that the compton cross section can depend strongly on the effective coupling constant @xmath0 and that a reduction of @xmath0 as small as @xmath11 may lead to an increase in the cross section by a factor of @xmath12 . moreover , the present results connect @xmath7 to a fundamental length @xmath1 and thus are consistent with the old hypothesis by born . the triple - differential scattering cross section for the process shown in fig . [ fig1 ] , which is the elementary step underlying compton scattering , is given by @xcite @xmath13 where @xmath14 is the scattering angle , @xmath15 is the fourier transform of the occupied dyson orbital @xmath16 with binding energy @xmath17 , @xmath18 is the momentum transferred to the final system , and @xmath19 and @xmath20 are , respectively , the energies of the photon before and after the collision . the ejected electron state is usually approximated by a plane wave with momentum @xmath21 and energy @xmath22 if @xmath23 . in this regime , compton scattering is a unique window on the electronic structure of matter because in contrast with most structural analysis techniques which can only deliver information on the total electron densities , this spectroscopy allows direct measurements in momentum space of the electron density associated with a single ionization channel ( i.e. a dyson orbital in a one - electron picture ) . in the low - energy limit ( i.e. @xmath24 ) , thirring @xcite has shown that the compton scattering cross section with all radiative corrections reduces in the non - relativistic expression given by eq . ( [ eqcross ] ) . the only effect of the vacuum or the medium is to renormalize the thomson scattering length @xmath7 . the _ thirring theorem _ is a consequence of lorentz and gauge invariance @xcite . we now turn to a general thermodynamic argument in order to derive how the electron volume @xmath25 depends on the effective coupling constant @xmath0 . since the classical electron radius @xmath7 is the length at which qed renormalization effects become important , our argument must be consistent with differential equations of the renormalization group @xcite . thermodynamics is widely considered as a fundamental theory , since it has a universal applicability @xcite . indeed it does not need any modification due to either relativity or quantum theory @xcite . the first law of thermodynamics gives the variation of internal energy @xmath26 where @xmath27 is the temperature , @xmath28 is the entropy and @xmath29 is a pressure imposed by a fictitious piston on the volume @xmath30 in order to set the units scale for a given @xmath0 @xcite . thus , the energy scale is characterized by @xmath31 , where @xmath32 represents a positive integer exponent to be determined . the negative sign of the pressure @xmath33 is explained by the fact that the electromagnetic vacuum fluctuation ( i.e. , the casimir effect ) tries to pull the piston back into the system . similar inward pressures are produced by cosmological constants @xcite . the third term in eq . ( [ eqde ] ) is similar to a chemical potential term , since the number of virtual photons is proportional to the effective coupling constant @xmath0 . thus , we are assuming that the electron mass @xmath4 determines the chemical potential of the virtual photons and that it is generated by the coulomb field of the electron . in adiabatic conditions , the term @xmath34 vanishes . moreover , at equilibrium , @xmath35 , thus the renormalization group @xmath36 function @xcite deduced from eq . ( [ eqde ] ) is given by @xmath37 the solutions for @xmath38 show that the electron localizes ( i.e. , @xmath39 becomes small ) when the interaction strength increases . when @xmath40 , the radius scales as @xmath41 and has a maximal finite size @xmath42 corresponding at @xmath43 while for @xmath44 , the scaling is @xmath45 where @xmath46 is radius corresponding at @xmath47 . the exponent @xmath48 is consistent with the qed @xmath36 function @xcite . the born hypothesis is also verified when @xmath48 , since the corresponding solution admits a minimal length @xmath1 different from zero . in this case , the thomson scattering length depends on @xmath49 by an exponential function @xmath50 where @xmath1 is a certain small length to be determined . moreover , the corresponding pressure @xmath51 sets the atomic energy units . in fact , the atomic units are as natural as the fundamental planck units @xcite : their ratios to the fundamental units can be explained within our present argument connecting the thomson scattering length to the fundamental scale . interestingly , the volume renormalization factor @xmath52 is @xmath53 for @xmath48 . this term is similar to the boltzmann distribution in statistical mechanics ( where @xmath0 plays the role of an effective temperature ) . as a function of the inverse of the effective coupling constant @xmath0 . both @xmath54 and @xmath0 are pure numbers ( units are dimensionless ) . ] the cross section enhancement defined by @xmath55^{2/3}\ ] ] is shown in fig . [ fig2 ] for the case @xmath48 : a reduction of @xmath0 by few percents induces a huge increase in @xmath54 . therefore , cross section enhancements obtained by tuning the incident photon energy near the binding energy of a deep core electron level can be described by the behavior for @xmath48 while the cases @xmath56 give negligible cross section enhancements for small variations of @xmath0 . the trend of @xmath54 illustrated in fig . [ fig2 ] can be produced by a change @xmath57 of the dielectric response near an absorption edge . standard inelastic x - ray scattering experiments without the measurement of the kinematics of the outgoing ( recoil ) electron contain many other processes in addition to the elementary scattering event of fig . [ fig1 ] . therefore , coincidence @xmath58 experiments @xcite are needed in order to separate the x - ray compton scattering with nearly free electrons from complicated processes . some @xmath58 spectrometers are already available for hard x - rays @xcite . unfortunately , standard @xmath58 experiments can be tremendously challenging . instead , one could use a soft - x - ray fluorescence spectrometer by carlisle _ _ @xcite . by tuning the incident photon at the @xmath59 edge of graphite , enhancement effects of the total cross section have been already observed . a coincidence measurement detecting the electrons escaping from the sample can then be used to separate compton from other types of inelastic scattering . in this much simpler setup multiple scattering of the electrons in the sample are not an impediment to extracting the compton contribution . realistic dielectric data for graphite provided by draine @xcite illustrate how tuning the incident photon energy near the binding energy of the @xmath59 core level changes the dielectric response and thus the effective coupling constant for the valence electrons . when @xmath48 , a compton cross - section enhancement @xmath54 of almost a factor @xmath12 is predicted in graphite by using draine s dielectric data as shown in fig . [ fig3 ] . ( ratio of the real part of the dielectric function @xmath6 and the vacuum permittivity @xmath60 ) near the @xmath59 edge of graphite . @xmath61 is the incident photon energy . the points are from draine @xcite . symbol size is representative of error bars . bottom : corresponding cross section enhancement @xmath54 . ] we note that a similar variation of the dielectric function for diamond has been previously reported by nithianandam and rife @xcite . besides , a calculation based on the finite difference method for near - edge structure ( fdmnes ) @xcite agrees with the dielectric data of draine . fdmnes shows that the anomalous scattering factor near the @xmath59 edge of graphite becomes greater in amplitude than the number of electrons , causing the real part of @xmath62 to be greater than unity ( @xmath60 is the vacuum permittivity ) . next , we justify a value for @xmath1 . according to veneziano @xcite , a consistent quantum gravitational theory should obey the born principle of reciprocity @xcite , a symmetry law under the interchange of space - time coordinates and the energy - momentum coordinates , which naturally leads to harmonic oscillators and to the normal modes of vibrating strings . in such theory it is natural to take as the action quantum the square of the planck length @xcite @xmath63 where @xmath64 is the planck constant and @xmath65 is the gravitational constant . indeed , by using eq . ( [ eqr0 ] ) with the planck length @xmath1 and @xmath66 , the calculated thomson scattering length is @xmath67 , which differs about @xmath68 from its exact value . minor renormalization effects of the gravitational constant @xmath65 could improve the agreement @xcite . finally , we could also reverse the logic . in our treatment the length @xmath1 is not fixed _ a priori_. therefore , we can use data from x - rays experiments in graphite to get information about the size of @xmath1 . this would strongly vivify a big portion of the existing literature in quantum gravity for which the presence of an effective minimal length is assumed to describe the discretization of a quantum space - time . presently , in the absence of any experimental signature for quantum gravity , such a minimal length is generically set between the electroweak scale of @xmath69 cm and the planck length . as a result we are opening the door to the possibility of determining an extreme energy effect with sophisticated low energy experiments . in addition since preliminary data seem to support the idea that @xmath70 up to 1 @xmath71 , we can get more stringent constraints about the extension of the conjectured additional spatial dimensions with respect to what we currently know from the observed short scale deviations of newton s law @xcite . in conclusion , we suggest that the low energy compton cross section for the valence ( i.e. , nearly free ) electrons of graphite can be described within the framework of the thirring theorem , implying that the only effect of the medium is to renormalize the thompson scattering length @xmath7 . besides , a general thermodynamic argument shows that the compton scattering cross section grows exponentially if the effective coupling constant @xmath0 decreases . in particular , a striking enhancement is predicted when the incident photon energy is tuned near the binding energy of the @xmath59 core level of graphite . the present enhancement effect is also consistent with the qed renormalization group . we are grateful to a. widom , p.m. platzman , g. barbiellini , b.t . draine , u. amaldi , b. tiburzi , p. nath and d. wood for useful discussions . is supported by the us department of energy , office of science , basic energy sciences contracts de - fg02 - 07er46352 and de - fg02 - 08er46540 ( cmsn ) and benefited from the allocation of computer time at nersc and northeastern university advanced scientific computation center ( nu - ascc ) . is supported by the helmholtz international center for fair within the framework of the loewe program ( landesoffensive zur entwicklung wissenschaftlich - konomischer exzellenz ) launched by the state of hesse . longair , in _ electron : a centenary volume _ , ed . m. springford , cambridge university press ( cambridge 1997 ) . e. d. isaacs and p.m. platzman , phys . today * 49 * ( 2 ) , 40 ( 1995 ) . m. cooper , p. mijnarends , n. shiotani , n. sakai , and a. bansil , _ x - ray compton scattering _ , oxford university press , ( oxford 2004 ) . w. schuelke , _ electron dynamics by inelastic x - ray scattering _ , oxford university press , ( oxford 2007 ) . isaacs , a. shukla , p.m. platzman , d.r . hamann , b. barbiellini , and c.a . tulk , phys . lett . * 82 * , 600 ( 1999 ) . k. nygard , m. hakala , s. manninen , m. itou , y. sakurai , and k. hamalainen , phys . 99 , 197401 ( 2007 ) . patrick h .- sit , c. bellin , b. barbiellini , d. testemale , j. l. hazemann , t. buslaps , n. marzari , and a. shukla , phys . b 76 , 245413 ( 2007 ) . j. kwiatkowska , b. barbiellini , s. kaprzyk , a. bansil , h. kawata and n. shiotani , phys . 96 * , 186403 ( 2006 ) . b. barbiellini , a. koizumi , p. e. mijnarends , w. al - sawai , hsin lin , t. nagao , k. hirota , m. itou , y. sakurai and a. bansil , phys . lett . * 102 * , 206402 ( 2009 ) ; y. sakurai , m. itou , b. barbiellini , p. e. mijnarends , r. s. markiewicz , s. kaprzyk , j .- gillet , s. wakimoto , m. fujita , s. basak , yung jui wang , w. al - sawai , h. lin , a. bansil , and k. yamada , science * 332 * , 698 ( 2011 ) . r. saniz , b. barbiellini and a. denison , phys . b * 65 * , 245310 ( 2002 ) . r. saniz , b. barbiellini , a. b. denison , a. bansil , phys . b * 68 * , 165326 ( 2003 ) . m. born , nature * 169 * , 1105 ( 1952 ) . i. g. kaplan , b. barbiellini , a. bansil , phys . b * 68 * , 235104 ( 2003 ) . b. barbiellini , a. bansil , j. phys . solids * 65 * , 2031 ( 2004 ) . w. thirring , phil . mag . * 41 * , 1193 ( 1950 ) . j. m. jauch and f. rohrlich , _ the theory of photons and electrons _ springer - verlag ( new york , heidelberg , berlin , 1976 ) . van haeringen has noted a problem of the thirring theorem related to the infrared divergence , which leads to a vanishing compton scattering amplitude ; this difficulty can be avoided by restricting the virtual photons to a reasonable frequency domain , see e.g. w. van heringen , physica 26 , 306 ( 1960 ) . arxiv:1001.0785 [ hep - th ] . nicolini , phys . d * 82 * , 044030 ( 2010 ) . for further arguments about the role of thermodynamics in physical theories see t. padmanabhan , _ gravitation : foundations and frontiers _ , cambridge university press ( cambridge 2010 ) , pp 670 - 672 and the references therein . a similar fictitious piston setting the system units is used in constant pressure molecular dynamics simulations , see e.g. h.c . andersen , j. chem . * 72 * , 2384 ( 1980 ) . http://www.astro.ucla.edu/wright/cosmo_constant.html yu . i. manin , _ mathematics and physics _ , birkhuser ( boston , basel , stuttgart 1981 ) . m. itou , s. kishimoto , h. kawata , m. ozaki , h. sakurai and f. itoh , j. phys . . jpn . * 68 * 515 ( 1999 ) . j. a. carlisle , eric l. shirley , e. a. hudson , l. j. terminello , t. a. callcott , j. j. jia , d. l. ederer , r. c. c. perera , and f. j. himpsel , phys . lett . * 74 * , 1234 ( 1995 ) . draine , the astrophysical journal * 598 * , 1026 ( 2003 ) ; http://www.astro.princeton.edu/ draine / dust / dust.diel.html . j. nithianandam and j.c . rife , phys . b * 47 * , 3517 ( 1993 ) . y. joly , phys . b * 63 * , 125120 ( 2001 ) . g. veneziano , europhys . * 2 * , 199 ( 1986 ) . m. born , rev . * 21 * , 463 ( 1949 ) . in ref . @xcite the planck length is smaller and it is given by @xmath72 . hamber and r.m . williams , phys . d * 75 * , 084014 ( 2007 ) . v. b. bezerra , g. l. klimchitskaya , v. m. mostepanenko and c. romero , phys . d * 81 * , 055003 ( 2010 ) .

a robust thermodynamic argument shows that a small reduction of the effective coupling constant @xmath0 of qed greatly enhances the low energy compton scattering cross section and that the thomson scattering length is connected to a fundamental scale @xmath1 . a discussion provides a possible quantum interpretation of this enormous sensitivity to changes in the effective coupling constant @xmath0 .

electron - electron interaction is responsible for very many phenomena studied by the different fields of physics ranging from astrophysics to biophysics . amongst them atomic physics and its part physics of ion - atom collisions often deal with most basic and clear manifestations of this interaction . atomic excitation and ionization @xcite-@xcite , projectile - electron excitation and loss @xcite-@xcite , @xcite , electron transfer ( capture ) @xcite , @xcite and pair production @xcite , @xcite-@xcite belong to the elementary reactions occurring when a projectile - ion collides with a target - atom . a combination of these reactions in a single - collision event is also possible and in such a case the electron - electron interaction during the collision when the external field is rapidly changing ( dynamic electron correlations ) is often crucial . in particular , mutual ionization in which both a target - atom and a ( partially stripped ) projectile - ion eject electrons , and transfer - ionization in which one of the atomic electrons is captured by a projectile - ion whereas another one is emitted , represent processes where dynamic electron correlations play a crucial role @xcite-@xcite , @xcite , @xcite , @xcite-@xcite . transfer - ionization in fast collisions of low charged ions ( mainly protons ) with helium is attracting much attention @xcite-@xcite . this process can be analyzed in terms of different reaction mechanisms which are characterized by distinct features in the electron emission pattern . depending on whether the electron - electron interaction plays in them a crucial role , these mechanisms can be termed `` correlated '' or `` uncorrelated '' . uncorrelated mechanisms are independent transfer - ionization ( iti ) and capture shake - off ( c - so ) . in the iti electron capture and emission occur due to `` independent '' interactions of the projectile with two target electrons . according to the c - so , a fast transfer of one electron from the atom to the ion leads to a `` sudden '' change of the atomic potential for another electron that leads to its emission . both iti and c - so result in emission of low - energy electrons . the correlated mechanisms include electron - electron thomas ( eet ) and electron - electron auger ( eea ) . within the eet transfer - ionization proceeds @xcite , @xcite-@xcite via a binary collision of the projectile with one of atomic electrons and a consequent rescattering of this electron on another atomic electron . after these two collisions one of the electrons moves together with the projectile ( that makes capture probable ) while the other is emitted perpendicular to the projectile motion . the functioning of the eea mechanism is based on the fact that merely the presence of the projectile makes the target unstable with respect to a kind of auger decay . indeed , viewing the collision in the rest frame of the projectile we can see that one of the electrons , belonging initially to a bound configuration of moving particles constituting the atom , can make a transition into a bound state of the ion by transferring the energy excess to another atomic electron which , as a result of this , is emitted from the atom in the direction of the atomic motion @xcite , @xcite . in the rest frame of the atom this electron finally moves in the direction opposite to the projectile velocity @xcite , @xcite , @xcite . the mechanisms , briefly discussed above , were proposed for describing transfer - ionization in collisions between a light atom and a low charged ion moving with a velocity @xmath0 , which is much higher than the typical orbiting velocities of the electron(s ) in their initial and final bound states : @xmath1 and @xmath2 , where @xmath3 and @xmath4 are the charges of the nuclei of the atom and ion , respectively . what , however , can one say about transfer - ionization in fast collisions with highly charged ions ( hcis ) when the charge @xmath5 of the ion is so large that @xmath6 ? one can expect that in such collisions , which are characterized by very strong fields generated by the hci , not only cross sections for transfer - ionization would be much larger than in collisions with equivelocity low charged ions but also new interesting features could arise in this process . therefore , in this article we explore transfer - ionization in fast collisions with hcis . it will be seen below that a strong field of hci has a dramatic effect on the correlated channels of transfer ionization : it weakens the eea , eliminates the eet and leads to qualitatively new structures in the emission spectrum . atomic units ( @xmath7 ) are used throughout except where the otherwise stated . we are mainly interested in the correlated transfer - ionization and begin with its treatment . this treatment will be semiclassical in which only the electrons are described quantum mechanically whereas the heavy particles ( the nuclei of the ion and atom ) are considered classically . in fast collisions the trajectories of the heavy particles are practically straight - line . it is convenient to make the basic consideration of the correlated transfer - ionization using the rest frame of the ion and to take its position as the origin . according to scattering theory the exact ( semiclassical ) transition amplitude can be written as @xmath8 here @xmath9 is an exact solution of the time - dependent schrdinger equation with the full hamiltonian @xmath10 which describes two electrons moving in the external field of the nuclei and interacting with each other , @xmath11 denotes the final state of the two electrons and @xmath12 is that part of @xmath10 which is not included into the wave equation for @xmath13 . since the contribution to transfer - ionization from collisions , in which electrons change spin , is negligible we shall disregard the spin parts of the states @xmath14 and @xmath13 . in the correlated transfer - ionization the velocities of the electrons with respect to the nucleus of the atom in the final state are of the order of @xmath0 @xcite . besides , in this state the relative velocity of the electrons is also of the same order . therefore , when this process occurs in fast collisions with hcis , for which one has @xmath15 but @xmath16 , the motion of both electrons in the final state @xmath13 is driven by the field of the hci whereas the interactions of the electrons with the atomic nucleus and with each other can be neglected . thus , we have @xmath17 where @xmath18 and @xmath19 are the coordinates of the electrons , @xmath20 is the bound state of an electron captured by the hci with an energy @xmath21 and @xmath22 is the state of emitted electron which moves in the hci s field and has asymptotically a momentum @xmath23 . when the state @xmath13 is taken in the form ( [ e2 ] ) the perturbation @xmath12 in eq.([e1 ] ) is equal to @xmath24 , where @xmath25 ( @xmath26 ) is the interaction between the @xmath27-th electron and the nucleus of the atom and @xmath28 is the electron - electron interaction . then we obtain @xmath29 it is the last term on the right - hand side of eq.([e3 ] ) which is relevant for the correlated transfer - ionization . in order to find a suitable approximation for @xmath30 let us note the following . in the process of correlated transfer - ionization occurring in very fast collisions ( @xmath31 ) both electrons undergo transitions in which the change in their momenta is much larger than their typical momenta in the initial atomic state . because of that , even if the projectile would have a low charge , the nucleus of the atom would be merely a spectator during this process @xcite . under such circumstances the so called impulse approximation , in which the role of the atomic nucleus is just to produce the momentum distribution ( and binding energy ) of the electrons in the initial state , can be used to treat transfer - ionization . in our case , in which transfer - ionization occurs in very asymmetric collisions where the charge of the hci is much higher than the charge of the atomic nucleus , the impulse approximation is even more appropriate because it enables one to fully account for the influence of the strong field of the hci on _ two electrons_. the initial ( undistorted ) atomic two - electron state is given by @xmath32 where @xmath33 is a straight - line classical trajectory of the atomic nucleus moving with a velocity @xmath34 and @xmath35 is the initial atomic state with an energy @xmath36 ( as viewed in the rest frame of the atom ) . now we perform the three - dimensional fourier transformation of the state @xmath37 , insert the obtained result into eq.([e4 ] ) and replace the plane - wave factors in the integrand by the corresponding coulomb waves in the field of the hci . in such a way we approximate the state @xmath30 in eq.([e3 ] ) by @xmath38 here @xmath39 is the fourier transform of the state @xmath35 , @xmath40 and @xmath41 are the initial momenta of the electrons with respect to the hci and @xmath42 ( @xmath26 ) are coulomb wave functions of electrons which move in the field of the hci being incident on it with asymptotic momenta @xmath43 . using eqs.([e2])-([e3 ] ) and ( [ e5 ] ) and assuming that the space part of the initial atomic state is symmetric with respect to the interchange of the electrons ( like it is the case for the ground state of helium ) one can show , after some algebra , that in the projectile frame the cross section for transfer - ionization differential in the momentum @xmath23 of the emitted electron is given by @xmath44 in this expression @xmath45 is the momentum transfer in the collision with @xmath46 being its transferse part ( @xmath47 ) and @xmath48 where @xmath49 and @xmath50 . expression ( [ e6 ] ) can be drastically simplified by using the fact that the fourier transform @xmath51 becomes very small when the absolute values of @xmath52 substantially exceed the typical electron velocities inside the atom . since we consider collision velocities , which are much higher than the latter ones , we may set in eq.([e8 ] ) @xmath53 and take @xmath54 in eq.([e6 ] ) out of the integrals . let us now say few words about the treatment of the uncorrelated channels of transfer - ionization . following @xcite , @xcite and @xcite we shall present the amplitude for this process as the product of single - electron transition amplitudes for capture and ionization which are obtained using three - body models : continuum - distorted - wave ( for capture ) and continuum - distorted - wave - eikonal - initial - state ( for ionization ) @xcite . because of a high charge of the projectile the c - so , compared to the iti , contributes negligibly . since the emission produced by these channels is localized in the same part of the momentum space the c - so can safely be neglected . in figures [ figure1]-[figure2 ] we present results @xcite for the momentum spectrum of electrons emitted in transfer - ionization in collisions of @xmath55 mev / u ca@xmath56 projectiles ( @xmath57 a.u . ) with helium atoms when capture occurs into the @xmath58 and @xmath59 shells . the spectrum is given in the rest frame of the target ( @xmath60 the laboratory frame ) in which the projectile moves with a velocity @xmath61 and is represented by the doubly differential cross section @xmath62 where @xmath63 and @xmath64 are the longitudinal and transverse parts , respectively , of the momentum @xmath65 of the emitted electron in the laboratory frame and @xmath66 . the integration in ( [ e9 ] ) runs over the transverse part of the momentum transfer and the azimuthal angle @xmath67 of the emitted electron . in figure [ figure1 ] the maximum at small momenta has its origin in the iti whereas the maxima at much larger @xmath68 appear due to the correlated channels . the maximum at small momenta yields a very important contribution to the total cross section . however , its structure is similar to that in fast collisions with low charged projectiles , studied in @xcite , and below will not be considered . the maxima at large @xmath68 deserve more attention . in order to see better their structure , in figure [ figure2 ] only the high - momentum part of the emission , which is produced solely via the correlated channels , is shown . in the rest frame of the ion the approximate energy balance for transfer - ionization is very simple : @xmath69 , where @xmath70 with @xmath71 being the principal quantum number of the bound state of the captured electron . since @xmath72 and @xmath73 , where @xmath74 and @xmath75 are respectively the transverse and longitudinal parts of @xmath23 , in the target frame the emission is concentrated on ridges located along rings with radii @xmath76 centered at ( @xmath77 , @xmath78 ) . the ridge with higher energy appears due to transfer - ionization with electron capture into the ground state of the projectile while the other ridge originates from capture into the projectile s @xmath59-shell . it is seen in the figure that each of these ridges has two distinct maxima : one centered at an emission angle of @xmath79 and the other at @xmath80 ( the angle @xmath81 is counted from the direction of the projectile motion ) . the shape of the correlated part of the spectrum in case of collisions with hcis is to be compared with that in collisions with low charged ions . the latter is displayed in figure [ figure3 ] for @xmath55 mev p@xmath82 + he(1s@xmath83 ) collisions . since at @xmath84 transfer - ionization is strongly dominated by capture into the ground state this spectrum @xcite is concentrated on a single ridge . compared to the corresponding ridge in [ figure2 ] the ridge in figure [ figure3 ] is shifted to lower @xmath68 because of the much smaller binding energy of the captured electron . it consists of two distinct parts : the maximum at @xmath79 is caused by the eea whereas the maximum at @xmath85 is the signature of the eet . comparing the spectra in figures [ figure2 ] and [ figure3 ] , one can attribute the maxima at @xmath79 in figure [ figure2 ] as arising due to the eea . however , in the strong - field regime there is no maximum at @xmath85 which is characteristic of the eet channel in collisions with fast low - charged ions . instead , a new maximum appears on each ridge at @xmath80 which is absent when the projectile has a low charge . this maximum is rather intense and its contribution to the total cross section is comparable to ( or even exceeds ) that of the eea maximum . in order to get more ideas about the correlated transfer - ionization in the strong - field regime , in figure [ figure4 ] we present the emission spectrum at even a stronger field when @xmath55 mev / u zn@xmath86 projectiles collide with helium . as before , we consider capture into the @xmath58 and @xmath59 shells only . from figure [ figure4]a it can be inferred that the structure of the spectrum is similar to that in collisions with ca@xmath56 ions : it has two ridges and each of them has two pronounced maxima centered at @xmath80 and @xmath87 . however , the relative intensity of these maxima is now different with the maxima at @xmath80 becoming noticeably more populated compared to those at @xmath88 . besides , the relative importance of the capture into the ground state descreases ( as expected ) . figure [ figure4 ] shows also three more results . the spectrum presented in [ figure4]b was obtained by approximating all the states @xmath89 , @xmath90 and @xmath91 by plane waves , i.e. assuming that both electrons in the initial channel as well as the emitted electron do not feel the hci s field . figure [ figure4]c displays results calculated when the states @xmath90 and @xmath91 are modelled by plane waves whereas the state @xmath89 is the coulomb one . thus , in this case we neglect the action of the hci s field only on that electron , which is not captured . finally , figure [ figure4]d shows the spectrum obtained if the state @xmath89 is taken as a plane wave but the states @xmath90 and @xmath91 are coulomb waves . when the action of the hci s field is neglected for both electrons ( of course , except in the final bound state @xmath20 ) the calculated spectrum has very pronounced maxima at @xmath88 and their intensity rapidly decrease when the transverse component @xmath92 of the electron momentum increases ( see figure [ figure4]b ) . if the field of the hci is neglected only for that electron , which is finally emitted , the maxima at @xmath88 become less pronounced but new maxima do not yet appear although the calculated spectrum has more extension in the direction of larger @xmath92 ( see figure [ figure4]c ) . if we neglect the action of the hci s field on that electron which is finally captured but take into account hci s action on the other electron , the calculated spectrum extends even more in the transverse direction but new maxima are still absent ( see figure [ figure4]d ) . and only when the action of the hci s field on the electrons is fully included , do new maxima appear at @xmath93 ( figure [ figure4]a ) . in this case the maxima at @xmath88 further loose in intensity and the ( relative ) extension of the spectrum in the direction of large @xmath92 is most pronounced . it follows from these results that the action of the hci s field on _ both _ electrons is necessary for the appearance of the second maxima in the emission spectrum . therefore , the physical origin of these maxima is qualitatively different from that of the eet mechanism , which proceeds via the interaction in the initial channel between the ion and only _ one _ of the atomic electrons . however , we could not find a simple physical picture for this strong - field mechanism . the maxima at @xmath79 for their appearance need , in principle , neither the interaction with the hci in the initial channel nor the interaction between the hci and the emitted electron . according to the results presented in figure [ figure4 ] the distortion of the initial two - electron state by the field of the hci and the action of this field on the emitted electron just diminish the eea mechanism . additional information about the transfer - ionization can be obtained by calculating cross sections for this process using the rest frame of the ion . let us briefly discuss the dependence of the transfer - ionization in this frame on the emission angle . this can be done by studying the cross section @xmath94 , where @xmath95 and @xmath96 are the electron emission energy and polar angle in this frame . in the rest frame of the ion the eea and eet mechanisms result in the maxima located at @xmath97 and @xmath98 , respectively ( see figure [ figure5 ] , @xcite ) , where the angle @xmath99 is counted from the velocity of the atom . in figures [ figure6]-[figure8 ] the energy - angular distribution is shown for the correlated transfer - ionization in collisions of 22.5 mev / u ne@xmath100 , ca@xmath56 and zn@xmath86 with helium resulting in electron capture into the ground state of the ion . in all these figures there is one maximum at @xmath97 corresponding to the eea , no maximum at @xmath98 and the second maximum whose center is located between @xmath101 and @xmath102 . it is seen in the figures that the relative intensity of this maximum ( compared to the eea ) strongly increases when the charge of the ion increases . the center of this maximum slightly moves to lower angles when we go from ne@xmath100 to ca@xmath56 ions but shifts somewhat upwards when ca@xmath56 is replaced by zn@xmath86 . thus , for a fixed impact velocity of @xmath57 a.u . the angular position of this center depends rather weakly ( and nonmonotonously ) on the charge of the ion when the latter varies in the ( quite broad ) range @xmath103 . in our calculations partial - wave expansions for the states @xmath89 , @xmath90 and @xmath91 are used @xcite . for @xmath84 the numbers of the partial waves become very large . besides , the radial integrands these waves are part of become highly oscillating . all this makes it difficult for us to perform the computation at @xmath84 using the coulomb form for all the states @xmath89 , @xmath90 and @xmath91 . therefore , at the moment we can not address an interesting point : whether with decreasing the ratio @xmath104 the second maximum smoothly goes over into the eet or it decreases in intensity and eventually disappears without moving into the region @xmath98 where the eet independently arises . we have considered transfer - ionization in collisions of helium with fast nuclei having so high charge @xmath5 that @xmath105 . in this consideration we focused on the correlated channels of this process in which the electron - electron interaction during the collision plays a crucial role . our results were obtained using a treatment which enables one to fully account for the action of the strong field generated by a highly charged ion on electrons , both in the initial and final states of the reaction . our consideration shows that the strong field has a very profound effect on the correlated transfer - ionization . compared to the weak - field regime realized in collisions with fast low charged ions the strong field weakens ( in relative terms ) the emission of high - energy electrons into the direction ( exactly ) opposite to the motion of the projectile and strongly suppresses the emission perpendicular to this motion . instead , a very substantial part of emission via the correlated channels goes now into the direction around @xmath80 and its relative importance increases with the strength of the perturbation . this is in contrast to collisions with fast low charged ions where this part of the momentum space is not important at all . the correlated transfer - ionization is intimately related to the process of radiative two - electron transfer in which two atomic electrons are captured by the projectile with emission of a single photon . indeed , the electron ejected in the transfer - ionization can undergo radiative recombination with the projectile leading to the radiative two - electron transfer . since the probability of radiative recombination is very low the latter process has much smaller cross sections than the transfer - ionization that makes it very difficult for experimental observation @xcite . therefore , further theoretical and experimental studies of transfer - ionization @xcite may also shed more light on the correlated two - electron one - photon capture .

transfer - ionization in fast collisions between a bare ion and an atom , in which one of the atomic electrons is captured by the ion whereas another one is emitted , crucially depends on dynamic electron - electron correlations . we show that in collisions with a highly charged ion a strong field of the ion has a very profound effect on the correlated channels of transfer - ionization . in particular , this field weakens electron emission into the direction opposite to the motion of the ion and strongly suppresses the emission perpendicular to this motion . instead , electron emission is redirected into those parts of the momentum space which are very weakly populated in fast collisions with low charged ions .

the problem of diffusion in a given velocity field has both theoretical and practical relevance in many different fields of science and engineering as , e.g. , transport processes in chemical engineering and combustion studies @xcite . the tracers transport , in particular the evolution of their concentration , plays an important role in many aspects of geophysics . for the oceanic flows , satellite data indicate that the mesoscale features , like eddies and cold filaments , advect temperature and nutrients over spatial and temporal scales longer than those of the geostrophic turbulence . the diffusion enhancement by a given velocity field has attracted a lot of works in the last years . in particular the role of the velocity field properties has been largely investigated while the effects of small scales parameterization are not understood . in this paper we will focus on the effects of a finite noise correlation time . this problem is relevant in studying the transport in the ocean since in this system the noise term comes from unresolved velocity scales which are correlated in time . in section 2 , by using the multiscale technique , we study the diffusion properties of the model proposed in ref . @xcite for transport in the upper mesoscale ocean . the transport is described by a langevin equation with a gaussian colored noise in time . the aim is to understand whether a finite noise correlation time @xmath2 enhances or depresses the dispersion process in a given velocity field @xmath6 with respect to the delta - correlated case ( @xmath7 ) . exploiting the scale separation in the dynamics we derive , using the multiscale technique @xcite , an effective diffusive equation for the macrodynamics , the calculation of the effective diffusivity second - order tensor is reduced to the solution of one auxiliary partial differential equation @xcite,@xcite,@xcite . in section 3 we consider a shear flow , in this case the diffusion coefficient increases with @xmath2 . the solution of the auxiliary equation is , in general , quite difficult , therefore , to investigate the role of the finite @xmath2 in section 4 we perform a small@xmath2 expansion . an alternative method is presented in the appendix a. in section 5 we study the case of two closed - streamlines fields that mimics the transport in the rayleigh - bnard system : the quasi - two - dimensional flow studied by shraiman in @xcite and the ab flow . in both the cases the presence of a small correlation time enhances the diffusion process . conclusions are reserved for the final section 6 . we consider large scale and long time dynamics of the model proposed in @xcite and already studied in @xcite for the transport of a fluid particle in the upper mesoscale ocean : @xmath8 where @xmath3 is a @xmath9-dimensional incompressible velocity field ( @xmath10 ) , for simplicity , periodic both in space and in time and @xmath11 is a gaussian random variable of zero mean and correlation function @xmath12 the term @xmath6 represents the part of the velocity field that one is able to resolve , i.e. , the larger scale mean flow , whereas @xmath13 represents the part of the velocity field containing the subgridscale flow processes , e.g. the small - scale turbulence . the plausibility of such a description is discussed in @xcite,@xcite,@xcite . + in the limit @xmath14 , resulting @xmath15 , the ( [ lan11 ] ) reproduces the widely studied delta - correlated case @xmath16 the diffusive properties of which we would like to compare with the @xmath2-correlated noise case . to study the dispersion of tracers evolving according to eqs . ( [ lan1 ] ) and ( [ lan11 ] ) on large scales and long times we use the multiscale technique . this is a powerful mathematical method , also known as homogenization , for studying transport processes on time and spatial scales much larger than those of the velocity field @xmath3 . it has been already applied to the delta - correlated case @xcite and it has been shown that the motion on large time and spatial scales is diffusive and it is described by an effective diffusion tensor which takes into account the effects of the advecting velocity on the bare diffusion coefficient @xmath17 . to apply this method to the case of gaussian colored noise , we first write eqs . ( [ lan1 ] ) and ( [ lan11 ] ) into a markovian process by enlarging the state space considering @xmath13 as a variable evolving according to the langevin equation : @xmath18 where now the noise @xmath19 is a white noise with correlation functions @xmath20 we have now a two - variable ( @xmath21 ) markovian process whose associated fokker - planck equation can be easily obtained . indeed introducing @xmath22 the equations ( [ lan1 ] ) and ( [ lan2 ] ) become @xmath23 the associated fokker - planck equation is @xmath24 where @xmath25 denotes the probability density . the doubling of the space dimension is the price to pay for having a fokker - planck equation . in the appendix a we discuss a different approach to the problem which does not double the dimension of the space , but leads in general to a non - markovian master equation . we can now apply the multiscale technique . following @xcite in addition to the _ fast _ variables @xmath26 and @xmath27 we introduce the _ slow _ variables defined as @xmath28 and @xmath29 where @xmath30 is the parameter controlling the separation between the small scales related to the velocity field @xmath3 and the large scale related to the @xmath31 variation . the two sets of variables are considered independent and so we have to make the substitution @xmath32 the solution of the fokker - planck equation ( [ fp ] ) is sought as a perturbative series @xmath33 where the functions @xmath34 depend on both fast and slow variables . by inserting ( [ newgrad ] ) and ( [ newteta ] ) into the fokker - planck equation ( [ fp ] ) , equating terms of equal powers in @xmath35 and choosing the solutions which have the same periodicities as the velocity field , we obtain a hierarchy of equations the first three of which are : @xmath36 where the operator @xmath37 is defined as @xmath38 in order to solve eq . ( [ eq0 ] ) we make use of scale separation and we write the solution @xmath39 as the sum of two terms : @xmath40 depending on the _ slow _ variables and @xmath41 depending on the _ fast variables_. here and in the following the @xmath42 indicates the average over the _ fast _ variables . equation ( [ eq0 ] ) then splits into the two equations @xmath43 @xmath44 one can show @xcite that the solution @xmath45 will relax to a constant with respect to fast variables , so we can simply take @xmath46 and write the solution of eq . ( [ eqmed ] ) as @xmath47 where @xmath48 is defined as @xmath49 with @xmath9 the dimension of the @xmath26-space . by using ( [ solmed ] ) we see that eq . ( [ eq1 ] ) can be written as @xmath50 with @xmath51 with solution @xmath52 the vector field @xmath53 is called the auxiliary field and it solves the auxiliary equation @xmath54 finally by averaging eq . ( [ eq2 ] ) over the fast variables , @xmath42 , and integrating over @xmath11 , @xmath55 , we obtain @xmath56 which , using eq . ( [ sol1 ] ) , becomes @xmath57 this is the diffusion equation describing the large scale dynamics , i.e. , the dynamics in the slow variables . the effective eddy diffusivity tensor @xmath58 is given by @xmath59 from the auxiliary equation ( [ eqaux ] ) one can show that the @xmath58 is positive definite . indeed , if we consider the @xmath60-th and the @xmath61-th component of ( [ eqaux ] ) , multiply by @xmath62 and @xmath63 respectively , sum the two terms and average the result over the periodicities , @xmath42 , and integrate over the random variable @xmath11 we end up with @xmath64 this result can be extended to non - periodic velocity field following the prescriptions in @xcite . the resolution of the auxiliary field equation for a generic @xmath3 is not an easy task . therefore not trivial solvable cases are useful to understand the properties of the solution . in particular the auxiliary equation can be resolved for parallel flows , which in two dimensions have the form @xmath65 where @xmath66 is an arbitrary function of @xmath67 . note that these flows automatically satisfy @xmath68 . to evaluate the effective diffusion coefficients we first write the solution of th auxiliary equation as @xmath69 where @xmath70 is the green function of the operator @xmath37 : @xmath71 inserting ( [ chigreen ] ) into eq . ( [ diff ] ) we have : @xmath72 now we note that the green function @xmath70 can be written as @xmath73 where the average is over the realizations of the white noise @xmath74 , @xmath75 and @xmath76 are the solutions of eqs . ( [ lan1 ] ) , ( [ lan2 ] ) and ( [ shear ] ) with initial condition @xmath77 and @xmath78 . for the velocity field ( [ shear ] ) the solutions of the equations ( [ lan1 ] ) and ( [ lan2 ] ) can be written as @xmath79 inserting ( [ green ] ) and ( [ solshe ] ) into ( [ diffgreen ] ) , after some straightforward straightforward algebra one obtains : @xmath80 \right ) \label{diffshe}\ ] ] and @xmath81 where @xmath82 is the fourier transform of @xmath66 . the same result can be obtained directly from the definition @xmath83 using ( [ solshe ] ) . now , because of the inequality @xmath84 \geq 1 \label{disugua}\ ] ] one has @xmath85 therefore for a stationary parallel flow a colored noise produces an enhancement of the dispersion . similar equations can be obtained for a time dependent shear flow . however in this case it is not simple to see the sign of the correction . the results will be reported elsewhere . it is trivial to show that the results in this section hold also for a @xmath86-d shear flow : @xmath87 by using multiscale technique , the calculation of the eddy diffusivities has been reduced to the solution of the auxiliary equation ( [ eqaux ] ) . numerical methods are generally needed to solve it but to do so we have to work in a @xmath88-dimensional space . in general , this is not feasible so , to get more insight the generic @xmath3 case we study the small @xmath2 case and expand the auxiliary field @xmath89 in a power series of @xmath2 . a typical time of the physical system which can be compared to @xmath2 is @xmath90 i.e. the average time it takes a particle to travel a characteristic length @xmath91 . taking @xmath92 we obtain the following expression for the eddy diffusivity tensor @xmath93 \label{difftau}\ ] ] and for the auxiliary equation : @xmath94 { \chi_i}^{(k)}= - \left ( v_i + \frac{\alpha_i}{\sqrt{\tau } } \right ) e^{-\frac{\alpha^2}{2 \sigma^2}}. \label{auxtau}\ ] ] from the expression ( [ difftau ] ) we see that in order to determine the correction of order @xmath2 to @xmath95 , we need the quantities @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 , @xmath101 , @xmath102 . the fields @xmath103 obey to the equations @xmath104 where the operator @xmath105 is defined as @xmath106 the solutions of eqs . ( [ opchi0 ] ) , ( [ opchi1 ] ) and ( [ opchi2 ] ) can be written in the form @xmath107 the coefficients @xmath108 and @xmath109 are functions of @xmath26 and @xmath27 while for the field @xmath110 it results that @xmath111 by inserting expressions ( [ solchi0 ] ) , ( [ solchi1 ] ) and ( [ solchi2 ] ) into eqs . ( [ opchi0 ] ) , ( [ opchi1 ] ) , ( [ opchi2 ] ) and ( [ opchi3 ] ) we can determine the coefficients @xmath108 for @xmath112 and @xmath109 for @xmath113 and then by integrating over @xmath114 the equations for @xmath115 , @xmath116 and @xmath117 respectively we finally have the equations for the remaining functions @xmath118 , @xmath119 and @xmath120 @xmath121 + 2 \ ; \sigma^2 \partial_{j } v_{m } \ ; \partial_{j m}^2 { \widetilde{\chi}_i}^{(0 ) } \label{eqcost2}\end{aligned}\ ] ] where @xmath122 the @xmath58 coefficients at the first order in @xmath2 read @xmath123 . \label{diffprimo}\end{aligned}\ ] ] we note that instead of the @xmath124-dimensional equation ( [ diff ] ) we have now a system of two @xmath9-dimensional equations ( [ eqcost1 ] ) and ( [ eqcost2 ] ) without the random variable @xmath11 . of course this is numerically much more convenient . we note that by defining new velocity and auxiliary fields as @xmath125 and negletting terms of @xmath126 equations ( [ eqcost1 ] ) , ( [ eqcost2 ] ) and ( [ diffprimo ] ) can be written as @xmath127 and @xmath128 formally equivalent to the gaussian white noise result . in the appendix a we use a different method to obtain the expression of the eddy diffusivity tensor for small @xmath2 up to @xmath126 . by starting from the master equation associated with the langevin equation ( [ lan1 ] ) and using a small @xmath2 expansion we derive the fokker - planck equation @xmath129 + \ ; \partial_{ij}^2 \left [ { \cal d}_{ij } ( { { \mbox{\boldmath $ x$}}},t ) \ ; \theta ( { { \mbox{\boldmath $ x$}}},t ) \right ] \label{fpapp}\ ] ] with @xmath130 \right ] . \label{dapp}\ ] ] now , by applying multiscale technique to ( [ fpapp ] ) we end up with the following equations @xmath131 \label{diffapp}\end{aligned}\ ] ] which differ from the previous ones . this result poses the question about the validity of the two sets of equations and hence of the expansions . in order to answer to this question we firstly note that the two sets can be considered as a particular case of the _ generalized _ equations @xmath132 \label{diffgen}\end{aligned}\ ] ] obtained starting from the master equation @xmath133 \label{fpg}\ ] ] where @xmath134 and by applying the multiscale technique . the role of the two free parameters @xmath135 and @xmath136 is of generalizing the master equation ( [ fpapp ] ) being our feeling that there exists a family of different microscopic process specified by @xmath137 and @xmath138 that correspond to the same macroscopic diffusive process specified by @xmath58 . the first set consisting of the equations ( [ eqcost1 ] ) , ( [ eqcost2 ] ) and ( [ diffprimo ] ) corresponds to @xmath139 and @xmath140 , whereas the second one consisting of the equations ( [ chi0app ] ) , ( [ chi1app ] ) and ( [ diffapp ] ) corresponds to @xmath141 and @xmath142 . now a closer analysis of the diffusion coefficients ( [ diffg ] ) reveals that it may take negative values , introducing unphysical singularities into the problem . moreover the multiscale technique requires a @xmath58 definite positive . this means that ( [ fpg ] ) do not constitute genuine markovian process with a well defined corresponding langevin equation driven by a gaussian white noise . in general this expansion do not converge uniformly in @xmath26 and the range of validity is restricted to @xmath143 , @xmath144 ( in dimensionless unit ) , thus they are asymptotic estimate for @xmath145 . in this range of equations validity we do not have any _ a priori _ arguments for choosing one of the two sets of equations and it is for this reason that we expect to obtain the same value of the eddy diffusivity tensor using the two sets . this is indeed the case for the parallel shear flows ( see appendix b ) . this result can be thought as a first indication of the equivalence ( with respect to the value of the diffusivity ) of the _ generalized _ equations . in general we do not expect that all possible choices of @xmath136 and @xmath135 give the same @xmath58 but we can think to select the class of equivalent process by applying the _ generalized _ equations to the parallel shear flow and imposing @xmath58 equal to the known expression ( [ zero ] ) independent on both @xmath136 and @xmath135 . this calculation is reported in appendix b and it ends up in the following condition @xmath146 all these considerations suggest that for short @xmath2 there exists a class of equivalent equations depending on the parameter @xmath135 that lead to the same eddy diffusivity tensor . if this is the case , among the all possible choices of @xmath136 and @xmath135 consistent with ( [ condshe ] ) , we can choose : @xmath147 and @xmath148 . in this case @xmath149 and ( [ fpapp ] ) reduces to a truly markovian process with the associated langevin equation @xmath150 where @xmath151 is a gaussian white noise . this is the microscopic markovian process which approximates the long time and large space transport properties of a colored noise process with short noise correlation time . in other words to study the diffusion properties of ( [ lan1])([lan11 ] ) for small @xmath2 we can replace the original colored noise process with the process described by ( [ lan32 ] ) . this will give the correct diffusion coefficients up to @xmath126 . we have checked numerically that for the ab flow the eddy diffusivity tensor assumes the same value for the three different choices of the parameters @xmath136 and @xmath135 consistent with ( [ condshe ] ) : @xmath140 and @xmath139 , @xmath142 and @xmath141 , @xmath147 and @xmath148 . this make us confident with our conclusions . we apply now our analysis to two models for the rayleigh - bnard steady convection : the first one consists of an horizontal extent of convection cells much larger than its height so that the flow can be considered quasi - two - dimensional ; the second one is the two - dimensional ab flow made of a structure periodically repeated in the space . we consider the flow discussed by shraiman @xcite . this is described by the stream function @xmath152 with @xmath153 being the characteristic velocity , @xmath154 the height of the cell ( @xmath155 $ ] ) and @xmath156 the @xmath157-periodicity of the roll pattern . the top and the bottom plates of the cell are assumed impermeable for the passive scalar so that the appropriate boundary conditions for the tracers density function @xmath31 are @xmath158 . the streamlines of the flow are illustrated in fig . ( [ fig_rb ] ) . using fokker - planck equation @xmath159 for large pclet number ) @xmath160 where @xmath17 is now the molecular diffusivity of the considered system . ] @xmath161 , @xmath162 and @xmath7 the eddy diffusivity coefficient @xmath163 has been calculated in @xcite and it is @xmath164 according to the results of the previous section , the colored noise case with small @xmath2 is described up to order @xmath126 by the same fokker - planck equation provided the velocity field is renormalized as eq . ( [ vgtilde ] ) with @xmath148 . taking into account that @xmath165 , @xmath166 , and using ( [ stream ] ) , we have @xmath167 therefore up to order @xmath126 the diffusion is described by @xmath168 since @xmath169 does not depend on @xmath26 and @xmath27 , we can repeat the calculation of shraiman and obtain under the same conditions of ( [ shradiff ] ) the expression for the eddy diffusivity coefficient @xmath170 we then conclude that , in this case , a _ small _ @xmath2 enhances the diffusion coefficient . the same result can be deduced from the multiscale equations ( [ auxtilde ] ) and ( [ difftilde ] ) : in fact because of the structure of these two equations it is not difficult to show that if we change only the module of the velocity field ( @xmath171 ) and we know the explicit form of @xmath172 as a function of @xmath3 we have @xmath173 . for large pclet number and @xmath7 the function @xmath174 is given by eq . ( [ shradiff ] ) from which ( [ shradifftau ] ) follows . the ab flow is given by the velocity field @xmath175 for @xmath176 the streamlines form a closed periodically repeated structure made of four cells as shown in fig . ( [ fig_ab ] ) . we expect that the diffusive behavior of such a system is similar to the previous case , in fact we know @xcite that for small pclet number @xmath177 the eddy diffusivity tensor is proportional to @xmath178 like in the quasi - two - dimensional case . in the figure ( [ fig1 ] ) the behavior of the quantity @xmath179 / [ d^e(0 ) \ ; \tau ] $ ] versus @xmath180 is shown . the correction @xmath181 has been calculated by integrating numerically the equations ( [ eqcost1 ] ) and ( [ eqcost2 ] ) and evaluating the quantity @xmath182\ ; = \ ; \frac{d^e(\tau)-d^e(0)}{d^e(0 ) \ ; \tau}=\delta \label{diffnum}\ ] ] for different values of @xmath17 . the equations are solved by using a pseudo - spectral method @xcite in the basic periodicity cell with a grid mesh of @xmath183 points . de - aliasing has been obtained by a proper circular truncation which ensures better isotropy of numerical treatment . it is evident that also in this case the numerical results follow a linear behavior with a positive angular coefficient and so we can conclude that the introduction of the colored noise leads to an enhancement of the diffusion . in particular we can see that the numerical results follow very well the line @xmath184 . this is not surprising because we know that for large @xmath177 @xmath185 therefore using the same arguments of the previous section we can deduce that @xmath186 in a very good agreement with the numerical results . in this paper we have studied the transport properties in velocity fields whose small scales are parameterized by gaussian colored noise . we analyzed in particular the effects of a finite noise correlation time @xmath2 on the diffusive properties for large time and spatial scales . in this limit , using the multiscale technique , we derive the diffusion equation ( [ macrodi ] ) and the associated effective diffusion tensor . the latter is obtained , once the velocity field @xmath3 is given , by the solution of an auxiliary equation , see ( [ eqaux ] ) and ( [ diff ] ) , of the same structure of the original fokker - planck equation @xcite . the former is , however , an exact result for the diffusive regime valid for very long times , thus avoiding all finite time effects of the fokker - planck equation or the associated langevin equation ( [ lan1]),([lan11 ] ) . the auxiliary equation can not be solved for a generic velocity field , nevertheless there are nontrivial flows for which the solution can be found . this is the case for the steady parallel flow for which the effective diffusion coefficient is an increasing function of @xmath2 . to study in more details the effects of a small noise correlation time for a generic @xmath3 we have performed a small-@xmath2 expansion and evaluated the first correction @xmath4 to the effective diffusion coefficient . this is done by using two different approach . we find that to order @xmath4 there exist a one - parameter family of flows with the same diffusion properties . this invariance can be used to pick up the most convenient microscopic dynamics , from both analytical and numerical porpoises . we apply the small-@xmath2 results to two two dimensional model flows with closed streamlines . in both the cases we find an enhancement of the diffusion . the enhancement of the diffusion for small @xmath2 has been interpreted in @xcite in terms of interference mechanism between turbulent and molecular diffusion . the colored noise makes the diffusion particles forgotten of their previous positions less rapidly than in the white noise case , thus the lagrangian correlation time increases and so does the eddy - diffusivity . the study of the problem for @xmath2 not small is object of current work we warmly thank angelo vulpiani , massimo vergassola and andrea mazzino for extensive , stimulating and very useful discussions and suggestions . pc is grateful to the european science foundation for a tao exchange grant . this work has been partially supported by infm ( progetto ricerca avanzata- turbo ) and by murst ( program no . 9702265437 ) . in this appendix we derive the master equation for the probability density @xmath187 in the limit of small @xmath2 for the process described by the langevin equation ( [ lan1 ] ) and ( [ lan11 ] ) . if the random variable @xmath11 is a gaussian white noise of zero mean , @xmath187 satisfies the fokker - planck equation . we address here the case of a colored noise . now the process @xmath26 is non - markovian and no exact simple equation for @xmath188 is known . let us consider an ornstein - uhlenbeck process @xmath189 , that is a zero mean gaussian process with correlations [ c.f.r eq . ( [ lan11 ] ) ] @xmath190 where @xmath2 is the correlation time . the probability density is given by @xmath191 where @xmath192 is a solution of ( [ lan1 ] ) for a given realization of @xmath11 and for a given initial condition . the average is taken over the noise realizations . taking the time derivative of ( [ densi ] ) and using eq . ( [ lan1 ] ) one gets @xmath193 + \partial_i \langle s_i(t ) \delta({{\mbox{\boldmath $ x$}}}(t)- { { \mbox{\boldmath $ x$}}})\rangle . \label{dedensi}\ ] ] taking advantage of the gaussian nature of @xmath11 the average in ( [ dedensi ] ) can be rewritten as @xmath194 + \ ; \partial_{ij}^2 \;\int_{t_0}^t dt ' \ ; \left ( \frac{\sigma^2}{\tau}\right ) e^{-\frac{t - t'}{\tau } } \left < \frac{\delta x_j(t ) } { \delta s_i(t ' ) } \ ; \delta({{\mbox{\boldmath $ x$}}}(t)-{{\mbox{\boldmath $ x$ } } } ) \right>. \label{fpc}\ ] ] because of the @xmath0-function a closed equation is possible only if the functional derivative either does not involve the process @xmath26 or depends on it solely at the `` markovian '' end - point @xmath195 . at this stage thus we can not simplify the master equation any further . in the limit of small correlation time @xmath2 a closed equation can be derived by performing the largely used small @xmath2 expansion . if the noise is close to the white noise limit ( @xmath7 ) it is reasonable to expand the functional derivative about its markovian value , i.e. , the one obtained for the @xmath0-correlated noise . the taylor expansion of @xmath196 around the markovian end point @xmath197 is @xmath198 inserting the expansion ( [ fpexpa ] ) into ( [ fpc ] ) , keeping only the first terms in @xmath2 and negletting the transients , i.e. the letting to @xmath199 we obtain after a straightforward algebra the small @xmath2 master equation : @xmath129 + \ ; \partial_{ij}^2 \left [ { \cal d}_{ij } ( { { \mbox{\boldmath $ x$}}},t ) \ ; \theta ( { { \mbox{\boldmath $ x$}}},t ) \right ] \label{fpfinale}\ ] ] where @xmath200 \label{d}\ ] ] and use of incompressibility has been made . we note that this expansion does not converge uniformly in @xmath26 , and the diffusion coefficient ( [ d ] ) may exhibit negative values , thereby introducing unphysical singularities into the problem . in other words ( [ fpfinale ] ) and ( [ d ] ) does not constitute a truly markovian process with well - defined corresponding langevin equation driven by white noise . in general these equations are valid only for @xmath201 and @xmath202 ( in dimensionless units ) . the exact expression of the eddy diffusivity tensor for a stationary bidimensional shear flow @xmath203 with @xmath66 a periodic function in @xmath67 can be deduced from ( [ diffshe ] ) and reads @xmath204 applying the generalized formulas ( [ chi0 g ] ) , ( [ chi1 g ] ) and ( [ diffgen ] ) to the shear flow we want to obtain the expressions ( [ esatta1 ] ) . thus eq . ( [ chi1 g ] ) becomes @xmath208 the stationary solutions are @xmath209 using eq . ( [ diffgen ] ) the eddy diffusivity tensor is @xmath210 and @xmath211 a comparison between ( [ gen1 ] ) and ( [ esatta1 ] ) shows that the generalized equations lead to the exact expression of the eddy diffusivity tensor for a stationary shear flow if @xmath212 the same condition is found if we consider a time - dependent shear flow . a. bensoussan , j .- l . lions and g. papanicolaou , _ asymptotic analysis for periodic structures _ ( north - holland , amsterdam , 1978 ) d. mc laughlin , g.c . papanicolaou and o. pironneau , _ siam journal of appl . _ , * 45 * , 780 , ( 1985 ) .

the diffusive properties in velocity fields whose small scales are parameterized by non @xmath0-correlated noise is investigated using multiscale technique . the analytical expression of the eddy diffusivity tensor is found for a @xmath1d steady shear flow and it is an increasing function of the characteristic noise decorrelation time @xmath2 . in order to study a generic flow @xmath3 , a small-@xmath2 expansion is performed and the first correction @xmath4 to the effective diffusion coefficients is evaluated . this is done using two different approaches and it results that at the order @xmath2 the problem with a colored noise is equivalent to the white - in - time case provided by a renormalization of the velocity field @xmath5 depending on @xmath2 . two examples of @xmath1d closed - streamlines velocity field are considered and in both the cases an enhancement of the diffusion is found . pacs number(s ) : ; +

several decades of experimental and theoretical work on electron scattering have provided a wealth of information on nuclear structure and dynamics @xcite . in these experiments the electron is the probe , whose properties are clearly specified , and the nucleus the target whose properties are under investigation . additional information on nuclear properties is available from @xmath0-nucleus scattering . neutrinos can excite nuclear modes unaccessible in electron scattering , can give information on the hadronic weak current and on the strange form factors of the nucleon . although of great interest , such studies are not the only aim of many neutrino experiments , which are better devised for a precise determination of neutrino properties . in neutrino oscillation experiments nuclei are used to detect neutrinos and a proper analysis of data requires that the nuclear response to neutrino interactions is well under control and that the unavoidable theoretical uncertainties on nuclear effects are reduced as much as possible . in recent years different models developed and successfully tested in comparison with electron scattering data have been extended to @xmath0-nucleus scattering . although the two situations are different , electron scattering is the best available guide to determine the prediction power of a nuclear model . nonrelativistic and relativistic models have been developed to describe nuclear effects with different approximations . they can be considered as alternative models , but only a relativistic approach is able to account for all the effects of relativity in a complete and consistent way . relativity is important at all energies , in particular at high energies , and in the energy regime of many neutrino experiments a relativistic approach is required . relativistic models for the exclusive and inclusive electron and neutrino scattering in the qe region @xcite are presented in this contribution . in the qe region the nuclear response is dominated by one - nucleon knockout processes , where the probe interacts with a quasifree nucleon that is emitted from the nucleus with a direct one - step mechanism and the remaining nucleons are spectators . in electron scattering experiments the outgoing nucleon can be detected in coincidence with the scattered electron . in the exclusive @xmath1 reaction the residual nucleus is left in a specific discrete eigenstate and the final state is completely specified . in the inclusive @xmath2 scattering the outgoing nucleon is not detected and the cross section includes all the available final nuclear states . for an incident neutrino or antineutrino nc and cc scattering can be considered @xmath3 in nc scattering only the emitted nucleon can be detected and the cross section is integrated over the energy and angle of the final lepton . also the state of the residual @xmath4-nucleus is not determined and the cross section is summed over all the available final states . the same situation occurs for the cc reaction if only the outgoing nucleon is detected . the cross sections are therefore semi - inclusive in the hadronic sector and inclusive in the leptonic one and can be treated as an @xmath1 reaction where only the outgoing proton is detected . the exclusive cc process where the charged final lepton is detected in coincidence with the emitted nucleon can be considered as well . the inclusive cc scattering where only the charged lepton is detected can be treated with the same models used for the inclusive @xmath2 reaction . for all these processes the cross section is obtained in the one - boson exchange approximation from the contraction between the lepton tensor , that depends only on the lepton kinematics , and the hadron tensor @xmath5 , that contains the nuclear response and whose components are given by products of the matrix elements of the nuclear current @xmath6 between the initial and final nuclear states , i.e. , @xmath7 where @xmath8 and @xmath9 are the energy and momentum transfer , respectively . different but consistent models to calculate @xmath5 in qe electron and @xmath0-nucleus scattering are outlined in the next sections . models based on the relativistic distorted - wave impulse approximation ( rdwia ) have been developed @xcite to describe the exclusive reaction where the outgoing nucleon is detected in coincidence with the scattered lepton and the residual nucleus is left in a discrete eigenstate @xmath10 . in rdwia the amplitudes of eq . [ eq.wmn ] are obtained in a one - body representation as @xmath11 where @xmath12 is the s.p . scattering state of the emitted nucleon , @xmath13 the overlap between the ground state of the target and the final state @xmath10 , i.e. , a s.p . bound state , and @xmath14 the one - body nuclear current . in the model the s.p . bound and scattering states are consistently derived as eigenfunctions of a feshbach - type optical potential @xcite . phenomenological ingredients are adopted in the calculations . the bound states are dirac - hartree solutions of a lagrangian , containing scalar and vector potentials , obtained in the framework of the relativistic mean - field theory @xcite . the scattering state is calculated solving the dirac equation with relativistic energy - dependent complex optical potentials @xcite . rdwia models have been quite successful in describing a large amount of data for the exclusive @xmath15 reaction @xcite . the transition amplitudes of the nc and cc processes where only the outgoing nucleon is detected are described as the sum of the rdwia amplitudes in eq . [ eq.dko ] over the states @xmath10 . in the calculations @xcite a pure shell - model ( sm ) description is assumed , i.e. , @xmath10 is a one - hole state and the sum is over all the occupied sm states . fsi are described by a complex optical potential whose imaginary part reduces the cross section by @xmath16 . a similar reduction is obtained in the rdwia calculations for the exclusive one - nucleon knockout . the imaginary part accounts for the flux lost in a specific channel towards other channels . this approach is conceptually correct for an exclusive reaction , where only one channel contributes , but it would be wrong for the inclusive scattering , where all the channels contribute and the total flux must be conserved . for the semi - inclusive process where an emitted nucleon is detected , some of the reaction channels which are responsible for the imaginary part of the potential are not included in the experimental cross section and , from this point of view , it is correct to include the absorptive imaginary part . numerical examples in different kinematics are given in @xcite . in the inclusive scattering where only the outgoing lepton is detected fsi are treated in the green s function approach ( gfa ) @xcite . in this model the components of the hadron tensor are written in terms of the s.p . optical model green s function . this is the result of suitable approximations , such as the assumption of a one - body current and subtler approximations related to the ia . the explicit calculation of the s.p . green s function is avoided by its spectral representation , which is based on a biorthogonal expansion in terms of a non hermitian optical potential @xmath17 and of its hermitian conjugate @xmath18 . calculations require matrix elements of the same type as the rdwia ones in eq . [ eq.dko ] , but involve eigenfunctions of both @xmath17 and @xmath18 , where the different sign of the imaginary part gives in one case an absorption and in the other case a gain of flux . thus , in the sum over @xmath10 the total flux is redistributed and conserved . the gfa guarantees a consistent treatment of fsi in the exclusive and in the inclusive scattering and gives a good description of @xmath2 data @xcite . an example is displayed in fig . [ fig1 ] , where the @xmath19 cross sections calculated in gfa are compared with the results of the relativistic plane wave ia ( rpwia ) , where fsi are neglected . the cross sections obtained when only the real part of the relativistic optical potential ( rrop ) is retained and the imaginary part is neglected are also shown in the figure . this approximation conserves the flux , but it is conceptually wrong because the optical potential has to be complex owing to the presence of inelastic channels . the partial contribution given by the sum of all the integrated exclusive one - nucleon knockout reactions , also shown in the figure , is much smaller than the complete result . the difference is due to the spurious loss of flux produced by the absorptive imaginary part of the optical potential . the analysis of data requires a precise knowledge of @xmath0-nucleus cross sections , where theoretical uncertainties on nuclear effects are reduced as much as possible . to this aim , it is important to check the consistency of different models and the validity of the approximations . the results of the relativistic models developed by our group and the madrid - sevilla group for the inclusive electron scattering are compared in @xcite . an example is shown in fig . [ fig2 ] for the @xmath20 cross sections calculated with different descriptions for fsi : rpwia , rrop , gfa ( with two parametrizations of the optical potential ) , and the relativistic mean field ( rmf ) @xcite , where the scattering wave functions are calculated with the same real potential used for the initial bound states . the differences between rmf and gfa increase with @xmath21 : they are small at @xmath21 = 500 mev@xmath22 and significant at @xmath21 = 1000 mev@xmath22 . the rmf cross section shows an asymmetry , with a long tail extending towards higher values of @xmath8 . a less significant asymmetry is obtained for both gfa cross sections , that at @xmath21 = 1000 mev@xmath22 are higher than the rmf one in the maximum region . the enhancement is different for the two optical potentials . the behaviour of the rmf and gfa results as a function of @xmath21 and @xmath8 can be understood if we consider that rmf is based on the use of strong energy - independent scalar and vector real potentials , while gfa on a complex energy - dependent optical potential . different values of @xmath21 and @xmath8 are sensitive to the behavior of the optical potential at different energies , and higher values correspond to higher energies . the gfa results are consistent with the general behavior of the optical potentials and are basically due to their imaginary part . different parameterizations give similar real terms and the rrop cross sections are practically insensitive to the choice of optical potential . the real part decreases increasing the energy and the rrop result approaches the rpwia one for large values of @xmath8 . in contrast , the imaginary part has its maximum strength around 500 mev and is sensitive to the parameterization of the rop . the imaginary part gives large differences between gfa and rrop in fig . [ fig2 ] , while only negligible differences are obtained in the different situation and kinematics of fig . [ fig1 ] . s. boffi , c. giusti , f. d. pacati , and m. radici , _ electromagnetic response of atomic nuclei _ , oxford studies in nuclear physics , vol . 20 ( clarendon press , oxford , 1996 ) ; s. boffi , c. giusti , and f. d. pacati , phys . rep . * 226 * 1 ( 1993 ) . w. pschl , d. vretenar , and p. ring , comput . * 103 * , 217 ( 1997 ) ; g.a . lalazissis , j. knig , and p. ring , phys . c * 55 * , 540 ( 1997 ) ; m.m . sharma , m.a . nagarajan , and p. ring phys . lett . * b312 * , 377 ( 1993 ) . clark , in _ proc . of the workshop on relativistic dynamics and quark - nuclear physics _ , ed . by m.b . johnson and a. picklesimer ( john wiley & sons , new york , 1986 ) , p. 302 ; e.d . et al . _ , c * 47 * , 297 ( 1993 ) .

relativistic models developed for the exclusive and inclusive quasielastic ( qe ) electron scattering have been extended to charged - current ( cc ) and neutral - current ( nc ) @xmath0-nucleus scattering . the results of different descriptions of final - state interactions ( fsi ) are compared .

the authors thank s. dittmaier for useful discussions . 99 s. l. glashow , nucl . * 22 * ( 1961 ) 579 ; s. weinberg , phys . * 19 * ( 1967 ) 1264 ; a. salam , proc . 8th nobel symposium , ed . n. svartholm ( almquits and wiksells , stockholm , 1968 ) , p. 367 s. dittmaier , d. schildknecht and g. weiglein , bi - tp 95/31 , hep - ph/9510386 , to be published in nucl . phys * b * m. bilenky , j. l. kneur , f. m. renard and d. schildknecht , nucl . phys . * b409 * ( 1993 ) 22 c. grosse - knetter , i. kuss and d. schildknecht , phys . lett . * b358 * ( 1995 ) 87 p. q. hung and j. j. sakurai , nucl . b143 * ( 1978 ) 81 j. maalampi , d. schildknecht and k.h . schwarzer , phys . lett . * b166 * ( 1986 ) 361 ; m. kuroda , j. maalampi , d. schildknecht and k. h. schwarzer , nucl . phys . * b284 * ( 1987 ) 271 ; phys . lett . * b190 * ( 1987 ) 217 h. c. corben and j. schwinger , phys . rev . * 58 * ( 1940 ) 953 ; t. d. lee , phys . rev . * 140 * ( 1965 ) b967 the cdf collaboration , phys . lett * 75 * ( 1995 ) 1017 ; the d0 collaboration , phys . lett * 75 * ( 1995 ) 1023 and 1034

in view of the forthcoming results on @xmath0 pair production at lep 2 , we emphasize that _ direct empirical evidence _ on non - trivial properties of the weak vector bosons can be obtained with relatively limited integrated luminosity . an integrated luminosity of @xmath1 at @xmath2 gev will be sufficient to provide direct experimental evidence for non - vanishing self - couplings of non - abelian type among the weak vector bosons . an integrated luminosity of @xmath3 at @xmath2 gev will provide direct evidence for the existence of an anomalous magnetic dipole moment of the charged vector bosons @xmath0 . bi - tp 96/11 + hep - ph/9603283 * i. kuss , d. schildknecht * + universitt bielefeld + fakultt fr physik + postfach 10 01 31 + d33501 bielefeld , germany + revised version , may 1996 [ [ section ] ] with respect to properties of the electroweak vector bosons , there are two salient predictions of the standard electroweak theory @xcite which lack direct experimental confirmation so far : the non - abelian coupling between the members of the @xmath4 triplet of vector bosons , and , closely connected to it , the anomalous magnetic dipole moment coupling of the charged vector bosons , @xmath0 . from the detailed analysis of the lep 1 precision data there is strong _ indirect _ evidence @xcite that indeed the standard couplings are realized in nature . the production of @xmath5 at lep 2 will nevertheless open up a new domain of investigations by providing _ direct _ tests of the non - abelian structure and the anomalous magnetic dipole moment prediction . limited luminosity will restrict the possibility of precision measurements of the trilinear @xmath6 and the @xmath7 couplings at lep 2 . it is the purpose of the present note to point out that a rather small integrated luminosity at lep 2 , however , will nevertheless be sufficient to provide empirical evidence for the existence of a non - abelian trilinear coupling among the massive vector bosons and for the existence of an anomalous magnetic dipole moment of the @xmath0 . we start from the phenomenological lagrangian for trilinear vector boson couplings widely used in the simulation of future data @xcite , @xmath8\nonumber\\ & & -iex_\gamma f_{\mu\nu}w^{+\mu}w^{-\nu}\nonumber\\ & & -ie(\frac{c_w}{s_w}+\delta_z)[z_\mu(w^{-\mu\nu}w^+_\nu - w^{+\mu\nu}w^-_\nu ) + z_{\mu\nu}w^{+\mu}w^{-\nu}]\nonumber\\ & & -iex_z z_{\mu\nu}w^{+\mu}w^{-\nu}. \label{xdellag}\end{aligned}\ ] ] it contains the three arbitrary parameters @xmath9 , @xmath10 and @xmath11 and reduces to the standard form for @xmath12 . the parameter @xmath9 is directly related to the anomalous magnetic dipole moment @xmath13 of the @xmath0 via @xmath14 where @xmath15 is the value in the standard model . this value follows from the linear realization of the @xmath16 symmetry in conjunction with the restriction to dimension-4 terms as embodied in the standard electroweak theory @xcite . realizing @xmath17 symmetry non - linearly , or else removing the restriction to dimension-4 terms in the linearized form , removes the restriction @xmath18 , e.g. @xcite . we also note that @xmath18 corresponds to a gyromagnetic ratio , @xmath19 , of the @xmath0 of magnitude @xmath20 , in units of the particle s bohr - magneton @xmath21 , while @xmath22 corresponds to @xmath23 as obtained for a classical rotating charge distribution . in general @xmath24 . the value of @xmath20 puts the electromagnetic properties of the @xmath0 in close analogy to the @xmath20 value of the spin-@xmath25 dirac theory . in order to relate @xmath10 and @xmath11 to the parameters in the lagrangian before diagonalization of the neutral sector , it is advantageous to rewrite ( [ xdellag ] ) in terms of unmixed neutral vector boson fields . for the present purpose it is useful to describe mixing in terms of current - mixing @xcite , i.e a mixing term between the third component of the weak isotriplet , @xmath26 , and the photon field tensor , which is denoted by @xmath27 in order to discriminate @xmath28 from the physical photon field tensor @xmath29 emerging upon diagonalization . for details we refer to @xcite . carrying out the transformtion to the unmixed fields , @xmath30 the lagrangian ( [ xdellag ] ) becomes @xmath31,\label{lint}\end{aligned}\ ] ] where the third component of @xmath32 is understood to be @xmath33 . the relations between @xmath34 , @xmath35 , @xmath13 and @xmath9 , @xmath11 and @xmath10 are given by ( [ kappagam ] ) and @xmath36 the standard model thus corresponds to @xmath37 , @xmath38 and @xmath15 . let us consider the lagrangian ( [ lint ] ) in some detail . it contains @xcite : * an @xmath4 symmetric interaction term of non - abelian form with arbitrary strength @xmath34 which coincides with the standard term for the special choice @xmath37 . note that this term is contained in the kinetic term for the @xmath32 fields , @xmath39 provided we use the non - abelian field tensor , @xmath40 for @xmath41 , the kinetic term reduces to a triplet of abelian vector boson fields , is obviously broken by the mass terms for the vector bosons . ] . accordingly , the case of @xmath41 may be referred to as the abelian triplet model . * a term of strength @xmath35 which violates @xmath4 symmetry independently of the presence of the photon field . imposing the constraint of no intrinsic @xmath4 violation@xcite , i.e. @xmath4 symmetry , when electromagnetism in ( [ lint ] ) is absent , we have to require @xmath38 , i.e. @xmath42 this requirement is abstracted from its empirical validity in the vector boson mass term and is sometimes called custodial @xmath43 symmetry . * the two terms describing the electromagnetic interactions . the term containing @xmath44 in ( [ lint ] ) is simply obtained by applying the minimal substitution principle , @xmath45 , @xmath46 being the charge operator , @xmath47 and @xmath48 , to the derivatives in the kinetic term ( [ wkin ] ) of the @xmath32 triplet . the anomalous magnetic moment term @xmath13 , @xmath49 in ( [ lint ] ) , on the other hand , need not necessarily be present , even though it is not excluded by the principle of minimal substitution @xcite . in fact , adding the @xmath4-invariant total derivative term @xmath50 to the kinetic term ( [ wkin ] ) and carrying out the minimal substitution prescription implies the electromagnetic interaction ( [ lint ] ) with an arbitrary value of @xmath51 . + we note that both the charge coupling , @xmath52 , as well as the magnetic moment coupling , @xmath13 , appearing in connection with the unmixed field @xmath44 , @xmath28 in ( [ lint ] ) , are identical to the couplings in the lagrangian ( [ xdellag ] ) for the physical fields , @xmath53 , @xmath29 . this is in contrast to the case of the massive neutral vector boson , where the third component of the triplet in ( [ lint ] ) couples with a strength @xmath34 different from the strength of the @xmath54 coupling , @xmath55 , in ( [ xdellag ] ) . we note in passing that the mentioned @xmath4-symmetry properties of the terms multiplied by @xmath34 and @xmath35 in ( [ lint ] ) are not a unique feature of the @xmath56 basis . the physical interpretation of the parameters @xmath34 and @xmath35 is thus not only valid in this basis . indeed , expressing ( [ xdellag ] ) in terms of the more conventional @xmath57 basis via @xmath58 one obtains the lagrangian in the form @xmath59.\label{lbw3}\end{aligned}\ ] ] this form of the lagrangian coincides with ( [ lint ] ) in the first two terms apart from @xmath60 being replaced by @xmath61 . this is a consequence of the fact that the second columns of the matrices in ( [ trafo1 ] ) and ( [ trafo2 ] ) are identical . obviously , however , the @xmath62-dependent terms in ( [ lbw3 ] ) differ from the @xmath63 terms in ( [ lint ] ) . in summary , it is a characteristic feature of a non - abelian theory to contain a coupling @xmath64 . experimental evidence for @xmath64 thus provides evidence for a non - abelian structure of the interactions among the members of the @xmath32 triplet and rules out a theory based on three abelian vector boson fields . it is a second characteristic feature of the non - abelian nature of the standard model interactions to contain an anomalous magnetic dipole moment term of definite strength , @xmath15 . while a precision measurement of @xmath13 will be difficult , empirical evidence for @xmath65 will nevertheless provide first direct experimental evidence for the existence of an anomalous magnetic dipole moment of the @xmath0 vector bosons . we turn to the experimental search for a non - abelian coupling , @xmath64 , and the search for an anomalous magnetic dipole moment , @xmath49 . we note that the present most stringent direct bound to the @xmath66-coupling @xmath9 @xcite is @xmath67 at 95% cl coupling . the bounds to the @xmath68 couplings reported in @xcite are based on specific model assumptions which are inconsistent with the present work . ] , corresponding to @xmath69 . we consider a measurement of the total cross section of @xmath70 at lep 2 with a cut @xmath71 on the scattering angle . formulae for the cross section in terms of the parameters @xmath72 and @xmath10 have been given in @xcite . we assume that the decay mode @xmath73 will be detected at lep 2 , where the lepton @xmath74 can be either an electron ( positron ) or a muon ( anti - muon ) . the branching ratio for this decay mode is 29.6% . we assume that future data are identical to the standard model predictions and calculate the lines of constant @xmath75 ( 86.5% cl ) and @xmath76 ( 39.4% cl ) in the @xmath9-@xmath10-plane according to such that the statistical error is given by @xmath77 . if one uses the error estimated from the experiment , @xmath78 , instead of the statistical error in the denominator of ( [ chi2 ] ) , the values of the parameter pair ( @xmath79 ) which have @xmath75 or @xmath76 change only little provided the number of measured events , @xmath80 , is much greater than one , @xmath81 . ] @xmath82 in ( [ chi2 ] ) , @xmath80 and @xmath83 denote the number of events in the sm and in the alternative model with the @xmath4 constraint ( [ su2c_rel ] ) and @xmath84 , respectively . in figure [ fig1 ] we show the @xmath85 and @xmath86 contours in the @xmath87 plane for an integrated luminosity of @xmath88 , at @xmath89 175 gev . we see that this very small value of @xmath90 , corresponding to a few weeks of running at lep 2 , is sufficient to detect a genuine non - zero non - abelian coupling , @xmath64 , at the @xmath86 level . likewise , a vanishing anomalous magnetic moment , @xmath22 , in conjunction with an abelian @xmath32 triplet , @xmath41 , is excluded . we note that similar results can be obtained with @xmath91 at @xmath92 gev and with @xmath93 at @xmath94 gev . to detect a non - vanishing magnetic dipole moment , @xmath49 , in the presence of a non - vanishing non - abelian coupling , @xmath64 , needs a somewhat higher integrated luminosity . the result in figure [ fig2 ] corresponds to a luminosity of @xmath95 at @xmath2 gev , thus providing direct evidence for a non - vanishing anomalous magnetic dipole moment of the @xmath96 , @xmath65 . since a sufficient number of standard events , @xmath97 , is expected in this case , the events can be arranged in 6 bins equidistant over the scattering angle @xmath98 , leading to the result also presented in figure [ fig2 ] . in the same figure we also show the theoretical prediction corresponding to a vanishing @xmath99 coupling , @xmath100 , corresponding to @xmath101 in ( [ xdellag ] ) , and @xmath102 , which is similarly ruled out . in conclusion , after a few weeks of running at @xmath2 gev at lep 2 , definite direct evidence for the existence of a genuine , non - vanishing coupling among the members of the @xmath103 triplet , characteristic of a non - abelian structure , can be obtained . likewise , after 7 months of running at lep 2 , definite evidence for a non - vanishing anomalous magnetic dipole moment of the charged vector bosons may be expected .

in recent years , gaussian filters have assumed a central role in image filtering and techniques for accurate measurement @xcite . the implementation of the gaussian filter in one or more dimensions has typically been done as a convolution with a gaussian kernel , that leads to a high computational cost in its practical application . computational efforts to reduce the gaussian convolution complexity are discussed in @xcite . more advantages may be gained by employing a _ spatially recursive filter _ , carefully constructed to mimic the gaussian convolution operator . + recursive filters ( rfs ) are an efficient way of achieving a long impulse response , without having to perform a long convolution . initially developed in the context of time series analysis @xcite , they are extensively used as computational kernels for numerical weather analysis , forecasts @xcite , digital image processing @xcite . recursive filters with higher order accuracy are very able to accurately approximate a gaussian convolution , but they require more operations . + in this paper , we investigate how the rf mimics the gaussian convolution in the context of variational data assimilation analysis . variational data assimilation ( var - da ) is popularly used to combine observations with a model forecast in order to produce a _ best _ estimate of the current state of a system and enable accurate prediction of future states . here we deal with the three - dimensional data assimilation scheme ( 3d - var ) , where the estimate minimizes a weighted nonlinear least - squares measure of the error between the model forecast and the available observations . the numerical problem is to minimize a cost function by means of an iterative optimization algorithm . the most costly part of each step is the multiplication of some grid - space vector by a covariance matrix that defines the error on the forecast model and observations . more precisely , in 3d - var problem this operation may be interpreted as the convolution of a covariance function of background error with the given forcing terms . + here we deal with numerical aspects of an oceanographic 3d - var scheme , in the real scenario of oceanvar . ocean data assimilation is a crucial task in operational oceanography and the computational kernel of oceanvar software is a linear system resolution by means of the conjugate gradient ( gc ) method , where the iteration matrix is relate to an errors covariance matrix , having a gaussian correlation structure . + in @xcite , it is shown that a computational advantage can be gained by employing a first order rf that mimics the required gaussian convolution . instead , we use the 3rd - rf to compute numerically the gaussian convolution , as how far is only used in signal processing @xcite , but only recently used in the field of var - da problems . + in this paper we highlight the main sources of error , introduced by these new numerical operators . we also investigate the real benefits , obtained by using 1-st and 3rd - rfs , through a careful error analysis . theoretical aspects are confirmed by some numerical experiments . finally , we report results in the case study of the oceanvar software . + the rest of the paper is organized as follows . in the next section we recall the three - dimensional variational data assimilation problem and we remark some properties on the conditioning for this problem . besides , we describe our case study : the oceanvar problem and its numerical solution with cg method . in section iii , we introduce the @xmath0-th order recursive filter and how it can be applied to approximate the discrete gaussian convolution . in section iv , we estimate the effective error , introduced at each iteration of the cg method , by using 1st - rf and 3rd - rf instead of the gaussian convolution . in section v , we report some experiments to confirm our theoretical study , while the section vi concludes the paper . the aim of a generic variational problem ( var problem ) is to find a best estimate @xmath1 , given a previous estimate @xmath2 and a measured value @xmath3 . with these notations , the var problem is based on the following regularized constrained least - squared problem : @xmath4 where @xmath1 is defined in a grid domain @xmath5 . the objective function @xmath6 is defined as follows : @xmath7 where measured data are compared with the solution obtained from a nonlinear model given by @xmath8 . + in ( [ ls pb ] ) , we can recognize a quadratic data - fidelity term , the first term and the general regularization term ( or penalty term ) , the second one . when @xmath9 and the regularization term can be write as : @xmath10 we deal with a three - dimensional variational data assimilation problem ( 3d - var da problem ) . the purpose is to find an optimal estimate for a vector of states @xmath11 ( called the analysis ) of a generic system @xmath12 , at each time @xmath13 given : * a prior estimate vector @xmath14 ( called the background ) achieved by numerical solution of a forecasting model @xmath15 , with error @xmath16 ; * a vector @xmath17 of observations , related to the nonlinear model by @xmath18 that is an effective measurement error : @xmath19 at each time t , the errors @xmath20 in the background and the errors @xmath21 in the observations are assumed to be random with mean zero and covariance matrices @xmath22 and @xmath23 , respectively . more precisely , the covariance @xmath24 of observational error is assumed to be diagonal , ( observational errors statistically independent ) . the covariance @xmath25 of background error is never assumed to be diagonal as justified in the follow . to minimize , with respect to @xmath11 and for each @xmath26 , the problem becomes : @xmath27 in explicit form , the functional cost of ( [ da pb ] ) problem can be written as : @xmath28 it is often numerically convenient to approximate the effects on @xmath29 of small increments of @xmath11 , using the linearization of @xmath30 . for small increments @xmath31 , follows @xcite , it is : @xmath32 where the linear operator @xmath33 is the matrix obtained by the first order approximation of the jacobian of @xmath30 evaluated at @xmath14 . + now let @xmath34 be the _ misfit_. then the function @xmath35 in ( [ da1 pb ] ) takes the following form in the increment space : @xmath36 at this point , at each time @xmath37 , the minimum of ( [ da2 pb ] ) is obtained by requiring @xmath38 . this gives rise to the linear system : @xmath39 or equivalently : @xmath40 for each time @xmath41 , iterative methods , able to converge toward a practical solution , are needed to solve the linear system ( [ system1 ] ) . however this problem , so as formulated , is generally very ill conditioned . more precisely , by following @xcite , and assuming that @xmath42 [ lambda_m ] is a diagonal matrix , it can be proved that the conditioning of @xmath43 is strictly related to the conditioning of the matrix @xmath22 ( the covariance matrix ) . in general , the matrix @xmath22 is a block - diagonal matrix , where each block is related to a single state of vector @xmath11 and it is ill conditioned . + this assertion is exposed in @xcite starting from the expression of @xmath22 for one - state vectors as : @xmath44 where @xmath45 is the background error variance and @xmath46 is a matrix that denotes the correlation structure of the background error . assuming that the correlation structure of matrix @xmath46 is homogeneous and depends only on the distance between states and not on positions , an expression of @xmath46 as a symmetric matrix with a circulant form is given ; i. e. as a toeplitz matrix . by means of a spectral analysis of its eigenvalues , the ill - conditioning of the matrix @xmath46 is checked . as in @xcite , it follows that @xmath22 is ill - conditioned and the matrix @xmath43 , of the linear system ( [ system1 ] ) , too . a well - known technique for improving the convergence of iterative methods for solving linear systems is to _ preconditioning _ the system and thus reduce the condition number of the problem . + in order to precondition the system in ( [ system1 ] ) , it is assumed that @xmath22 can be written in the form @xmath47 , where @xmath48 is the square root of the background error covariance matrix @xmath22 . because @xmath22 is symmetric gaussian , @xmath49 is uniquely defined as the symmetric ( @xmath50 ) gaussian matrix such that @xmath51 . + as explained in @xcite , the cost function ( [ da2 pb ] ) becomes : @xmath52 \ \ \quad \!=\ ! \frac{1}{2 } ( d_t \!- \ ! { \bf h } \delta x_t)^t { \bf r}^{-1 } ( d_t \!- \ ! { \bf h } \delta x_t ) + \frac{1}{2 } \delta x_t^t ( { \bf v^t})^{-1 } { \bf v}^{-1}\delta x_t \end{array}\ ] ] now , by using a new control variable @xmath53 , defined as @xmath54 , at each time @xmath26 and observing that @xmath55 we obtain a new cost function : @xmath56 equation ( [ da3 pb ] ) is said the _ dual problem _ of equation ( [ da2 pb ] ) . finally , to minimize the cost function @xmath57 in ( [ da3 pb ] ) leads to the new linear system : @xmath58 upper and lower bounds on the condition number of the matrix @xmath59 are shown in @xcite . in particular it holds that : @xmath60 moreover , under some special assumptions , it can be proved that @xmath61 is very well - conditioned ( @xmath62 ) . as described in @xcite , at each time @xmath26 , oceanvar software implements an oceanographic three - dimensional variational da scheme ( 3d var - da ) to produce forecasts of ocean currents for the mediterranean sea . the computational kernel is based on the resolution of the linear system defined in ( [ system2 ] ) . to solve it , the conjugate gradient ( cg ) method is used and a basic outline is described in * algorithm 1*. [ cg ] @xmath63 ; @xmath64 , the initial guess ; + @xmath65 ; + @xmath66 ; + @xmath67 ; @xmath68 ; @xmath69 ; + @xmath70 ; @xmath71 ; + @xmath72 ; @xmath73 ; + [ cga ] we focus our attention on step 5 . : at each iterative step , a matrix - vector product @xmath74 is required , where @xmath75 @xmath76 is the residual at step @xmath77 and @xmath78 depends on the number of observations and is characterized by a bounded norm ( see @xcite for details ) . more precisely , we look to the matrix - vector product @xmath79 which can be schematized as shown in * algorithm 2*. [ five_gaussian ] @xmath80 ; + @xmath81 ; + @xmath82 ; + @xmath83 ; [ pcg ] the steps 1 . and 3 . in * algorithm 2 * consist in a matrix - vector product . these products , as detailed in next section , can be considered discrete gaussian convolutions and the matrix @xmath49 , for one - dimensional state vectors , has gaussian structure . even for state vectors defined on two ( or more ) dimensions , the matrix @xmath49 can be represented as product of two ( or more ) gaussian matrices . since a single matrix - vector product of this form becomes prohibitively expensive if carried out explicitly , a computational advantage is gained by employing gaussian rfs to mimic the required gaussian convolution operators . + in the previous oceanvar scheme , it was implemented a 1st - rf algorithm , as described in @xcite . here , we study the 3rd - rf introduction , based on @xcite . + the aim of the following sections is to precisely reveal how the @xmath0-th order recursive filters are defined and , through the error analysis , to investigate on their effect in terms of error estimate and perfomences . in this section we describe gaussian recursive filters as approximations of the discrete gaussian convolution used in steps 1 . and 3 . of * algorithm 2*. let denote by @xmath84 the normalized gaussian function and by @xmath85 the square matrix whose entries are given by @xmath86 now let be @xmath87 a vector ; the discrete gaussian convolution of @xmath88 is a new vector @xmath89 defined by means of the matrix - vector product @xmath90 the discrete gaussian convolution can be considered as a discrete representation of the continuous gaussian convolution . as is well known , the continuous gaussian convolution of a function @xmath88 with the normalized gaussian function @xmath91 is a new function @xmath92 defined as follows : @xmath93(x ) = \int_{-\infty}^{+\infty } g(x- \tau ) s^0(\tau ) d \tau.\ ] ] discrete and continuous gaussian convolutions are strictly related . this fact could be seen as follows . let assume that @xmath94 is a grid of evaluation points and let set for @xmath95 @xmath96 by assuming that @xmath88 is @xmath97 outside of @xmath98 $ ] and by discretizing the integral ( [ convgauss ] ) with a rectangular rule , we obtain @xmath99 @xmath100 @xmath101 an optimal way for approximating the values @xmath102 is given by gaussian recursive filters . the @xmath0-order rf filter computes the vector @xmath103 as follows : @xmath104 s_i^k=\beta_i p_i^{k}+ \displaystyle{\sum_{j=1}^n } \alpha_{i , j } s_{i+j}^k \,\,\,\ i = m,\ldots,1 \end{array } \right .. \ ] ] the iteration counter @xmath77 goes from @xmath105 to @xmath106 , where @xmath106 is the total number of filter iterations . observe that values @xmath107 are computed taking in the sums terms @xmath108 provided that @xmath109 . analogously values @xmath110 are computed taking in the sums terms @xmath111 provided that @xmath112 . the values @xmath113 and @xmath114 , at each grid point @xmath115 , are often called _ smoothing coefficients _ and they obey to the constraint @xmath116 in this paper we deal with first - order and third - order rfs . the first - order rf expression ( @xmath117 ) becomes : @xmath118 s_m^k=\beta_m p_m^{k } , \\ \ , s_i^k=\beta_i p_i^{k}+ \alpha_i s_{i+1}^k \qquad i = m-1,\ldots,1 . \end{array } \right.\ ] ] if @xmath119 is the correlation radius at @xmath115 , by setting @xmath120 coefficients @xmath121 e @xmath122 are given by @xcite : @xmath123 the third - order rf expression ( @xmath124 ) becomes : @xmath125 s_i^k=\beta_i p_i^{k}+ \displaystyle{\sum_{j=1}^3 } \alpha_{i , j } s_{i+j}^k \,\,\,\ i = m,\ldots,1 . \end{array } \right.\ ] ] third - order rf coefficients @xmath126 and @xmath122 , for one only filter iteration ( @xmath127 ) , are computed in @xcite . if @xmath128 the coefficients expressions are : @xmath129 \alpha_{i,2}=-(3.382473 \sigma_i^2 + 3 \sigma_i^3)/a_i\\[.2 cm ] \alpha_{i,3}= \sigma_i^3 /a_i\\[.2 cm ] \beta_i=1-(\alpha_{i,1}+\alpha_{i,2}+\alpha_{i,3})=3.738128/a_i . \end{array}\ ] ] in @xcite is proposed the use of a value @xmath130 instead of @xmath131 . the @xmath132 value is : @xmath133 3.97156 - 4.14554 \sqrt{1 - 0.26891\sigma_i } \quad \text{oth . } \end{array } \right.\ ] ] in order to understand how gaussian rfs approximate the discrete gaussian convolution it is useful to represent them in terms of matrix formulation . as explained in @xcite , the @xmath0-order recursive filter computes @xmath134 from @xmath135 as the solution of the linear system @xmath136 where matrices @xmath137 and @xmath138 are respectively lower and upper band triangular with nonzero entries @xmath139 by formally inverting the linear system ( [ sysl - u ] ) it results @xmath140 where @xmath141 . a direct expression of @xmath142 and its norm could be obtained , for instance , for the first order recursive filter in the homogenus case ( @xmath143 ) . however , in the following , it will be shown that @xmath142 has always bounded norm , i.e. @xmath144 observe that @xmath142 is the matrix operator that substitutes the gaussian operator @xmath49 in ( [ gauss_oper ] ) , then a measure of how well @xmath134 approximates @xmath92 can be derived in terms of the operator distance @xmath145 ideally one would expect that @xmath146 goes to @xmath97 ( and @xmath147 ) as @xmath106 approaches to @xmath148 , yet this does not happen due to the presence of edge effects . in the next sections we will investigate about the numerical behaviour of the distance @xmath149 for some case study and we will show its effects in the cg algorithm . here we are interested to analyze the error introduced on the matrix - vector operation at step 5 . of * algorithm 1 * , when the gaussian rf is used instead of the discrete gaussian convolution . as previously explained , in terms of matrices , this is equivalent to change the matrix operator , then * algorithm 2 * can be rewritten as shown in * algorithm 3*. [ five_rf ] @xmath150 ; + @xmath151 ; + @xmath152 ; + @xmath153 ; [ pcg2 ] now we are able to give the main result of this paper : indeed the following theorem furnishes an upper bound for the error @xmath154 , made at each single iteration @xmath77 of the cg ( * algorithm 1 * ) . this bound involves the operator norms @xmath155 the distance @xmath156 and the error @xmath157 accumulated on @xmath158 at previous iterations . + let be @xmath158 , @xmath159 , @xmath160 , @xmath161 as in * algorithm 2 * and * algorithm 3*. let be @xmath162 and let denote by @xmath163 the difference between values @xmath158 and @xmath164 . then it holds @xmath165 \ \ + \vert { \bf f_n^{(k ) } } \!-\!{\bf v } \vert \!\cdot\ ! \vert{\bf \psi}\vert \!\cdot \!\big(\vert{\bf v } \vert \!+\!\vert { \bf f_n^{(k)}}\vert\big ) \!\cdot\ ! \vert\widetilde{\rho}_k\vert.\end{aligned}\ ] ] * proof : * a direct proof follows by using the values @xmath166 and @xmath167 introduced in algorithm 2 and in algorithm 3 . it holds : @xmath168 then , for the difference @xmath169 , we get the bound @xmath170 hence , for the difference @xmath171 , we obtain @xmath172 in the second - last inequality we used the fact that @xmath173 finally , observing that @xmath174 and taking the upper bound of @xmath175 , the thesis is proved . + @xmath176 previous theorem shows that , at each iteration of the cg algorithm , the error bound on the computed value @xmath177 at step 5 . , is characterized by two main terms : the first term can be considered as the contribution of the standard forward error analysis and it is not significant , if @xmath178 is small ; the second term highlights the effect of the introduction of the rf . more in detail , at each iteration step , the computed value @xmath177 is biased by a quantity proportional to three factors : * the distance between the original operator ( the gaussian operator * v * ) and its approximation ( the operator @xmath179 ) ; * the norm of @xmath78 ; * the sum of the operator norms @xmath180 and @xmath181 . * table 1 * : operator norms + [ cols="<,^,^",options="header " , ] these case study shows that , neglecting the edge effects , the 3-rd rf filter is more accurate the the 1st - rf order with few iterations . this fact is evident by observing the results in figure [ fig : pr ] and the operator norms in table 3 . finally , we remark that the 1-st order rf has to use 100 iteration in order to obtain the same accuracy of the 3-rd order rf . this is a very interesting numerical feature of the third order filter . the theoretical considerations of the previous sections are useful to understand the accuracy improvement in the real experiments on ocean var . the preconditioned cg is a numerical kernel intensively used in the model minimizations . implementing a more accurate convolution operators gives benefits on the convergence of gc and on the overall data assimilation scheme @xcite . here we report experimental results of the 3rd - rf in a global ocean implementation of oceanvar that follows @xcite . these results are extensively discussed in the report @xcite . in real scenarios @xcite scientific libraries and an high performance computing environments are needed . the case study simulations were carried - out on an ibm cluster using 64 processors . the model resolution was about @xmath182 degree and the horizontal grid was tripolar , as described in @xcite . this configuration of the model was used at cmcc for global ocean physical reanalyses applications ( see @xcite ) . the model has 50 vertical depth levels . the three - dimensional model grid consists of 736141000 grid - points . the comparison between the 1st - rf and 3rd - rf was carried out for a realistic case study , where all in - situ observations of temperature and salinity from expendable bathythermographs ( xbts ) , conductivity , temperature , depth ( ctds ) sensors , argo floats and tropical mooring arrays were assimilated . the observational profiles are collected , quality - checked and distributed by @xcite . the global application of the recursive filter accounts for spatially varying and season - dependent correlation length - scales ( clss ) . correlation length - scale were calculated by applying the approximation given in @xcite to a dataset of monthly anomalies with respect to the monthly climatology , with inter - annual trends removed . the analysis increments from a 3dvar applications that uses the 1st - rf with 1 , 5 and 10 iterations and the 3rd - rf are shown in figure 5 with a zoom in the same area of western pacific area as in figure 5 , for the temperature at 100 m of depth . the figure also displays the differences between the 3rd - rf and the 1st - rf with either 1 or 10 iterations . the patterns of the increments are closely similar , although increments for the case of 1st - rf ( k=1 ) are generally sharper in the case of both short ( e.g. off japan ) or long ( e.g. off indonesian region ) clss . the panels of the differences reveal also that the differences between 3rd - rf and the 1st - rf ( k=10 ) are very small , suggesting once again that the same accuracy of the 3rd - rf can be achieved only with a large number of iterations for the first order recursive filter . finally , in [ arxiv ] was also observed that the 3rd - rf compared to the 1st - rf ( k=5 ) and the 1st - rf ( k=10 ) reduces the wall clock time of the software respectively of about 27% and 48% . recursive filters ( rfs ) are a well known way to approximate the gaussian convolution and are intensively applied in the meteorology , in the oceanography and in forecast models . in this paper , we deal with the oceanographic 3d - var scheme oceanvar . the computational kernel of the oceanvar software is a linear system solved by means of the conjugate gradient ( gc ) method . the iteration matrix is related to an error covariance matrix , with a gaussian correlation structure . in other words , at each iteration , a gaussian convolution is required . generally , this convolution is approximated by a first order rf . in this work , we introduced a 3rd - rf filter and we investigated about the main sources of error due to the use of 1st - rf and 3rd - rf operators . moreover , we studied how these errors influence the cg algorithm and we showed that the third order operator is more accurate than the first order one . finally , theoretical issues were confirmed by some numerical experiments and by the reported results in the case study of the oceanvar software . 99 m. abramowitz , i. stegun - _ handbook of mathematical functions_. dover , new york , 1965 . m. belo pereira , l. berre - _ the use of an ensemble approach to study the background - error covariances in a global nwp model_. mon . 2466 - 2489 , 2006 . c. cabanes , a. grouazel , k. von schuckmann , m. hamon , v. turpin , c. coatanoan , f. paris , s. guinehut , c. bppne , n. ferry , c. de boyer montgut , t. carval , g. reverding , s. puoliquen , p.y . l. traon - _ the cora dataset : validation and diagnostics of in - situ ocean temperature and salinity measurements_. ocean sci 9 , pp . 1 - 18 , 2013 . s. cuomo , a. galletti , g. giunta and a. starace-_surface reconstruction from scattered point via rbf interpolation on gpu _ , federated conference on computer science and information systems ( fedcsis ) , 2013 , pp . 433 - 440 . g. dahlquist and a. bjorck - _ numerical methods_. prentice hall , 573 pp . j. derber , a. rosati - _ a global oceanic data assimilation system_. journal of phys . 1333 - 1347 , 1989 . l. d amore , r. arcucci , l. marcellino , a. murli- _ hpc computation issues of the incremental 3d variational data assimilation scheme in oceanvarsoftware_. journal of numerical analysis , industrial and applied mathematics , 7(3 - 4 ) , pp 91 - 105 , 2013 . r. deriche - _ separable recursive filtering for efficient multi - scale edge detection_. proc . workshop machine vision machine intelligence , tokyo , japan , pp 18 - 23 , 1987 s. dobricic , n. pinardi - _ an oceanographic three - dimensional variational data assimilation scheme_. ocean modeling 22 , pp 89 - 105 , 2008 . r. farina , s. cuomo , p. de michele , f. piccialli-_a smart gpu implementation of an elliptic kernel for an ocean global circulation model _ , applied mathematical sciences , 7 ( 61 - 64 ) , 2013 pp.3007 - 3021 . r. farina , s. dobricic , s. cuomo-_some numerical enhancements in a data assimilation scheme _ , aip conference proceedings 1558 , 2013 , doi : 10.1063/1.4826017 . n. ferry , b. barnier , g. garric , k. haines , s. masina , l. parent , a. storto , m. valdivieso , s. guinehut , s. mulet - _ nemo : the modeling engine of global ocean reanalysis_. mercator ocean quaterly newsletter 46 , pp 60 - 66 , 2012 . s. haben , a. lawless , n. nichols - _ conditioning of the 3dvar data assimilation problem_. university of reading , dept . of mathematics , math report series 3 , 2009 ; s. haben , a. lawless , n. nicholos - _ conditioning and preconditioning of the variational data assimilation problem_. computers and fluids 46 , pp 252 - 256 , 2011 . l. haglund - _ adaptive multidimensional filtering_. linkping university , sweden , 1992 . lorenc - _ iterative analysis using covariance functions and filters_. quartely journal of the royal meteorological society 1 - 118 , pp 569 - 591 , 1992 . lorenc - _ development of an operational variational assimilation scheme_. journal of the meteorological society of japan 75 , pp 339 - 346 , 1997 . g. madec , m. imbard - _ a global ocean mesh to overcome the north pole singularity_. clim . dynamic 12 , pp 381 - 388 , 1996 . purser , w .- s . parish , n.m . roberts - _ numerical aspects of the application of recursive filters to variational statistical analysis . part ii : spatially inhomogeneous and anisotropic covariances_. monthly weather review 131 , pp 1524 - 1535 , 2003 . c. hayden , r. purser - _ recursive filter objective analysis of meteorological field : applications to nesdis operational processing_. journal of applied meteorology 34 , pp 3 - 15 , 1995 . a. storto , s. dobricic , s. masina , p. d. pietro - _ assimilating along - track altimetric observations through local hydrostatic adjustments in a global ocean reanalysis system_. mon . 139 , pp 738 - 754 , 2011 . vliet , i. young , p. verbeek - _ recursive gaussian derivative filters_. international conference recognition , pp 509 - 514 , 1998 . van vliet , p.w . verbeek - _ estimators for orientation and anisotropy in digitized images_. proc . asci95 , heijen ( netherlands ) , pp 442 - 450 , 1995 . a. t. weaver , p. courtier - _ correlation modelling on the sphere using a generalized diffusion equation_. quarterly journal of the royal meteorological society 127 , pp 1815 - 1846 , 2001 . a. witkin - _ scale - space filtering_. proc . joint conf . on artificial intelligence , karlsruhe , germany , pp 1019 - 1021 , 1983 . young , l.j . van vliet - _ recursive implementation of the gaussian filter_. signal processing 44 , pp 139 - 151 , 1995 .

computational kernel of the three - dimensional variational data assimilation ( 3d - var ) problem is a linear system , generally solved by means of an iterative method . the most costly part of each iterative step is a matrix - vector product with a very large covariance matrix having gaussian correlation structure . this operation may be interpreted as a gaussian convolution , that is a very expensive numerical kernel . recursive filters ( rfs ) are a well known way to approximate the gaussian convolution and are intensively applied in the meteorology , in the oceanography and in forecast models . in this paper , we deal with an oceanographic 3d - var data assimilation scheme , named oceanvar , where the linear system is solved by using the conjugate gradient ( gc ) method by replacing , at each step , the gaussian convolution with rfs . here we give theoretical issues on the discrete convolution approximation with a first order ( 1st - rf ) and a third order ( 3rd - rf ) recursive filters . numerical experiments confirm given error bounds and show the benefits , in terms of accuracy and performance , of the 3-rd rf .

in the last few years there has been another wave of excitement regarding the question of neutrino masses . this is largely due to the many new experiments testing neutrino oscillations , most notably the positive indications obtained by super kamiokande on atmospheric neutrino oscillations @xcite . similar indications come from other experiments @xcite . the solar neutrino experiments have for many years provided independent evidence for neutrino oscillations @xcite . accelerator and reactor experiments have also played an important role . they have furnished strict bounds on neutrino oscillation parameters @xcite . in the case of the lsnd experiment @xcite at los alamos evidence for @xmath2 oscillation has been reported . see refs @xcite for recent reviews . it is hoped that new experimental results can be used to determine the neutrino squared mass differences and mixing angles . in turn , these may help to infer the neutrino mass matrix . this is presumably a possible gateway to a more fundamental theory beyond the standard model . of course this is a highly speculative area , and even though there are many imaginative proposals @xcite , it seems fair to say that the the true answer is essentially unknown . in order to make progress in this direction , it seems useful to investigate various plausible ansatze for the neutrino mass matrix . from this point of view we propose the ansatz for the 3 generation neutrino mass matrix , @xmath3 : @xmath4 and investigate its consequences . we are considering the neutrinos to be represented by 2-component spinors so that , in the most general situation , @xmath3 is an arbitrary symmetric complex matrix . as we will see in section ii , eq . ( 1.1 ) can be motivated from an so(10 ) grand unified model @xcite , in which it may be derived with some assumptions . physically , eq . ( 1.1 ) corresponds to the well known approximate signature of grand unification that @xmath5 . furthermore we will see in sections iv and v that eq . ( 1.1 ) can be straightforwardly combined with experimental information to get an idea of the neutrino masses themselves as well as the `` texture '' of @xmath6 . relevant matters of notation are discussed in section iii while a summary is presented in section vi . in the so(10 ) grand unification model each generation contains one light massive two component neutrino and also a very heavy one which is `` integrated out '' according to the `` seesaw mechanism '' @xcite . the effective @xmath7 neutrino mass matrix takes the form : @xmath8 where @xmath9 , @xmath10 and @xmath11 are respectively the mass matrices of the light neutrinos , heavy neutrinos and heavy - light mixing ( or `` dirac matrix '' ) . generally the second , seesaw , term is considered to dominate . here however we shall assume the first term to be the dominant one . this is necessary for the present derivation of eq . ( 1.1 ) to hold . also , a rough order of magnitude estimate for the second term would be @xmath12 or about @xmath13 ev . thus , the seesaw term could be negligible if neutrino masses turn out to be appreciably larger than this value . now in so(10 ) , higgs mesons belonging to the 10 , 120 and 126 representations can contribute to the fermion masses at tree level . one has @xcite for the down quark , charged lepton and light neutrino mass matrices , @xmath14 where @xmath15 , @xmath16 , @xmath17 , @xmath18 , @xmath19 are numbers representing higgs meson vacuum values . s(10 ) , a(120 ) and s(126 ) are the matrices of the yukawa type constants which couple the fermions to the 10 , 120 and 126 higgs mesons respectively ; the matrices s(10 ) and s(126 ) must be symmetric while a(120 ) is antisymmetric . finally , @xmath20 is a renormalization factor for comparing the quark masses with the charged lepton masses at a low energy scale rather than at the grand unified scale ; @xmath21 is a similar factor for the neutrino masses . with the stated assumption that the @xmath22 term dominates in eq . ( 2.1 ) we get @xmath23 which clearly also holds when any number of 10 s or 120 s are present but only a single 126 . the matrices appearing in eq . ( 2.3 ) are so far essentially unrestricted complex ones . to proceed , we make the further assumption that the matrices are hermitian . then @xmath24 and @xmath25 may each be brought to diagonal form by unitary transformations . thus the right hand side of eq . ( 2.3 ) may be evaluated to yield approximately , @xmath26 according to a well known numerical success , based on the observation that @xmath27 , of grand unification @xcite . note that we have not needed to assume that the mass matrix has any zero elements . where , in addition , a special combined fritzsch - stech ansatz was used . here we are not making any special ansatz of this type for the mass matrices . ] even if the cancellation on the right hand side of eq . ( 2.4 ) is not perfect , it should still be a good approximation . in an so(10 ) model where the mass matrices are hermitian , @xmath28 will be real symmetric . we will investigate this case and also the possibility that the more general case holds . our plan is to combine the ansatz eq . ( 1.1 ) with experimentally obtained results on neutrino oscillations in order to learn more about @xmath29 itself . for this purpose it may be helpful to set down our notation @xcite for the pieces of the effective @xmath30 theory involving neutrinos and to make some related remarks . the free lagrangian containing three two component massive fields is : @xmath31 where @xmath32 is the ( not yet diagonalized ) neutrino mass matrix of the underlying theory to be identified with the matrix in eq . note that we are free to multiply the first mass term in eq . ( [ lfree ] ) by an overall arbitrary phase which is a matter of convention . it is possible@xcite to find a unitary matrix @xmath33 which brings @xmath0 to real , positive , diagonal form in the following way : @xmath34 the mass diagonal fields @xmath35 are then @xmath36 similarly , the column vector of left handed negatively charged leptons in the underlying theory , @xmath37 is related to the mass diagonal fields @xmath38 by @xmath39 where @xmath40 . combining factors from eq . ( 3.3 ) and eq . ( 3.4 ) we obtain the unitary mixing matrix , @xmath41 for the charged current weak interaction , @xmath42 this appears in the lagrangian term , @xmath43 where a conventional four component dirac notation with @xmath44 diagonal is being employed and @xmath35 has only the first two components non zero . next we parameterize @xmath41@xcite . it is possible to restrict @xmath45 by adjusting an overall phase which can be absorbed in @xmath46 . then we write @xmath47 where @xmath48 with @xmath49 and , for example @xmath50 . \label{3.9}\ ] ] eq . ( 3.7 ) contains the eight parameters needed to characterize an arbitrary unitary unimodular matrix . from the standpoint of eq . ( 3.6 ) it can be further simplified by using the freedom to rephase @xmath51 without changing the free part of the charged lepton lagrangian . on the other hand , the form of the mass terms in eq . ( [ lfree ] ) shows that the neutrino fields can not be rephased . thus a suitable minimal parameterization for @xmath41 in ( 3.6 ) is @xmath52 involving three angles " , @xmath53 and three phases " , @xmath54 . note the identity @xmath55 this identity may be used to transfer two of the phases @xmath54 in eq . ( 3.7 ) to a diagonal matrix on the right of @xmath41 as , for example , @xmath56 where @xmath57 , which may be used instead of eq . ( 3.10 ) . we also need the formula for the amplitude of neutrino oscillation . for the case when a neutrino , produced by a charged lepton of type @xmath15 , oscillates " to make at time @xmath58 , a charged lepton of type @xmath16 , we have @xmath59 where the sum goes over the neutrinos of definite mass , @xmath60 . inserting the parameterization eq . ( 3.12 ) into eq . ( 3.13 ) shows that the effect of the factor @xmath61 cancels out . thus for ordinary oscillations , @xmath41 is parameterized by three angles and one cp violating phase as for the ckm quark mixing matrix . on the other hand , the two additional cp violating phases @xmath62 and @xmath63 show up if one considers neutrino - antineutrino oscillations @xcite or neutrinoless double beta decay @xcite . the formula for the probability , @xmath64 is gotten by taking the squared magnitude of eq . ( 3.13 ) and replacing the exponential factor @xmath65 by @xmath66 , where @xmath67 is the neutrino energy and @xmath68 is the oscillation distance . for practical reasons it is very important to take account of the experimental uncertainties in @xmath67 and @xmath68 . the simplest approximation@xcite is to define @xmath69 and assume that one can smear @xmath64 with a gaussian distribution in @xmath16 . @xmath70 is defined as the mean value and @xmath71 as the standard deviation appropriate to the particular physical setup . then we find for the smeared probability @xmath72 , \label{3.14}\end{aligned}\ ] ] where @xmath73 and @xmath74 . notice that when @xmath75 , @xmath76 is independent of the sign of @xmath77 . since @xmath78 provides only two real equations for 12 real parameters , it is clear that it has a relatively small amount of predictivity . in particular it can not say much about the texture ( e.g. possible zeroes ) of @xmath0 which is suppposed to derive from a deeper theory than the standard model . on the other hand , we shall see that our ansatz is complementary to the results which should emerge from analysis of neutrino oscillation experiments . together , they should enable us to actually ( with some conditions ) reconstruct @xmath0 . first , in this section , we shall consider @xmath0 to be hermitian so that the argument in favor of @xmath78 presented in section ii holds without any further assumptions . since @xmath0 is symmetric it must be real . it can be brought to real diagonal form via a real rotation @xmath79 as @xmath80 . however there is no guarantee that all eigenvalues of @xmath0 will be positive . we can make them all positive by rephasing the diagonal fields with negative eigenvalues by a factor @xmath81 ( see eqs . ( 3.1 ) and ( 3.2 ) ) . this means that the general diagonalizing matrix @xmath33 in eq . ( 3.2 ) now takes the form @xmath82 where @xmath83 with @xmath84 for a positive eigenvalue and @xmath85 for a negative eigenvalue . we notice that eq . ( 4.1 ) is of the form eq . ( 3.12 ) for which we already noticed that the factor @xmath86 cancels out in the neutrino oscillation formula eq . ( 3.14 ) . furthermore only the square of the mass is relevant in eq . ( 3.14 ) . thus we choose to work in this section with some negative masses and no factor @xmath86 in eq . ( 4.1 ) . to avoid confusion , we remark that the factor @xmath86 in eq . ( 4.1 ) does _ not _ introduce any cp violation in the theory @xcite since @xmath0 is real in any event . now let us suppose that an experimental analysis of all neutrino oscillation experiments is made based on a formula like eq . ( 3.14 ) ( or one which treats the experimental uncertainties in a more sophisticated way ) . furthermore assume that the cp violating phase @xmath87 in eq . ( 3.12 ) is negligible . then we should know the _ magnitudes _ of the squared neutrino mass differences @xmath88 where @xmath89 and @xmath90 can be either positive or negative . then , assuming the leptonic theory to be cp conserving , our ansatz would imply @xmath91 where eq . ( 3.2 ) was used . ( 4.2 ) and ( 4.3 ) comprise three equations for the three neutrino masses @xmath92 , @xmath93 and @xmath94 . we can solve to get : @xmath95 , \\ \nonumber m_2 & = & \frac { { b } - { \left ( m_1 \right)}^2}{2m_1 } , \\ \nonumber m_3 & = & - m_1 - m_2 . \label{4.4}\end{aligned}\ ] ] this leads to a limited number of solutions , depending on sign choices . if we make the further assumption that the charged lepton mixing matrix @xmath96 in eq . ( 3.4 ) is approximately the unit matrix ( this is expected to be a reasonable but not perfect approximation ) we can identify @xmath97 which would be obtained from experiment . then , using the masses found in eq . ( 4.4 ) , we could reconstruct @xmath0 as @xmath98 to proceed , we need only insert the experimental results for @xmath89 , @xmath90 and @xmath41 in ( 4.4 ) and ( 4.5 ) . of course , it is presumably the task of the next decade to solidify the experimental determination of these quantities . we can , at the moment , only give a preliminary discussion . for this purpose we will use the results of a recent preliminary analysis of all neutrino experiments by ohlsson and snellman @xcite . these authors found , by a least square analysis , a best fit for ( our notation ) @xmath99 , @xmath100 and the leptonic mixing matrix @xmath41 . they used the formula eq . ( 3.14 ) with a suitable choice of @xmath70 and @xmath101 for each experiment . furthermore , they made the simplifying assumption that @xmath41 is real . finally they only searched for a fit in the range @xmath102 , @xmath103 . this range corresponds to mass difference choices for which the msw effect @xcite for solar and atmospheric neutrinos is not expected to be important and so greatly simplifies the analysis . thus there is no guarantee that the solution of @xcite is unique . altogether they fit sixteen different solar neutrino , atmospheric neutrino , accelerator and reactor experiments , including lsnd . the best fit is : @xmath104 for the squared mass differences and @xmath105 \label{4.7}\ ] ] for the lepton mixing matrix @xmath41 . as discussed above we will identify @xmath106 here , keeping @xmath41 real but allowing for negative masses . the best fit matrix k was obtained to be similar but not identical to the `` bimaximal mixing '' matrix @xcite . with the best fit squared mass differences in eq . ( 4.6 ) , our model predicts , from the first of eq . ( 4.4 ) , eight different possibilities . these correspond to four different sign configurations for @xmath89 and @xmath90 times the two possible signs for @xmath92 . however , only two of these eight are essentially different ; these are @xmath107 the other solutions correspond to interchanging whichever of @xmath108 and @xmath109 is greater ( which has only a negligible effect since they are almost degenerate ) or reversing the signs of all masses . physically it is clear what is happening : the smallness of @xmath110 compared to @xmath111 in eq . ( 4.2 ) forces @xmath112 . then we have either @xmath113 with , using the constraint eq . ( 4.3 ) , @xmath114 or @xmath115 with @xmath94 very small . since we have assumed the neutrinos to be of majorana type for our plausibility argument in section ii , their interactions will violate lepton number . then they should mediate neutrinoless double beta decay @xmath116 @xcite . such a process has not yet been observed and an upper bound has been set for the relevant quantity @xmath117 the best upper bound at present is @xcite @xmath118 ev , reflecting some uncertainty in the estimation of the needed nuclear matrix elements . substituting the best fit for the matrix k from eq . ( 4.7 ) together with our results in eqs . ( 4.8 ) and ( 4.9 ) into eq . ( 4.10 ) yields predictions for the two cases : @xmath119 both solutions seem to be acceptable , the type i case marginally but the type ii case definitely . note that the small value for @xmath120 in the type ii case is due to the best fit prediction @xcite @xmath121 and also to the fact that @xmath93 is negative . the same value would clearly result if we made @xmath93 positive and set @xmath122 as discussed around eq . ( 4.1 ) above . finally , let us reconstruct the underlying neutrino mass matrices for each of the two cases . we use eq . ( 4.5 ) based on the assumption that @xmath0 is real and also our ansatz to find ( in units of ev ) : @xmath123 , \\ { \rm type } \hskip 0.2 cm { \rm ii } : \quad m_\nu & = & m_\nu^t \approx \left [ \begin{array } { c c c } 6.7 \times 10^{-5 } & -0.9199&0.5078\\ -0.9199&0.0654&0.0410\\ 0.5078&0.0410&-0.0654\\ \end{array } \right ] . \label{4.13}\end{aligned}\ ] ] the type i matrix does not have an excellent candidate for a `` texture '' zero . however the small value of @xmath124 in the type ii case is certainly suggestive . these matrices lead to neutrino masses and a mixing matrix which give a best fit to all present data . it will be interesting to see if either of them hold up in the future . incidentally , on comparing eqs . ( 4.13 ) and ( 4.14 ) it is amusing to observe the large difference in two mass matrices `` generated '' in the same way except with respect to how @xmath125 is satisfied . it seems interesting to also investigate the ansatz @xmath126 when @xmath0 is no longer restricted to be real . this also raises the problem of constructing the unitary diagonalizing matrix @xmath33 in eq.(3.2 ) , in terms of the experimentally measured lepton mixing matrix @xmath127 . for simplicity , as before , we will make the approximation that the charged lepton diagonalizing matrix @xmath96 is the unit matrix . apart from an overall ( conventional ) phase we may write @xmath128 where the 2-parameter quantity @xmath129 was defined in eq . ( 3.8 ) . since @xmath127 has four parameters @xmath33 in eq . ( 5.1 ) is described by eight parameters . as mentioned before , the two parameters in @xmath61 are not measurable in neutrino oscillation experiments but show up when one considers @xmath130 . the two parameters in @xmath131 may be eliminated , for experimental purposes , by rephasing the charged leptons . however , for the theoretical purpose of reconstructing the underlying neutrino mass matrix @xmath0 , their existence can not be ruled out . ( they also do not contribute to @xmath132 . ) for the purpose of relating the ansatz on @xmath0 to the physical neutrino masses in @xmath133 , we note @xmath134 for further simplicity of the analysis we adopt the special case @xmath135 and also identify @xmath127 with the real best fit in eq . ( 4.7 ) ; our ansatz now reads @xmath136 with the redefinitions @xmath137 and @xmath138 , eq . ( 5.3 ) becomes @xmath139 this may be conveniently visualized as the vector triangle shown in fig . [ trianglefig ] . combining eqs . ( 5.4 ) and ( 4.2 ) gives four real equations for the five unknown quantities ( @xmath140 ) . thus we have ( for each set of ( @xmath141 ) sign choices ) a one parameter family of solutions . it is convenient to choose this parameter to be @xmath94 . then @xmath92 and @xmath93 may be found from the equations ( 4.2 ) , provided that solutions exist . in this way all three sides of the triangle in fig.[trianglefig ] are determined . the angles may finally be found as @xmath142 we also need to investigate the constraint arising from the non - observation of @xmath132 . ( 4.10 ) now becomes , with eq . ( 5.1 ) as the mixing matrix @xmath143 using the ansatz constraint eq . ( 5.4 ) , eq . ( 5.6 ) may be rewritten as @xmath144 m_2 e^{-i\beta_2 } + \left [ ( k_{exp\hskip .2 cm 13})^2 - ( k_{exp\hskip .2 cm 11})^2 \right ] m_3 \right|.\ ] ] this form is very convenient when identifying @xmath127 with the best fit solution in eq . ( 4.7 ) . in the present context such an identification corresponds to cp violation for the @xmath132 process but not for usual neutrino oscillations . since the ( 11 ) and ( 12 ) matrix elements are equal in eq . ( 4.7 ) we find the simple result @xmath145 thus if the upper bound on @xmath146 is conservatively identified as in the @xmath147 ev range , we should have in this case @xmath148 in the present complex case there is a continuum of possible solutions labelled by those values of @xmath94 satisfying eq . ( 5.9 ) , rather than just the two possibilities found in eq . ( 4.8 ) and eq . ( 4.9 ) . actually , the continuum separates roughly into two classes similar to either eq . ( 4.8 ) or eq . ( 4.9 ) . in the generalized type i class , @xmath94 is of the order @xmath149 ev while @xmath92 and @xmath93 are related to nearly oppositely directed vectors in fig . [ trianglefig ] and are also of the order @xmath90 . in the generalized type ii class , @xmath90 is negative ; @xmath92 and @xmath93 correspond to vectors of order @xmath150 which are oppositely directed to each other , while @xmath94 ranges from very small to order @xmath100 . given the bound eq . ( 5.9 ) from the non - observation of @xmath130 , there are important limitations on the allowed @xmath94 values for type i solutions . in this case @xmath90 is positive so the equation @xmath151 will only allow solutions for @xmath152 ev . this range is barely compatible with eq . thus the type ii case where @xmath153 and @xmath154 seems most probable . as an example of a solution for complex @xmath0 , consider choosing @xmath155 ev . then a solution is obtained with ( compare with fig . [ trianglefig ] ) @xmath156 the matrix @xmath33 , which diagonalizes @xmath0 is obtained from eq . ( 5.1 ) , with the approximation @xmath157 , and with now : @xmath158 this factor introduces cp violation in the @xmath132 process but not in ordinary neutrino oscillations . finally the underlying neutrino mass matrix , @xmath159 is `` reconstructed '' as ( in units of ev ) : @xmath160.\ ] ] this is structurally similar to the real type ii solution displayed in eq . ( 4.14 ) , although the suppression of the ( 11 ) element is not so pronounced . notice that @xmath94 is considerably smaller than the almost degenerate pair @xmath92 and @xmath93 . furthermore @xmath92 and @xmath93 are large enough to possibly play some role in astrophysics . we investigated the ansatz @xmath78 for the underlying ( pre - diagonal ) three generation neutrino mass matrix . it was motivated by noting that in an so(10 ) grand unified model where @xmath0 was taken to be real ( cp conserving ) , it corresponds to the well known unification of b quark and @xmath161 lepton masses . while not very predictive by itself it yields information complementary to what would be gotten from a complete three flavor analysis of all lepton number conserving neutrino oscillation experiments . specifically from the specification of the magnitudes of two neutrino squared mass differences and also of the leptonic mixing matrix we can , with some assumptions , find the neutrino masses themselves and `` reconstruct '' @xmath0 . this determination can be sharpened by consideration of the constraints imposed by non - observation of neutrinoless double beta decay . for the purpose of testing our ansatz we employed the results of a reasonable best fit to all present neutrino experiments ( including lsnd ) by ohlsson and snellmann @xcite . this fit will inevitably be improved in the next few years as new experiments are completed . they were able to fit the data without assuming any cp violation . this agrees with assuming @xmath0 to be real . we found two essentially different solutions in that case . the first features two neutrinos having approximately equal mass 0.608 ev and a third neutrino of mass 1.217 ev . this solution is on the borderline of being ruled out by non - observation of @xmath162 . the second solution has two neutrinos with approximately degenerate mass 1.054 ev and a third neutrino with a mass @xmath163 ev . this solution is very safe from being ruled out by @xmath164 experiments . it also features a reconstructed @xmath0 which has an extremely small ( 11 ) element . note that , for both solutions , even though @xmath0 is real there are some ( cp conserving ) factors of @xmath81 in the mixing matrix when all masses are taken to be positive . alternatively one may have no @xmath81 s in the mixing matrix while allowing some masses to be negative . the latter form is useful for seeing intuitively how @xmath165=0 is possible . the case of matching the above best fit data to a complex @xmath0 was also considered . in this situation there are cp violating phases in the lepton mixing matrix which affect the @xmath132 process but do not affect ordinary total lepton number conserving neutrino oscillations . such phases could also be measurable in principle with the observation of a decay like @xmath166 . the case of complex @xmath0 allows a larger number of solutions . with a simplifying assumption there is a one parameter family of allowed neutrino mass sets . roughly , these fall into one of the two types already encountered for real @xmath0 . a question of some interest is whether the neutrinos are massive enough to play a role in cosmology . the relevant criterion @xcite for this to occur is usually stated as @xmath167 ev . for the type ii solutions with complex @xmath0 we have found the largest mass sum to be about 4.5 ev corresponding to @xmath168 ev and @xmath169 ev . however this is on the very border of acceptability for non observation of @xmath132 . super - kamiokande collaboration , y. fukuda _ et al . * 81 * , 1562 ( 1998 ) . super - kamiokande and kamiokande collaborations , t. kajita , in _ proceedings of the xviiith international conference on neutrino physics and astrophysics ( neutrino 98 ) _ , takayama , japan , 1998 [ nucl . phys . b ( proc . suppl . ) * 77 * , 123 ( 1999 ) ] . soudan collaboration , w.w.m . et al . _ , b * 391 * , 491 ( 1997 ) ; kamiokande collaboration , y. fukuda _ et al . lett . * 77 * , 1683 ( 1996 ) ; chooz collaboration , m. apollonio _ et al . _ , b * 420 * , 397 ( 1998 ) ; chooz collaboration , c. bemporad , in proceedings of the xviiith international conference on neutrino physics and astrophysics ( neutrino 98 ) , nucl . b ( proc . suppl . ) * 77 * 159 ( 1999 ) ; e. k. akhmedov , hep - ph/0001264 ; j. n. bahcall , nucl . . suppl . * 48 * , 309(1996 ) ; s. bilenky , hep - ph/990246 ; p. ficher , b. kayser and k. s. mcfarland , anual review of nuclear and particle physics , * 49 * ( 1999 ) ; d. r. o. morrison , part . world * 3 * , 30(1992 ) ; a. rubbia , acta . b * 30 * , 2351(1999 ) ; a. y. smirnov , hep - ph/9901208 ; j. w. f. valle , hep - ph/9911224 , hep - ph/9906539 ; j. d. vergados , phys . * 133 * ( 1986 ) . c. h. albright and s. m. barr , phys . lett . b**461 * * , 218(1999 ) ; v. barger , s. pakvasa , t. j. weiler and k.whisnant , phys . lett . b**437 * * , 107(1998 ) ; h. b. benaoum and s. nasri ; phys . rev . d**60 * * , 113003(1999 ) ; m. jezabek and y. sumino , phys . lett . b**457 * * , 139(1999 ) ; s. k. kang and c. s. kim , phys . rev . d**59 * * , 091302(1999 ) ; r. n. mohapatra and s. nussinov , phys . d**60 * * , 013002(1999 ) , r. n. mohapatra , a. perez - lorenzano and c .a . de sousa pires , phys . b**474 * * , 355(2000 ) ; h. georgi and s. l. glashow , phys . rev . d*61 * , 097301 ( 2000 ) . t. yanagida , proc . of the workshop on unified theory and baryon number in the universe , ed . by o. sawada and a. sugamato ( kek report 79 - 18,1979 ) , p 95 ; m. gell - mann , p. ramond and r. slansky in supergravity , eds p. van niewenhuizen and d. z. freedman ( north holland , 1979 ) ; r. n. mohapatra and g. senjanovic , phys . * 44 * , 912 ( 1980 ) . see for example k. matsuda , t. fukuyama and h. nishiura , phys . d * 61 * , 053001(2000 ) ; a. bottino , c. w. kim , h. nishiura and w. k. tse , phys . d * 34 * , 862 ( 1986 ) ; k. s. babu and r. n. mohapatra , phys . rev * 70 * , 2845 ( 1993 ) .

we propose a simple ansatz for the three generation neutrino mass matrix @xmath0 which is motivated from an so(10 ) grand unified theory . the ansatz can be combined with information from neutrino oscillation experiments and bounds on neutrinoless double beta decay to determine the neutrino masses themselves and to reconstruct , with some assumptions , the matrix @xmath1 . = 17pt # 1=#1pt = by -by -makecaption#1#2 tempboxa tempboxa > = = # 1 # 2 to = 13pt su-4240 - 719 + hep - ph/0004105 + * complementary ansatz for the neutrino mass matrix * = 12pt deirdre black amir h. fariborz + 0.5 cm salah nasri joseph schechter + _ department of physics , syracuse university , syracuse , ny 13244 - 1130 , usa . _ + 0.5 cm * abstract *

when matter is exposed to intense laser fields , high harmonics ( hhs ) of the incident radiation may be produced . usually , only odd harmonics are obtained even when the laser pulses are short ( for theoretical and experimental work which demonstrates this see @xcite and @xcite , respectively ) . since the duration of the pulse in time is inversely proportional to its width in energy space , one may find this result surprising , as one may expect to obtain also a large distribution of frequencies in the scattered field . why are only odd harmonics obtained even when the laser pulses are short ? for cw lasers ( and symmetric field - free potential ) using the non - hermitian floquet theory it was proved that only odd harmonics are obtained when the dynamics is controlled by a single resonance floquet quasienergy ( qe ) state @xcite,@xcite . when laser pulses are used it was argued that this proof holds since usually the populated resonance states are associated with very different lifetimes and the dynamics is controlled by the resonance state which has the longest lifetime . however , this argument may hold only when the duration of the laser pulses is large enough to enable the decay of the short lived resonances . indeed numerical simulations showed that the harmonic generation spectra ( hgs ) as obtained from a single non - hermitian ( complex scaled ) resonance floquet state is in a remarkable agreement with the results obtained from conventional ( i.e. , hermitian ) time dependent simulation @xcite . the question that is addressed in this work is weather an analytical criteria for the shape and duration of the laser pulse for which the system is controlled by a single resonance floquet state can be given . it is obvious that the question regarding the possibility of population of a single resonance state is connected with the question regarding the degree of adiabaticity of the process . the question is therefore under which conditions can a short laser pulse be defined as adiabatic one . the answer to this question is important not only to harmonic generation ( hg ) studies but also for other , more general studies where lasers are used to control the dynamics , for example adiabatic stirap procedures @xcite . in order to answer this question we use the ( t , t ) formalism @xcite together with the non - hermitian quantum mechanics ( nhqm ) formalism . the use of nhqm formalism to describe the dynamics of atoms / molecules subjected to cw laser fields is essential , since only then can the dynamics be described in terms physical , square integrable , resonance floquet states . otherwise , the description of the dynamics in terms of hermitian floquet states results in very little physical insight on the problem , as well as numerical problems , not to mention that it is limited to the description of bound systems only . in the hermitian case the spectrum is continuous and becomes discrete only due to the use of finite box quantization ; moreover , a single floquet state in hermitian qm can not describe neither the resonance phenomena nor the field ionization phenomena @xcite . our strategy is as follows . in section 2 we give a brief review of the formalism used in our derivation [ namely the ( t , t ) formalism for hermitian and non - hermitian hamiltonians ] . in section 3 we introduce the new derivation of the adiabatic theorem for open quantum systems in strong laser pulses . in section 4 we apply the adiabatic theorem as derived in section 3 to a test - case model hamiltonian which describes a 1d xe atom subjected to a sin - square pulse of monochromatic laser . in section 4 we conclude . the ( t , t ) formalism enables one to obtain analytical solutions for any time - dependent schrdinger equation ( tdse ) with time - dependent hamiltonians . the formalism rests on lifting of the tdse to an extended hilbert space , propagation of the wavefunction there and finally projecting back to the physical hilbert space . the solution of a general tdse is given as usual by : @xmath0 with the initial condition @xmath1 and the vector operator @xmath2 describes the internal degrees of freedom . it has been shown by peskin and moiseyev that by regarding time as an extra coordinate @xmath3 , one can obtain another schrdinger equation with a time independent hamiltonian in the extended hilbert space @xmath4 , whose solution @xmath5 has an analytical time dependence ( given by the analytical time evolution operator associated with time - independent hamiltonians ) @xcite . our desired wavfunction @xmath6 can be deduced from this wavfunction by a simple operation ( that will be shown later ) . the advantage is that efficient propagation schemes designed for solving the tdse with time - independent hamiltonians could then be used also for time dependent hamiltonians , thus releasing one from the difficulties associated with time ordering . let us define the following floquet - type operator @xmath7 where @xmath3 should be regarded as a * coordinate * and @xmath8 is the final time of propagation . when the hamiltonian @xmath9 is time periodic with period @xmath10 , the operator is the floquet operator and @xmath11 . provided that correct boundary conditions are chosen for the @xmath3 coordinate , this operator is hermitian since it is the sum of two hermitian operators . this operator has eigenstates and eigenvalues which are given by the eigenvalue equation : @xmath12 and the set of eigenstates is complete in the extended hilbert space @xmath4 with respect to the inner product @xmath13 say the following tdse with time - independent hamiltonian need to be solved with the initial state @xmath14 : @xmath15 using the definition of @xmath16 this equation reads @xmath17 by setting the cut @xmath18 on eq.[eq6 ] one gets : @xmath19|_{t'=t}=h(\mathbf{r},t ) [ \overline{\psi}(\mathbf{r},t',t)|_{t'=t}]=i\hbar \frac{\partial}{\partial t } [ \overline{\psi}(\mathbf{r},t',t)|_{t'=t}]\label{eq7}\ ] ] where the following property @xmath20= [ ( \frac{\partial}{\partial t}+\frac{\partial}{\partial t ' } ) \overline{\psi}(\mathbf{r},t',t)]|_{t'=t}\label{eq8}\ ] ] ( which holds true for * any * function of @xmath3 and @xmath21 . the relation between the solution of eq.[eq7 ] and the solution of the original tdse ( eq.[eq1 ] ) is given by @xmath22 provided that these two differential equations have the same initial condition . hence , @xmath23 it is seen from eq.[eq10 ] that apparently the initial condition is nt unique . while this holds true in case that one is interested only in the physical wavefunction @xmath6 , it should be noted that if one wishes to calculate physical quantities in the extended hilbert space using the function @xmath5 and then go back to the original hilbert space , and get the correct results , the initial condition in the extended hilbert space should behaves as a delta - function in @xmath3 . the correct initial condition will therefore be @xmath24 . the main advantage of the ( t , t ) formalism from a numerical point of view is that it enables the use of an analytical expression for the time evolution operator , without the necessity of time ordering , even when the hamiltonian is strongly time - dependent . any tdse with time - dependent hamiltonian could be replaced by a different tdse , with time - independent hamiltonian , for which a greater number of accurate integration schemes exist and the solution is given formally by @xmath25 ) . the price one pays however is that the new tdse need to be integrated over one more dimension . the main advantage of the _ ( t , t ) _ formalism from a conceptional point of view is that it enables to describe any time - dependent dynamics in terms of stationary eigenstates and eigenvalues . so far the derivation has been carried out within the framework of the conventional ( i.e. , hermitian ) quantum mechanics . using the complex - scaling ( cs ) transformation @xcite @xmath26 the quasienergy spectrum of the floquet hamiltonian becomes complex and square - integrable resonance states , that were embedded in the continuum in the unscaled problem , are uncovered . for sake of simplicity we drop the index @xmath27 in all @xmath27-dependent expressions ( operators , eigenvalues , eigenvectors etc . ) in the proceeding text . the floquet operator @xmath28 can be represented with the orthogonal fourier basis set @xmath29 as a square matrix @xmath30_{n',n}=\frac{1}{t}\int_{0}^{t}dt'e^{-i\omega n ' t ' } h_f(\mathbf{r},t')e^{-i\omega n t } \label{eq11}\ ] ] where @xmath31 . the left and right eigenvectors ( bi - orthogonal set of eigenvectors , @xcite ) of this floquet matrix are the fourier components of the floquet states as defined in eq.[eq3 ] , that is @xmath30^{t}\overrightarrow{\varphi}^{l}_{\alpha}(\textbf{r})=e_{\alpha}\overrightarrow{\varphi}^{l}_{\alpha}(\textbf{r } ) \label{eq11a}\ ] ] @xmath30\overrightarrow{\varphi}^{r}_{\alpha}(\textbf{r})=e_{\alpha}\overrightarrow{\varphi}^{r}_{\alpha}(\textbf{r } ) \label{eq11b}\ ] ] where @xmath32_{n}\equiv\varphi^{l / r}_{\alpha , n}(\textbf{r } ) . \label{eq11c}\ ] ] since in our case @xmath33^{t}=[\underline{\underline{h_f(\mathbf{r})}}]$ ] then @xmath34 . there are two sets of eigenfunctions of the floquet operator @xmath35 @xmath36 and @xmath37 as pointed out in @xcite the c - product , which is associated with the non usual inner - product in linear algebra ( see for example wilkinson s text book @xcite ) , reads @xmath38 where @xmath39 . however , as proposed recently by moiseyev and lein ( @xcite ) , the inner - product should be modified even further when time - dependent functions are used as basis set due to the time - asymmetry problem in nhqm . if we define @xmath40 and @xmath41 to be the floquet eigenfunctions and eigenvalues respectively , then the time - dependent basis functions could be @xmath42 ( which are solutions of the tdse ) and also the left functions @xmath43 following the modified definition of the inner product ( `` finite - range '' product , `` f - product '' @xcite ) @xmath44 it implies that the @xmath45 quasienergy state decays exponentially in time . for more detailed discussion see @xcite,@xcite . the adiabatic theorem for time - dependent * bound * systems was derived in 1997 by kohn _ _ @xcite and in 1999 by holthaus _ @xcite who used the ( t , t ) formalism to describe the evolution of a system subjected to chirped laser pulses . as discussed by kohn and co - workers , the adiabatic approach is * not * applicable for open systems since the quasienergy level spacing reduce to zero as the number of basis functions used in the numerical calculation is increased . to avoid this difficulty baer et al . @xcite applied the adiabatic theorem to time - dependent * open * systems in the high frequency regime where the system was stabilized and the resonances ( which were embedded in the continuum part of the floquet spectra ) became so narrow that they could be practically treated as bound states . the purpose of our work is to derive the adiabatic theorem for general time - dependent open systems where there are no bound states and the resonances are nt necessarily narrow . we are using the non - hermitian floquet formalism ( through the cs formalism ) which allow us to describe the dynamics in term of non - hermitian resonance states ( see for example @xcite-@xcite , and also the work of day et al . @xcite who used the nh floquet multistate method to study the applicability of the single floquet resonance approximation in the description of the dynamics of h atom subjected to intense laser fields of various strengths ) . below we derive the adiabatic theorem for time - dependent open systems using the extended ( t , t ) formalism . by the term `` extended '' _ ( t , t ) _ formalism we mean that in the same manner presented , one may add any number of time `` coordinates '' to the schrdinger equation as one wishes , if by this a better understanding or easier solution of the problem is achieved . here we found that by addition of 2 time `` coordiantes '' to the tdse , we simplified the derivation of the adiabaticity criteria for photo - induced dynamical systems . in this sense , we are using a _ ( t , t,t ) _ formalism . we would like to study the dynamics of a single active electron in an atom or molecule , subjected to a pulse of strong monochromatic linearly polarized laser radiation . in the dipole approximation the tdse which describes this process is : @xmath46 where , @xmath47 and , @xmath48 here @xmath49 is the function which describes the envelope of the laser pulse and @xmath50 is a vector as defined in eq.[eq18 ] ; @xmath51 is the laser s amplitude , @xmath52 is a unit vector in the direction of the electric component of the laser field , @xmath53 is the laser s frequency , with @xmath54 the optical period . @xmath55 is the field - free hamiltonian and the vector operator @xmath2 describes the internal degrees of freedom ( the coordinates are complex - scaled throughout ) . in the same spirit of subsection ( 2.1 ) , we define the following operator @xmath56 where @xmath57 and @xmath58 should be regarded as additional * coordinates*. upon complex - scaling @xcite @xmath59 becomes non - hermitian . therefore the inner c - product should be used as mentioned before . the quasi - energy solutions of this complex - scaled floquet - type operator are : @xmath60 @xmath61 where the symbol @xmath62 does nt stand for an operator but for the transpose of the matrix representing the operator @xmath63 . the eigenfunctions form a complete set in the extended hilbert space @xmath64 . say we want to solve the following tdse with the initial state @xmath65 : @xmath66 the solution to this equation is @xmath67 and a function @xmath68 , which is * not * a solution of any schrdinger equation , is defined as @xmath69 where by taking the cut @xmath70 on eq.[eq24 ] it is easily seen that the expansion coefficients are @xmath71 . [ @xmath72 is a solution of the original tdse ( apply the cut t=t=t on eq.[eq21 ] and compare the result to the result obtained when the function @xmath72 is substituted in the original tdse ) ; therefore , any linear combination of these solutions is also a solution ] . let us now return to the main purpose of this article , the derivation of the adiabatic theorem for non - hermitian open systems . we would like to treat @xmath73 as an adiabatic coordinate ( this is the `` coordinate '' associated with the pulse envelope ) in the same way that the electronic motion is separated from the nuclear one in the treatment of molecules within the born - oppenheimer approximation . first we define the following operator @xmath74 where @xmath73 should be regarded as a parameter now . this means that this hamiltonian is a floquet hamiltonian describing the interaction of the atom with * cw * laser of strength @xmath75 where following eq.[eq18 ] @xmath76 the eigenstates of this operator form a complete basis ( in the @xmath77 space ) , for every value of the parameter @xmath73 : @xmath78 notice that due to the complex scaling , @xmath79 and @xmath80 get complex values . we can expand each eigenstate of the complete problem ( eq.[eq21 ] ) in this basis : @xmath81 substituting eq.[eq28 ] into eq.[eq21 ] , multiplying the obtained equation from the left hand side by @xmath82 and integrating over @xmath2 and @xmath3 one gets , in matrix notation , the equality : @xmath83\overrightarrow{\chi}_{k}(t'')=\lambda_{k}\overrightarrow{\chi}_{k}(t '' ) \label{eq29}\ ] ] where @xmath84_{\alpha,\alpha'}=\varepsilon^{ad}_{\alpha}(t'')\delta_{\alpha,\alpha'},\ ] ] @xmath85_{\alpha,\alpha'}=((\psi^{ad , l}_{\alpha}(\mathbf{r},t',t'')|-i\hbar \frac{\partial}{\partial t '' } @xmath86_{\alpha}=\chi_{\alpha , k}(t '' ) . \label{eq30}\ ] ] notice that in case that the matrix on the left hand side of eq.[eq29 ] is diagonal , a homogeneous systems of uncoupled equations is obtained . in such a case one should solve each equation separately . therefore , the sum in eq.[eq28 ] reduces to a single product . this is exactly the adiabatic approximation as appears in the born - oppenheimer context . the next step in our derivation is to represent the matrix @xmath87 by its spectral decomposition @xmath88\underline{\underline{d}}^{r}(t'')=\underline{\underline{d}}^{r}(t'')\underline{\underline{w}}(t '' ) \label{eq31}\ ] ] @xmath88^{t}\underline{\underline{d}}^{l}(t'')=\underline{\underline{d}}^{l}(t'')\underline{\underline{w}}(t '' ) \label{eq32}\ ] ] the matrix of eigenvalues @xmath89 is diagonal and the right and left eigenvectors are normalized with respect to each other in order to maintain the correct inner product : @xmath90^{t}\underline{\underline{d}}^{r}(t'')=\underline{\underline{i } } \label{eq33}\ ] ] in the case that the matrix @xmath87 is not strictly diagonal we can use first - order perturbation theory to get the first - order deviation from diagonal . if we treat the matrix @xmath91 as perturbation , we get @xmath92_{\alpha',\alpha}=\delta_{\alpha',\alpha}+\frac{[\underline{\underline{v}}(t'')]_{\alpha',\alpha}}{\varepsilon^{ad}_{\alpha}(t'')-\varepsilon^{ad}_{\alpha'}(t '' ) } \label{eq34}\ ] ] the matrix will be diagonal to a good approximation if @xmath93_{\alpha',\alpha}|\ll 1 \label{eq35}\ ] ] which produces the following adiabaticity criteria , @xmath94 using the specific form of the hamiltonian of the problem given in eq.[eq17 ] and the hellman - feynman theorem one gets the adiabatic condition for time - dependent open systems : @xmath95 where @xmath96 the index @xmath97 is a super - index ; since ( eq.[eq27 ] ) it is easily seen that not only is @xmath98 a solution of the eigenvalue equation , with eigenvalue @xmath80 , but also @xmath99 is a solution , with the eigenvalue @xmath100 , for any integer @xmath101 . let us take all the states whose corresponding eigenvalues lie in the interval @xmath102 $ ] ( the first brillouin zone ) and define them to have index @xmath103 ; we call these states @xmath104 , @xmath105 and the corresponding eigenvalues @xmath106 and get : @xmath107 @xmath108 @xmath109 where @xmath110 thus , the index @xmath97 actually counts both the position of the quasienergy within the first brillouin zone ( the index @xmath111 ) and the brillouin zone itself ( the index @xmath112 ) . with respect to the generalized inner product , two states with one or more of the indices ( @xmath111,@xmath112 ) different are orthogonal . going back to eq.[eq37 ] now , it is seen that the probability to couple an initial adiabatic state @xmath113 to any other adiabatic state @xmath114 is given by , @xmath115 where @xmath116 @xmath117 and the functions @xmath118 , that will be termed as _ `` adiabatic cross terms '' _ from now on , are given by @xmath119 since the energies @xmath120 are complex ( the hamiltonian is non - hermitian ) then for @xmath121 it is most unlikely that @xmath122 . it is clear that the denominator hardly ever vanishes even when @xmath103 . this holds true even when @xmath123 but @xmath124 . _ therefore , the criteria for a pulse to be considered adiabatic is that the condition _ @xmath125 _ be fulfilled . _ who is the adiabatic state @xmath113 whose couplings to all other states @xmath114 should remain small in the adiabatic limit ? assuming that the system is in a stationary state of the field - free problem @xmath126 before the action of the field ( relaying on the superposition principle of the solutions of the tdse generality is not lost by this assumption ) and provided that the field is switched adiabatically , this floquet resonanace state @xmath113 is the state which is `` born '' from the stationary state @xmath126 as the field is switched on . if the process is not done adiabatically many floquet resonanace states will be populated , resulting in considerable couplings of @xmath113 to them and collapse of the adiabatic condition in eq.[eq46 ] . the only adiabatic check which is physically meaningful is one in which _ @xmath97 denotes a resonance state _ ( which is associated with square - integrable function ) . @xmath127 however _ stands for both resonances and rotated continuum states_. one should notice that the derivation of the adiabatic theorem presented above holds for many electron systems . in order to avoid complicated notation the symbol @xmath2 can stand for many electrons . it also holds for polychromatic radiation , where the cw field is a collection of monochromatic fields with frequencies @xmath128 and phases @xmath129 . notice that for a given problem ( given spectral profile of the cw field and given field - free potential ) , the sum over absolute value of the adiabatic cross terms should be calculated as function of the effective cw - field intensity ( which is symbolized here through @xmath73 ) * only once*. the adiabatic cross terms should then be converted to be functions of time , through the explicit time dependence of the pulse envelope and then the sum of their absolute values should be multiplied by the time derivative of the pulse envelope and by the maximal field intensity to obtain the final expression which indicates whether the adiabatic criteria is fulfilled or not . it is easily seen in eq.[eq43 ] that for a given system , the shape and intensity of the laser pulse determines its adiabaticity since these parameters influence the shape - derivative term . a short pulse which is switched on or off abruptly and has a high maximal intensity , will most likely not be adiabatic . in the case that the adiabaticity criteria is fulfilled , the sum in eq.[eq28 ] could be reduced to a single product : @xmath130 and the adiabatic states are assigned with two good quantum numbers @xmath131 and @xmath111 . the solution to the eigenvalue equation eq.[eq29 ] is now ( @xmath132_{\alpha,\alpha'}\approx 0 $ ] ) @xmath133 by using eq.[eq24],[eq51],[eq52 ] we get that the adiabatic solution , in the ( t , t,t ) formalism , is given by , @xmath134\phi^{ad , r}_{j}(\mathbf{r},t',t'')e^{-\frac{i}{\hbar}\int^{t''}d\tau e^{ad}_{j}(\tau)}\label{eq53}\ ] ] now , applying the cut @xmath135 in order to obtain the physical solution of eq.[eq16 ] one gets ( eliminating the phase factor @xmath136 ) : @xmath137 this is the adiabatic solution of the tdse associated with initial state @xmath126 . let us summarize and clarify the procedure that need to be made in order to determine if a pulse is adiabatic or not ; the determination is carried out through the calculation of the expression @xmath138 ( it is t now , not t ! ): \(1 ) perform non - hermitian adiabatic floquet simulations ( eq.[eq27 ] ) with * cw * field , for a range of intensities @xmath75 which covers all intensities between zero and the maximal intensity @xmath51 that the studied laser pulse reaches . the adiabatic floquet hamiltonian is therefore @xmath139 and the eigenvalue equation to be solved is @xmath140 obtain the quasienergy spectrum @xmath141 and all the adiabatic cross terms @xmath142 as defined in eq.[eq45],[eq49 ] as function of the intensities . this stage is done only once , for a given system . \(2 ) for a given laser pulse @xmath49 with a maximal intensity @xmath51 , evaluate the effective cw - field intensity as function of time @xmath143 ( eq.[eq18 ] ) . then , convert the adiabatic cross terms to be functions of time via the equality ( eq.[eq26a ] ) @xmath144 using @xmath142 as calculated in step 1 @xmath145 where @xmath146 is the transformation which fulfills @xmath147=t$ ] . \(3 ) for each given resonance state @xmath148 calculate @xmath149 using @xmath150 from step 2 where notice that here @xmath73 in eqs.([eq42]-[eq45 ] ) is replaced by @xmath21 . if for a given resonance state @xmath148 the corresponding expression @xmath138 is smaller then unity at every instant , it is guaranteed that the system initially at the bound state which corresponds to this resonance , will evolve adiabatically to that resonance . in this case the hgs spectra will show only odd harmonics and the ionization probability as function of time will have a simple form that will be shown . we studied a single - electron 1d xe atom subjected to a single sin - square pulse of strong monochromatic laser field in two approaches . in the first one hermitian simulations were carried out whereas in the second non - hermitian floquet simulations based on the complex scaling method were carried out . the hermitian simulations were carried out by solving the following tdse @xmath151\psi(x , t)=i\hbar \frac{\partial}{\partial t}\psi(x , t ) \label{eq55}\ ] ] with a sine - square envelope @xmath152 and the field - free effective potential @xmath153 was an inverse gaussian @xmath154 which supports two bound states that mimic the two lowest electronic states of xe , with energies @xmath155 , @xmath156 and a third weakly bound state with energy @xmath157 the wave function was taken initially at the ground state ( g.s . ) of the field - free hamiltonian of the system : @xmath158 and was calculated for times @xmath159 ( single sine - square pulse , @xmath160 was the number of optical cycles that entered the pulse ) . using this wavefunction the ionization probability at times @xmath161 was calculated ( at these times the potential felt by the electron was the field - free potential @xmath153 ) @xmath162 where @xmath129 were the bound states of the field - free problem ( 3 bound states over which we summed in this example ) . also hgs was calculated , which following the classical - quantum correspondence principle ( larmor formula , ) equals the modulus - square of the fourier - transformed time - dependent acceleration expectation value . this is actually the intensity of the radiation emitted by the oscillating electron as presented in energy space . @xmath163 the non - hermitian simulations were floquet simulations which were carried out for different field intensities . the quasienergy spectrum of complex energies @xmath164 and the adiabatic cross terms @xmath142 were calculated for each intensity . then , these quantities were expressed as function of time through eq.[eq26a ] , i.e. @xmath165 . it was verified that in the cases where the adiabatic criteria was fulfilled , the hgs obtained from the hermitian propagation simulation contained only odd harmonics . a more quantitative measure of the existence of the adiabatic criteria was obtained through the comparison of ionization probabilities as obtained from the hermitian propagation simulation and the non hermitian simulation . in order to get non - hermitian floquet hamiltonian , the complex coordinate method was used . the floquet hamiltonian ( eq.[eq48 ] ) @xmath166 was diagonalized . provided that the scaling parameter @xmath27 was sufficiently large , the resonance quasienergy states were @xmath27 independent : @xmath167 @xmath168 being the position of the state , @xmath169 being the width of the state and @xmath170 its lifetime . since the resonance states had finite lifetimes , and since the resonances are the states which are associated with the dynamics , these resonance lifetimes should have fingerprints in the hermitian propagation simulation . indeed it was found to be so when the ionization probabilities , as computed in the two simulations , were compared ; the ionization probability at each instant was given by the following expression , which was obtained using eq.[eq54 ] and the f - product definition for the inner product @xcite : @xmath171 where @xmath172 was associated with the floquet resonanace state that was `` born '' from the ground stationary state @xmath173 as the field was turned on . the resonances that were `` born '' from the field - free hamiltonian bound states were identified by plotting the quasienergy spectrum as function of the effective field intensity @xmath75 . the resonance complex quasienergy trajectories started from the field - free hamiltonian bound states real energies and formed continuous trajectories in the complex energy plane as function of the intensity . how was this expression for the ionization probability obtained ? according to the f - product definition when the complex energy given in eq.[eq62 ] is substituted in eq.[eq54 ] , the `` ket '' ( right ) solution of the non - hermitian tdse is obtained : @xmath174 and it is easily seen that this function decays with time . the `` bra '' ( left ) solution is not a solution of a schrdinger equation but is derived from the `` ket '' solution ( the explanation of this point is beyond the scope of this work ; for an explanation see @xcite ) @xmath175 and also this function decays with time . when the f - product of @xmath176 and @xmath177 is calculated ( this is an overlap integral without complex conjugation of the left state ; in the usual dirac - product notation this reads @xmath178 ) the terms containing the real part of the energy cancel each other and the overlap integral ( c - product ) of the adiabatic floquet states gives unity ( remember the completeness property of floquet states also in coordinate space alone ) . the expression in eq.[eq63 ] is obtained for the specific case that only the floquet resonance state which is `` born '' from the field - free bound state is populated . when several floquet states are populated the same type of calculation could be repeated , where this time @xmath176 and @xmath177 are given by liner combination of terms as in eq.[eq64 ] with the proper coefficients . since the adiabatic floquet states are orthogonal to each other , a generalization of the result of eq.[eq63 ] is obtained : @xmath179 notice that in the non - hermitian formalism the norm is not conserved but decays with time ; it contains the knowledge about decay processes inherently . the two functions as given in eq.[eq59],[eq63 ] were compared and it was found that the resemblance between the functions increased as the adiabatic limit was increasingly reached by the laser pulse parameters . the numerical method used to solve the tdse with hermitian hamiltonian was the split operator forest - ruth algorithm with 7 points @xcite ; the grid size , time step and/or grid step were adjusted as required to achieve convergence . for the non - hermitian floquet simulation , the ( t , t ) formalism was used , with the complex coordinate method . the number of basis functions , box length and scaling angle were adjusted as required to achieve convergence . in fig.[fig1 ] the hgs as obtained for pulse strength of @xmath180 ( corresponding to intensity of @xmath181 ) , pulse - durations of @xmath182 optical cycles , laser frequencies of @xmath183 ( corresponding to energies of @xmath184 and wavelengths of @xmath185 , @xmath186 , respectively ) and initial state @xmath187 are shown . the appearance of odd harmonics and the absence of even or non - integer harmonics , is in general the main feature which appears for long pulses , regardless of the frequency . the odd harmonics are obtained even for the shortest laser pulses although the existence of odd - symmetry selection rules is sensitive to the frequency : some deviation appears at frequency of @xmath188 and the obtained spectra is more complicated . in fig.[fig2 ] the complex quasienergies of the 3 resonances which are `` born '' from the 3 bound states of the field - free hamiltonian ( will be given indices 1 - 3 from now on ) , as obtained from adiabatic floquet simulations ( eq.[eq49 ] ) for @xmath189 are shown as function of the cw - field strength @xmath75 . notice that the lifetimes of these resonances are not monotonic functions of the field intensity . it should be noted that these were not the only resonances that appeared in the quasienergy spectrum ; there were also other resonances which did nt emerge from the field - free bound states . however , as will be shown in fig.[fig4 ] , in this case for not too large field intensities resonances 1 - 3 were the only important resonances and they alone determined the dynamical behavior of the system . in fig.[fig3 ] the ionization probability as obtained from the hermitian simulation ( eq.[eq59 ] ) with pulse strength of @xmath180 , pulse - durations of @xmath182 optical cycles , laser frequency of @xmath189 and initial state @xmath187 , is compared to the ionization probability as obtained from the expression given in eq.[eq63 ] , which is derived from the f - product formalism together with the resonances quasienergies obtained from the non - hermitian simulation . it is seen that as the pulse becomes longer the results obtained from the two simulations become identical . both results of hgs and ionization probabilities showed that the time - dependent wavefunction of the studied systems could be well approximated by the adiabatic expression given in eq.[eq54 ] , even for short pulses . the values of the terms which check the adiabatic criteria , as seen in fig.[fig6 ] , gave the explanation why this was so . in fig.[fig4 ] the expressions @xmath190 ( eq.[eq44 ] ) which describe the couplings between every two resonances from the set of 3 tracked resonances is shown as function of the cw - field strength @xmath75 , for the case @xmath189 . it can be seen that the coupling between resonances 1 and 2 is strong for field strength of @xmath191 , @xmath192 and @xmath193 . this could be partially explained on the basis of the values of the quasienergies , as seen in fig.[fig2 ] , at least for two field strengths out of the three . it is seen that for field strength of @xmath194 and @xmath192 the real parts of the quasienergies cross , resulting in small value of the denominator in the expression given in eq.[eq45 ] for @xmath195 ( at least for one @xmath101 term ) . in the same way the strong couplings between resonances 1 and 3 at field strength of @xmath196 , and between resonances 2 and 3 at field strength of @xmath197 could be explained on the basis of the quasienergy values at these field - strengths . in particular it should be noticed that the couplings between resonances 1 and 3 at @xmath196 are the strongest among all 3 resonances due to the close values of both real and imaginary parts of the quasienergies . however , it should be noted that crossings in the quaisenergy plot are not always indications of large couplings since also the overlap between the wavefunctions ( the nominator of the expression for @xmath198 ) is important . in fig.[fig5 ] the sum @xmath199 which describes the couplings between * all * quasistates of the system @xmath200 to resonance 1 @xmath201 is shown as function of the cw - field strength @xmath75 , for the case @xmath189 . in addition , also the partial sum @xmath202 which describes the couplings between resonances 2 and 3 to the resonance state @xmath201 is shown . it can be shown that up to a moderate intensity of @xmath203 the first resonance is mainly coupled only to the other 2 resonances and not to other , higher resonances or continuum states . the entire dynamics is governed almost solely by the 3 resonances which are `` born '' from the 3 bound states of the field - free hamiltonian . the sums described above , which are functions of a cw - field strength @xmath75 , are converted to be explicit functions of time for the specific sine - square pulse with maximal intensity @xmath180 used in the simulation ( fig.[fig5 ] , upper part ) . for this purpose for each time @xmath21 @xmath159 the effective cw - field strength @xmath204 is calculated and the values of the 2 functions shown in fig.[fig5 ] which fit this effective cw - field strength are taken . hence the 2 sum functions are converted to be explicit functions of time . in the middle part of fig.[fig6 ] the full term @xmath205 , which represents the degree of adiabaticity in the process of shining a 1d xe atom initially at the ground state with a sine - square laser pulse supporting @xmath206 optical cycles of monochromatic radiation with frequency @xmath189 and strength @xmath180 , is shown as function of time . it is seen that the term @xmath205 is bounded by the value of @xmath207 for all times , whether it is calculated by coupling of the first resonance to other 2 resonances only or to all other states . the structure of the hhs for @xmath206 seen in fig.[fig1 ] implies that this value is indeed small and the process is adiabatic . in the lower part of fig.[fig6 ] the same quantity is shown , but for @xmath188 . here it is seen the terms @xmath205 are bounded by a much larger value of @xmath208 for all times and the more complex structure of the hhs for @xmath206 seen in fig.[fig1 ] implies that this value is not small enough to indicate the appearance of an adiabatic process . it should be noted that for a given system with field - strength @xmath51 , frequency @xmath53 and sine - square pulse envelope , we have @xmath209 . therefore , as the number of optical cycles the pulse supports increases the general shape of the terms @xmath205 is kept the same but is attenuated . therefore , for @xmath188 as @xmath160 gets bigger , the pulse s envelope varies more slowly and the process becomes more and more adiabatic , as seen in the hhs spectra for @xmath210 for example . with the help of the ( t , t ) formalism we derive an adiabatic theorem for open systems . the use of the complex scaling transformation plays a key role in our derivation . for example , the spectrum of the floquet hamiltonian of an open system is changed dramatically . rather than a continuous spectrum that is responsible for the absence of an adiabatic limit for @xmath160(number of basis functions)@xmath211 in the conventional qm , the resonances are associated with a point spectrum and are separated from the continuum which is rotated into the lower half of the complex energy plane . an interesting important numerical result of our derivation is that the calculation of the effect of the pulses s shape on the dynamics does not require heavy computations . the entire effect of the laser pulse is embedded in a multiplication factor of @xmath212 where @xmath213 is the variation of the maximum field amplitude as function of time . as a numerical example we applied the adiabatic theorem we derived to a model hamiltonian of xe atom ( with symmetric field - free potential ) which interacts with strong , monochromatic laser pulses . we have shown that the generation of odd - order harmonics and the absence of even - order harmonics , even when the pulses are extremely short , can be explained with the help of the adiabatic theorem we derived . the use of a single - electron 1d model to describe a realistic atom is justified since it has been shown before in many cases that all the main strong field effects are reproduced . therefore the conclusions obtained with this model are also valid for a realistic atom . this work was supported in part by the by the israel science foundation and by the fund of promotion of research at the technion . milan indelka is acknowledged for most helpful and fruitful discussions . s. x. hu and z. z. xu , _ phys . rev . a _ * 56 * , 3916 ( 1997 ) ; s. dionissopoulou , t. mercouris and c. a. nicolaides , _ ibid . _ * 61 * , 063402 ( 2000 ) ; a. di piazza and e. fiordilino , _ ibid . _ * 64 * , 013802 ( 2001 ) ; v. vniard , r. taeb and a. maquet , _ ibid . _ * 65 * , 013202 ( 2001 ) . j. a. fleck , j. r. morris and m. d. feit , _ appl . * 10 * , 129 ( 1976 ) ; m. d. feit , j. a. fleck and a. steiger , _ j. comput . phys . _ * 47 * , 412 ( 1982 ) ; e. forest and r. d. ruth , _ physica d _ * 43 * , 105 ( 1990 ) ; m. campostrini and p. rossy , _ nucl . b _ * 329 * , 753 ( 1990 ) ; j. candy and w. rozmus , _ j. comput . phys . _ * 92 * , 230 ( 1991 ) .

in the conventional quantum mechanics ( i.e. , hermitian qm ) the adiabatic theorem for systems subjected to time periodic fields holds only for bound systems and not for open ones ( where ionization and dissociation take place ) [ d. w. hone , r. ketzmerik , and w. kohn , phys . rev . a * 56 * , 4045 ( 1997 ) ] . here with the help of the ( t , t ) formalism combined with the complex scaling method we derive an adiabatic theorem for open systems and provide an analytical criteria for the validity of the adiabatic limit . the use of the complex scaling transformation plays a key role in our derivation . as a numerical example we apply the adiabatic theorem we derived to a 1d model hamiltonian of xe atom which interacts with strong , monochromatic sine - square laser pulses . we show that the generation of odd - order harmonics and the absence of hyper - raman lines , even when the pulses are extremely short , can be explained with the help of the adiabatic theorem we derived . department of chemistry and minerva center of nonlinear physics in complex systems technion israel institute of technology haifa 32000 , israel . + 03.65.-w , 42.50.hz , 42.65.-ky , 32.80.rm

there has been considerable interest in the localized solutions of the einstein s equations with nonlinear field sources in recent years ( @xcite , @xcite , @xcite , @xcite , @xcite ) . gravitating non - abelian solitons and black holes with yang - mills fields is investigated in @xcite . such problems were not investigated earlier in the history of gr , mainly because of two reasons : 1 . it was widely accepted that the gravitational effects are too weak to affect -in an essential way- the properties of soliton solutions of nonlinear field theories . 2 . the resulting equations are usually formidable such that the ordinary analytical approaches become idle . more recently , however , the availability of high speed computers and advanced numerical methods have changed the case , and extensive numerical attempts have been made in this direction ( see e.g. 387n term project in @xcite ) . it has emerged from recent studies that the effects due to the inclusion of gravity are not always negligible . consider , for example , the einstein - yang - mills ( eym ) system . it has been shown that the eym equations have both soliton and black hole solutions ( @xcite , @xcite and @xcite ) . this is in contrast to the fact that vacuum einstein and pure yang - mills equations do not have by themselves soliton solutions . we can therefore conclude that gravity may have dramatic effects on the existence or non - existence of soliton solutions of nonlinear field equations . another interesting example is the discovery that black hole solutions may have skyrmion hair @xcite . it was previously believed that stationary black holes can only have global charges given by surface integrals at spatial infinity ( the so - called no - hair theorem ) . in the ordinary o(3 ) model , spherically symmetric solutions have an energy density which behave like @xmath0 at large distances ( @xcite ) . when formulated in a curved background , this model leads to a spacetime which is not asymptotically flat , and the adm mass is not well defined ( @xcite ) . a nonlinear o(3 ) model ( thereafter referred to as the isovector model ) was introduced in ( @xcite ) , which possesses spherical , soliton - like solutions with a @xmath1 energy density at large distances . such a model , is therefore expected to be well behaved in an asymptotically flat background . in the present paper , we examine this model , and discuss its self - gravitating solutions . these new solutions are compared with those obtained previously in a flat spacetime . the present manuscript is organized in the following way . in section [ sec2 ] , we will review the isovector model of @xcite . in section [ sec3 ] , flat - space solitons of the isovector model and their resemblence to charged particles are introduced . in section [ sec4 ] , the isovector model will be reformulated in a curved background . the resulting differential equations for a spherically symmetric ansatz will be introduced in this section , together with the necessary boundary conditions . these equations will be solved numerically , for several choices of the coupling constant . we will compare the self gravitating solutions with those obtained for a flat spacetime . soliton masses using the tolman formalism will be discussed in section [ sec5 ] , together with the behavior as a function of the model parameter . section [ sec6 ] will contain the summary and conclusion . consider an isovector field @xmath2 ( @xmath3 ) with a @xmath4 vacuum at @xmath5 each component @xmath2 is a pseudo - scalar under spacetime transformations , and @xmath6 is a constant . a topological current can be defined for such a field according to ( @xcite ) @xmath7 for the time being , spacetime is assumed to be the flat minkowski spacetime and @xmath8 with @xmath9 ( @xmath10 is assumed throught this paper ) . @xmath11 and @xmath12 are the totally anti - symmetric tensor densities in 4 and 3 dimensions , respectively . it can be easily shown that the current ( [ topcur ] ) is identically conserved ( @xmath13 ) , and the total charge is quantized @xmath14 where @xmath15 . in this equation , @xmath16 and @xmath17 are area elements of @xmath4 surfaces in the @xmath18-space ( as @xmath19 ) and @xmath20-space ( as @xmath21 ) , respectively . the current ( [ topcur ] ) can identically be written as the covariant divergence of an anti - symmetric , second - rank tensor @xmath22 where @xmath23,\ ] ] in which @xmath24 is an auxiliary vector field . the dual field @xmath25 with the tensorial components @xmath26 satisfies the equation @xmath27 provided that the vector field @xmath28 is a solution of the following wave equation @xmath29 it can be easily shown that the right hand side of this equation defines another conserved current @xmath30 with @xmath31 using the language of differential forms , ( [ feq ] ) can be written in the following form @xmath32 where the components of the 2-forms @xmath33 and @xmath34 are given by @xmath35 and @xmath36 we now have @xmath37 and @xmath38 the 2-form @xmath39 is therefore hodge - decomposable , and cohomologous with @xmath33 ( i.e. they belong to the same cohomology class , since they differ only by an exact form ) . the resemblance of equations ( [ max1 ] ) and ( [ max2 ] ) to the maxwell s equations and the capability of this model to provide non - singular solutions behaving like charged particles were discussed in @xcite . in the next section , we will only outline the main results valid in a flat spacetime . the requirement of having non - singular , finite energy and stable solitons , severely restrict the possible choices of the lagrangian density of the isovector field . let us follow @xcite , and adopt the following lagrangian density which satisfies the above requirements : @xmath40 with @xmath41 , and @xmath42 real constants . the potential @xmath43 satisfies the following conditions @xmath44 which leads to the spontaneous breaking of the ( global ) @xmath45 symmetry of the system . the dynamical equation for the isovector field is easily obtained , using the variational principle @xmath46 , which leads to @xmath47 similar to the ansatz used in the skyrme model ( @xcite ) , we start with the so - called hedgehog ansatz @xmath48 where @xmath49 , @xmath3 represent the euclidean coordinates @xmath18 , @xmath50 , and @xmath51 , respectively . this ansatz immediately leads to @xmath52 and @xmath53 where @xmath54 is the unit vector in the radial direction , and @xmath55 is the elementary topological charge defined in ( [ topcharge ] ) . note that @xmath56 and @xmath57 . the charge density corresponding to this ansatz is easily obtained to be @xmath58 which leads to @xmath59 showing that the ansatz bears unit topological charge . note that in deriving this result , we have used the boundary conditions @xmath60 and @xmath61 it can be shown that the following asymptotic solutions are valid : @xmath62 , \ ] ] close to the center of the soliton ( @xmath63 ) , where @xmath64 is a dimensionless constant , and @xmath65 and @xmath66 far from the soliton ( i.e. @xmath19 ) . it can be seen that the total soliton energy @xmath67 4\pi r^2 dr=5.21\phi_o^3\left [ b_o(-4\lambda)^3\right]^{1/4}.\ ] ] using the scale transformation @xmath68 ( while keeping @xmath20 unchanged ) , it can be shown that @xmath69 has a minimum at @xmath70 , which is a signature of the stability of the soliton under radial perturbations . by self - gravitating isovector solitons , we mean static solutions of the coupled isovector - gravitational equations which are everywhere regular and represent localized lumps of energy . by numerically integrating the coupled nonlinear equations , we will show that such solutions do arise depending on the value of the model parameters . based on or results , we will also criticize the widely expressed view that gravity has only a minute effect on the structure and properties of extended solitons . let us start with the action@xmath71 in which @xmath33 is the gravitational constant , @xmath72 is the curvature scalar , @xmath73 is the lagrangian density of the matter source , and @xmath74 is the determinant of the metric tensor . as the source of gravity , we consider the isovector field ( [ isolag ] ) . by varying the action ( [ action ] ) with respect to @xmath75 and @xmath2 , we obtain the corresponding field equations : @xmath76 , \ ] ] and @xmath77 by contracting equation ( [ einseq ] ) , we obtain the following equation for the curvature scalar:@xmath78 this equation is useful , since it expresses a simple relation between the curvature scalar and the self - interaction potential of the isovector field , and provides a means to check some of the calculations . we employ the coordinates @xmath79 , and the general , spherically symmetric , static metric @xmath80 . for the hedgehog ansatz ( [ hedge ] ) , the independent field equations become @xmath81 @xmath82 and @xmath83 note that equation ( [ isograv ] ) reduces to the flat - space equation ( [ isoeq ] ) , by putting @xmath84 . equations ( [ met1 ] ) and ( [ met2 ] ) , however , become inconsistent for obvious reasons ( the spacetime can not be flat in the presence of matter sources ) . let us introduce the following three length scales @xmath85 @xmath86 and @xmath87 which appear naturally in equations ( [ met1 ] ) to ( [ isograv ] ) . it can be seen that @xmath88 is proportional to the geometric mean of @xmath89 and @xmath90 : @xmath91 emergence of new length scales is similar to what happens in non - abelian gauge fields ( @xcite ) , and leads to the appearance of new branches of spherical , static solitons . having additional length scales , one expects departures from the einstein - maxwell solutions and the appearance of more subtle features . using these two length scales , equations ( [ met1 ] ) to ( [ isograv ] ) can be made dimensionless and suitable for numerical integration ; @xmath92 @xmath93 and @xmath94 in these equations , @xmath95 , @xmath96 , and @xmath97 which is the dimensionless parameter of the system : @xmath98 and @xmath99 the asymptotic behavior of the non - singular solutions can be found by using the following series ansatze : @xmath100 for @xmath101 , and @xmath102 for @xmath103 . the unknown coefficients are calculated by inserting these ansatze into equations ( [ metdim1 ] ) to ( [ isogravdim ] ) and balancing the terms of the same order in @xmath18 . table [ coeff ] shows the leading coefficients for the above asymptotics . .leading , non - vanishing coefficients of the asymptotic solutions . [ cols="<,<",options="header " , ] in order to solve the coupled nonlinear des numerically , we used the gerald s shooting method , which is based on two guesses for the initial slope of the unknown functions . using these initial guesses , the equations are integrated via the runge - kutta - fehlberg method , reaching the end point of the independent variable . a better guess for the initial slope is then found by comparing the end point values with the boundary conditions , and interpolating for the initial slope ( @xcite ) . the boundary conditions for asymptotically flat spacetime read @xmath104 and @xmath105 the procedure described above is iterated until the correct boundary values are reached with a reasonable accuracy . in order to test the method , flat space solutions were first computed and compared to the solutions obtained via energy minimization algorithms ( @xcite ) . figure [ fig1 ] shows numerical variations of the @xmath106 function for several values of the parameter @xmath107 . the corresponding results for the metric coefficients @xmath108 and @xmath109 are shown in figures [ fig2 ] and [ fig3 ] , respectively . it is seen that for @xmath107 of the order of unity , the self - gravitating solutions differ only slightly from the flat - space solution . the difference vanishes completely as @xmath110 . as @xmath107 becomes much smaller than 1 , significant differences with the flat - space solution emerge . for example , the metric signature changes at some @xmath111 for sufficiently small @xmath107 . although there are still controversies about an exact definition for the total mass of a self - gravitating system ( @xcite , @xcite,@xcite ) , we adopt the tolman formalism for computing the total mass of the gravitating solitons ; @xmath112 where @xmath113 and @xmath114 .\ ] ] in most cases ( including the model we are discussing here ) , the second term is absent since the matter lagrangian does not contain derivatives of the metric tensor . for the isovector model , we have @xmath115.\ ] ] using the transformations ( [ epseq ] ) and ( [ weq ] ) , the total mass can be written in terms of a dimensionless integral : @xmath116^{1/4}s_t,\ ] ] where @xmath117 @xmath118 and @xmath119 . the dimensionless integral @xmath120 were computed numerically for several values of the parameter @xmath107 . the results are shown in figure [ massfig ] . it is seen that ( 1 ) @xmath121 as @xmath122 , leading to the asymptotic ( flat space ) mass , ( 2 ) there is a maximum mass around @xmath123 , and ( 3 ) the total mass decreases as @xmath124 , with @xmath125 at @xmath126 . it is also interesting to note that in the asymptotic series solutions summarized in table [ coeff ] , the coefficients @xmath127 and @xmath128 vanish , which imply vanishing total mass of the soliton as deduced from the asymptotic form of the metric @xcite . we extended the isovector model to incorporate the effects of gravity . the resulting equations were integrated numerically for spherically symmetric ansatz , using the gerald s shooting method . it was found that for large values of the dimensionless parameter of the system the effect of gravity is negligible . for small values of @xmath107 , however , gravity has a considerable effect on the qualitative and quantitative behavior of the solutions . such dramatic changes in the behavior of the spherical solutions in the presence of gravity were also reported in the framework of eym equations ( @xcite , @xcite , and @xcite ) . gravitating solitons of the isovector model in an asymptotically flat background bear quantized topological charges , exactly similar to the flat - space solitons . the quantization is due to a @xmath129 homotopy between the boundary of the curved space ( @xmath4 at @xmath130 ) , and the vacuum @xmath4 of the isovector field ( @xmath131 ) . this is in analogy with the quantization of the magnetic pole intensity in the thooft polyakov monopoles ( @xcite ) . solutions presented in this paper ( figures [ fig1 ] to [ fig3 ] ) do not exhibit horizons . using the well - known result from general relativity ( @xcite ) , @xmath132 , where @xmath133 and @xmath134 are the emitted ( at @xmath135 ) and detected ( at @xmath136 ) periods of photons , event horizons correspond to @xmath137 , which are not fulfilled by the present solutions . however , the appearance of horizons and signature changes is not ruled out and will be addressed elsewhere . in particular , it should be interesting to know whether isovector black holes have hair . as pointed out in @xcite , solitons can make bound states with ordinary black holes to form hairy black holes . in such a case , the total adm mass of the hairy black hole is the sum of the mass of the bare black hole , the mass of the soliton , and the gravitational binding energy between the two ( @xcite ) . 99 r. bartnik and j. mckinnon , _ phys . _ , * 61 * , 41 , ( 1988 ) . p. bizon , _ phys . _ , * 64 * , 2844 , ( 1990 ) . p. bizon and t. chmaj , _ phys . _ , * b297 * , 55 , ( 1992 ) . y. kim and s. moon , _ gravitating @xmath138 model solitons _ , gr - qc/9803003 . u. nucamendi and d. sudarsky , _ class . quantum grav . _ , * 17 * , 4051 , ( 2000 ) . m.s . volkov and d.v . galtsov , _ phys . _ , * 319 * , 1 , ( 1999 ) . ` http://wwwrel.ph.utexas.edu/~p378g2 ` p. bizon , preprint uwthph-1994 - 5 , ( 1994 ) . m. barriola and a. vilenkin , _ phys . _ , * 63 * , 341 , ( 1989 ) . a. vasheghani and n. riazi , _ int . _ , * 35 * , 578 , ( 1996 ) . i.g . moss , n. shiiki , and e. winstanley , _ class . _ , * 17 * , 4161 , ( 2000 ) skyrme , _ proc . _ , * a260 * , 127 , ( 1960 ) . a. ashtekar , a. corichi , and d. sudarsky , gr - qc/0011081 v3 , ( 2001 ) . f. gerald and p.o . wheatley , _ applied numerical analysis _ addison - wesley , ( 1994 ) . bonnor , _ class quantum grav . _ * 9 * , 269 , ( 1992 ) . l. herrera and j. ibanez , _ class . quantum grav . _ , * 10 * , 535 , ( 1993 ) . askari and n. riazi , _ int . _ , * 34 * , 417 , ( 1995 ) . p. goddard and d.i . olive , _ rep . _ , * 41 * , 1357 , ( 1978 ) . s. weinberg , _ gravitation and cosmology _ , john wiley and sons , usa , ( 1972 ) .

we formulate the nonlinear isovector model in a curved background , and calculate the spherically symmetric solutions for weak and strong coupling regimes . the usual belief that gravity does not have appreciable effects on the structure of solitons will be examined , in the framework of the calculated solutions , by comparing the flat - space and curved - space solutions . it turns out that in the strong coupling regime , gravity has essential effects on the solutions . masses of the self - gravitating solitons are calculated numerically using the tolman expression , and its behavior as a function of the coupling constant of the model is studied .

stars form in dense , cold molecular clouds . measuring the molecular gas content of the clouds is very important if we are to estimate the star formation efficiency and relate it to the properties of the clouds and to their environments . the total mass , however , can not be measured directly because the lowest levels of from which the observable emission can arise have excitation energies ( e.g. , @xmath2 @xmath3 500 k , @xmath4 = @xmath5 ) too high to be thermally excited in the cold ( @xmath6 k ) molecular clouds . in the milky way , the @xmath1co line . ] ( hereafter ) traces the molecular gas content . the conversion factor ( @xmath7 ) between the column density and the velocity integrated intensity of co has been measured via the virial theorem ( @xmath7 = @xmath8 / ( ) , solomon et al . 1987 ; digel et al . 1997 and references therein ) , or via gamma - ray emission ( @xmath7 = @xmath9 / ( ) , bloemen et al . 1986 ; digel et al . 1997 and references therein ) . the metallicity dependence of the conversion factor has been an issue . cohen et al . ( 1988 ) and wilson ( 1995 ) used cloud masses determined using the virial theorem to argue that the value of @xmath10 increases as the metallicity of the individual galaxy decreases . arimoto , sofue , & tsujimoto ( 1996 ) extend this conclusion to argue that there are radial increases in @xmath10 in the milky way and m51 corresponding to radial decreases in metallicity . by contrast , taylor , kobulnicky , & skillman ( 1996 ) showed that some low abundance galaxies have lower @xmath10 , suggesting that a factor other than the abundance ( e.g. , temperature ) can affect the measured value of @xmath10 . far - uv photons from massive young stars strike the surfaces of nearby molecular clouds and produce photodissociation regions or photon - dominated regions ( hereafter pdrs , tielens & hollenbach 1985 , 1997 ) . in these surface layers , the far - uv photons dominate the ionization of atoms , the formation and destruction of molecules , and the heating of the gas . inside the pdr , absorption by dust , c , and diminishes the far - uv field . several authors have constructed pdr models appropriate to conditions in the magellanic clouds , with particular emphasis on the c@xmath0/c / co transition ( maloney & black 1988 ; van dishoeck & black 1988b ; lequeux et al . 1994 ; maloney & wolfire 1997 ) . in irregular galaxies , where metallicities and dust - to - gas ratios are lower than those in the galaxy , far - uv photons penetrate deeper into clouds , and dissociate co molecules to greater depths ( israel et al . therefore , for a cloud with a given column density , the co column density should be lower at lower metallicity . if the column density is high enough for the to self - shield against photodissociation ( @xmath11 @xmath12 @xmath13 , van dishoeck & black 1988a ) , the co column density will also be high enough for the line to be optically thick , and the line intensity ( @xmath14 ) will not depend strongly on the metallicity . in that case , lower intensities can only stem from geometrical or beam - filling effects . on the other hand , if the cloud column density is not high , most of the co will be dissociated and the resulting line will be optically thin and very weak . on the surface of the clouds , the destruction and formation of molecules are also affected by the change of metallicity , but the mechanism is different from that for co molecules . the molecules are dissociated by far - uv photons attenuated by dust or by self - shielding . if self - shielding dominates over dust attenuation , the destruction rate is independent of the dust abundance . on the other hand , the formation rate is proportional to the dust abundance , because reforms on the surfaces of dust grains . the magellanic clouds are the best targets to test pdr models that include metallicity effects because of their proximity ( @xmath15 = 50.1 kpc and @xmath16 = 60.3 kpc , westerlund 1990 ) , their low metal abundance ( @xmath17 = 0.28 , @xmath18 = 0.54 , @xmath19 = 0.050 , and @xmath20 = 0.21 , where @xmath21 is normalized to the galactic value ; dufour 1984 ) , and their low dust - to - gas ratio ( @xmath22 = 0.25 and @xmath23 = 0.059 , where @xmath24 is normalized to the galactic value ; koornneef 1984 ) . in this paper , we observed the magellanic clouds in the near - ir emission lines and in the line ( see sections [ sec : obs5 ] and [ sec : results ] ) . we compare the line intensities of , , and [ ] 158 emission from the pdrs in the magellanic clouds with those from galactic star formation regions ( see section [ sec : compare ] ) . section [ sec : models ] discusses the numerical pdr models which we compare to the observed data to learn how metallicity changes affect the chemical structure of the galactic clouds and the clouds in the magellanic clouds . some limited regions in the magellanic clouds were previously observed in the lines ( koornneef & israel 1985 ; israel & koornneef 1988 ; kawara , nishida , & taniguchi 1988 ; israel & koornneef 1992 ; krabbe et al . 1991 ; poglitsch et al . however , the published [ ] and data ( johansson et at . 1994 ; poglitsch et al . 1995 ; israel et al . 1996 ) cover more extended regions than the existing maps . we observed near - ir emission lines from the magellanic clouds with the university of texas near - ir fabry - perot spectrometer whose equivalent - disk size , is the diameter of a cylindrical beam whose solid angle is same as the integrated solid angle of the actual beam pattern . ] ( @xmath25 = 81 ) is comparable to those of the existing [ ] data ( @xmath25 = 68 ) and data ( @xmath25 = 54 ) . we also observed emission at positions where no emission had been detected at the sensitivity of the existing surveys . we observed the and lines in 1994 december , and and lines in 1995 october , at the cerro tololo inter - american observatory 1.5 m telescope using the university of texas near - infrared fabry - perot spectrometer ( ut fps , luhman et al . the instrument was designed to observe extended , low surface brightness line emission , and has a single 1 mm diameter insb detector to maximize the area - solid angle product . this product , @xmath26 , depends on the telescope coupling optics but can be made as large as @xmath27 @xmath28 sr . the beam from the @xmath29 secondary of the telescope is collimated and guided to an h - band or k - band fabry - perot interferometer and to a @xmath30 k ( solid nitrogen temperature ) dewar in which a doublet camera lens ( focal length = 20 mm ) focuses onto the detector . a 0.5 % ( for the line ) or a 1 % ( for the and lines ) interference filter in the dewar selects a single order from the fabry - perot interferometer . we used a collimator of focal length = 838 mm in 1994 december run and a collimator of focal length = 686 mm in 1995 october . the change of the collimator affected the beam size and spectral resolution ( table [ tbl-1 ] ) . since the coupling of the 686 mm collimator with the telescope optics is better , the beam profile in 1995 october is much closer to a box - function than the 1994 december beam profile ( figure [ fig-01 ] ) . an automatic alignment routine aligned the fabry - perot etalon by executing every @xmath31 minutes ( luhman et al . the fabry - perot etalon maintained alignment for @xmath32 minutes , but the ambient temperature changes caused the plate separation to drift by the equivalent of @xmath33 per minute . using telluric oh lines ( oliva and origlia 1992 ) , we calibrated the wavelength scale to within @xmath34 . we operated the fabry - perot interferometer in _ scanning mode _ at selected positions and in _ frequency switching mode _ at most positions . in the scanning mode , the plate separation of the etalon was varied to cover @xmath35 centered at the line in 15 sequential steps . figure [ fig-02 ] shows observed and lines at the 30 doradus ( 0 , 0 ) position ( see the object list in table [ tbl-3 ] ) , and telluric oh lines . the oh ( 9,7 ) r@xmath36(2 ) ( @xmath37 = 2.12267 , wavelength in air , oliva & origlia 1992 ) and the ( 9,7 ) r@xmath38(1 ) ( @xmath37 = 2.12440 ) lines are within the scan range . figure [ fig-02 ] shows the red - wing of the oh ( 4,2 ) p@xmath38(3 ) ( @xmath37 = 1.61242 ) line , and the oh ( 4,2 ) p@xmath38(2 ) ( @xmath37 = 1.60264 ) line which was displaced by one free spectral range and penetrated through the blue side of the order sorting filter at @xmath39 @xmath3 420 . the typical intensity of the oh ( 4,2 ) p@xmath38(3 ) and ( 4,2 ) p@xmath38(2 ) lines is @xmath40 which is more than @xmath41 times the intensity . the oh intensity fluctuates spatially and temporally . the @xmath42 power spectrum of the temporal fluctuations limits the sensitivity of the system , e.g. , in k - band , the oh noise becomes important at 30 seconds ( coadded integration time ) per step when the switching interval between the source position and an off - source `` sky '' position is two minutes ( luhman et al . 1995 ) . in the observations of the magellanic clouds , we chopped the secondary mirror from the source to @xmath43 = @xmath44 or @xmath45 at 0.25 hz in h - band and 0.5 hz in k - band . each spectral step ( at the same etalon plate separation ) has one chopping cycle which consists of four exposures : @xmath46 @xmath47 @xmath48 @xmath47 @xmath48 @xmath47 @xmath46 . each exposure had an integration time of 0.5 or 1 second . in the frequency switching mode , we tuned the fabry - perot interferometer to the line wavelength , @xmath49 , and a nearby wavelength free of line emission , @xmath50 . one observing cycle consists of four steps : @xmath49 @xmath47 @xmath50 @xmath47 @xmath50 @xmath47 @xmath49 . as is the case in the scanning mode , each step consists of one chopping cycle . we tried to place @xmath50 away from wings of the instrument spectral profile for @xmath49 and at positions away from telluric oh lines ( see table [ tbl-2 ] ) . for flux calibration , we measured hr 1713 ( b8i , @xmath51 = @xmath52 mag ) in 1994 december and hr 8728 ( a3v , @xmath53 = @xmath54 mag and @xmath51 = @xmath55 mag ) in 1995 october . even though the beam sizes of the 1994 december run and the 1995 october run are different from each other , intensities measured at the same positions on both runs agree to within the errors . the absolute flux calibration is accurate to @xmath56 % . we observed line in 1995 december at the sest located on la silla in chile . the beam size ( fwhm ) of the sest is 45 . the intensities presented in this paper are the main beam brightness temperatures , @xmath58 ( see the convention by mangum 1993 ) . the online temperature , @xmath59 , has been converted to @xmath58 by dividing by 0.7 ( which is the product of the forward spillover and scattering efficiency , and the main beam efficiency ) . we used frequency switching to gain a factor of 2 in observing time and to avoid possible emission from reference positions . since this method can leave residual ripples in the spectra , the detection limit for weak signals is determined by the ripple . for some positions , we complemented the frequency switched data with beam switched data with a reference beam about 12 away in azimuth . for example , in the 30 doradus ( 0 , @xmath60 ) region , it turned out to be impossible to use the frequency switched data . we mapped spectra around positions where we detected emission and where the co emission was below the lowest contour level ( 3 ) of previously published co maps ( booth 1993 ; johansson et al . the co maps are fully sampled on 20 grids within each 81 beam . in a single 45 beam , the typical root - mean - square noise is 0.07 k for a channel spacing of 0.45 , and the typical @xmath61 statistical uncertainty of the intensity @xmath62 = @xmath14 , integrated over a velocity range of 15 , is 0.2 . in low metallicity objects , the low dust abundance allows far - uv photons to penetrate for long distances beyond the /h boundary . one possible scenario is that the transparency to uv photons leads to substantial regions of neutral gas where far - uv photons have dissociated co but where self shielding permits hydrogen to be primarily in the form of . especially in the n159/n160 complex , the published maps show the [ ] 158 emission ( israel et al . 1996 ) extending farther than the emission ( johansson et al . these existing observations make such a possibility seem reasonable in the lmc . near - ir emission in response to far - uv radiation provides a direct test for the presence of molecular gas near cloud boundaries , albeit with no information about column density or total abundance . we selected observing targets in regions of the lmc where high spatial resolution ( @xmath63 ) [ ] maps and maps existed ( booth 1993 ; johansson et al . 1994 ; poglitsch et al . 1995 ; israel et al . 1996 ) . in the smc , our observations cover positions for which there were data available ( rubio et al . 1993 ) but do not coincide with the observed [ ] positions ( israel & maloney 1993 ) . table [ tbl-3 ] lists the observed sources and their reference ( @xmath64 ) positions . the line intensity in the surveyed regions is @xmath65 @xmath66 , except in the central regions of 30 doradus and n160 . about 70% of the observed points have detections of the line with a significance of @xmath67 or more . the intensities are listed in table [ tbl-4 ] , and the and intensities in table [ tbl-5 ] . reddening toward the stars in the lmc has been measured using ( ) and ( ) colors . the galactic foreground extinction , , toward the lmc is @xmath68 mag ( greve , van genderen , & laval 1990 ; lee 1991 ) . applying = 0.112 ( rieke & lebofsky 1985 ) , the foreground implies a negligible of @xmath69 mag . the molecular clouds in the lmc may , however , extinguish the emission from their own back sides . section [ sec : without_epsilon ] discusses the emission from the back side of the clouds . table [ tbl-4 ] lists the and [ ] intensities convolved to the beam size of the observations ( @xmath25 = 81 ) . we assumed that the beam profile of the university of texas fabry - perot spectrometer is a box function . the columns for far - ir intensity and dust temperature in table [ tbl-4 ] are from the iras data processed at ipac using standard hires processing . the hires processor enhances the spatial resolution of iras images using a maximum correlation method algorithm ( aumann , fowler , & melnyk 1990 ) . the effective resolutions are @xmath70 at 60 and @xmath71 at 100 . we calculate the integrated far - ir intensity using the approximation of helou et al . ( 1988 ) : @xmath72 where @xmath73 and @xmath74 are in units of mjy / sr and is in . this equation is valid within 20 % errors , when @xmath75 @xmath65 @xmath76 @xmath65 58 k. we deduce the dust temperature , @xmath76 , from the ratio of 60 to 100 intensity : @xmath77 where @xmath78 is the planck s constant , @xmath79 is the boltzmann s constant , and @xmath80 is the frequency in units of hz . in the above equation , we assumed that the 60 emission is not dominated by small ( @xmath81 nm ) , thermally - spiking , grains but by large grains in steady state temperatures ( draine & lee 1984 ) , and that the dust emissivity ( @xmath82 ) is proportional to @xmath83 where the dust emissivity index , @xmath84 , is 1 . in the smc , we detected emission , with @xmath85 , near iras sources in the dem s 16 , dem s 37 , and li - smc 36 regions . toward supernova remnants , e.g. , dem s 37 ( @xmath86 ) and li - smc 36 ( @xmath87 ) , we did not detect emission at a @xmath61 level of @xmath88 . figure [ fig-03 ] shows the far - ir , , and intensity maps of the 30 doradus region . at the 30 doradus ( @xmath64 ) point , the intensity of the line , , is @xmath89 in our 81 beam ( hereafter , denotes the intensity of the line ) . poglitsch et al . ( 1995 ) , using their imaging nir - spectrometer fast , observed intense emission around the central cluster r136 and showed that the source appears highly fragmented ( @xmath90 or 1 pc scale ) with a typical intensity of @xmath91 . our observations show that the emission in 30 doradus is very extended , @xmath92 pc , compared to the known extent of the co j=1@xmath470 emission ( @xmath93 pc , see the right map in figure [ fig-03 ] ) . at 15 ( or 22 pc ) from the ( @xmath64 ) position , the intensity is only a factor of two lower than at the peak ( see figure [ fig-03 ] ) . we also detected faint emission at ( @xmath94 ) , = @xmath95 , where emission had not been detected during the survey of the eso - sest key program ( booth 1993 ) . the n159/n160 complexes are 40 south of 30 doradus . the [ ] line ( israel et al . 1996 ) and the far - ir continuum distributions ( left map in figure [ fig-04 ] ) show that the far - uv fields are strong both near n159 and n160 . on the other hand , the co intensity around n159 is more than four times stronger than that around n160 . the h@xmath36 line had been detected at `` n159 blob '' ( a compact source with a size of @xmath96 , heydari - malayeri & testor 1985 ) by several groups . in table [ tbl-6 ] , we compare the previous data with our new results . the flux increases as the beam size increases . this suggests that , if there is a single source , the emission region is more extended than @xmath97 . alternatively , there could be clumpy emission filling our @xmath98 beam with an area covering factor @xmath99% of the covering factor in the inner @xmath100 region . in our survey , we observed very extended ( @xmath101 or 70 pc ) emission from the n159/n160 complex ( see the right map in figure [ fig-04 ] ) . in the n159 region , we detected emission where the co cloud complex is bright and extended ( see figure 1 in johansson et al . 1994 ) . in the n160 region , however , we also detected emission beyond the lowest contour level ( 3 ) in the map by johansson et al . ( 1994 ) . in spite of the weak or absent co emission in the n160 region , the observations indicate that the size of the molecular cloud complex is as big as that in the n159 region . [ [ section ] ] ~~ at several positions in the outer regions of 30 doradus and n160 , we detected emission where earlier co surveys failed to detect the line . in order to determine if all co is dissociated at these positions , we observed the regions again in the line ( see section [ sec : obs - co ] ) with higher sensitivity ( @xmath102 ) than the previous observations of which the lowest contour level was 3 ( booth 1993 ; johansson et al . 1994 ) . we made a fully sampled map inside the ut fps beam at ( @xmath94 ) in 30 doradus , and detected at a level of 0.5 ( see figure [ fig-03 ] ) . we also made maps in the outer regions of n160 where the previous map did not show any emission , and sampled some positions in n159 to confirm the flux calibration of the new observations ( see figures [ fig-04 ] and [ fig-05 ] ) . we plot the contours at logarithmic intervals to emphasize the edges of the molecular cloud complexes ; note the dense contour lines on the northwestern side and on the western side of the n160 complex . the emission regions cover the emission regions except the ( @xmath103 ) and the ( @xmath104 ) positions where the co cloud complex fills @xmath105 % of the beams . table [ tbl-5 ] lists the observed / and / line ratios in the magellanic clouds . observations of the lines in archetypal shocked regions , e.g. , orion bn - kl , hh 7 , and the supernova remnant ic 443 , show that the ratios of / are almost constant @xmath106 or @xmath107 k ( burton et al . 1989 ; richter , graham , & wright 1995 ) . assuming that the excited levels are in lte and @xmath108 k , the / ratio should be only @xmath109 . molecular hydrogen in pdrs absorbs @xmath110 nm photons in the @xmath111 lyman and @xmath112 werner bands . about 15% of the electronically excited molecules are dissociated ( draine & bertoldi 1996 ) . the remaining 85% of the excitations result in populations of various ro - vibration levels of the ground electronic state . if @xmath65 @xmath113 , the relative line intensities arising in uv - excited are insensitive to density or to uv field strength ( black & van dishoeck 1987 ) . in this pure - fluorescent transition case , the ratio of / is 0.56 and / is 0.38 . at densities @xmath114 @xmath113 ( luhman et al . 1997 ) , the collisional de - excitation of uv - pumped begins to affect the ro - vibrational level populations . if the pdr boundaries become sufficiently warm ( @xmath115 k ) and dense ( @xmath114 @xmath113 ) , collisional excitation thermalizes the low vibrational ( @xmath116 ) level populations ( sternberg & dalgarno 1989 ; luhman et al . 1997 ) . the detections of line in the magellanic clouds ( table [ tbl-5 ] ) verify the uv - excitation of . the observed line ratios from the n160 region are those expected for pure - fluorescence . the ratios toward other regions show that the ro - vibration levels may be affected somewhat by collisions . in 30 doradus , the peak intensity is @xmath117 ( poglitsch et al . 1995 ) which is brighter than the maximum predicted by our pdr models ( @xmath118 @xmath119 @xmath120 , see figure [ fig-12 ] ) . the models may underestimate the intensity by as much as a factor of two in regions with high densities and high uv fields by neglecting the large increase in uv pumping in the very warm ( 420 - 600 k ) region at the cloud edge . alternatively , there may be some collisional excitation of v=1 as well as collisional de - excitation of the high v states , as one finds in clouds subjected to intense far - uv fields ( luhman et al . we compare the , [ ] 158 , , and far - ir emission from the clouds in the magellanic clouds ( see table [ tbl-4 ] ) to the emission from star forming clouds in orion and ngc 2024 for which we have comparable data sets ( see table [ tbl-7 ] ) . in the lmc , we select positions where complete , [ ] , and data sets exist : 5 positions in the 30 doradus region , 4 positions in the n159 region , and 8 positions in the n160 region . in the smc , we use 4 positions with only and data . for the galactic cloud data , we make use of the published , [ ] 158 and data in schloerb & loren ( 1982 ) , stacey et al . ( 1993 ) , jaffe et al . ( 1994 ) , luhman et al . ( 1994 ) , luhman & jaffe ( 1996 ) , and luhman et al . the far - ir data are from the hires processed iras data ( see section [ sec : far - ir ] ) . in the orion molecular cloud , stacey et al . ( 1993 ) made two [ ] strip maps in right ascension , both of which pass close to the peak . the [ ] flux along the strips was `` bootstrapped '' , integrated by assuming zero flux at the ends of the cut and summing the chopped differences . the data from _ cut 1 _ ( observed west to east ) and _ cut 2 _ ( observed east to west ) are not in complete agreement ; therefore , we take only those data with : @xmath121 we also exclude positions where the iras data are saturated , i.e. , @xmath73 @xmath122 1500 mjy / sr , and where the ro - vibrational level populations begin to show effects of collisional de - excitation ( / @xmath65 0.15 , see table 2 in luhman & jaffe 1996 ) . some positions where collisional de - excitation was unlikely were not observed in the line but are included in the data we use ( see the discussion in section [ sec : hh_cii ] ) . table [ tbl-7 ] lists the compiled data sets in the galactic clouds : 4 positions from the orion a molecular cloud ( hereafter `` orion '' ) , and 11 positions from the cloud associated with the ngc 2024 region ( hereafter `` ngc 2024 '' ) . figure [ fig-06 ] compares the data from 15 positions in the galactic clouds with the data by jaffe et al . ( 1994 ) who presented and [ ] data from ngc 2024 , and showed that the different parts of the source have very different distributions on a @xmath123 versus @xmath124 plot . the left plot in figure [ fig-06 ] shows the distinction between _ the cloud proper zone _ ( @xmath43 @xmath122 @xmath125 with respect to ngc 2024 irs 1 , shown as open circles ) and _ the western edge zone _ ( @xmath43 @xmath65 @xmath125 , shown as open triangles ) in ngc 2024 . data from the cloud proper zone agree with the pdr models of wolfire et al . ( 1989 ) , while many of the / ratios toward the western edge zone are much higher than any of the model ratios . jaffe et al . ( 1994 ) interpreted the western edge zone results as implying that the mean column density of clouds decreases to the west . the right plot in figure [ fig-06 ] shows the location in the @xmath126 space of the data sets in table [ tbl-7 ] , which will be used to compare with those from the magellanic clouds . the intensity of the interstellar far - uv radiation field is usually expressed in terms of the scaling factor , the mean intensity in the solar neighborhood ( draine 1978 ; also see footnote 2 in black & van dishoeck 1987 ) . the critical part of the uv range for fluorescence , c ionization , and co dissociation is 91.2 @xmath127 113 nm ( black & van dishoeck 1987 ; van dishoeck & black 1988a ) . the intensity of the @xmath128 field integrated over 91.2 @xmath127 113 nm is @xmath129 . note that some authors ( e.g. , tielens & hollenbach 1985 ) use habing s ( 1968 ) determination of the local far - uv field , @xmath130 , for the integrated intensity between 6 @xmath131 13.6 or 91.2 @xmath127 206.6 nm , and use the symbol @xmath132 to indicate the degree of enhancement over this standard integrated intensity . the corresponding intensity of the draine field integrated over the same band as used for @xmath132 is @xmath133 ( for a detailed comparison between the draine field and the habing field , see draine & bertoldi 1996 ) . most of the far - uv energy is absorbed by grains and reradiated in the far - ir : @xmath134 where the factor of 2 accounts for incident radiation at longer wavelengths , @xmath135 nm ) ( wolfire , hollenbach , & tielens 1989 ) . we can use the far - ir emission to normalize the various line intensities to compensate for beam filling factor effects . before we do so , however , we need to understand how the far - ir emission arises . figure [ fig-07 ] shows versus @xmath76 for the galactic star forming cloud positions and the magellanic clouds positions in our sample . we derive from the measured 60 and 100 intensities using equation [ eq : i_fir ] , and derive @xmath76 using equation [ eq : t_dust ] . assuming that the dust is heated by an external far - uv field , we can approximate the results of pdr models : @xmath136 where @xmath137 is the equivalent stellar surface temperature which would produce the incident uv field ( hollenbach , takahashi , & tielens 1991 ; spaans et al . 1994 ) . assuming @xmath137 = @xmath138 k , the expected relation between @xmath76 and is : @xmath139 which is shown as a dotted line in figure [ fig-07 ] . the observed @xmath76 and distributions in orion and ngc 2024 agree with the model in equation [ eq : tfir_model ] . is proportional to the beam filling factor , while @xmath76 , which was deduced from the ratio of @xmath140 , is independent of the beam filling factor . with a beam size of @xmath63 ( or @xmath141 pc at the distance of orion ; 415 pc , anthony - twarog 1982 ) , the projected beam filling factor for the clouds in orion and ngc 2024 is @xmath142 , which explains the agreement between the observed @xmath143 relation and the model . if we assume that the dust size distribution is independent of metallicity and that the clouds are optically thick in the far - uv , the @xmath76-@xmath144 relation should be independent of the dust abundance ( see section [ sec : self - shielding ] for more discussion ) . for the magellanic clouds , the observed values are an order of magnitude weaker than the values predicted by the @xmath143 relation . this difference implies a beam filling factor in the magellanic clouds of @xmath145 ( @xmath146 kpc , westerlund 1990 ) . the line and continuum emission we describe here arises in the layers of molecular clouds where uv photons influence the chemistry and the physical conditions . various parameters affect the emergent intensity : @xmath147 where @xmath148 is the _ observed _ , [ ] 158 , or emission line intensity , and is the _ observed _ far - ir continuum intensity in units of . @xmath149 is the combined intensity arising from the front and back sides of a plane - parallel model cloud which fills the beam . the front and back surfaces of the cloud are exposed to external far - uv radiation , and are perpendicular to the line - of - sight . @xmath150 depends on the hydrogen number density , @xmath151 = @xmath152 + @xmath153 , where @xmath152 and @xmath154 are the atomic and molecular hydrogen density , respectively ; the far - uv field strength , @xmath155 ; the doppler velocity dispersion , @xmath156 ( or half - width at 1/e point ) ; and the metal abundance , @xmath21 . @xmath157 is a beam filling factor and @xmath158 is a geometric correction factor which we describe below . unlike the substructure of the fully molecular interior of clouds , which may have many size scales , the penetration scale length of far - uv photons and uv photochemistry insure that structures within giant molecular cloud ( gmc ) complexes with uv - illuminated surfaces and co - bearing cores have a minimum size determined by their density and metallicity . measurements and theoretical arguments show pdr structures outside the dense cores of molecular clouds are clumpy on a size scale of @xmath142 pc ( burton , hollenbach , & tielens 1990 ; jaffe et al . 1994 ) , comparable to or larger than the orion beam but very much smaller than the beams in the lmc . since the measured intensity is a beam average over any source structure , we have to correct for the effects of different beam filling factors in order to directly compare the emitted intensities from the lmc with those from the galactic clouds . we define the beam filling factor , @xmath157 , as the fraction of the observed beam area filled by a single cloud ( not an ensemble of clouds ) , using the outermost edge as the cloud boundary ( or the boundary between and regions ) : @xmath159 where @xmath160 is the radius of the cloud to the outer ( /h ) boundary , @xmath161 is the distance to the cloud , and @xmath25 is the diameter of the telescope equivalent disk in radians . @xmath157 is same for all types of emission in the same cloud . in order to simplify model simulations in section [ sec : models ] , we assume that only one cloud is in the telescope beam , the telescope has a box - function beam profile , and the telescope beam area is always larger than the cloud size , i.e. , @xmath162 @xmath114 @xmath163 or @xmath157 @xmath119 1 . it may be more realistic to consider three dimensional rather than planar clouds . in case of a three dimensional cloud with an external far - uv field whose intensity is uniform on the surface of the cloud , the far - ir , [ ] , and emission arise in the outer shells of the cloud , while the emission arises from the surface of the co core inside the cloud . we use a `` geometric correction parameter '' , @xmath164 or @xmath165 , in equation [ eq : intensities ] to simulate the _ observed _ intensity of a three dimensional cloud . the spherical geometry parameter accounts for limb - brightening and for the size differences between the co , [ ] , , and far - ir emission regions in a given spherical cloud . when examining our three dimensional model clouds , we can eliminate the beam filling factor , @xmath157 , by dividing @xmath166 by : @xmath167 we will therefore divide the observed intensities by the far - ir intensity at each position as a means of removing distance - related beam filling factor effects in our subsequent analysis of the data from the galaxy and the magellanic clouds . table [ tbl-8 ] shows the average ratios between the observed , , and intensities , and the standard deviations of the ratio distributions in the galactic clouds and in the magellanic clouds . figures [ fig-08 ] , [ fig-09 ] , and [ fig-10 ] show plots of @xmath169 versus @xmath170 , @xmath170 versus @xmath171 , and @xmath169 versus @xmath171 , for the galactic clouds and for the clouds in the magellanic clouds . from these figures and table [ tbl-8 ] , the line ratios of @xmath172 between the galaxy and the magellanic clouds are in good agreement , and the ratios of @xmath173 and @xmath174 in the magellanic clouds are slightly higher ( but in agreement within the standard deviations ) than those in the galaxy . section [ sec : models ] will discuss our pdr models and compare the observed data with pdr models with various input parameters . using plane - parallel codes , the metallicity dependence of pdr structure and of emergent line intensities has been calculated and discussed by several authors ( maloney & black 1988 ; van dishoeck & black 1988b ; wolfire , hollenbach , & tielens 1989 ; maloney & wolfire 1997 ) . the emergent intensities of the , [ ] , and lines from pdrs depend on @xmath151 , , @xmath156 , and @xmath21 ( see section [ sec : dividefir ] ) . as long as the line is optically thick , the resulting intensities of , [ ] , and lines are not very sensitive to the metallicity . a simplified analysis can illustrate the reason for this insensitivity . in the outer part of the pdr , gas - phase carbon is in the form of . when the metallicity is lowered , the number density of ions drops , but the far - uv photons penetrate deeper into the clouds because the dust - to - gas ratio is also lower . thus , the column density of in the pdr is almost independent of the metallicity . the [ ] line is optically thin , and the intensity is proportional to the column density of . the intensity depends much more weakly on the co column density because the line is optically thick and most of emission arises on the surface of the co region inside the cloud . therefore , the intensity outside of very high column density cloud cores does not depend strongly on the metallicity when one considers emission from a plane - parallel slab . we will discuss the metallicity dependence of in sections [ sec : self - shielding ] and [ sec : without_epsilon ] . as we discussed in section [ sec : dividefir ] , in the case of three dimensional clouds , we use the parameter , @xmath175 or @xmath165 , to account for effects such as limb - brightening and the size differences between the outer shells from which far - ir , [ ] , and emission arises and the inner co core from which emission arises . the depths from the cloud surface to the /c transition layer and to the h/ transition layer are about inversely proportional to the dust abundance , so the metal and dust abundance are more important parameters in spherical - shell pdrs than in plane - parallel models . strzer , stutzki , & sternberg ( 1996 ) and mochizuki ( 1996 ) modeled a pdr on the surface of a spherical cloud with galactic metallicity . this section presents our own models of spherical - shell pdrs and calculate the emergent line intensities for a range of densities , uv fields , and metallicities . we first ran a plane - parallel pdr code ( van dishoeck & black 1986 ; black & van dishoeck 1987 ; van dishoeck & black 1988a ; jansen et al . 1995 ) with a range of densities ( @xmath151 = @xmath176 , @xmath177 , and @xmath113 ) , uv fields ( @xmath178 = @xmath179 , @xmath180 , @xmath41 , and @xmath181 ) , and metallicities ( for the galaxy , the lmc , and the smc ) . in this code , one side of the model cloud is exposed to uv radiation , and the cloud is divided into 200 slabs , each of which is in chemical steady state . since the code only includes inelastic collisions of within v=0 , the ro - vibrational level populations of are not correct at @xmath151 @xmath114 @xmath182 . the pdr code calculates the gas temperature and chemical abundances in each slab . since there is no detailed information on the grain properties in the lmc and smc , we assume that the formation rate of on grains scales linearly with the dust abundance ( section [ sec : self - shielding ] ) and that the heating efficiencies of the grains are the same as in the galactic case . in practice , the lower grain formation rate is accomplished in the models by decreasing the parameter @xmath183 ( black & van dishoeck 1987 ) from 1.0 to 0.25 and 0.1 in the lmc and smc , respectively . table [ tbl-9 ] lists input parameters for the code . we employ spherically symmetric cloud models to study the effects arising in three dimensional chemical structures . the outputs , e.g. , chemical abundances and kinetic temperatures , from the plane - parallel code are applied to the spherical shells of the cloud used in the radiative transfer model . we map the temperatures and abundances derived at each distance from the / interface by the plane - parallel model into the spherical radiation transfer model , ignoring any changes in the chemistry or thermal balance due to the difference in geometry . the monte carlo code of choi et al . ( 1995 ; hereafter mc ) calculates the level populations of the atoms and molecules . the mc code simulates photons in a one - dimensional ( spherical ) cloud , and adjusts the level populations according to the result of simulations until the populations converge . we assume a purely turbulent velocity field with a doppler velocity dispersion of 1 ( @xmath156 or half - width at 1/e point ) with no systematic motion . since the mc calculation only includes 40 slabs , the 200 slab plane parallel model was smoothed in such a way as to retain high resolution at the transition regions ( li 1997 ) . we use the output of mc to calculate emission line profiles using a virtual telescope code ( choi et al . 1995 ; hereafter vt ) . the vt code convolves the integrated emission from each spherical shell along the line - of - sight with a virtual telescope beam profile to simulate observations . while it would be better to examine the geometric effects using fully self - consistent models which calculate the detailed excitation , photodissociation , and and line intensities , our present three - step procedure should be good enough to allow us to analyze the global properties of galactic and lmc clouds and to establish relative trends . inside neutral clouds , the external far - uv field is attenuated by dust absorption , and by c and absorption . co absorption of far - uv is negligible in the outer part of the cloud because most of the carbon is ionized . can survive in the outer parts of the cloud , either as a result of self shielding or as a result of shielding by dust . we can analyze the conditions under which self - shielding from far - uv photons is dominant over shielding by dust ( adapted from burton , hollenbach , & tielens 1990 ) . within a plane - parallel cloud , the formation rate , @xmath184 , at depth @xmath185 measured from the surface of the cloud toward the center is @xmath186 where @xmath151 is the hydrogen number density , @xmath151 = @xmath187 + @xmath188 , assumed to be constant over the cloud ; @xmath187 and @xmath189 are number densities , at @xmath185 , of h atoms and molecules respectively ; @xmath190 is the value of the formation rate coefficient ( a quantity which depends linearly on the dust abundance ) for the galactic dust abundance ; @xmath191 is the dust abundance normalized to the galactic value , i.e. , @xmath192 = 1 . @xmath190 is a slowly varying function of the gas temperature ( @xmath190 @xmath193 @xmath194 ) , so we take an average value ( @xmath195 @xmath196 s@xmath197 , burton , hollenbach , & tielens 1990 ) for this analysis . if we assume that , at the position we are considering , the dust optical depth is negligible and the absorption is governed by the square root portion of the curve of growth ( jura 1974 ) the far - uv field is attenuated by @xmath198 , where @xmath199 the destruction rate , @xmath200 , at depth @xmath185 is @xmath201 where @xmath202 is the unshielded dissociation rate of at = 1 ( @xmath203 s@xmath197 , black & van dishoeck 1987 ) ; and @xmath204 is the self - shielding parameter ( @xmath205 cm@xmath197 , jura 1974 ) . we can integrate the steady state equation , @xmath184 = @xmath200 , over @xmath185 : @xmath207 by substituting equation [ eq : n_hh ] into the above equation , we get : @xmath208 self - shielding is more important than dust shielding , when the far - uv field attenuation by dust is still negligible , @xmath209 at the point , @xmath210 , where the molecular hydrogen column density becomes equal to the atomic hydrogen column density , @xmath211 @xmath212 is the optical depth of dust at @xmath37 = 100 nm . even though the ratio of @xmath213 depends on the chemical composition and the size distribution of the dust , the change of this ratio from the galaxy to the lmc and the smc is negligible compared to that of the dust - to - gas ratio , e.g. , the ratios of @xmath214 in the galaxy , the lmc , and the smc are 3.4 , 4.1 , and 5.1 , respectively ( pei 1992 ) , while @xmath215 changes by a factor of 10 ( table 9 ) . to simplify our analysis , we assume that the ratio of @xmath213 is constant and that only the dust - to - gas ratio ( adopted from bohlin et al . 1983 and black & van dishoeck 1987 ) depends on the metallicity : @xmath216 and @xmath217 where @xmath218 is the dust - to - gas ratio normalized to the galactic value ( see table 9 ) , and @xmath219 is the hydrogen column density , @xmath219 = @xmath220 + @xmath221 . substituting equation [ eq : n_h_condition ] into equation [ eq : form_dest ] , we obtain the following relationship between the column density at which self shielding becomes effective , and the density , metallicity , and strength of the incident uv field : @xmath222 from equations [ eq : tau_condition ] , [ eq : tau_dust ] , [ eq : rho_dust ] , and [ eq : n_h_result ] , we derive the conditions under which the self - shielding is dominant over shielding by dust : @xmath223 where we assume that the optical absorption properties and the efficiency for formation of the dust per unit h atom vary in the same way as with dust abundance : @xmath224 . figure [ fig-11 ] shows the linear size of the region ( ) and the region ( , denotes vibrationally excited ) from the pdr model results : and are the distances from the surface of the cloud to the inner edges of the region where @xmath225 = @xmath226 , and of the region where @xmath227 = @xmath228 respectively . the plot of in figure [ fig-11 ] shows how self - shielding affects the depth of the region from the surface of the cloud . based on equation [ eq : self - shielding ] , the dotted lines divide the ( , @xmath151 , ) space into a region where dust absorption is dominant and a self - shielding dominant region . we can predict the behavior of as a function of , @xmath151 , and @xmath218 , using the relations derived in this section . when dust absorption is more important than self - shielding , the zone ends when the far - uv field is attenuated to a fixed level , independent of the incident field . as a result , changes in result in a variation of @xmath229 according to : @xmath230 where @xmath231 is a constant . substituting equations [ eq : tau_dust ] and [ eq : rho_dust ] into the above equation gives that @xmath232 where @xmath233 is the hydrogen column density at . in other words , if we increase by an order of magnitude , @xmath233 increases only by an additional additive factor of @xmath234 @xmath235 . when self - shielding is dominant , from equation [ eq : n_h_result ] , @xmath236 e.g. , if decreases by an order of magnitude , @xmath233 decreases by two orders of magnitude . the above equation explains , why , when self - shielding dominates , the depth of the shell decreases rapidly as decreases or as @xmath151 increases ( see figure [ fig-11 ] ) . in the lmc , @xmath237 @xmath238 ( see equation [ eq : self - shielding ] ) , and the dotted line is shifted toward lower by only a factor of 2.2 . even in the case of the smc , @xmath239 @xmath240 . variations in metal abundance , therefore , do not significantly affect the self - shielding criterion . we first ran the mc and vt codes to obtain results where the spherical natures of the model clouds , e.g. , limb - brightening and the geometrical size differences ( see section [ sec : dividefir ] ) , are not important by setting the virtual telescope beam size ( @xmath241 ) much smaller than the cloud size ( @xmath163 ) . this permits us , in effect , to obtain the line intensities from a plane - parallel cloud with a finite thickness whose front and back surfaces are exposed to external far - uv radiation . the surfaces of the cloud are perpendicular to the line - of - sight . these models are equivalent to setting @xmath157 = 1 and @xmath164 = 1 in equation [ eq : intensities ] : @xmath166 = @xmath149 . the resulting [ ] intensity is from both the front and back surfaces , because the [ ] line emission is nearly optically thin ( stacey et al . the line is optically thick , so the resulting emission comes predominantly from the front surface . the vt code does not calculate the ro - vibrational lines of . we derive the emission from the column density , @xmath242 , which results from the plane - parallel pdr model . the electronically excited molecules relax radiatively to the ground electronic state and then cascade through the vibrational energy levels by emitting ro - vibrational lines . in this process , the line ratios are insensitive to @xmath151 and , and we can use a constant conversion factor between @xmath242 and intensity ( black & van dishoeck 1987 ) : @xmath243 where @xmath244 is the intensity from the front surface of the cloud . while the ro - vibrational lines of are optically thin , the emission from the back surface , @xmath245 , is affected by the extinction , , through the cloud itself . using = @xmath246 ( rieke & lebofsky 1985 ) and equation [ eq : rho_dust ] , we estimate the observed intensity from the back surface : @xmath247 as we will discuss in section [ sec : with_epsilon ] , the cloud size ( in units of hydrogen column density @xmath248 ) has a lower limit to keep the co molecules from being dissociated completely in an intense far - uv field ( @xmath122 @xmath41 ) . figure [ fig-13 ] shows that @xmath249 should be larger than @xmath250 ; therefore , @xmath251 we will neglect the emission from the back side of the cloud in the following discussion : @xmath252 @xmath3 @xmath244 . in some cases , there may be a bit of additional flux from the back side which can penetrate through the thinner parts of the cloud . figure [ fig-12 ] shows the [ ] , , and emission intensities from the pdr code and the mc / vt code for the two sided planar clouds . when dust absorption dominates ( equation [ eq : self - shielding ] ) , the intensity increases as @xmath151 increases . on the other hand , when self - shielding dominates , the intensity increases as increases . in the lmc model ( the right plot in figure [ fig-12 ] ) , the intensity is enhanced by a factor of @xmath253 over the intensity in the galactic model with the same @xmath151 and . the enhancement is mainly due to the different self - shielding criteria in the galactic model and in the lmc model . in order to understand the effect of limb - brightening and of differences in the physical sizes of the , , and co zones in spherical clouds of varying metallicity , we ran the vt code , setting the virtual telescope beam size ( @xmath25 ) to match the cloud size ( @xmath163 ) . the resulting intensities are equivalent to setting @xmath157 = 1 in equation [ eq : intensities ] : @xmath166 = @xmath254 . in figure [ fig-13 ] , we plot the model values of , , and versus cloud size . in figures [ fig-06 ] , [ fig-09 ] , and [ fig-10 ] , we overlay the results on the observed data . and are obtained directly from the vt code , and the far - ir continuum emission and are from equation [ eq : intensities ] with @xmath165 and @xmath255 derived as follows : as we discussed in section [ sec : dividefir ] , the effects of spherical geometry include both limb - brightening and size differences between the different emission regions in the cloud . the far - ir emission has very low optical depth and the emission is optically thin . the far - ir and emission regions lie on the surface of the cloud and fill the telescope beam . we can analyze the spherical geometry effects for optically thin lines by projecting the three dimensional emission shell onto a two dimensional emission disk : @xmath256 where @xmath257 is the depth from the cloud surface to the transition regions , e.g. , and which are defined in section [ sec : self - shielding ] ; and @xmath160 is the radius of the cloud . in the above equation , we assume that @xmath258 and @xmath259 are constant in the corresponding shells . for the far - ir emission , if we consider that the incident far - uv energy is conserved as the output far - ir energy ( see equation [ eq : g0-i_fir ] ) , we can write : @xmath260 we use equation [ eq : eps1 ] to obtain @xmath255 , and equation [ eq : eps2 ] to obtain @xmath165 = 2 . the line arises on the surface of the co core . because external uv radiation provides most of the heating , the gas temperature gradient is positive , @xmath261 @xmath122 0 ( @xmath262 at the cloud center ) . as we observe an optically thick line , the edge of the projected co disk has a higher brightness temperature than the center , and we see limb - brightening . on the other hand , the size of the co core ( @xmath263 ) is smaller than the cloud size ( @xmath160 ) which is affected by and @xmath151 . when @xmath160 is larger than , and is optically thick : @xmath264 when we compare the spherical cloud models to plane - parallel models , the change in the co core size ( @xmath263 ) is more significant than the limb - brightening effect . we plot the co intensity ( by setting @xmath157 = 1 ) resulting from the numerical code in dotted lines in figure [ fig-13 ] . this method also takes into account any additional effect of opacity variations on the local co emissivity . as @xmath263 @xmath47 0 , the resulting decreases rapidly . as we discussed in sections [ sec : without_epsilon ] and [ sec : with_epsilon ] , the observed far - ir , [ ] , and line intensities depend almost exclusively on @xmath151 and , while the observed line intensity depends on @xmath151 , , @xmath21 , and @xmath160 . in this section , we apply the model results to the data and estimate @xmath151 , , and @xmath160 in the regions we observed . in table [ tbl-8 ] , the standard deviations of the observed @xmath172 are smaller than those of other ratios ( @xmath265 versus @xmath266 ) , because the values of @xmath267 and @xmath252 depend similarly on @xmath151 and ( see figure [ fig-12 ] ) , and in the case of spherical clouds , the [ ] and intensities are not very sensitive to variations in cloud size , i.e. , 2/3 @xmath65 @xmath268 @xmath65 2 ( see equation [ eq : epsilon ] and figure [ fig-13 ] ) . in figure [ fig-08 ] , most of the data from ngc 2024 are within the model grids at 3.2 @xmath269 4.2 , and 1.5 @xmath270 3 . there are positions ( at @xmath43 = @xmath271 , @xmath45 , and @xmath272 ) in ngc 2024 , however , where the intensities lie at least a factor of 2.5 above the model grids . at these positions , the clouds may be small enough to be transparent for emission from the back side of the cloud ( see equation [ eq : ih2back ] ) . based on our experience with other positions , it is extremely unlikely that the level populations at these positions are thermalized by effects present in high density pdr s ( @xmath151 @xmath273 ) or by shocks ( see section [ sec : excitation ] ) . the ratios of @xmath172 in orion are larger than those in ngc 2024 . the orion data in figure [ fig-08 ] lie slightly below and slightly above models for @xmath274 = 4.7 , implying that the gas density in orion is higher than that in ngc 2024 . the mean ratios of @xmath172 in 30 doradus , n159 , and n160 are similar to each other , suggesting that the observed clouds in the lmc have similar densities , and that collisional de - excitation does not affect the level populations ( see table [ tbl-8 ] ) . in figure [ fig-08 ] , most of the data in the lmc match models in the range of 3.7 @xmath269 4.7 and 2 @xmath270 3 . by comparing the observed data and the model results in figure [ fig-08 ] , we measured the median density and far - uv field for each source . with this derived @xmath151 and , we found the region depth , , using figure [ fig-11 ] . the values of @xmath151 , , and determined in this way are listed in table [ tbl-10 ] . both cloud size differences and metallicity differences affect the observed / and the / ratios . for the clouds in ngc 2024 , the results of jaffe et al . ( 1994 ) implied that the size of the clouds ( @xmath160 ) in the western edge zone is smaller than in the cloud proper zone ( see section [ sec : galactic_data ] ) , while the region depth , , in the cloud is not much different . as we discussed in section [ sec : with_epsilon ] , for a given @xmath151 and , is proportional to @xmath275 . in the lmc , the dust - to - gas ratio , @xmath22 , is lower than in the galaxy by the factor of @xmath276 , and the far - uv radiation penetrates deeper into the clouds ( figure [ fig-14 ] ) : @xmath277 if we assume that the cloud sizes in the galaxy and in the lmc are same , then @xmath278 is less than @xmath279 . if @xmath278 @xmath3 @xmath279 , the cloud size in the lmc must be larger than the cloud size in the galaxy ( see figure [ fig-14 ] ) . the / ratio is also affected by the strength of the uv field . mochizuki et al . ( 1994 ) compared the @xmath173 ratios in the lmc with the ratios in the galactic plane . their log - ratio in the lmc , @xmath280 , agrees well with our ratio in 30 doradus , n160 , and n159 ( @xmath281 , see table [ tbl-8 ] ) ; however , their ratio in the galactic plane , @xmath282 , is much smaller than the ratio ( @xmath283 , see table [ tbl-8 ] ) in orion and ngc 2024 . we argue that the _ galactic plane _ at @xmath284 @xmath285 @xmath286 has a different physical environment from active star formation regions like orion and ngc 2024 . most of the @xmath287 ratios at the positions in the galactic plane are between 0.2 and 0.3 ( see figure 2 in nakagawa et al . applying these ratios to equations [ eq : t_dust ] , [ eq : g0-i_fir ] , and [ eq : tfir_model ] ( see also equation [ eq : i_uv - t_dust ] ) , we can estimate that the far - uv fields in the galactic plane are in the range of 1.5 @xmath270 1.7 . the far - uv field is , therefore , much lower than toward the positions we observed in orion and ngc 2024 ( 1.9 @xmath270 3.2 ) . our model results in figure [ fig-12 ] show that , at @xmath288 = 1.6 and @xmath274 = 3.7 , the @xmath173 ratio is 3.2 , which agrees with the observed value of the galactic plane in mochizuki et al . on the other hand , the far - uv fields in 30 doradus , n160 , and n159 are very high ( 2.0 @xmath270 3.8 ) . even the extended regions in the lmc in which mochizuki et al . ( 1994 ) measured the @xmath173 ratio show higher far - uv fields ( 1.8 @xmath270 2.7 , derived from the @xmath287 ratios ) than those in the galactic plane . the more intense far - uv fields , e.g. , in orion , ngc 2024 , 30 doradus , n160 , and n159 , dissociate more co molecules and have deeper regions , , in equation [ eq : epsilon ] ; therefore , the / ratio becomes larger . figure [ fig-09 ] plots the observed data , / versus / , and overlays the results from the planar cloud models and the spherical cloud models . for the spherical cloud models , the tick marks along dashed lines mark the position in the space of models with different column densities , @xmath289 . the locus of each model comes from the value of = @xmath290 and = @xmath291 determined for that column density . the solid lines and dotted lines show the results of the planar cloud model ( = @xmath267 and = @xmath292 ) . only the data from the cloud proper zone in ngc 2024 are within the planar cloud model space , and most of the observed data deviate systematically toward lower / or higher / from the model results . the models do not include spherical geometry effects ( see section [ sec : without_epsilon ] ) , and equation [ eq : eps1 ] implies that the observed data should not be significantly affected by going from plane - parallel to spherical geometry . therefore , the offset of observed points from the models on the versus plot in figure [ fig-09 ] is solely a result of the effect of spherical geometry on the observed emission ( @xmath275 ) . given the observed value of / , we measured how much the observed / is shifted from the point expected from the models for a given density and far - uv intensity on the model grids in figure [ fig-09 ] : @xmath293 applying the measured and @xmath275 to equation [ eq : epsilon ] , we derived the column densities through the clouds ( @xmath289 in table [ tbl-10 ] ) under the assumption that the clouds are spherical . for the data from the cloud proper zone in ngc 2024 , we use the uncertainty in the observations to set the lower limit , i.e. , @xmath275 @xmath114 @xmath294 . madden et al . ( 1997 ) also estimated @xmath275 of the nearby irregular galaxy , ic 10 , using far - ir , [ ] , and data . in table [ tbl-10 ] , we list the adopted values of and @xmath275 at positions a and b in ic 10 . using the dust - to - gas ratio of ic 10 , @xmath295 @xmath3 0.2 , and the pdr models , we estimate @xmath296 and then @xmath289 . models indicate that , like , is not affected strongly by changing from planar to spherical geometry . therefore , the similar offset of the data from models in figure [ fig-10 ] is probably due to the inappropriateness of applying the plane - parallel models for predicting . as were the data in figure [ fig-09 ] , only the data from the cloud proper zone in ngc 2024 are within the model space , and most of the observed data deviate systematically leftward from the model results . the data from the smc , which are not available in figure [ fig-09 ] , also deviate toward lower @xmath297 or higher @xmath298 from the lmc model results , if we assume that the smc models without spherical geometry effects are not very different from the lmc models . in the smc , the observed emission could be higher than the model results as a result of the unreddened emission from the back side of the cloud . because [ ] data are not available and the data are not certain , we can not measure @xmath299 from the versus plane . we can still estimate a lower limit on the column density of clouds in the smc , assuming @xmath275 @xmath300 . applying the dust temperature of @xmath301 k from iras 60/100 ratios ( see figure [ fig-07 ] ) to equations [ eq : g0-i_fir ] and [ eq : tfir_model ] , @xmath302 we obtain a far - uv field ( ) of @xmath303 at the observed positions in the smc . assuming @xmath274 = 3.7 , we get @xmath304 = @xmath305 . therefore , the central column density ( @xmath248 ) of the clouds in the smc should be at least @xmath306 ( see table [ tbl-10 ] ) . we estimate cloud column densities in the molecular complexes in the galaxy and in the magellanic clouds . we first derive @xmath151 and by comparing the observed data and the model results in the / versus / plot ( figure [ fig-08 ] ) . we take a median value of the data in each object , except in n159 where we consider each position separately . using the values of @xmath151 and derived from figure [ fig-08 ] , we get , the depth from the surface of the cloud to the -to - c - to - co transition layer , in figure [ fig-11 ] . we also get @xmath275 , the parameter that accounts for spherical geometry , by comparing the data and models in figure [ fig-09 ] . when we do not have or data , we assume that @xmath274 = 3.7 , and estimated from dust color temperature using @xmath287 and equations [ eq : t_dust ] and [ eq : i_uv - t_dust ] . we finally derive the central column densities of clouds using equation [ eq : epsilon ] , and list them in table [ tbl-10 ] ( see also figure [ fig-15 ] ) . note that the column densities derived in this way are a measure of the number of molecules between the /h boundary and the cloud center , not of the overall column denisyt through the gmc complexes . the cloud column density measured in the western edge zone in ngc 2024 is at least a factor of 5 smaller than the column density measured in the cloud proper zone in ngc 2024 . at the four data positions in n159 , the variations of @xmath151 and are not significant , e.g. , 3.7 @xmath269 4.1 , and 2.3 @xmath270 2.7 ; however , the intensities change by a factor of @xmath307 , indicating that the column density varies more than a factor of 5 within the n159 region . the cloud sizes represented by the hydrogen column density for a given density , @xmath151 = @xmath177 , in the lmc ( @xmath308 = @xmath309 ) , the smc ( @xmath310 ) , and ic 10 ( @xmath311 ) are bigger than those in the galaxy ( @xmath312 = @xmath313 ) . table [ tbl-10 ] and figures [ fig-14 ] and [ fig-15 ] show that the column density and , for constant density , cloud size depend roughly linearly on metallicity and therefore the visual extinction through clouds is constant as one goes from the galaxy to the lmc and smc . mckee ( 1989 ) presented a theory of _ photoionization - regulated star formation _ ( see also bertoldi & mckee 1996 ) . according to his theory , the rate of low - mass star formation is governed by ambipolar diffusion in magnetically supported clouds , and the ionization of gas is mostly by far - uv photons . he suggested that the mean extinction of molecular clouds in dynamical equilibrium is in the range of = @xmath314 mag . using equation [ eq : rho_dust ] , the range of extinction corresponds to : @xmath315 this range of column density roughly agrees with the empirical values of galactic gmcs using and virial theorem : @xmath248 = @xmath316 ( scoville et al . 1987 ; solomon et al . 1987 ) . in regions with higher incident uv fields , the equilibrium range should move to higher column density . our results from the galaxy , the lmc , the smc , and ic 10 are also within = @xmath317 mag , and agree with the theory to within a factor of two for the range of observed incident fields . we observed , , and lines from the magellanic clouds , and detected emission even in regions where co had not been seen during the eso - sest key program , implying that is present in these regions . with deeper co observations , we also detected weak ( @xmath318 ) emission in pdrs where the previous eso - sest program did not detect . we conclude that , in all regions where we could detect , co is not completely dissociated and that the emission is still optically thick in the magellanic clouds . our observations may have consequences for studies using co to trace molecular mass . our spherical - shell cloud model demonstrates the effects of limb - brightening and differences in the physical sizes of the , , and co zones on the observed line strengths . the , [ ] , and line intensities from the plane - parallel cloud model are not very sensitive to the metallicity . when geometric effects are taken into account using spherical models , the , and [ ] line intensities normalized to the far - ir intensity are also not very sensitive to the metallicity ; however , the intensity is proportional to the surface area of the co core . as we decrease the cloud size ( or total column density , @xmath248 ) , the co intensity normalized to the far - ir intensity decreases . we compiled data of , , [ ] , and far - ir in the star formation regions in the magellanic clouds and in the galaxy , and compared them with simulated line intensities from a plane - parallel cloud . the data in the / versus / plot agree with the model results , and show no significant difference between the magellanic clouds and the galaxy . we use the / versus / plot to estimate @xmath151 and at the positions we observed . in the / versus / plot and the / versus / plot , the data from the western edge zone in ngc 2024 and from the magellanic clouds are shifted to the lower than the model results , which can be explained as an effect of spherical geometry . we measured the amount of the effect , and used the result to derive the column density through the clouds . on galaxy scales , the derived average column density and therefore size of the clouds appears to increase as the metallicity of the galaxy decreases . our observational result , that the mean extinction of clouds is constant and independent of their metallicity , is consistent with the idea that photoionization may play a role in regulating star formation . we thank luke keller , michael luhman , and thomas benedict for contributions to the fabry - perot spectrometer project , and jonathan elias , brooks gregory , and the staff of the ctio for their assistance in setting up our instrument . we also thank minho choi for providing the mc / vt code , wenbin li for helping calculations , kenji mochizuki , david jansen , and marco spaans for helpful discussions of pdr models , and harriet dinerstein , bruce draine , john black , and the referee , marc kutner , for critical comments on this paper . this work was supported by the david and lucile packard foundation and by nsf grants ast 9117373 and ast9530695 . anthony - twarog , b. j. 1982 , , 87 , 1213 arimoto , n. , sofue , y. , & tsujimoto , t. 1996 , , 48 , 275 aumann , h. h. , fowler , j. w. , & melnyk , m. 1990 , , 99 , 1674 bertoldi , f. , & mckee , c. f. 1996 , in amazing light : a volume dedicated to c. h. townes on his 80th birthday , ed . r. y. chiao ( new york : springer ) black , j. h. , & van dishoeck , e. f. 1987 , , 322 , 412 bloemen , j. b. g. m. , strong , a. w. , blitz , l. , cohen , r. s. , dame , t. m. , grabelsky , d. a. , hermsen , w. , lebrun , f. , mayer - hasselwander , h. a. , & thaddeus , p. 1986 , , 154 , 25 bohlin , r. c. , hill , j. k. , jenkins , e. b. , savage , b. d. , snow , t. p. , jr . , spitzer , l. , jr . , & york , d. g. 1983 , , 51 , 277 booth , r. s. 1993 , in the second european meeting on the magellanic clouds , new aspects of magellanic cloud research , eds . b. baschek , g. klare , & j. lequeux ( springer - verlag ) , 26 burton , m. g. , brand , p. w. j. l. , geballe , t. r. , & webster , a. s. 1989 , , 236 , 409 burton , m. g. , hollenbach , d. j. , & tielens , a. g. g. m. 1990 , , 365 , 620 choi , m. , evans , n. j. , ii , gregersen , e. m. , wang , y. 1995 , , 448 , 742 cohen , r. s. , dame , t. m. , garay , g. , montani , j. , rubio , m. , thaddeus , p. 1988 , , 331 , l95 digel , s. et al . 1997 , in proc . of the 170th iau symp . , co : twenty - five years of millimeter - wave spectroscopy , eds . w. b. latter , s. j. e. radford , p. r. jewell , j. g. mangum , & j. bally ( kluwer academic pub . ) , 22 draine , b. t. 1978 , , 36 , 595 draine , b. t. , & bertoldi , f. 1996 , , 468 , 269 draine , b. t. , & lee , h. m. 1984 , , 285 , 89 dufour , r. j. 1984 , in iau symp . 108 , structure and evolution of the magellanic clouds , eds . s. van den bergh & k. s. de boer ( dordrecht : reidel ) , 353 greve , a. , van genderen , a. m. , laval , a. 1990 , , 85 , 895 habing , h. j. 1968 , bull . netherlands , 19 , 421 helou , g. , khan , i. r. , malek , l. , & boehmer , l. 1988 , , 68 , 151 heydari - malayeri , m. , & testor , g. 1985 , , 144 , 98 hollenbach , d. j. , takahashi , t. , & tielens , a. g. g. m. 1991 , , 377 , 192 israel , f. p. 1988 , in proc . of an international summer school held at stirling in scotland , millimetre and submillimetre astronomy , eds . r. d. wolstencroft & w. b. burton ( kluwer academic publishers ) , 281 israel , f. p. , de graauw , th . , van de stadt , h. , & de vries , c. p. 1986 , , 303 , 186 israel , f. p. , & koornneef , j. 1988 , , 190 , 21 israel , f. p. , & koornneef , j. 1992 , , 250 , 475 israel , f. p. , & maloney , p. r. 1993 , in the second european meeting on the magellanic clouds , new aspects of magellanic cloud research , eds . b. baschek , g. klare , & j. lequeux ( springer - verlag ) , 44 israel , f. p. , maloney , p. r. , geis , n. , hermann , f. , madden , s. c. , poglitsch , a. , & stacey , g. j. 1996 , , 465 , 738 jaffe , d. t. , zhou , s. , howe , j. e. , herrmann , f. , madden , s. c. , poglitsch , a. , van der werf , p. p. , stacey , g. j. 1994 , , 436 , 203 jansen , d.j . , van dishoeck , e.f . , black , j.h . , spaans , m. , & sosin , c. 1995 , a&a , 302 , 223 jefferys , w. h. 1990 , biometrika , 77 , 597 johansson , l. e. b. , greve , a. , booth , r. s. , boulanger , f. , garay , g. , de graauw , th . , israel , f. p. , kutner , m. l. , lequeux , j. , murphy , d. c. , nyman , l. - . , & rubio , m. 1998 , , in press johansson , l. e. b. , olofsson , h. , hjalmarson , . , gredel , r. , black , j. h. 1994 , , 291 , 89 jura , m. 1974 , , 191 , 375 kawara , k. , nishida , m. , taniguchi , y. 1988 , , 100 , 458 koornneef , j. 1984 , in iau symp . 108 , structure and evolution of the magellanic clouds , eds . s. van den bergh & k. s. de boer ( dordrecht : reidel ) , 333 koornneef , j. , & israel , f. p. 1985 , , 291 , 156 krabbe , a. , storey , j. , rotaciuc , v. , drapatz , s. , & genzel , r. 1991 , in iau symp . 148 , the magellanic clouds , eds . r. haynes & d. milne ( kluwer academic publishers ) , 205 lee , m. g. , 1991 , in iau symp . 148 , the magellanic clouds , eds . r. haynes & d. milne ( kluwer academic publishers ) , 207 lequeux , j. , le bourlot , j. , pineau des forts , g. , roueff , e. , boulanger , f. , & rubio , m. 1994 , a&a , 292 , 371 li , w. 1997 , ph.d . thesis , university of texas at austin ( in preparation ) luhman , m.l . , & jaffe , d.t . 1996 , , 463 , 191 luhman , m. l. , jaffe , d. t. , keller , l. d. , & pak , s. 1994 , , 436 , l185 luhman , m. l. , jaffe , d. t. , keller , l. d. , & pak , s. 1995 , , 107 , 184 luhman , m. l. , jaffe , d. t. , sternberg , a. , herrmann , f. , & poglitsch , a. 1997 , , 482 , 298 madden , s. c. , poglitsch , a. , geis , n. , stacey , g. j. , & townes , c. h. 1997 , , 483 , 200 maloney , p. , & black , j. h. 1988 , , 325 , 389 maloney , p. , & wolfire , m. g. 1997 , in proc . of the 170th iau symp . , co : twenty - five years of millimeter - wave spectroscopy , eds . w. b. latter , s. j. e. radford , p. r. jewell , j. g. mangum , & j. bally ( kluwer academic pub . ) , 299 mangum , j. g. 1993 , , 105 , 117 mckee , c. f. 1989 , , 345 , 782 mochizuki , k. , nakagawa , t. , doi , y. , yui , y. y. , okuda , h. , shibai , h. , yui , m. , nishimura , t. , & low , f. j. 1994 , , 430 , l37 mochizuki , k. 1996 , ph.d . thesis , university of tokyo nakagawa , t. , doi , y. , yui , y. y. , okuda , h. , mochizuki , k. , shibai , h. , nishimura , t. , low , f. 1995 , , 455 , l35 pei , y. c. 1992 , , 395 , 130 poglitsch , a. , krabbe , a. , madden , s. c. , nikola , t. , geis , n. , johansson , l. e. b. , stacey , g. j. , & sternberg , a. 1995 , , 454 , 293 oliva , e. , & origlia , l. 1992 , , 254 , 466 richter , m. j. , graham , j. r. , & wright , g. s. 1995 , , 454 , 277 rieke , g. h. , & lebofsky , m. j. 1985 , , 288 , 618 rubio , m. , lequeux , j. , boulanger , f. , booth , r. s. , garay , g. , de graauw , th . , israel , f. p. , johansson , l. e. b. , kutner , m. l. , & nyman , l .- a . 1993 , , 271 , 1 schloerb , f. p. , & loren , r. b. 1982 , in ann . , 395 , symposium on the orion nebula to honor henry draper , eds . a. e. glassgold , p. j. huggins , & e. l. schucking , 32 scoville , n. z. , yun , m. s. , clemens , d. p. , sanders , d. b. , & waller , w. h. 1987 , , 63 , 821 solomon , p. m. , rivolo , a. r. , barrett , j. , & yahil , a. 1987 , , 319 , 730 spaans , m. , tielens , a. g. g. m. , van dishoeck , e. f. , & bakes , e. l. o. 1994 , , 437 , 270 stacey , g. j. , jaffe , d. t. , geis , n. , genzel , r. , harris , a. i. , poglitsch , a. , stutzki , j. , & townes , c. h. 1993 , , 404 , 219 stacey , g. j. , townes , c. h. , poglitsch , a. , madden , s. c. , jackson , j. m. , herrmann , f. , genzel , r. , & geis , n. 1991 , , 382 , l37 sternberg , a. , & dalgarno , a. 1989 , , 338 , 197 strzer , h. , stutzki , j. , & sternberg , a. 1996 , , 310 , 592 taylor , c. , kobulnicky , h. , & skillman , e. , 1996 , , 189 , 122.01 tielens , a. g. g. m. , & hollenbach , d. 1985 , , 291 , 722 tielens , a. g. g. m. , & hollenbach , d. 1997 , araa , in press van dishoeck , e. f. , & black , j. h. 1986 , , 62 , 109 van dishoeck , e. f. , & black , j. h. 1988a , , 334 , 771 van dishoeck , e. f. , & black , j. h. 1988b , in molecular clouds in the milky way and external galaxies , eds . r. l. dickman , r. l. snell , & j. s. young ( springer - verlag ) , 168 westerlund , b. e. 1990 , , 2 , 29 wilson , c. d. 1995 , , 448 , l97 wolfire , w. g. , hollenbach , d. , & tielens , a. g. g. m. 1989 , , 344 , 770 lccccccc & 1.61308 & 1995 oct & 81.0 & 23.0 & 125 & 104 & 163 & 2.12125 & 1995 oct & 81.0 & 25.4 & 94 & 125 & 196 & & 1994 dec & 88.9 & 26.4 & 94 & 120 & 189 & 2.24710 & 1994 dec & 88.9 & 26.4 & 89 & 127 & 199 lccccc smc & & 1.61386 & @xmath319 & 1.61506 & @xmath320 & & 2.12219 & @xmath321 & 2.12069 & @xmath322 lmc & & 1.61464 & @xmath323 & 1.61584 & @xmath324 & & 2.12321 & @xmath325 & 2.12171 & @xmath326 & & 2.24935 & @xmath327 & 2.24785 & @xmath326 lccc dem s 16 & 00 43 33.4 & @xmath328 39 05 & 1 dem s 37 & 00 46 16.2 & @xmath328 32 47 & 1 li - smc 36 & 00 44 50.5 & @xmath328 22 23 & 1 30 dor & 05 39 11.5 & @xmath329 06 00 & 2 n159/n160 & 05 40 18.2 & @xmath329 47 00 & 3 lcccccccccl dem s 16 & @xmath330&@xmath331&@xmath332&0.66 & 2.19 & 0.11&&0.43&49&2 ; & @xmath330&@xmath333&@xmath334&0.66 & 2.39 & 0.11&&0.27&42&2 ; dem s 37 & @xmath335&@xmath336&@xmath337&2.20 & & & & & & & @xmath338&@xmath339&@xmath340 & 1.1&0.584&0.073&&0.65&45&1 ; li - smc 36&@xmath341&@xmath342&@xmath343&1.77 & & & & & & & @xmath342&@xmath342&@xmath344 & 1.3 & 6.20&0.060&&0.33&42&1 ; 30 dor&@xmath342&@xmath345&@xmath346 & 2.7 & 2.87&&2.6&14 & 45 & 3;3 & @xmath345&@xmath342&@xmath347 & 2.2 & 3.45&&4.2&18 & 51 & 3;3 & @xmath342&@xmath342&@xmath348&0.93 & 10.5&&8.1&35 & 75 & 3;3 & @xmath349&@xmath342&@xmath350 & 2.4 & 2.23&&3.7&7.8 & 44 & 3;3 & @xmath342&@xmath349&@xmath351 & 2.9 & 3.39&&2.9&6.6 & 42 & 3;3 & @xmath352&@xmath353&@xmath354&0.75 & & 3 & & 1.4&43 & 4 ; & @xmath342&@xmath355&@xmath356&0.78&0.746&0.052&&2.6&43 & 1 ; & @xmath355&@xmath342&@xmath357&0.53 & & 3 & & 1.2 & 56 & 4 ; n159 & @xmath358&@xmath359&@xmath360 & 1.5 & 25.0 & 0.14&&0.31&34&5 ; & @xmath361&@xmath359&@xmath354&0.90 & 46.0 & 0.14&&0.32&35&5 ; & @xmath362&@xmath342&@xmath363 & 1.9 & 29.4 & 0.14&0.77 & 1.8 & 31 & 5;6 & @xmath345&@xmath342&@xmath364 & 1.4 & 47.3 & 0.14 & 2.3 & 7.1 & 46 & 5;6 & @xmath342&@xmath342&@xmath365&0.99 & 7.70&0.052 & 2.5 & 8.2 & 42 & 1;6 & @xmath361&@xmath361&@xmath366 & 1.3 & 19.8&0.052 & 2.9 & 8.7 & 56 & 1;6 & @xmath342&@xmath367&@xmath368&0.72&0.634&0.052 & 1.0 & 1.7 & 46 & 1;6 n160 & @xmath342&@xmath369&@xmath370 & 1.2 & & 3 & 0.49 & 1.2 & 37 & 5;6 & @xmath362&@xmath371&@xmath372&0.80 & 1.59&0.073&0.77 & 2.6 & 36 & 1;6 & @xmath373&@xmath371&@xmath374&0.91 & 9.89&0.073 & 2.0 & 15 & 76 & 1;6 & @xmath342&@xmath371&@xmath375&0.71 & 5.82&0.073 & 1.7 & 6.5 & 54 & 1;6 & @xmath376&@xmath377&@xmath378&0.64 & 4.25&0.073 & 1.1 & 2.9 & 39 & 1;6 & @xmath342&@xmath377&@xmath379&0.56 & 6.76&0.073 & 1.1 & 1.8 & 34 & 1;6 & @xmath341&@xmath380&@xmath381&0.65 & & 3 & 0.68 & 0.97&35 & 5;6 & @xmath373&@xmath380&@xmath382&0.59 & 2.76&0.073&0.75 & 1.1 & 34 & 1;6 & @xmath376&@xmath383&@xmath384&0.51 & 1.64&0.073&0.55 & 1.0 & 46 & 1;6 & @xmath342&@xmath383&@xmath385&0.52 & 3.86&0.073&0.36 & 0.75&43 & 1;6 & @xmath373&@xmath386&@xmath387&1.3&4.49&&&0.71&41 & 5 ; lcccccc dem s 16 & + 0.11&@xmath331 & & & @xmath388&@xmath389 & + 0.11&@xmath333 & & & @xmath390&@xmath391 30 dor & 0 & 0 & @xmath392&@xmath393&@xmath394&@xmath395 & @xmath349 & 0 & & & @xmath396&@xmath397n159 & @xmath398 & @xmath398 & & & @xmath399&@xmath400n160 & @xmath401 & @xmath402 & @xmath403&@xmath404&@xmath405&@xmath406 lccccccccl orion & @xmath413 & 0 & 23 & 6 & 34.0&9.9 & 56 & 55 & 1;2 + & @xmath414 & 0 & 28 & 5 & 29.0&6.1 & 20 & 39 & 1;2 + & @xmath415 & 0 & 5.2&1.6&29.5&4.3 & 8.3 & 36 & 1;2 + & @xmath416&+9.83&27 & 4 & 60.7&4.5 & 41 & 42 & 1;3 + ngc 2024:edge&@xmath417&0 & 9 & 1.9&13.4&5.6 & 3.3 & 36 & 4;4 + & @xmath418 & 0 & 12 & 3 & 18.6&6.7 & 4.7 & 37 & 4;4 + & @xmath419 & 0 & 13 & 2 & 14.4&8.6 & 4.9 & 37 & 4;4 + & @xmath420 & 0 & 3.3&1.2&6.70&5.0 & 4.3 & 37 & 4;4 + & @xmath421 & 0 & 4.5&1.8&15.6&4.7 & 4.4 & 35 & 4;4 + & @xmath422 & 0 & 6.4&1.9&51.0&5.6 & 6.6 & 38 & 4;4 + & @xmath423 & 0 & 8 & 2 & 84.6&6.3 & 10 & 41 & 4;4 + ngc 2024:prop&@xmath424 & 0&9 & 3 & 110 & 8.8 & 19 & 49 & 4;4 + & @xmath425 & 0 & 6 & 3 & 132 & 11 & 15 & 43 & 4;4 + & @xmath426 & 0 & 8 & 3 & 154 & 10 & 26 & 47 & 4;4 + & @xmath427 & 0 & 11 & 4 & 130 & 9.9 & 55 & 52 & 4;4 + lccccccc the galaxy & 15 & -1.85 & 0.28 & 4.04 & 0.37 & 2.19 & 0.47 + ... orion & 4 & -1.53 & 0.31 & 4.01 & 0.25 & 2.48 & 0.32 + ... ngc 2024:prop & 4 & -2.08 & 0.13 & 3.68 & 0.03 & 1.61 & 0.14 + ... ngc 2024:edge & 7 & -1.91 & 0.15 & 4.26 & 0.37 & 2.35 & 0.38 + the lmc & 17 & -1.70 & 0.17 & 4.37 & 0.44 & 2.71 & 0.40 + ... 30 dor & 5 & -1.81 & 0.11 & 4.82 & 0.14 & 3.01 & 0.23 + ... n160 & 8 & -1.59 & 0.18 & 4.18 & 0.21 & 2.59 & 0.20 + ... n159 & 4 & -1.78 & 0.07 & 4.19 & 0.65 & 2.41 & 0.62 + the smc & 4 & & & & & 2.89 & 0.40 + lccccccc the galaxy & 1 & @xmath428 & & & & & + ... galactic plane & & & ( 3.7 ) & 1.6 & 0.29 & @xmath429 & @xmath430 + ... orion & & 4 & 4.7 & 2.5 & 0.27 & 0.44 & 1.6 + ... ngc 2024:prop & & 4 & 3.7 & 2.5 & 0.45 & @xmath429 & @xmath431 + ... ngc 2024:edge & & 7 & 3.7 & 1.8 & 0.32 & 0.083 & 0.89 + the lmc & 0.25 & 17 & & & & & + ... 30 dor & & 5 & 4.0 & 2.7 & 1.5 & 0.048 & 3.9 + ... n160 & & 8 & 4.2 & 2.7 & 1.5 & 0.17 & 5.0 + ... n159@xmath432 & & 1 & 4.1 & 2.7 & 1.5 & @xmath429 & @xmath428 + ... n159@xmath433 & & 1 & 3.9 & 2.7 & 1.7 & 0.36 & 8.4 + ... n159@xmath434 & & 1 & 3.7 & 2.7 & 1.8 & 0.15 & 5.8 + ... n159@xmath435 & & 1 & 4.1 & 2.3 & 1.2 & 0.063 & 3.2 + the smc & 0.1 & 4 & ( 3.7 ) & 2.6 & 4.5 & @xmath436 & @xmath437 + ic 10 & 0.2 & 2 & & & & & + ... position a & & 1 & ( 3.7 ) & 2.7 & 2.2 & 0.3 & 9.9 + ... position b & & 1 & ( 3.7 ) & 2.6 & 2.0 & 0.1 & 5.8 +

the chemical structure of neutral clouds in low metallicity environments is examined with particular emphasis on the h to and c@xmath0 to co transitions . we observed near - ir , , and lines and the @xmath1co line from 30 doradus and n159/n160 in the large magellanic cloud and from dem s 16 , dem s 37 , and li - smc 36 in the small magellanic cloud . we find that the emission is uv - excited and that ( weak ) emission always exists ( in our surveyed regions ) toward positions where and [ ] emission have been detected . using a pdr code and a radiative transfer code , we simulate the emission of line radiation from spherical clouds and from large planar clouds . because the [ ] emission and emission arise on the surface of the cloud and the lines are optically thin , these lines are not affected by changes in the relative sizes of the neutral cloud and the bearing core , while the optically thick emission can be strongly affected . the sizes of clouds are estimated by measuring the deviation of co emission strength from that predicted by a planar cloud model of a given size . the average cloud column density and therefore size increases as the metallicity decreases . our result agrees with the photoionization regulated star formation theory by mckee ( 1989 ) .

the _ @xmath0-cube _ @xmath4 , or @xmath0-dimensional hypercube , has been extensively studied . nevertheless , many open questions remain . et al . _ wrote a comprehensive survey on hypercube graphs @xcite . recall that the _ @xmath0-cube _ @xmath4 has vertex set @xmath5 and @xmath0-tuples representing vertices are adjacent if and only if they differ in exactly one coordinate . then , @xmath4 is an @xmath0-regular bipartite graph with @xmath6 vertices and it is natural to consider its vertex set as partitioned into @xmath7 layers , the _ layer _ @xmath8 consisting of the @xmath9 vertices containing exactly @xmath1 @xmath10s , @xmath11 . seeing the vertices of @xmath4 as the characteristic vector of subsets of @xmath12=\{1,2,\dots , n\}$ ] , the vertices of layer @xmath8 correspond to the subsets of cardinality @xmath1 , while the adjacencies correspond to the inclusion relation . if @xmath0 is odd , @xmath13 , the middle two layers @xmath14 and @xmath15 of @xmath4 have the same number @xmath16 of vertices . then the middle cube graph , denoted by @xmath17 , is the graph induced by these two layers . it has been conjectured by dejter , erds , havel @xcite among others , that @xmath17 is hamiltonian . it is known that the conjecture holds for @xmath18 ( see savage and shields @xcite ) , and it was _ almost _ solved by robert johnson @xcite . in this paper we study some of the properties of the middle cube graphs in the light of the theory of distance - regular graphs . in particular , we completely determine their spectra ( eigenvalues and their multiplicities , and associated eigenvectors ) . in this context , qiu and das provided experimental results for eigenvalues of several interconnection networks for which no complete characterization were known ( see @xcite ) . before proceeding with our study , we fix some basic definitions and notation used throughout the paper . we denote by @xmath19 a ( simple , connected and finite ) _ graph _ with vertex set @xmath20 an edge set @xmath21 . the _ order _ of the graph @xmath22 is @xmath23 and its _ size _ is @xmath24 . we label the vertices with the integers @xmath25 . if @xmath26 is adjacent to @xmath27 , that is , @xmath28 , we write @xmath29 or @xmath30 . the _ distance _ between two vertices is denoted by @xmath31 . we also use the concepts of _ even distance _ and _ odd distance _ between vertices ( see bond and delorme @xcite ) , denoted by @xmath32 and @xmath33 , respectively . they are defined as the length of a shortest even ( respectively , odd ) walk between the corresponding vertices . the set of vertices which are _ @xmath34-apart _ from vertex @xmath26 , with respect to the usual distance , is @xmath35 , so that the _ degree _ of vertex @xmath26 is simply @xmath36 . eccentricity _ of a vertex is @xmath37 and the _ diameter _ of the graph is @xmath38 . given @xmath39 , the _ distance_-@xmath34 graph @xmath40 has the same vertex set as @xmath22 and two vertices are adjacent in @xmath40 if and only if they are at distance @xmath34 in @xmath22 . an _ antipodal graph _ @xmath22 is a connected graph of diameter @xmath41 for which @xmath42 is a disjoint union of cliques . in this case , the _ folded graph of _ @xmath22 is the graph @xmath43 whose vertices are the maximal cliques of @xmath44 and two vertices are adjacent if their union contains and edge of @xmath22 . if , moreover , all maximal cliques of @xmath44 have the same size @xmath45 then @xmath22 is also called an _ antipodal @xmath45-cover _ of @xmath43 ( double cover if @xmath46 , triple cover if @xmath47 , etc . ) . recall that a graph @xmath22 with diameter @xmath41 is _ distance - regular _ when , for all integers @xmath48 ( @xmath49 ) and vertices @xmath50 with @xmath51 , the numbers @xmath52 do not depend on @xmath53 and @xmath54 . in this case , such numbers are called the _ intersection parameters _ and , for notational convenience , we write @xmath55 , @xmath56 , and @xmath57 ( see brower _ et al . _ @xcite and fiol @xcite ) . the odd graph , independently introduced by balaban _ et al . _ @xcite and biggs @xcite , is a family of graphs that has been studied by many authors ( see @xcite ) . more recently , fiol _ et al . _ @xcite introduced the twisted odd graphs , which share some interesting properties with the odd graphs although they have , in general , a more involved structure . for @xmath58 , _ the odd graph _ @xmath59 has vertices representing the ( @xmath3)-subsets of @xmath60=\{1,2,\dots,2k-1\}$ ] , and two vertices are adjacent if and only if they are disjoint . for example , @xmath61 is the complete graph @xmath62 , and @xmath63 is the petersen graph . in general , @xmath59 is a @xmath1-regular graph on @xmath64 vertices , diameter @xmath65 and girth @xmath66 if @xmath67 , @xmath68 if @xmath69 , and @xmath70 if @xmath71 ( see biggs @xcite ) . the odd graph @xmath59 is a distance - regular graph with intersection parameters @xmath72 , \quad c_j=\left[\frac{j+1}{2}\right ] \quad ( 0\le j\le k-1).\ ] ] with respect to the spectrum , the distinct eigenvalues of @xmath59 are @xmath73 , @xmath74 , with multiplicities @xmath75 let @xmath19 be a graph of order @xmath0 , with vertex set @xmath76 . bipartite double _ graph @xmath77 is the graph with the duplicated vertex set @xmath78 , and adjacencies induced from the adjacencies in @xmath22 as follows : @xmath79 thus , the edge set of @xmath80 is @xmath81 . from the definition , it follows that @xmath80 is a bipartite graph with stable subsets @xmath82 and @xmath83 . for example , if @xmath22 is a bipartite graph , then its bipartite double graph @xmath80 consists of two non - connected copies of @xmath22 ( see fig . [ fig.p4 ] ) . and its bipartite double graph . ] the bipartite double graph @xmath80 has an involutive automorphism without fixed edges , which interchanges vertices @xmath26 and @xmath84 . on the other hand , the map from @xmath80 onto @xmath22 defined by @xmath85 is a @xmath86-fold covering . if @xmath22 is a @xmath87-regular graph , then @xmath80 also is . moreover , if the degree sequence of the original graph @xmath22 is @xmath88 , the degree sequence for its bipartite double graph is @xmath89 . the distance between vertices in the bipartite double graph @xmath80 can be given in terms of the even and odd distances in @xmath22 . namely , @xmath90 note that always @xmath91 even if @xmath92 . actually , @xmath80 is connected if and only if @xmath22 is connected and non - bipartite . more precisely , it was proved by bond and delorme @xcite that if @xmath22 is a non - bipartite graph with diameter @xmath41 , then its bipartite double graph @xmath80 has diameter @xmath93 , and @xmath94 if and only if for some vertex @xmath95 the subgraph induced by the vertices at distance less than @xmath41 from @xmath26 , @xmath96 , is bipartite . in figs . [ fig.lessdiam]-[fig.dcpetersen ] , we can see the bipartite double graph of three different graphs . the cycle @xmath97 and petersen graph both have diameter @xmath98 , and their bipartite double graphs have diameter @xmath99 , while in the first example ( fig . [ fig.lessdiam ] ) @xmath80 has diameter @xmath100 . has diameter 2 and @xmath80 has diameter 3 . ] and its bipartite double graph . ] and @xmath101 as another view of its bipartite double graph . ] the _ extended bipartite double _ graph @xmath102 of a graph @xmath22 is obtained from its bipartite double graph by adding edges @xmath103 for each @xmath95 . note that when @xmath22 is bipartite , then @xmath102 is the direct product @xmath104 . let us now recall a useful result from spectral graph theory . for any graph , it is known that the components of its eigenvalues can be seen as charges on each vertex ( see fiol and mitjana @xcite and godsil @xcite ) . let @xmath19 be a graph with adjacency matrix @xmath105 and @xmath106-eigenvector @xmath107 . then , the charge of vertex @xmath95 is the entry @xmath108 of @xmath107 , and the equation @xmath109 means that the sum of the charges of the neighbors of vertex @xmath26 is @xmath106 times the charge of vertex @xmath26 : @xmath110 in what follows we compute the eigenvalues of the bipartite double graph @xmath80 and the extended bipartite double graph @xmath102 as functions of the eigenvalues of a non - bipartite graph @xmath22 . we also show how to obtain the eigenvalues together with the corresponding eigenvectors of @xmath80 and @xmath102 . first , we recall the following technical result , due to silvester @xcite , on the determinant of some block matrices : [ lema.blocks ] let @xmath111 be a field and let @xmath112 be a commutative subring of @xmath113 , the set of all @xmath114 matrices over @xmath111 . let @xmath115 , then @xmath116 for example , if @xmath117 , where @xmath118 are @xmath114 matrices over @xmath111 which commute with each other , then theorem [ lema.blocks ] reads @xmath119 now we can use the above theorem to find the characteristic polynomial of the bipartite double and the extended bipartite double graphs . let @xmath22 be a graph on @xmath0 vertices , with the adjacency matrix @xmath105 and characteristic polynomial @xmath120 . then , the characteristic polynomials of @xmath80 and @xmath102 are , respectively , @xmath121 from the definitions of @xmath80 and @xmath102 , their adjacency matrices are , respectively , @xmath122 thus , by ( [ determinant ] ) , the characteristic polynomial of @xmath80 is @xmath123 whereas , the characteristic polynomial of @xmath102 is @xmath124 as a consequence , we have the following corollary : [ prop.dc.espectrum ] given a graph @xmath22 with spectrum @xmath125 where the superscripts denote multiplicities , then the spectra of @xmath80 and @xmath102 are , respectively , @xmath126 just note that , by ( [ policargtitlla ] ) and ( [ policargbarret ] ) , for each root @xmath106 of @xmath120 , @xmath127 are roots of @xmath128 , whereas @xmath129 are roots of @xmath130 . note that the spectra of @xmath80 and @xmath102 are symmetric , as expected , because both @xmath80 and @xmath102 are bipartite graphs . in the next theorem we are concerned with the eigenvectors of @xmath80 and @xmath102 , in terms of the eigenvectors of @xmath22 . the computations also give an alternative derivation of the above spectra . let @xmath22 be a graph and @xmath107 a @xmath106-eigenvector of @xmath22 . let us consider the vector @xmath131 with components @xmath132 , and @xmath133 , with components @xmath134 and @xmath135 , @xmath136 . then , * @xmath131 is a @xmath106-eigenvector of @xmath80 and a @xmath137-eigenvector of @xmath102 ; * @xmath133 is a @xmath138-eigenvector of @xmath80 and a @xmath139-eigenvector of @xmath102 . in order to show that @xmath131 is a @xmath106-eigenvector of @xmath80 , we distinguish two cases : * for a given vertex @xmath26 , @xmath140 , all its adjacent vertices are of type @xmath141 , with @xmath142 . then @xmath143 * for a given vertex @xmath84 , @xmath140 , all its adjacent vertices are of type @xmath27 , with @xmath142 . then @xmath144 by a similar reasoning with @xmath145 , we obtain @xmath146 therefore , @xmath133 is a ( @xmath147)-eigenvector of the bipartite double graph @xmath80 . in the same way , we can prove that @xmath131 and @xmath133 are eigenvectors of @xmath102 with respective eigenvalues @xmath148 and @xmath149 . notice that , for every linearly independent eigenvectors @xmath150 and @xmath151 of @xmath22 , we get the linearly independent eigenvectors @xmath152 and @xmath153 of @xmath80 . as a consequence , the geometric multiplicity of eigenvalue @xmath106 of @xmath22 coincides with the geometric multiplicities of the eigenvalues @xmath106 and @xmath147 of @xmath80 , and @xmath148 and @xmath149 of @xmath102 . for @xmath154 and @xmath155 , the _ middle cube graph _ @xmath17 is the subgraph of the @xmath0-cube @xmath4 induced by the vertices whose binary representations have either @xmath3 or @xmath1 number of @xmath10s . then , @xmath17 has order @xmath156 and is @xmath1-regular , since a vertex with @xmath3 @xmath10s has @xmath1 zeroes , so it is adjacent to @xmath1 vertices with @xmath1 @xmath10s , and similarly a vertex with @xmath1 @xmath10s has @xmath1 adjacent vertices with @xmath3 @xmath10s ( see figs . [ fig.3midcub ] and [ fig.5midcub ] ) . the middle cube graph @xmath17 is a bipartite graph with stable sets @xmath157 and @xmath158 constituted by the vertices whose corresponding binary string has , respectively , even or odd _ hamming weight _ , that is , number of @xmath10s . the diameter of the middle cube graph @xmath17 is @xmath159 . notice that , if @xmath160 and @xmath161 are both subsets of @xmath60 $ ] , @xmath162 if and only if @xmath160 and @xmath163 are disjoint . moreover , if @xmath164 then @xmath165 . this gives the following result . the mapping from @xmath166 to @xmath17 defined by : @xmath167 & \to & v[mq_k ] \\ & \u & \mapsto & \u \\ & \u ' & \mapsto & \overline{\u } \end{array}\ ] ] is clearly bijective . moreover , according to the definition of bipartite double graph in eq.([def.doble.bipartit ] ) , if @xmath168 and @xmath169 are two vertices of @xmath166 , then @xmath170 which is equivalent to say that if @xmath171 , in @xmath166 , then @xmath172 , in @xmath17 . for example , the middle cube graph @xmath173 contains vertices with one or two @xmath10s in their binary representation . the adjacencies give simply a 6-cycle ( see fig . [ fig.3midcub ] ) , which is isomorphic to @xmath174 . as another example , @xmath175 has 20 vertices because there are @xmath176 vertices with two @xmath10s , and @xmath177 vertices with three @xmath10s in their binary representation ( see fig . [ fig.5midcub ] ) . compare the figs . [ fig.dcpetersen ] and [ fig.5midcub ] in order to realize the isomorphism between the definitions of @xmath175 and @xmath178 . according to the result of corollary [ prop.dc.espectrum ] , the spectrum of the middle cube graph @xmath182 can be obtained from the spectrum of the odd graph @xmath183 . the distinct eigenvalues of @xmath17 are @xmath184 and @xmath185 , @xmath74 , with multiplicities @xmath186 the eigenvalues of the @xmath175 are @xmath190 and the zeroes of polynomial ( [ eq.hoffmann ] ) : @xmath191 and their multiplicities , @xmath192 , can be computed using the highest degree polynomial @xmath193 , according to the result by fiol @xcite : @xmath194 where @xmath195 . of course , this expression yields the same result as eq . ( [ mult_vaps_micu ] ) . namely , @xmath196 , @xmath197 . let @xmath22 be a graph with diameter @xmath41 and distinct eigenvalues @xmath204 , where @xmath205 . a classical result states that @xmath206 ( see , for instance , biggs @xcite ) . other results related to the diameter @xmath41 and some ( or all ) different eigenvalues have been given by alon and milman @xcite , chung @xcite , van dam and haemers @xcite , delorme and sol @xcite , and mohar @xcite , among others . et al . _ @xcite showed that many of these results can be stated with the following common framework : if the value of a certain polynomial @xmath207 at @xmath208 is large enough , then the diameter is at most the degree of @xmath207 . more precisely , it was shown that optimal results arise when @xmath207 is the so - called _ @xmath1-alternating polynomial _ , which in the case of degree @xmath209 is characterized by @xmath210 , and satisfies @xmath211 , where @xmath212 . in particular , when @xmath22 is a regular graph on @xmath0 vertices , the following implication holds : @xmath213 this result suggested the study of the so - called _ boundary graphs _ @xcite , characterized by @xmath214 fiol _ et al . _ @xcite showed that extremal ( @xmath215 ) boundary graphs , where each vertex has maximum eccentricity , are 2-antipodal distance - regular graphs . as we show in the next result , this is the case of the middle cube graphs @xmath216 where the antipodal pairs of vertices are @xmath217 , with @xmath218 and @xmath219 . recall that the eigenvalues of @xmath216 are @xmath220 that is , @xmath221 , @xmath222 . now , according to eq . ( [ boundary ] ) , we have to prove that @xmath223 . computing @xmath224 , for @xmath225 , we get @xmath226 this implies @xmath227 giving exactly the multiplicities of the corresponding eigenvalues , as found in eq . [ mult_vaps_micu ] . by summing up we get @xmath228 but @xmath229 and @xmath230 where we have used eq . ( [ eq.calcul ] ) changing @xmath1 by @xmath3 . thus , replacing the above values in eq . ( [ eq.calcul0 ] ) , we get the result . research supported by the _ ministerio de ciencia e innovacin _ , spain , and the _ european regional development fund _ under project mtm2011 - 28800-c02 - 01 , and the _ catalan research council _ under project 2009sgr1387 . i. havel , semipaths in directed cubes , in m. fiedler ( ed . ) , _ graphs and other combinatorial topics _ , teunebner texte math . , teubner , leipzig ( 1983 ) . a. j. hoffman , on the polynomial of a graph , _ amer . monthly _ * 70 * ( 1963 ) 3036 .

we study a family of graphs related to the @xmath0-cube . the middle cube graph of parameter @xmath1 is the subgraph of @xmath2 induced by the set of vertices whose binary representation has either @xmath3 or @xmath1 number of ones . the middle cube graphs can be obtained from the well - known odd graphs by doubling their vertex set . here we study some of the properties of the middle cube graphs in the light of the theory of distance - regular graphs . in particular , we completely determine their spectra ( eigenvalues and their multiplicities , and associated eigenvectors ) .

here we give the details of the calculation of the proximity induced amplitudes @xmath146 . because the superconductor is only weakly coupled through the tunnel barrier @xmath52 , we can derive an effective 1d model via low - order quasi - degenerate perturbation theory . we split @xmath147 into two parts , where @xmath59 is diagonal in the eigenbasis @xmath37 , and @xmath148 $ ] . since @xmath62 is diagonal in spin and valley , we suppress the indices @xmath149 in the following . to first order in @xmath62 , @xmath150 and to second order , @xmath151 \notag\\&\hspace{.5cm}\times \int dx{\tilde\phi}^{0\dagger}_{p_x , p_y}(x)h_1(x ) \phi^{0,n'}_{p_y}(x ) , \end{aligned}\ ] ] where @xmath152 are the unperturbed free states above the gap @xmath153 with real @xmath154 and @xmath32 at energy @xmath155 . we impose the quantization condition @xmath156 and normalize the extended wavefunctions according to @xmath157 . the quantization length @xmath101 and the highest momentum @xmath154 are increased until the second order matrix elements converge . to study cooper pair transport only the parts of @xmath158 , @xmath159 are relevant which are proportional to @xmath160 , i.e. , they mix electron and hole states and therefore change the particle number . the relevant momenta @xmath32 are close to the crossing of the respective electron and hole band ( see the discussion on approximate momentum conservation in the main text ) . this can involve one band , @xmath161 and @xmath162 , or both , @xmath163 , where @xmath164 are the fermi points of the unperturbed dispersion , eq . . the linearized subgap dispersion , eq . , reads @xmath165 around the fermi points and @xmath166 around zero momentum . the coefficients @xmath167 and @xmath168 used in the transport calculation , e.g. , eq . , can be read off immediately . in the most general case the incoming holes in a nsn junction can be transmitted @xmath169 , reflected @xmath170 , or undergo local ( @xmath171 or crossed ( @xmath172 ) andreev reflection . the outgoing state is @xmath173 rewriting the hole operators @xmath174 in terms of electron operators @xmath175 , and the fermi sea @xmath131 in terms of the lowered fermi sea @xmath128 as explained in eq . in the main text , we arrive at the first line contains the product state contributions , the second line local pairs , and the third line nonlocal pairs . in the conventional reflection - dominated case , @xmath177 , realized in y - junction cooper pair splitters , the leading order contributions are @xmath178\ket{}_{\delta\mu } , \end{aligned}\ ] ] i.e. , lar produces local pairs and car produces nonlocal pairs . in the transmission - dominated situation , @xmath179 , the situation is reversed : the leading order is @xmath180\ket{}_{\delta\mu } , \end{aligned}\ ] ] so lar produces nonlocal pairs and car produces local pairs . in the situation discussed in the main text , both car and reflection are forbidden , ruling out local pairs to all orders , as long as the valley symmetry is obeyed . generally speaking it is undesirable to have simultaneously strong ordinary reflection and strong lar or to have simultaneously strong transmission and strong car to build a cooper pair splitter useful to create spin entanglement . the notation becomes more cumbersome , when both subgap bands are considered but the considerations are completely analogous . without superconductivity the outgoing scattering state is @xmath181 where @xmath182 is the band index . in the presence of the superconductor , the transmitted holes can change the subgap band from @xmath183 to @xmath184 with an amplitude @xmath185 . like in the one - band case , whenever the energy of an incoming electron is such that the spectrum of the s region has a gap , the transmission amplitude @xmath186 is exponentially suppressed with the length of the proximity region , and due to unitarity there is a finite amplitude @xmath187 for the spin-@xmath0 hole to be andreev reflected locally as a spin-@xmath0 electron at energy @xmath134 : the higher order terms in @xmath138 contain multiple cooper pairs and are not necessarily entangled , e.g. , the @xmath190 contribution is a pure product state in which all states in the left / right lead at energy @xmath191 are occupied . 40ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/physrevb.63.165314 [ * * , ( ) ] link:\doibase 10.1007/s10051 - 001 - 8675 - 4 [ * * , ( ) ] link:\doibase 10.1103/physrevb.65.165327 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.89.037901 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.91.267003 [ * * , ( ) ] link:\doibase 10.1038/nphys621 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.100.147001 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.226401 [ * * , ( ) ] link:\doibase 10.1038/nature08432 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.104.026801 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.157002 [ * * , ( ) ] link:\doibase 10.1038/ncomms2169 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.114.096602 [ * * , ( ) ] http://dx.doi.org/10.1038/ncomms8446 [ * * , ( ) ] link:\doibase 10.1103/physreva.57.120 [ * * , ( ) ] link:\doibase 10.1007/s100510051010 [ * * , ( ) ] link:\doibase 10.1209/epl / i2001 - 00303 - 0 [ * * , ( ) ] @noop link:\doibase 10.1103/physrevb.84.115420 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.226802 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.120403 [ * * , ( ) ] link:\doibase 10.1103/physrevb.91.085415 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.036802 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.100.036804 [ * * , ( ) ] link:\doibase 10.1073/pnas.1308853110 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.80.1337 [ * * , ( ) ] http://dx.doi.org/10.1038/nphys547 [ * * , ( ) ] http://dx.doi.org/10.1038/nature14364 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.086805 [ * * , ( ) ] @noop _ _ ( , ) @noop link:\doibase 10.1103/physrevb.90.184517 [ * * , ( ) ] link:\doibase 10.1140/epjb / e2004 - 00284 - 8 [ * * , ( ) ] http://stacks.iop.org/1367-2630/7/i=1/a=176 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.91.157002 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.83.407 [ * * , ( ) ] link:\doibase 10.1103/physrevb.61.r16303 [ * * , ( ) ] link:\doibase 10.1103/physrevb.81.125435 [ * * , ( ) ] http://stacks.iop.org/1367-2630/12/i=8/a=083063 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.115423 [ * * , ( ) ]

bilayer graphene hosts valley - chiral one dimensional modes at domain walls between regions of different interlayer potential or stacking order . when such a channel is brought into proximity to a superconductor , the two electrons of a cooper pair which tunnel into it move in opposite directions because they belong to different valleys related by the time - reversal symmetry . this is a kinetic variant of cooper pair splitting , which requires neither coulomb repulsion nor energy filtering but is enforced by the robustness of the valley isospin in the absence of atomic - scale defects . we derive an effective model for the guided modes in proximity to an @xmath0-wave superconductor , calculate the conductance carried by split and spin - entangled electron pairs , and interpret it as a result of _ local _ andreev reflection processes , whereas crossed andreev reflection is absent . creating mobile nonlocal spin - entangled electrons in a transport experiment with the help of superconductor normal junctions has attracted a lot of attention in theory @xcite and experiment @xcite because the spin degree of freedom of the electron could serve as a solid - state qubit @xcite . in the existing experiments , the envisaged process where a cooper pair is split over two normal leads is crossed andreev reflection ( car ) @xcite , which is enhanced by the repulsive electron - electron interaction on two quantum dots weakly coupled to the superconductor @xcite or by energy filtering @xcite . the basic mechanism of these entanglers is not very sensitive to the specific material used , i.e. , the underlying band structure . it has been shown that characteristic features of new materials exhibiting dirac - cones like graphene or topological insulators can be useful for splitting cooper pairs @xcite . in these proposals , the efficiency of the splitting process , in the absence of interactions , relies on non - protected resonance conditions or the split cooper pair is not spin - entangled due to spin - helicity or spin - polarization of the leads . helical edge states of the quantum spin hall regime have , however , been proposed to detect spin entanglement @xcite . or different stacking order a topological valley - chiral channel forms . cooper pairs tunneling into it from a nearby @xmath0-wave superconductor ( s ) are split because the two electrons belong to opposite valleys @xmath1 and thus have opposite velocities . they remain spin entangled and propagate to separate normal leads ( n ) . ( b ) in each valley two subgap modes along the domain wall emerge . energy and momentum conservation along the ns interface single out four points in the subgap spectrum at which cooper pairs are injected . ] here , we propose to exploit the valley degree of freedom in bilayer graphene ( bg ) , where valley - chiral , spin - degenerate one - dimensional ( 1d ) channels are formed at domain walls . such domain walls can be engineered by switching the sign of an interlayer voltage or by reversing the stacking order @xcite . if brought into proximity to a superconductor , the pairs emitted into the channel are split , i.e. , two electrons propagate in different directions but remain spin - entangled since , as required by time - reversal symmetry , the two electrons forming the cooper pair in the superconductor are from different valleys @xcite . as long as the valley degree of freedom is robust , the splitting efficiency is unity , independent of resonance conditions . the device extends the upcoming `` valleytronics '' in graphene @xcite to nonlocal einstein - podolsky - rosen pairs . a 1d channel defined by opposite stacking order has recently been created experimentally in bg @xcite with mean free paths over several 100 nm , demonstrating weak intervalley scattering . in this scenario , the normal reflection of an incoming hole ( or electron ) and car are absent . in the limit of a weak proximity effect , where the normal transmission through the proximity region of the channel is almost perfect , the spin - entangled pair emission with electrons moving to opposite normal terminals ( fig . [ fig - setup ] ) is equivalent to local andreev reflection ( lar ) processes , opposite to the normal reflection - dominated case , where car produces entangled pairs . we analyze the setup of fig . [ fig - setup ] in two steps : first , we investigate the influence of the superconductor on the 1d channel by solving a bogoliubov - de gennes ( bdg ) equation , and derive an effective 1d model to describe the proximity effect in the channel . second , we calculate the subgap conductance when applying a bias voltage between the superconductor and the channel using a rate equation approach . we interpret the subgap transport in a scattering matrix picture and show that the outgoing scattered state is a two - particle spin - entangled state on top of a filled normal - state fermi sea with a chemical potential lowered by the bias voltage . to leading order its weight is given by the lar amplitude . _ model._ we consider a bg sheet with bernal @xmath2 stacking in the presence of an interlayer voltage @xmath3 @xcite . we model the superconductor region as bg in which the bands are shifted by a scalar potential @xmath4 due to doping and which has an induced @xmath0-wave pairing amplitude @xmath5 . we employ the low - energy approximation for bg , valid at energies and ( inter)layer voltages smaller than the interlayer hopping @xmath6 ev . without the superconductor , the valley index @xmath7 distinguishing the two @xmath8-points @xmath9 and the electron spin @xmath10 are good quantum numbers and we write the bogoliubov - de gennes equation as @xmath11 @xmath12\tau_z + \delta({\bm r})\tau_x , \label{eq - hbdg } \end{aligned}\ ] ] where @xmath13 . the pauli matrices @xmath14 act in the pseudospin @xmath15 space and @xmath16 in electron - hole space and we set the fermi energy @xmath17 . the 4-component spinor is @xmath18 where we have introduced the electron @xmath19 and hole components ( @xmath20 ) on the two sublattices . excitations with energy @xmath21 are then expanded as @xmath22 with the vector of field - operators @xmath23 . in the absence of the superconductor ( @xmath24 ) and assuming the modes to propagate along the @xmath25-direction along a domain wall at @xmath26 , i.e. , @xmath27 with @xmath28 , the topologically confined modes can be found analytically @xcite . the electron and hole sectors in eq . decouple . in the electron sector , the solutions in each half space have the form @xmath29 where @xmath30 with @xmath31 for any fixed energy @xmath21 and momentum @xmath32 there are four allowed values @xmath33 which become complex when @xmath34 , i.e. there are no propagating modes in the bulk at energies below @xmath35 . matching the wavefunctions decaying away from the domain wall and their derivatives , one obtains the two electronic subgap solutions @xmath36 in each valley , @xmath37 , with the dispersion relation @xmath38 with velocities opposite in the two valleys @xcite . the solutions for the hole sector @xmath39 where @xmath40 at energy @xmath41 are obtained from eqs . and by setting @xmath42 . the relevant momenta @xmath32 are close to the @xmath8 points : taking @xmath43 , we obtain from eq . the momentum scale @xmath44 , on which the k - points are located at @xmath45 for @xmath46 . the guided modes decay into the bulk on a length scale of @xmath47 , which then is on the order of several @xmath48 nm . this sets the scale of the separation between the guided mode and a superconductor required to obtain a proximity effect . _ perturbation theory for superconducting pairing._ assuming a superconductor / normal interface with translational invariance along the @xmath25-direction , there are three distinct areas : in the superconductor area , @xmath49 , the pairing amplitude @xmath50 is finite and @xmath51 is negative . the area @xmath52 is in the normal state as before , @xmath53 , but the interlayer voltage is finite , @xmath54 . this region is a tunnel barrier between the superconductor and the domain wall at the interface to the third region @xmath55 , where @xmath53 and @xmath56 . in this situation guided modes exist at @xmath57 because states above @xmath35 can propagate in the normal regions and states above @xmath58 can propagate in the superconductor . because of the tunnel barrier the guided modes are only weakly affected by the superconductor and we can apply standard quasidegenerate perturbation theory @xcite , for which the unperturbed hamiltonian @xmath59 is obtained from @xmath60 by setting @xmath61 everywhere and so the perturbation @xmath62 , which adds the missing parts , is finite only at @xmath49 . as a result of the perturbation the electron and hole states of the channel acquire a finite overlap @xmath63 , where @xmath64 label the subgap bands . this allows for particle number non - conserving processes , i.e. , cooper pair transport . to first order only the electron and hole states belonging to the same subgap band mix , @xmath65 , @xmath66 this agrees with the result one expects when introducing superconductivity phenomenologically by constructing the bdg equation directly from the guided modes with a uniform pairing @xmath67 . the second order corrections , which take into account the modification of the wavefunctions due to the superconductor , however , reveal that the situation is different in the geometry we consider . the electron hole overlap differs in both bands , @xmath68 , and band mixing is finite , @xmath69 ( fig . [ fig - pert ] ) . this is confirmed by the full dispersion relation of @xmath60 we obtain by matching the 4-component spinor and its derivatives at both interfaces numerically ( fig . [ fig - pert ] , inset ) : two gaps of different size open at zero energy ( @xmath70 and @xmath71 ) and two gaps open at zero momentum where electron and hole states from different subgap bands cross ( @xmath72 ) . this means that there is cooper pair transport at zero energy as well as at the finite energies @xmath73 . ( @xmath71 ) in the 1d channel at the respective fermi point @xmath74 and induced interband superconductivity @xmath75 at the band crossing @xmath76 . for illustrative purposes we choose the bulk superconducting gap @xmath77 and the doping @xmath78 ( @xmath79 is equally feasible ) . the amplitudes decay exponentially with the separation @xmath80 between the superconductor and the channel because the @xmath81 region acts as a tunnel barrier . inset : in the normal state dispersion ( dashed ) two different - sized gaps open at the fermi energy because @xmath68 , shown for @xmath82 . additionally , @xmath75 opens a gap at the electron - hole crossing at @xmath76 . this point contributes significantly to cooper pair transport because compared to the fermi points the normal density of states is higher and because the energy is larger such that the bound states extend further into the bulk , increasing the coupling to the superconductor . ] _ cooper pair transport._ we use fermi s golden rule to calculate the cooper pair current @xmath83 , where @xmath84 is the transition rate from an initial state @xmath85 with probability @xmath86 at energy @xmath87 to the final state @xmath88 with 2 more ( less ) electrons at energy @xmath89 . the tunnel hamiltonian @xmath90 comprises the particle number non - conserving terms of the second - quantized perturbative model with electron operators @xmath91 and hole operators @xmath92 , where @xmath93 , @xmath94 because the superconductor interface has a finite width @xmath95 , we restrict the pairing amplitude in real space @xmath96 to @xmath97 $ ] . in momentum space ( suppressing all indices ) this amounts to @xmath98 with @xmath99 from the microscopic calculation and @xmath100}{(l - k)\frac{l}{2 } } \frac{\sin\big[(l+k')\frac{w}{2}\big]}{(l+k')\frac{l}{2 } } , \end{aligned}\ ] ] where @xmath101 is the total length of the system , which does not enter the final results , and we have exploited that the integrand is peaked around @xmath102 . with this , the rates for removing ( adding ) a cooper pair , @xmath103 , become @xmath104 , where at low temperatures the occupation probability @xmath105 with @xmath106 the voltage applied between the superconductor and the channel . rewriting the sum over momenta as energy integrals , the current becomes @xmath107 the combination of energy conservation and approximate momentum conservation implies that the pair tunneling probability @xmath108 has a single peak as a function of @xmath21 for each pair @xmath64 . injection into the same subgap band , @xmath109 , happens near @xmath110 , and into different subbands , @xmath111 , near @xmath112 ( fig . [ fig - setup]b ) . linearizing the dispersion around these points @xcite , @xmath113 , the tunnel amplitude becomes @xmath114 /l(\varepsilon-\varepsilon_0^{nn'})$ ] and we obtain the conductance @xmath115 , \label{eq - cond } \end{aligned}\ ] ] where @xmath116 is the conductance quantum , @xmath117/(\pi w\varepsilon^2)$ ] becomes the delta function for @xmath118 , and @xmath119 is the effective tunneling strength . note that the conductance grows with the length of the interface . this is in contrast to conventional cooper pair splitters , which suffer from an exponential suppression in the spatial size . the reason is that here cooper pairs are split kinematically only after having tunneled locally into the channel , a process which can happen simultaneously along the whole interface . the conductance contains a central zero - bias peak and two characteristic side peaks [ fig . [ fig - conductance](a ) ] , which arise because of the special subgap band structure and which correspond to the injection points marked in fig . [ fig - setup](b ) . the peak height is proportional to the induced superconducting pairings . a factor of 4 arises due to the spin and valley degeneracy and a factor of 2 due to pair transport . and cooper pair current @xmath120 of a @xmath121 long interface between the superconductor and the 1d channel at a distance @xmath122 with @xmath123 . the peak structure reflects simultaneous energy and approximate momentum conservation . the oscillations are caused by the sharp boundary of the superconductor region and vanish if an exponential cutoff is used instead ( dashed ) . ( b ) interpretation of cooper pair splitting in terms of andreev processes . incoming holes ( open circles ) filled up to the bias @xmath106 are either transmitted ( t ) or locally andreev - reflected ( lar ) . ordinary reflection and crossed andreev reflection ( car ) are zero by the valley chirality . a lar process creates an outgoing electron ( filled circle ) on the same side and no outgoing hole on the opposite side , which corresponds to an electron of opposite spin , momentum and energy ( dashed arrow ) . these two electrons are spin entangled ( text ) . ] _ local andreev reflection and cooper pair splitting._ in eq . the singlet nature of the injected cooper pairs is manifest . it is well established that cooper pair splitting is closely related to car @xcite . this applies if the dominant process is ordinary reflection . in our device , the 1d channel with a proximity - induced superconducting region is a nsn junction , in which only transmission through the s - region with amplitude @xmath124 and _ local _ andreev reflection ( an incoming quasiparticle in valley @xmath1 is reflected as an outgoing antiparticle with opposite velocity in valley @xmath125 ) with amplitude @xmath126 are possible . we consider the general scattering problem with finite car and normal reflection in the supplemental material @xcite . when a voltage bias @xmath106 is applied between the superconductor and the channel to extract cooper pairs , the incoming modes are filled with holes up to @xmath127 . without the superconductor , all are transmitted and fill the outgoing modes up to @xmath127 , which is equivalent to a fermi sea for electrons @xmath128 with the fermi energy @xmath129 @xcite @xmath130 here , @xmath131 is the quasiparticle vacuum with respect to the fermi level @xmath132 of the superconductor and @xmath133 creates outgoing holes with spin @xmath0 at energy @xmath134 in the left / right lead , which is the same as annihilating an electron with opposite spin @xmath135 at energy @xmath136 . we drop the valley index which is fixed by the requirement that outgoing modes move away from the superconducting region and the band index for simplicity @xcite . due to the proximity effect lar becomes finite . the key observation is that when lar occurs , no hole with spin @xmath0 at energy @xmath134 is transmitted to the other side . the outgoing mode is therefore occupied by a spin @xmath135 electron at energy @xmath136 [ fig . [ fig - conductance](b ) ] . to see this , we use eq . to write the outgoing state in terms of @xmath128 @xcite , @xmath137\ket{}_{\delta\mu}. \end{aligned}\ ] ] if @xmath138 is small , it becomes @xmath139 \ket{}_{\delta\mu}$ ] , where the desired nonlocal singlet state is explicit . this corresponds to a situation , where individual splitting events are well separated and it is meaningful to talk about pairs . in this regime of interest the perturbative result from the previous section holds . only the emitted pairs contribute to the shot noise of the scattering state . in the opposite limit of perfect lar with @xmath140 , @xmath141 , the outgoing state is a nonentangled product state . lar is most pronounced at energies @xmath142 and @xmath143 [ fig . [ fig - conductance](a ) ] where the superconductor opens gaps @xmath144 in the spectrum for the case of an infinitely long ( @xmath118 ) tunnel - junction ( fig . [ fig - pert ] ) . the lar process becomes weak for all energies , when @xmath95 falls below the coherence lengths @xmath145 . _ conclusion._ our setup allows for highly efficient creation of nonlocal spin - entangled electrons without the need for repulsive interaction or energy filters . we note that the topological channel can be created electrically in the bulk of the bg sample , completely avoiding sharp sample edges , the main source of intervalley scattering @xcite , which could reduce the splitting efficiency . moreover , using an electrically tunable channel geometry ballistic beamsplitters could be created to prove the spin entanglement via noise @xcite , so far an elusive goal . the spin relaxation and decoherence in bg are expected to be weak due to the small spin - orbit coupling @xcite and the sparsity of nuclear spins . we thank a. baumgartner , p. samuelsson , c. schnenberger , and a. levy yeyati for helpful discussions and acknowledge support from the eu - fp7 project se2nd , no . 271554 , the dfg , grant no . re 2978/1 - 1 and research training group grk1952/1 `` metrology for complex nanosystems '' , and the braunschweig international graduate school of metrology b - igsm .

the chemical peculiar ( cp ) stars of the upper main sequence display abundances that deviate significantly from the standard ( solar ) abundance distribution . the existence of strong global magnetic field specifies a subset of this class , the magnetic chemically peculiar ( mcp ) stars . the periodic variability of mcp stars is explained in terms of the oblique rotator model , according to which , the period of the observed light , spectrum , and magnetic field variations is identical to the rotational period . the photometric changes are due to variations of global flux redistribution caused by the phase - dependent line blanketing and continuum opacity namely in the ultraviolet part of stellar spectra @xcite . the amplitude of the photometric variability is determined by the degree of non - uniformity of the surface brightness ( spots ) , the used passband , and the line of sight . the observed light amplitudes are up to a few tenths of magnitudes , standardly in the milky way , we know of a statistically significant number of rotational periods for mcp stars deduced from photometric and/or spectroscopic variability studies @xcite . nevertheless , also extragalactic mcp stars were found in the meanwhile . after the first photometric detection of classical chemically peculiar ( cp ) stars in the large magellanic cloud ( lmc ) @xcite , a long term effort was spent to increase the sample @xcite . finally , we were able to verify our findings with spectroscopic observations @xcite . in this paper , we present the tools of the time series analysis developed for finding and testing of the expected periodic light variations of mcp candidates in the lmc @xcite . the list of targets @xcite was compared with the ogle database @xcite for corresponding measurements . in total , fourteen common objects were found and the @xmath0 and @xmath1 light curves analysed . the description of methods are also published in the textbook by @xcite . the basic tool to search for periodic variations of a measured value ( intensity , magnitude , and radial velocity ) are so called periodograms . these plot some quantities expressing the significance of the phase sorting of searched data according to an individual angular ( circular ) frequency @xmath2 , where @xmath3 is a frequency , @xmath4 , @xmath5 is a period . the extremes of periodograms then indicate the best arrangement of observed data versus a period and hence the possible periodicity of a signal . the basic linear regression models of periodic variations are harmonic polynomials of the @xmath6-order : @xmath7 where @xmath8 is the model of detrended measured quantities @xmath9 corrected for their mean , @xmath10 are @xmath11 harmonic coefficients . the harmonic coefficients for the best fit of model function @xmath8 : @xmath12 for the fixed @xmath13 can be determined by the standard least square method technique allowing to count with uneven uncertainties of individual measurements @xmath14 . the simplest way how to construct lsm spectral periodogram is to plot scalar value @xmath15 versus @xmath13 or @xmath4 , where @xmath16}^2= \sum_{i=1}^n\ , { \left[\frac{y_i^2}{\sigma_i^2}- \frac{f^2(\omega , t_i)}{\sigma_i^2}\right]}.\ ] ] now we can find and discuss the frequencies for which the value @xmath15 reach their minima . this method is fairly general because it can be applied to any kind of time series ( magnitudes , intensities , spectral line equivalent widths , or radial velocities ) . nevertheless for data of the same type ( magnitudes , intensities ) we recommend to use the following modification with some value added. the first sum of equation ( [ chikva ] ) where the first sum on the right is a constant that not depends on the frequency , while the second is the weighted sum of the squares of the model prediction given by the function @xmath17 . therefore , instead of the minimum of the @xmath15 we can find the maximum of the modulated amplitude @xmath18 @xmath19 which is nearly equal to the effective amplitude @xmath20 of a periodic function ( see in * ? ? ? for the first estimate of the variation period it is sufficient to use the simplest possible option : @xmath21 which gives also the most explicit results . then @xmath22}^2.\ ] ] . periodograms are displayed in arbitrary units normalized to the height of the period peak . the results of particular periodograms are comparable , nevertheless both of the new tools of periodograms ( ( b ) and ( c ) ; sec.[sumator ] and sec.[modulator ] ) are a slightly better than the results of the notorious lomb - scargle periodogram ( ( a ) , sec.[scargle]).,scaledwidth=98.0% ] d@xmath23 ( rotation period ) and subsidiary peaks at @xmath24,scaledwidth=97.0% ] the second lsm type of periodograms uses as a criterion of the significance of individual peaks a robust `` signal - to - noise '' @xmath25 criterion which is defined as : @xmath26 @xmath27 is an estimate of the uncertainty of the quantity @xmath28 for a particular angular frequency . the model function @xmath29 is assumed in the form of the harmonic polynomial of the @xmath6-th order ( see eq.[harm ] ) . the detailed description of both lsm novel periodogram criteria can be found in the textbook by @xcite . we tested the properties of the @xmath30 criterion on thousands samples with sine @xmath31 signals scattered by randomly distributed noise . we found that if there is no periodic signal in such data , the median of the maximum @xmath30 value in a periodogram is 4.52 , in 95% of cases we find a @xmath25 value between 4.2 and 5.4 . consequently , the occurrence of peaks definitely higher than 6 indicates possible periodic variations . the periodogram @xmath30 is very sensitive ( see fig.[fig1]b ) and suppresses the false periods which results from usually bad time distribution of astrophysical observational data , very well . during the treatment of ogle - iii time series @xcite , we concluded that both types of periodograms correlate very well with other time - proven periodograms as e.g. the lomb - scargle ( see e.g. * ? ? ? so we are able to consider them as generally interchangeable ( see also fig.[fig1 ] ) . the lomb - scargle method assumes that the found changes have sine / cosine type phase curves . the method uses , as the measure of individual periods significance , the quantity @xmath32 : @xmath33 } ; \quad \displaystyle \omega=\frac{2\,\pi}{p};\\ & q_{\mathrm{ls}}(\omega)=\displaystyle \frac { \sum_{i=1}^n { \left\{y_i\ , \cos[\omega\,(t_i-\tau)]\right\}}^2}{\sum_{i=1}^n \cos^2[\omega\,(t_i-\tau)]}+\displaystyle \frac{\sum_{i=1}^n { \left\{y_i \sin[\omega\,(t_i-\tau)]\right\}}^2}{\sum_{i=1}^n \sin^2[\omega\,(t_i-\tau)]}.\nonumber\end{aligned}\ ] ] the definition of the significance indicator @xmath34 ( see eq.[lomb ] ) can be easily modified to the more general case of data with uneven uncertainties . bs cir ( hd 125630 ) is a moderately cool mcp star with the rotation period @xmath35 . there are many observations of the double - wave light curves available ( for details see in * ? ? ? * ; * ? ? ? although light curves taken in various colours differ , those ones obtained in the spectral region 500560 nm can be considered as almost identical . the periodogram measurements taken from @xcite . ] presented in fig.[fig2 ] shows the spectrum of the square of the modulated amplitude @xmath36 , @xmath37 , see eqs.[harm],[amgen ] ) dominated by the basic period with frequency @xmath38d@xmath23 , and subordinate peaks at @xmath39 . this spectrum results from the fact that the light curves in various filters are not exactly the same . the mean modulated amplitude of the maximum peak correspond to 0.11 mag , the mean weighted @xmath25 of used 4766 individual photometric observations is about 20 . we obtain such a periodogram only in very rare cases when we study an object with the pronounced variability documented by thousands of observations not suffering from bad sampling . nevertheless , frequently we have to analyze data of much worse quality and time sampling . cq uma ( hr5153 = hd 119213 ) , is also a moderately cool mcp star , similar to bs cir , with a strong variation in the @xmath40 color and @xmath41 index . the period of variations of all kind is constant , @xmath42 @xcite . . the low amplitude of it and the absence of any other pronounced period peaks mean that the periodogram of the star is formed namely by aliases of the basic ( true ) period . , scaledwidth=80.0% ] in 1983 - 4 , we obtained nine measurements of the effective magnetic field ( @xmath43 ) using the hydrogen magnetograph of the 6-m sao reflector ( see full dots in fig.[fig3 ] ) . the expected synchronous periodic variations in the light and magnetic field ( @xmath43 ) were confirmed @xcite . however , are these nine magnetic observations of cq uma sufficient for the period determination ? definitely not , the period spectrum is overcrowded by plenty of aliases ( fig.[fig4 ] ) . several of them are so much dominant that they exceed even the peak of the real period . the aliases can be suppressed by a continuation observations . we can simulate it by adding of 27 ` new ' ( artificial ) magnetic measurements obtained during the following six years ( see open circles in fig.[fig3 ] ) . the new periodogram ( fig.[fig5 ] ) undoubtedly indicates the only dominant period at the true placement ] . the phase diagram plotted with this period displays simple sinusoidal variations with a @xmath44 . if we remove the sinusoidal signal from our magnetic measurements , we can search for other possible relic periods using the same periodogram technique - see the bottom of fig.[fig4 ] . it seems that there is no periodic signal in the remaining data meaning that the found period is unique . the remaining data correspond to pure noise which is not very prominent . the characteristics of the periodogram is determined mainly by aliases originated from the very low number of measurements ( 36 ) and the short time basis of observations ( 8 years ) . the aliases can be lowered if we use , besides not very accurate magnetic measurements of @xmath43 , more precise photometric data ( @xmath25 usually up to 20 ) . then we are able to achieve the pure periodogram similar to the periodogram of bs cir displayed in fig.[fig2 ] . several quite different problems were encountered when we analyzed the light curves of mcp candidates in the large magellanic cloud . the number of individual ogle observations in the @xmath0 and @xmath1 colours are @xmath45 and the duration of observations ( 8 years ) were satisfactory . the effect of aliases in such data should be insignificant . unfortunately , the periodograms suffer here from relatively large scatter due to observational noise . the ratio @xmath25 of an individual observation was very low because of the weakness of the signal . we shall now discuss the period analysis of a typical representative of the mcp candidates in lmc - the star ogle lmc136.7 16501 ( 04 50 46.10 -69 59 16.7 , 2000.0 ) , denoted in @xcite as a star no . 4 , with a mean magnitude in @xmath46 mag , the absolute magnitude @xmath47 mag . and @xmath0 light curves of ogle lmc136.7 16501 plotted versus the ` best ' period @xmath48.,scaledwidth=100.0% ] the periodogram depicted in fig.[fig6 ] is quite dissimilar to the periodograms of mcp stars discussed above ( compare with fig.[fig2 ] and fig.[fig5 ] ) . we do not see here any prominent period peak . the characteristics of the period spectrum is mainly determined by the stochastic scatter that competes with several possible periods of the the principal aliases conjugated with the basic sampling period of one sidereal day . the frequencies of aliases obey to notorious @xcite relation : @xmath49 where @xmath50 is the frequency of the alias of the @xmath51-th order , @xmath52 is the frequency of the maximum period peak @xmath53 , and @xmath54 is the sampling frequency , here it is the frequency of a sidereal day @xmath55 . the most significant period in the periodogram ( fig.[fig6 ] ) is 4373 , with amplitudes of the light curves @xmath56 mag , both in @xmath0 and @xmath1 . the ratio of @xmath57 for this peak is very low . the light curves in @xmath0 and @xmath1 are depicted in fig.[fig7 ] . nevertheless , all predicted aliases : @xmath58 are indicated . does it mean that the period 4373 is real ? unfortunately not . the occurrence of aliases at their ` right positions ' given by the tanner relation eq.[tanner ] is only the confirmation of the distinctiveness of expected sampling period or periods in observational data . because the ogle data were obtained at a ground based station , the pronounced one day sampling period can be easily explained . the observed ordering of observation into a phase curve with the period 4373 can be a mere coincidence . that is why we shall solve the crucial question : are the found periodic variations real ? some partial answers may give us several simple tests of variability of general use we have developed @xcite . the heuristic _ shuffling method _ is able to test the ` pessimistic ' hypothesis that the distribution of the data is only random . we can analyze the data of our object displaying weak periodical variations again by the same way as it was described above , only we randomly shuffled all the individual observations ( magnitudes and their uncertainties ) , the times of the observations remained the same . we are convinced that this derogation will destroy any periodic signal and randomize the data . for the shuffled version of data we then find the maximum modulated amplitude @xmath59 and the maximum of @xmath25 . then we repeat the same procedure many times and compare the results with @xmath59 and @xmath25 for the original unshuffled data set . if the results are nearly the same , we may conclude that the periodic variability of the object ( if any ) is undetectable in the investigated data . we found for the mcp candidate ogle lmc136.7 16501 , the median of amplitudes in period peaks of shuffled data @xmath60 mag and a ratio @xmath61 , what is very close to the values found above @xmath56 mag , both in @xmath0 and @xmath1 and @xmath57 . consequently , the lmc136.7 16501 photometric data are very probably random . the same results we obtained also for the other 11 mcp candidates , only for stars no 12 and no 14 ( according to the list of lmc mcp cadidates in * ? ? ? * ) the periods seem significant . we deduce a @xmath62 for significance . the following tests will help to quantify the significance of the periods . the technique of bootstrap @xcite has proved to be very useful for testing the statistical significance and therefore the reality of found periods . it helped us to quantify this reality as a probability that the periodogram of a randomly created bootstrap subset has its dominant peak at the same frequency as the standard periodogram . we tested it with one hundred of bootstrap subsets for each star of our sample . we consider a period as statistically significant , if the maximum peak occurs at one of the aliased frequency because during the bootstrap choice aliases often exceed the basic peak . the results for found periods of all 14 mcp candidates were dismal : the bootstrap significance never exceeds 50% . especially , the bootstrap significance of the period 437 in the case of the star 4 was only 26% ! the periodicity of stars no . 12 and 14 , mentioned in sec.[shuple ] , is disputable . for data , including observations in two or more spectral regions their individual periodograms should indicate nearly the same peak period . it can be tested , e.g. by the periodogram of the product of individual @xmath18 or @xmath25 values . all mentioned newly developed methods are open source . for details see in @xcite or the textbook of @xcite . this paper uses observations made at the south african astronomical observatory ( saao ) . this work was funded by the grant gar p209/12/0217 and supported by the somopro ii programme ( 3sga5916 ) , co - financed by the european union and the south moravian region , the grant ga r 7amb12at003 , and the financial contributions of the austrian agency for international cooperation in education and research ( bg-03/2013 and cz-09/2014 ) . mn acknowledges the support by the grant 14 - 26115p of the czech science foundation.this work reflects the opinion of the authors and the european union is not responsible for any possible application of the information included in the paper . catalano , f. a. , leone , f. 1993 , , 100 , 319 esa , 1998 , the hipparcos and tycho catalogs , celestia 2000 , sp1220 krtika , j. , mikulek , z. , henry , g. w. , et al . 2009 , a&a , 499 , 567 krtika , j. , mikulek , z. , lftinger , t. et al . 2012 , a&a , 537 , a14 krtika , j. , mikulek , z. , zverko , j. , & iovsk , j. 2007 , , 470 , 1089 maitzen , h. m. , paunzen , e. , & pintado , o. i. 2001 , , 371 , l5 manfroid , j. , renson , p. 1983 , , 51 , 267 mas - hesse , j.m . , gimnez , a. , domingo , a. , rsquez , d. , caballero , m.d . , gutirrez , r. , solano , e. and the omc team , 2004 . 5th integral workshop on the integral universe , 552 , 729 mikulek , z. , glagolevskij , yu . v. , romanyuk , i. i. , shtol , v. g. , bychkov , v. d. 1984 , in magnetic stars , 13 mikulek , z. , jank , j. , zverko , j. , iovsk , j. , netolick , & vako , m. 2007 , astronomische nachrichten , 328 , 1 , 10 mikulek , z. , krtika , j. , jank , j. et al . 2011 , in magnetic stars , proceedings of the international conference , sao ras 2010 , eds : i. i. romanyuk and d. o. kudryavtsev , 52 mikulek , z. , krtika , j. , jank , j. et al . 2013arxiv1310.6640 mikulek , z. , jank , j. , krtika , j. , zejda , m. , & jagelka , j. this contributions 2015 mikulek , z. & zejda , m. , in vod do studia promnnch hvzd , isbn 978 - 80 - 210 - 6241 - 2 , masaryk university , brno 2013 mikulek , z. , zverko , j. , krtika , j. et al . 2007 , in magnetic stars , eds . iosif romanyuk , dmitry kudryavtsev 2007 , 352 paunzen , e. , maitzen , h. m. , pintado , o. i. , et al . 2006 , , 459 , 871 paunzen , e. , mikulek , z. , poleski , krtika , j. , netopil , & m. zejda , m. 2013 , , 556 , a12 paunzen , e. , netopil , m. , bord , d. j. 2011 , , 411 , 260 pojmaski , g. 2001 , asp conference series vol . 246 , iau colloquium 183 . b. paczynski , w - p . chen , and c. lemme , 53 renson , p. , catalano , f. a. 2001 , , 378 , 113 press , w. h. , & rybicki , g. b. 1989 , , 338 , 277 tanner , r. w. 1948 , journal of the royal astronomical society of canada , 42 , 177 udalski , a. , syzmanski , m. k. , soszynski , i. , & poleski , r. 2008 , acta astronomica , 58 , 69 vogt , n. , faundez , a. m. 1979 , 36 , 477

our paper presents new methods for finding and testing of weak periodic variability of stellar objects developed for the purpose of detecting expected regular light variations of magnetic chemically peculiar ( mcp ) candidates in the large magellanic cloud . we introduce two new periodograms of the mcp star , bs cir ( hd 125630 ) , appropriate for rotating spotted variables and compare the results with those obtained by the well - known lomb - scargle periodogram . the usage of periodograms and the testing of the significance of the found period candidates are demonstrated with two examples : the observed and simulated observations of the magnetic field of the mcp star cq uma ( hd 119213 ) and the mcp candidate ogle lmc136.7 16501 . three newly developed tests of the periodic variability the shuffling , bootstrap and subsidiary ones , are presented . we demonstrate that the found periodic variations known with signal - to - noise ratio larger than 6 can be approved as real .

considering observable effects from wormholes or ringholes is not new . more than a decade ago some of such effects were already predicted by several author [ 1,2 ] . in particular , it was shown that these tunnelings can induce lensing effects from luminous sources . more precisely , by embedding these tunnelings in friedmann space , it was seen [ 2 ] that , besides the expected lensing , at least ringholes were able to induce other potentially observable effects such as frequency shifting of the emitting sources , discontinuous change of background temperature , broading and intensity enhancement of the spectral lines , as well as a rather dramatic increase in the luminosity of any objects at the tunnel throat . moreover , the precise form of the lensing signature left by wormholes and ringholes has been quite more recently seen to consist of a single ring and a double concentric ring , respectively [ 3,4 ] . in spite of the interest that all of such results may have from a pedagogical standpoint , none of these induced phenomena has by itself any practical usefulness in order to identify the existence of space - time tunnels in the universe because there are other observable objects in the universe able to produce similar effects . in particular , some hope was raised in using wormholes and specially ringholes to get a direct evidence of the existence of other universes to which they could be connected , so traversing some information from them to ours own . nevertheless , such a hope is easily shown to vanish when the following two arguments are taken into account . first of all , the above alluded uncertainty that the effects be produced by these tunnelings and not by more familiar astronomical objects such as galaxies , stars black holes or quasars . on the other hand , even though some authors have tried to consider models where other universes were made observable to us through collision with our own [ 5 ] . moreover , from the very definition of universe it follows that provided that there can not be any space - time connections between whichever pair of universes , less yet it is allowed the possibility for any well - defined relation between the spaces or times of them . it could be argued that if one allows the connecting wormholes or ringholes to be converted into time machines with completely unspecified mutual velocities between the mouths one might create a whole space - time that would represent two universes . however , such a possibility can not be entirely implemented because such a whole space - time would be orientable against the opposite evidence that it would necessarily violate orientability . thus , rather than using orientable space - time tunnels such as wormhole or ringholes , it appears that two universes can be mutually tunneled to each other while preserving their property of still being well - defined universes if one by instance used non orientable klein bottle holes converted into time machines by allowing their mouths to move relative to one another at completely unspecified speeds [ 6 ] . in this paper we shall consider the effects that such connections would produce and the possibility to using them to check the existence of universes other than ours . one of the most important revolutions in cosmology is taking place right now , that of the so called multiverse [ 7 ] . in order to convert this revolution in more than just a speculative idea , providing it with a physical content , one should try to consider alternate ways which would for example include the effects that time machines derived from nonorientable klein bottle holes [ 6 ] may have on the luminous sources placed beyond the limit of our universe . something that any of the existing multiverse theories is very needed of . in fact , linde ideas coming about string theories and alike [ 8 ] or other s ideas [ 9 ] may make the notion of a multiverse more plausible , but they do not prove that other universes are really out there . the staggering challenge is to think of a conceivable experiment or observation confined to our own universe based on looking for some footprints left by nonorientable tunnelings connecting our universe to other universes . general principles of physics cast in fact serious doubts on whether it make sense to talk about other universes if they can never be detected . rees , an early supporter of multiverse idea , agrees [ 10 ] that it may never be possible to observe other universes directly , but he argues that scientists may still be able to make a convincing case for their existence . to do that , he says , physicists will need a theory of the multiverse that makes new but testable predictions about properties of our own universe . if , similarly to as current observations have confirmed big bang as a well established model , new experiments coming perhaps from large hadron collider or the planck satellite space mission , indirectly confirmed such a theory predictions about the universe we can see , rees believes , they would also make a strong case for the reality of those we can not . string theory is still very much a work in progress , but it could form the basis for the sort of theory that rees is calling for . however , the very essentials of quantum theory show some great , almost insurmountable odds against the rees philosophy , specially if one adheres to the quantum - cosmological ideas that support the principle according to which physical reality should be directly observable or it vanishes into nothing . in this way , it appears a casus of full necessity to explore the existing ways that may lead us to directly observe a property or characteristic of a universe other than ours , such as the main objective of the present paper is aiming at exploring the possible existence of other universes through a search for the lensing signature of orientable ringholes and non orientable klein bottle holes connecting us to other universe . it was first noted by shatskiy [ 3 ] that wormholes , which are usually disguised as black holes , can be made observable and recognizable in terms of bright , glowing rings originating from the necessary flaring out of the embedding surface around their throat which is produced by the presence of the so called phantom stuff [ 11 ] . the really most devastating argument against the proposed wormhole distinguishable character of the shatskiy rings is that , even if exotic matter does exist , other many objects are able to create a similar lensing light signature [ 12 ] . in particular , while the orientability of the wormhole manifold makes it impossible using these solutions for observing any thing from a universe other than ours , it is hard to see how the resulting lensing rings could be differentiated from the astronomical blueprint left by negative energy stars and , mainly , from all those massive astronomical objects , such as galaxies , black holes or quasars , whose gravitational lensing effects appear as the so called einstein rings [ 11 ] . the actual problems are with the symmetry and the orientability of the throat . wormholes are orientable manifolds which are characterized by a spherically symmetric throat and , therefore , there will be a diverging lensing effect undergone by any bundle of light rays coming from a luminous source placed behind the furthest wormhole mouth which would necessarily manifest to observers on earth as single perfectly circular rings , such as it was indicated by shatskiy [ 3 ] . this pattern could well be misinterpreted as being originated from a star or other massive astronomical object necessarily placed in our own universe , instead of a wormhole , with a radius quite smaller than that for that wormhole throat radius . gravitational lensing effect produced by a ringhole from a single luminous source . ( a ) parameters defining the toroidal ringhole throat in terms of which metric ( 1 ) is defined . ( b ) rays passing near the outer and inner surfaces respectively flare outward and inward , leading to a image from a luminous point placed behind the ringhole which is made of two concentric bright rings . the relative mutual positions of these rings would depend on the distance between the ringhole and the luminous source . if that distance is small enough then the larger outer ring comes from the flaring inward surface , and conversely , if the distance source - ringhole is increased then the outer ring comes from the outward surface , the larger that distance the greater the difference between the two bright rings . ] an inner tunneling symmetry which would give rise to a more distinguishable lensing pattern is that of a ringhole [ 13 ] , that is , an orientable space - time tunnel whose throat has the toroidal symmetry , instead of the spherical symmetry . using the set of geometrical parameters specified in the upper part of fig . ( 1 ) we can derive the metric for a ringhole to be [ 13 ] @xmath0d\varphi_2 ^ 2 + m^2 d\varphi_1 ^ 2\ ] ] where @xmath1 with @xmath2 and @xmath3 arbitrary integration constants , and @xmath4 and @xmath5 the radius of the circumference generated by the circular axis of the torus and that of a torus section , respectively , with @xmath6 . metric ( 1 ) is defined for @xmath7 , @xmath8 and the angles ( see fig . 1 ( a ) ) @xmath9 , @xmath10 . a thorough analysis [ 4,13 ] leads then us to deduce that one would expect lensing effects to occur at or near the ringhole throat , that is to say , the mouths would act like a diverging lens for world lines along @xmath11 , and like a converging lens for world lines along @xmath12 . no lensing actions would therefore take place at @xmath13 and @xmath14 . in fact , in the case of ringholes , instead of producing just a single flaring outward for light rays passing through the wormhole throat , this multiply connected topology , in addition to that flaring outward ( diverging ) effect , also produces a flaring inward ( converging ) effect@xmath15 on the light rays that pass through its throat , in such a way that an observer on earth would interpret light passing through the ringhole throat from a single luminous source as coming from two bright , glowing concentric rings , which form up the distinctive peculiar pattern from ringholes ( see fig . 1 ( b ) ) . that pattern can not be generated by any other possible disturbing astronomical object other than a very implausible set of three luminous massive objects ( let us say galaxies ) which must be so perfectly aligned along the sigh line that its occurrence becomes extremely unlikely . it is readily inferred from fig . 1 ( b ) that , for a reasonably large ringhole sufficiently far from the luminous source , the inner bright ring would correspond to the flaring inward ( converging ) surfaces . if we keep the ringhole size invariant and the distance between the ringhole and the luminous source is decreased drastically , then the inner bright ring would turn to be produced by the flaring outward ( diverging ) surface . such a ringhole signature may have been already observed though it has been so far attributed to the combined effect of two einstein rings originated from the above - considered to be extremely unlikely superprecise alignment of three galaxies . in fact , at the beginning of 2008 the nasa / esa hubble space telescope revealed [ 14 ] a never - before - seen phenomenon in space : a pair of glowing rings , one nestled inside the other like a bulls - eye pattern . this double - ring pattern was interpreted as a double einstein ring being caused by the complex bending of light from two distant galaxies strung directly behind a foreground massive galaxy , like three beads on a string along the line of sight , simply because at the time there were no other available interpretations for what was being observed . being more than just a novelty , this very rare phenomenon found with the hubble space telescope could , moreover , eventually offer insight into dark matter , dark energy , the nature of distant galaxies , and even the curvature of the universe . as previously stated , for that interpretation to be feasible , the massive foreground galaxy had to be almost perfectly aligned in the sky with two background galaxies at different distances to justify the finding . the foreground galaxy is 3 billion light - years away . now , in order to justify the ratio between the two ring radii , the inner ring and outer ring would be comprised of multiple images of two galaxies at a distance of some 6 billion and approximately 11 billion light - years . however , the odds of observing the required extremely precise alignment of the three galaxies are so small ( an estimated 1 in 10,000 ) that even some of the discoverers of that astronomical phenomena said that they had hit the jackpot with the discovery . at the time , the authors of ref . 14 had no alternative other than accepting that quite improbable interpretation of the result . nevertheless , having we uncovered that such concentric rings may well be also interpreted as the blueprint of the presence of a ringhole in the direction in space where the double bright ring system was discovered , we may also adopt the latter interpretation in terms of a ringhole as an alternate possible explanation for that phenomenon , taking now the luminous sources at redshifts corresponding to 3 and 6 billion light - years as measuring the positions of the two ringhole mouths on the sky ( provided the mouths are surrounded by some sufficiently large quantities of luminous material ) , and their respective luminosities as stemming from the respective light deflections along the angle @xmath16 caused by the combined effect of the size of the throat radius and the relative distance between the two mouths . even in the absence of sufficiently high matter densities around the two ringhole mouths in a still large enough tunneling , since the two rings have the same spectra in the case of a ringhole , it could well be that the unseen second source might be too faint so that these spectra actually be the same which corresponds to matter placed anywhere along the ringhole tunneling . the feature that the surface brightness of the two rings would be different may also be justified by the above - suggested different distribution of ordinary and exotic matter leading to distinct absorption and dispersion effects of the incoming light along the two horizon separated @xmath16-angular regions around the throat . in any event , because of the orientability of the single ringhole manifold or the composite manifold formed by the aligned galaxies , which is required to get the observed concentric double ring bright image , the luminous source originating the lensing effect can never be placed in a universe other than ours , making in this way a ringhole a completely useless tool to check the possible existence of other universes . the use of inter - universe tunnelings in order to check the existence of other universes should actually require choosing spacetime holes which be ( i ) non orientable and ( ii ) convertible into a time machine with completely unspecified relative speed between its mouths . one such space - time construct has in fact been already studied , in the shape of what is dubbed a klein bottle hole [ 6 ] . in this case , non orientability is guaranteed by the existence of a throat with the topology of a klein bottle and it was also shown that this space - time can be stable to vacuum fluctuations and is also convertible into a time machine with fully arbitrary inter mouths speed . employing the set of geometrical parameters specified in fig . 2 and in the lower part of fig . 3 we can derive the space - time metric for one of such klein bottle holes to be [ 6 ] @xmath17 where the @xmath18 s are the step heaviside function [ 15 ] , with @xmath19 for @xmath20 and @xmath21 for @xmath22 , @xmath23 s for @xmath24 is given by @xmath25\right\}d\varphi_1 ^ 2\nonumber\\ & & b_1 ^ 2d\varphi_2 ^ 2-b_1\sqrt{(a_1-c_1)(a_1-a_1)}\sin\varphi_2d\varphi_1 d\varphi_2 , \end{aligned}\ ] ] in which @xmath26 @xmath27 @xmath28 now , for @xmath29 @xmath30\right\}d\varphi_2 ^ 2\nonumber\\ & & b_2 ^ 2d\varphi_2 ^ 2-b_2\sqrt{(a_2-a_2)(c_2-a_2)}\sin\varphi_2 d\varphi_1 d\varphi_2 , \end{aligned}\ ] ] where in this case @xmath31 @xmath32 finally we have @xmath33 @xmath34 where we have extended the range of the angular coordinate @xmath35 to also encompass values continuously running from @xmath36 to @xmath37 and @xmath38 , @xmath39 , @xmath40 and @xmath41 , @xmath42 , are given sets of adjustable parameters which are arbitrary unless for the conditions @xmath43 , @xmath44 , @xmath45 and @xmath46 for the angular interval @xmath47 , whereas for @xmath29 we must have @xmath48 , @xmath49 , @xmath50 and @xmath51 , with @xmath52 , @xmath53 and @xmath54 , with @xmath55 . @xmath56 and @xmath57 are the radius of the circumference generated by the circular axis of the klein bottle torus and that of a klein bottle section , respectively , with @xmath58 . metric ( 3 ) is defined for @xmath7 , @xmath59 and the angles ( see figs . 2 and 3 ) @xmath9 , @xmath10 . pictorial representation of a klein bottle hole showing the non orientable topology of its throat and some of the parameters in terms of which metric is defined ( see also fig . 3 ) . ] in order to check the properties of a klein bottle hole as a lens , we now write the static space - time metric of a single , traversible klein bottle hole in the form @xmath60\nonumber\\ & & + \theta(\varphi_1 -2\pi)\left[\left(\frac{n_2(\ell_2)}{r_2(\ell_2)}\right)^2 d\ell_2 ^ 2 + d\omega_2 ^ 2(\ell_2)\right ] , \end{aligned}\ ] ] where @xmath61 , with @xmath62 , and @xmath42 . here , @xmath57 is replaced for @xmath63 , @xmath64 being the proper radial distance of each transversal section of the klein bottle on the respective @xmath35 interval for @xmath65 , and @xmath66 @xmath67 @xmath68 in which @xmath69 is as given from @xmath70 and @xmath71 for constant parameter adjusted to be the radius of the throat of the klein bottle hole at @xmath72 . as @xmath64 increases from @xmath73 to 0 , @xmath57 decreases monotonously from @xmath74 to its minimum value @xmath69 at the throat radius , and as @xmath64 increases onward to @xmath74 , @xmath57 increases monotonously to @xmath74 again . now , for metrics ( 3 ) and ( 13 ) to describe a klein bottle hole we must embed it in a three - dimensional euclidean space with cylindrical coordinates at fixed time @xmath75 [ 6 ] whose metric can be written as @xmath76dr^2 + r^2 d\phi^2 , \ ] ] however , since @xmath77 and @xmath35 are no longer independent to each other , one can always convert metric ( 3 ) into a line element which is embeddible in the cylindrical space ( 17 ) . that conversion can be made by first obtaining the expression for the variation of @xmath77 with respect to @xmath35 , that is @xmath78\sin(i\varphi_1/2)}{2ir_i } .\ ] ] from which we get @xmath79 with @xmath80 . taking now @xmath81 , for @xmath82 and @xmath83 , for @xmath84 we can derive for any allowed value of @xmath16 , @xmath85 hence it follows that the metric for the non orientable klein bottle hole which is embeddible in flat space is described by eq . ( 19 ) with @xmath86 as given by ( 20 ) and @xmath87 . using these coefficients , metric ( 17 ) will be the same as metric ( 19 ) for constant values of @xmath16 if we identify the coordinates @xmath88 , @xmath89 of the embedding space with either coordinates @xmath90 , @xmath91 , for @xmath82 , or the coordinates @xmath92 , @xmath35 , for @xmath84 , and if we require the function @xmath93 to satisfy @xmath94 for any value of @xmath35 , @xmath95\left[r_1-\rho(\phi_2)_1\right]},\ ] ] for @xmath82 , and @xmath96\left[r_2-\rho(\phi_2)_2\right]},\ ] ] for @xmath84 , where @xmath97 and @xmath98 from these expressions and the requirement that non orientable klein bottle holes be connectible to asymptotically flat space - time , one can deduce how the embeddible surfaces would flare at or around the hole throat . thus , from eq . ( 21 ) , one obtains @xmath99 which is positive for @xmath100 and negative for @xmath101 . thus , exactly to as it happens in the case of toroidal ringholes@xmath102 , the embedding surface flares outward for @xmath103 and flares inward @xmath104 . it is for this reason that a klein bottle hole would generally behave like a diverging lens for @xmath105 , and like a converging lens for @xmath106 , even though , unlike in the ringhole case , these behaviors will also depend on the value of the angle @xmath35 due to the non orientable character of the klein bottle hole space - time , such as we shall show in some detail in what follows . in order to investigate how the embedding surface flares at or around the throat as the angle @xmath35 is varied , so making the manifold non orientable , we distinguish two cases . the first case corresponds to condition ( 22 ) , from which we can derive @xmath107\left[r(\varphi_2)_1- \rho(\varphi_2)_1\right]\sin(\varphi_1/2)}{2\left\{\left[r(\varphi_2)_1- r_1\right]\left[r_1 -\rho(\varphi_2)_1\right]\right\}^2 } .\ ] ] now , since @xmath58 for @xmath24 one obtains that this expression vanishes at @xmath108 and becomes negative for @xmath109 , on which values the embedding surface flares toward larger values of the radius @xmath110 , and negative for @xmath111 on whose values the embedding surface flares toward smaller values of @xmath110 . the second case corresponds to condition ( 23 ) for which we get @xmath112\left[r(\varphi_2)_2- \rho(\varphi_2)_2\right]\sin(\varphi_1)}{4\left\{\left[r(\varphi_2)_2- r_2\right]\left[r_2 -\rho(\varphi_2)_2\right]\right\}^2 } .\ ] ] the critical value of @xmath35 becomes then @xmath113 . for @xmath114 , eq . ( 28 ) becomes negative so that the embedding surface flares toward smaller values of @xmath115 , while it becomes positive for @xmath116 , for which the embedding surface flares toward larger values of @xmath115 . the above analysis leads us to expect lensing effects to occur on the mouths of the nonorientable klein bottlehole with respect to a bundle of light rays , at or near the throat , coming from the distribution of negative and positive values for the energy density [ 6 ] ; that is , the mouths would act like a diverging lens for world lines along the values of the coordinates , at or near the throat , which correspond to negative energy density , and like a converging lens for world lines passing through regions with positive energy density . in order to confirm with full accuracy which regions around the throat behave like a converging or diverging lens , one must consider the null - ray propagation governed by the integral of the stress - energy tensor . from the einstein equations [ 6 ] it can be obtained that the mouths focus or defocus a bundle of rays , depending on the sign of the integral [ 6 ] @xmath117\cos^2\left(\frac{\varphi_1}{4}\right)\right\ } } + \frac{2\left(1+\sin\varphi_2\right)}{n_1 b_1 } -\frac{4cos\left(\frac{\varphi_1}{2}\right)\sin\varphi_2}{(a_1 -c_1)m_1 \sin^2\left(\frac{\varphi_1}{2}\right)}\right\}\end{aligned}\ ] ] in what follows we shall restrict ourselves to consider the case @xmath118 and the still realistic situation where @xmath119 and @xmath120 . we obtain then @xmath121\end{aligned}\ ] ] where @xmath122}\ ] ] @xmath123 we can conclude in this way that the integral @xmath124 is always definite positive , irrespective of the value of the angle @xmath35 . it follows that the dependence of @xmath124 on @xmath35 will only contribute the strength of the action of the two different gravitational lenses that can be distinguished at and around the klein - bottle hole throat , so just quantitatively modifying the behavior governed by the angle @xmath16 . inspection on the above equations leads to the final result which is twofold . on the one hand , we can derive that the surface gravity @xmath125 is definite positive for @xmath126 and definite negative for @xmath127 , and hence , since generally we have @xmath128 , that the klein bottle hole emits thermal ( phantom ) radiation at negative temperature from the first of these regions and thermal radiation at positive temperature from the second region [ 16 ] . on the other hand , a ready calculation leads us to obtain ( see fig . 3 ) that the signature that a klein bottle hole would in any event leave from a luminous source placed beyond it for an observer on earth is that of two concentric truncated double spiral . gravitational lensing effect produced by a klein bottle hole from a single luminous source . on the lower part of the figure the parameters defining the klein bottle hole throat in terms of which metric ( 3 ) is defined . the rays passing near the outer and inner surfaces respectively flare outward and inward , leading to a image from a luminous point placed behind the klein bottle hole which is made of two concentric truncated bright spirals . ] in this case , besides valuable information on dark matter , dark energy and universe curvature or the very existence of klein bottle holes in our universe , what could eventually be most astonishing in its implications would be an unprecedented insight into some of the contents of other universes linked to ours by means of these klein bottle holes . after all , a klein bottle hole is a perfectly valid solution to the einstein equations for stress - energy tensor containing a given proportion of exotic stuff - possibly phantom energy- which is becoming more and more familiar in the full context of current cosmology [ 17 ] . the potentially attainable insight from such an interpretation is twofold . on the one hand , we would get a direct evidence for the existence of klein bottle holes and , by the way , possibly of wormholes and ringholes , and on the other hand , we could have found a way to open the door to observe a parallel universe , and hence got a first direct evidence for the existence of the multiverse scenario . there is an observation which may in principle distinguish a static klein bottle hole staying within our own universe and having its two mouths at rest with respect to each another , from a klein bottle hole that connect our universe to a parallel universe or , in general , to other universe of a multiverse scenario . in the latter case since there is no common space - time for the two universes ( parallel or not ) , the two mouths should necessarily be in perpetual quasi periodic relative random motion with completely unspecified speed . this would make the time and space for the two universes at all independent because the relative motion of the two mouths converts the klein bottle hole in a time machine that contains completely arbitrary closed timelike curves . in the case of the inner static klein bottle hole , if the luminous source is kept motionless and the klein bottle hole does not behave like a time machine , the double truncated spirals would be well resolved and defined on the pattern . however , if the positions of the two mouths continuously vary relative to each other in a random though quasi periodic way then the width of each of the truncated spirals would be stretched out and their resolution spoiled and clearly blurred due to the continuous and completely arbitrary changes of distance between the two mouths , thus leading to a glowing background around the spirals of fig . 3 , showing just a maximum of intensity at the average relative position of the mouths , provided the relative motion keep a sufficiently high degree of periodicity . in the latter case , the metric of the klein bottle hole would change to be given by a line element that describes arbitrary time travel induced by a nearly periodic relative motion between the two mouths . using arguments similar to those employed in refs . [ 4 ] and [ 5 ] we finally get for that line element @xmath129 ^ 2\right\}.\nonumber\\ & & \left.\times e^{2\phi}dt^2 + c(1)d\ell_1 ^ 2+d(1)q(1)dr_1 d\varphi_1 + d\omega_1 ^ 2\right)\nonumber\\ & & + \theta(\varphi_1 -\pi)\left(\left\{-[1+\bar{g}f(\ell_2)\ell_2\sin\varphi_2\right]^2 + 1\right\}\nonumber\\ & & \left.\times e^{2\phi}dt^2 + d(2)q(2)dr_2 d\varphi_1 + d\omega_2 ^ 2\right),\end{aligned}\ ] ] where @xmath130 is the average acceleration of the moving mouth , with @xmath131 the corresponding average velocity , and @xmath132 is the average on the fuzzy relativistic factor ; finally @xmath133 is a form factor that vanishes in the half of the klein bottle hole which is assumed to be kept motionless , and rises up on average from 0 to 1 as one moves along the direction of the moving mouth . we must finally point out that any ringhole which is a time machine even within our universe will also show an defocused double truncated spiral pattern though not so blurred perhaps as that corresponding to an inter - universe ringhole . we finally briefly discuss the odds of finding a macroscopic klein bottle hole which is kept stable . it was first argued [ 18 ] that only quantum wormholes , and hence quantum ringholes and klein bottle holes , with nearly the planck size can be stable , with larger tunnelings being violently destabilized by quantum effects produced by catastrophic particle creation taking place near the chronology horizons . actually , hawking even advanced his chronology protection conjecture [ 19 ] for wormholes which can also be applied to ringholes and klein bottle holes , preventing the appearance of closed timelike curves , so making the universe safe for historians and free of the occurrence of the kind of phenomena dealt with in this paper . thus , neither wormholes or ringholes nor klein bottle holes could exist due to these quantum fluctuation instabilities . however , besides some counter - examples to the hawking s conjecture that includes e.g. some compelling argument by li and gott [ 20 ] , it has been shown [ 21 ] that macroscopic wormholes , ringholes and klein bottle holes can be stabilized after the cosmic coincidence time taking place at the onset of the dark energy era by the accelerating expansion of the universe which induces their throat to quickly growing comovingly to the super - luminal universal expansion . on the other hand , similarly to as it happens with wormholes [ 22 ] , accretion of phantom energy onto the ringholes and klein bottle holes should also induce in them a ultra rapid swelling up that would circumvent the kind of quantum effects considered by hawking so that , such as it also happens with their above - mentioned size increasing which is comoving to the universal expansion , the destabilizing quantum effects here can not act in time to destroy the tunnel during the current speeding - up of the universe . therefore , the odds for all of these tunnelings to exist and gravitationally act on the light coming from luminous sources the way we showed before appear to be good enough in the context of our accelerating universe as for allowing the kind of interpretation considered in this letter . on the other hand , in spite of the feature that for ringholes this interpretation just requires two objects aligned on axis , as opposed to requiring three objects to be on axis as the interpretation first suggested in ref . [ 14 ] did , one could likewise think that possibly ringholes are quite rarer than galaxies so that the former interpretation is far from being quite more likely than the latter interpretation either . thus , at least for the time being , it appears hard to decide which of these two interpretations should be chosen . only more accurate analysis on the involved spectra and on the relative brightness of the two rings , and mainly the discovery of other double rings systems , could be used to finally choose which among these two interpretations is more likely to hold . as to using klein bottle holes to check the physical existence of other universes , it appears just a matter of time to find a double truncated spiral blurred enough to clearly show a connection with other universes . morris , k.s . thorne , and u. yurtsever , phys . 61 , 1446 - 1449 ( 1988 ) ; m.s . morris and k.s . thorne , am . 56 , 395 - 412 ( 1988 ) ; j.g . cramer , r.l . forward , m.s . morris , m. visser , g. benford and g.a . landis , phys . d51 ( 1995 ) 3117 . gonzlez - daz , phys . d56 , 6293 - 6297 ( 1997 ) . shatskiy , astro - ph/0712.2572 ; astro - ph/0809.0362 ; astron . 51 , 81 - 86 ( 2007 ) [ astron . 84 , 99 - 104 ( 2007 ) ; a.a . shatskiy , i.d . novikov , and n.s . kardashev , phys .- usp 51 , 457 - 464 ( 2008 ) . gonzlez - daz , _ lensing effects in ringholes and the multiverse _ in : proceedings of the workshop on `` cosmology , quantum vacuum , and zeta functions '' , ice , csic , uab , barcelona ( 2010 ) ( to be published ) ; p.f . gonzlez - daz , _ is the 2008 nasa / esa double einstein ring actually a ringhole signature ? _ , astro - ph/1001.3778 . whereas , as reported by m. chown in new scientist , 24 sept . 2007 , l. mersini - houghton and r. holman attributed an observed big void in our universe to an entanglement effect caused by other universe within the context of string theory , a. aguirre , m.c . johnson and a. shomer considered the observable signatures left by other bubble universe which collided with our universe during their primordial evolution , in phys . d76 ( 2007 ) 063509 . gonzlez - daz and l.j . garay , phys . d59 ( 1999 ) 064026 . see contributions in _ universe or multiverse _ , edited by b. carr ( cambridge univ . press , cambridge , uk , 2007 ) . linde and v. vanchurin , _ how many universes are in the multiverse _ hep - th/0910.158 v2 ; a. linde , mod . phys . lett a1 , 81 - 85 ( 1986 ) ; phys . b175 , 395 - 400 ( 1986 ) . h. everett , rev . 29 , 454 - 462 ( 1957 ) ; see papers by h. everett iii , j. a.wheeler and b. s. de- witt in : _ the many - worlds interpretation of quantum mechanics _ , edited by b. s. dewitt and n. graham . ( princeton univ . press , princeton , usa , 1973 ) ; k. ashok and m. douglas , jhep 060 - 065 ( 2004 ) 0401 [ arxiv : hep - th/0307049 ] ; l. susskind , [ arxiv : hep - th/0302219 ] ; s. robles - prez , p. mart - moruno , a. rozas - fernndez and p.f . gonzlez - daz , class . 24 , f41 - f44 ( 2007 ) . see paper by martin rees in ref . 7 . r. r. caldwell , phys . b 545 , 23 - 29 ( 2002 ) ; s. m. carroll , m. hoffman and m. trodden , phys . d 68 ( 2003 ) 023509 ; r. r. caldwell , m. kamionkowski and n. n. weinberg , phys . 91 ( 2003 ) 071301 ; p.f . gonzlez - daz , phys . b 586 , 1 - 4 ( 2004 ) ; p.f . gonzlez - daz , phys . d 69 ( 2004 ) 063522 . a. einstein , science 84 , 506 - 507 ( 1936 ) . gonzlez - daz , phys . d54 , 6122 - 6131 ( 1996 ) . r. gavazzi , t. treu , l.v . e. koopmans , a.s . bolton , l.a . moustakas , s. burles and p.j . marshall , _ the sloan lens acs survey . vi : discovery and analysis of a double einstein ring _ , astro - ph/0801.1555 . m. abramowitz and i.a . stegun , _ handbook of mathematical functions _ , ( dover , new york , usa , 1972 ) . the ultimate physical reason why a klein bottle hole may radiate phantom energy at a negative temperature lies on the feature that the surface embedding the klein bottle hole space - time flares outward at or near the throat on the region @xmath134 , see p. martn - moruno and p.f . gonzlez - daz , phys . d80 , 024007 ( 2009 ) ; clas . quant 26 , 215010 ( 2009 ) . d. j. mortlock and r. l. webster , the statistics of wide - separation lensed quasars , mon . not . * 319 * , 872 - 878 ( 2000 ) [ arxiv : astro - ph/0008081 ] ; a. g. riess _ et al . _ [ supernova search team collaboration ] , astron . j. * 116 * ( 1998 ) 1009 - 1038 [ arxiv : astro - ph/9805201 ] ; s. perlmutter _ et al . _ [ supernova cosmology project collaboration ] , astrophys . j. * 517 * , 565 - 586 ( 1999 ) [ arxiv : astro - ph/9812133 ] ; j. l. tonry _ et al . _ [ supernova search team collaboration ] , astrophys . j. * 594 * , 1 - 24 ( 2003 ) ; d. n. spergel _ et al . _ [ wmap collaboration ] , astrophys . j. suppl . * 148 * , 175 - 194 ( 2003 ) ; c. l. bennett _ et al . _ , astrophys . j. suppl . * 148 * , 1 - 27 ( 2003 ) ; p.f . gonzlez - daz , phys . d58 ( 1998 ) 124011 . s.w . hawking , d46 , 603 - 611 ( 1992 ) . li - xin - li and j. r. gott , phys . 80 ( 1998 ) 2980 - 2983 ( 1998 ) . gonzlez - daz , phys . d68 ( 2003 ) 084016 . gonzlez - daz , phys . 93 ( 2004 ) 071301 .

it is argued that whereas the shatskiy single rings produced by the gravitational inner field of a spherically symmetric wormhole and the concentric double einstein rings generated by a toroidal ringhole could not be used without some uncertainty to identify the presence of such tunnelings in the universe or the existence of a parallel universe , the image which the inner gravitational field of a non orientable klein - bottle hole tunneling would leave by lensing a single luminous source is that of a truncated double spiral , which is a signature that can not be attributed to any other single or composite astronomical object in whichever universe it may be placed . in this report we argue some more reasons to predict that such a signature would imply the discovery of one such non orientable tunneling in our or other universe . after all , a nonorientable klein - bottle hole is also a perfectly valid solution to the einstein equations and the stuff which would make it feasible is becoming more and more familiar in cosmology .

electron spins confined in carbon nanotube@xcite ( cnt ) quantum dots@xcite ( qd ) are considered attractive for quantum information storage and processing due to the absence of the hyperfine interaction@xcite which is the main source of decoherence in iii - v nanostructures . the spin - orbit ( so ) coupling that is intrinsically present in cnts due to s - p hybridization accompanying the curvature of the graphene plane @xcite paves the way for electrical control of the confined carrier spins . in particular the so interaction allows for spin flips induced by ac electric fields @xcite according to the mechanism of the electric - dipole spin resonance as studied earlier for iii - v quantum dots.@xcite in nanotube quantum dots the so coupling splits the four - fold degeneracy of energy levels with respect to the spin and valley into kramers doublets with spin - orbit coupling energy varying from a fraction of mev @xcite to several mev . @xcite in this work we study the states confined in a qd defined electrostatically within the cnt and simulate spin and valley transitions driven by ac electric field between the quadruple of nearly degenerate energy levels in external magnetic field . for clean cnts the coupling between the @xmath0 and @xmath1 valleys is absent which motivates ideas to use the valley degree of freedom as a carrier of the quantum information alternative for the electron spin . in the transport experiments the valley filters and valves were proposed @xcite for clean samples in which the inter - valley scattering can be neglected . for clean cnt double quantum dots the phenomenon of valley blockade has been demonstrated in experiment @xcite and studied theoretically @xcite as the equivalent of the pauli spin blockade . @xcite a theory for rabi inter - valley resonance for cnt has also been presented @xcite within a continuum approximation of the tight - binding hamiltonian . in this work we report on time - dependent tight - binding simulations for the spin - valley transitions driven by ac field . in the present model the electron confinement within the dot , the lattice disorder , and the spin - valley dynamics are monitored at the atomic scale . we work with a direct solution of the time dependent schrdinger equation which allows us to resolve not only the rabi oscillations corresponding to the first order transition but also the fractional resonances in higher - order transitions observed in edsr experiments on iii - v @xcite as well as cnt @xcite qds . we discuss the effects driving the spin - flips with a particular focus on the electric field component that is perpendicular to the axis of the cnt , and which is bound to appear in experimental setups with cnts deposited or suspended above the gates.@xcite we show that a very similar dynamics of transitions is obtained for a bent cnt . the bend of the nanotube for electric dipole spin resonance in nanotubes was previously proposed@xcite but in the context of the electron motion along the bend in the external magnetic field and the resulting variation of the effective zeeman splitting . in the present system the motion of the electron is limited to the qd area and has a secondary effect on the transitions , still the bend of the nanotube in external _ electric _ field lowers the symmetry of the eigenstates which allows for the spin flips . we discuss the consequences of the perpendicular electric field , disorder and the bend of the cnt for selection rules and transition times . [ cols= " < , < " , ] in summary , we presented simulations of the spin flip and inter - valley transitions in a quantum dot defined within a semiconducting carbon nanotube . we considered a single excess electron in the quantum dot and evaluated the dynamics of the spin and valley transitions driven by external ac electric field . time - dependent calculations used the basis of localized eigenstates as determined by the tight - binding approach . for a straight and clean cnt the spin - flips are forbidden even for strong so coupling . the spin transitions are triggered by electric field perpendicular to the axis of the cnt . we demonstrated that the spin - flip transition times are inversely proportional to the value of the perpendicular electric field component . we demonstrated that the bend of the cnt in external electric field allows for the spin - flips due to lifting of the selection rules by lowering the angular symmetry of the eigenstates with the spin - flip transition times scaling linearly with @xmath2 . we demonstrated that when so coupling is present the atomic disorder alone allows for all types of transitions including spin flips . we discussed the disorder introduced by a vacancy which even when far from the qd perturbs the angular symmetry of the eigenstates lifting the selection rules prohibiting the inter - valley transitions . the inter - valley transitions when allowed by the lattice disorder appear roughly 100 to 1000 times faster than the spin flips and are insensitive to the electric fields perpendicular to the axis of the cnt . this work was supported by national science centre according to decision dec-2013/11/b / st3/03837 , by pl - grid infrastructure and by ministry of science and higher education within statutory tasks of the faculty . calculations were performed in ack cyfronet agh on the rackserver zeus . j. c. charlier , x. blase , and s. roche , rev . phys 79 , 677 ( 2007 ) . s. sapmaz , p. jarillo - herrero , l. p. kouwenhoven and h. s. j. van der zant , semicond . technol . * 21 * s52 ( 2006 ) . h. o. h. churchill , f. kuemmeth , j. w. harlow , a. j. bestwick , e. i. rashba , k. flensberg , c. h. stwertka , t. taychatanapat , s. k. watson , and c. m. marcus , phys . lett . * 102 * , 166802 ( 2009 ) . t. ando , j. phys . . jpn . * 69 * , 1757 ( 2000 ) . d. huertas - hernando , f. guinea , and a. brataas , phys . b * 74 * , 155426 ( 2006 ) . d. v. bulaev , b. trauzettel , and d. loss , phys . rev . b * 77 * , 235301 ( 2008 ) . m. del valle , m. margaska , and m. grifoni , phys . b * 84 * , 165427 ( 2011 ) . f. kuemmeth , s. ilani , d. c. ralph , and p. l. mceuen , nature * 452 * , 448 ( 2008 ) . j. klinovaja , m. j. schmidt , b. braunecker , and d loss , phys . b * 84 * , 085452 ( 2011 ) . e. a.laird , f. pei , and l. p. kouwenhoven , nature nano . * 8 * , 565 ( 2013 ) . k. c. nowack , f. h. l. koppens , yu . v. nazarov , and l. m. k. vandersypen , science * 318 * , 1430 ( 2007 ) . j. r. petta , a. c. johnson , j. m. taylor , e. a. laird , a. yacoby , m. d. lukin , c. m. marcus , m. p. hanson , and a. c. gossard , science * 309 * , 2180 ( 2005 ) . f. h. l. koppens , c. buizert , k. j. tielrooij , i. t. vink , k. c. nowack , t. meunier , l. p. kouwenhoven , and l. m. k. vandersypen , nature ( london ) * 442 * , 766 ( 2006 ) . s. nadj - perge , s. m. frolov , e. p. a. m. bakkers , and l. p. kouwenhoven , nature * 468 * , 1084 ( 2010 ) . s. nadj - perge , v. s. pribiag , j. w. g. van den berg , k. zuo , s. r. plissard , e. p. a. m. bakkers , s. m. frolov , and l. p. kouwenhoven , phys . * 108 * 166801 ( 2012 ) . t. s. jespersen , k. grove - rasmussen , j. paaske , k. muraki , t. fujisawa , j. nygrd , and k. flensberg , nature physics * 7 * , 348 ( 2011 ) . g. a. steele , f. pei , e. a. laird , j. m. jol , h. b. meerwaldt , and l. p. kouwenhoven , nature communications * 4 * , 1573 , ( 2013 ) . a. rycerz , j. tworzydo , c.w.j . beenakker , nature physics * 3 * , 172 ( 2007 ) . f. pei , e. a. laird , g. a. steele , and l. p. kouwenhoven , nature nano . * 7 * , 630 ( 2012 ) . a. plyi and g. burkard , phys . b * 82 * , 155424 ( 2010 ) . m. r. buitelaar , j. fransson , a. l. cantone , c. g. smith , d. anderson , g. a. c. jones , a. ardavan , a. n. khlobystov , a. a. r. watt , k. porfyrakis , and g. a. d. briggs , phys . b * 77 * , 245439 ( 2008 ) . a. plyi and g. burkard , phys lett . * 106 * , 086801 ( 2011 ) . j. stehlik , m. d. schroer , m. z. maialle , m. h. degani , j. r. petta , phys . lett . * 112 * , 227601 ( 2014 ) ; e. i. rashba , phys . b * 84 * , 241305 ( 2011 ) ; m. p. nowak , b. szafran , and f. m. peeters , phys . rev . b * 86 * , 125428 ( 2012 ) ; e. n. osika , b. szafran , and m. p. nowak , phys . rev . b * 88 * , 165302 ( 2013 ) ; g. szchenyi , and a. plyi , phys . rev . b * 89 * , 115409 ( 2014 ) . k. flensberg and c.m . marcus , phys . b * 81 * , 195418 ( 2010 ) . d. h. cobden , and j. nygrd , phys . * 89 * , 046803 ( 2002 ) . w. liang , m. bockrath , h. park , phys . lett . * 88 * , 126801 ( 2002 ) . p. jarillo - herrero , j. kong , h. s. j. van der zant , c. dekker , l. p. kouwenhoven , and s. de franceschi , phys . lett . * 94 * , 156802 ( 2005 ) . s. moriyama , t. fuse , m. suzuki , y. aoyagi , and k. ishibashi , phys . 94 * , 186806 ( 2005 ) .

we describe dynamics of spin and valley transitions driven by alternating electric fields in quantum dots defined electrostatically within semiconducting carbon nanotubes ( cnt ) . we use the tight - binding approach to describe the states localized within a quantum dot taking into account the circumferential spin - orbit interaction due to the s - p hybridization and external fields . the basis of eigenstates localized in the quantum dot is used in the solution of the time - dependent schrdinger equation for description of spin flips and inter - valley transitions that are driven by periodic perturbation in the presence of coupling between the spin , valley and orbital degrees of freedom . besides the first order transitions we find also fractional resonances . we discuss the transition rates with selection rules that are lifted by atomic disorder and the bend of the tube . we demonstrate that the electric field component perpendicular to the axis of the cnt activates spin transitions which are otherwise absent and that the resonant spin - flip time scales with the inverse of the electric field .

physics of ultracold atoms is fascinating in many respects . the ways how these systems can be manipulated in magnetic and optical traps open possibilities to an enormous number of physical effects which can be realized and studied in these systems @xcite , hofstetter . thanks to a large variety of experimental techniques one can manipulate atomic gases by magnetic and electromagnetic fields giving rise to realization of various macroscopic quantum effects . first of all , a number of effects known in standard superfluids and superconductors were already realized in ultracold atomic gases @xcite , hofstetter . in that sense a number of already developed theoretical methods and ideas in solid state physics were used in physics of ultracold atoms . on the other hand this fascinating field gives us new possibilities in studying many aspects of the bose - einstein condensation ( bec ) and of the superfluidity of fermi condensates ( bcs ) which are difficult to realize in solid state physics . for instance in an ultracold gas of bosonic atoms , such as @xmath12 or @xmath13 , one can study bose - einstein condensation(bec ) . if instead we deal with a gas of _ fermionic atoms _ , such as @xmath14 and @xmath15 , one can study not only fermionic superfluidity ( bcs ) but also the transition bec - bcs by directly controlling the interaction ( scattering length ) between atoms @xcite . then by tuning magnetic field in magnetic traps and/or electromagnetic field in optical lattices one can vary the atomic scattering length @xmath16 in a broad range ( especially for energies just near the feshbach resonance ) , that even bcs - like pairing of ultracold atoms can be realized with @xmath17 . it seems that this possibility is already realized with ultracold fermionic alkali gases @xmath18 and @xmath15 @xcite , @xcite . the cooling of a magnetically trapped spin - polarized @xmath15 fermi gas up to @xmath19 atoms at @xmath20 @xmath21 and up to @xmath22 atoms at @xmath23 @xmath21 is already realized . the strength of the atomic interaction was controlled by applying a magnetic field and tuning to feshbach resonances , which occur when the total energy of interacting particles in open channels is the same as the energy of a bound molecular state in closed channels . for such a two - level fermionic system it was proposed by the present author and his collaborator in the paper `` inhomogeneous superconducting phase in absence of paramagnetic effect '' a bcs - like model for the _ `` isotopic spin '' pairing _ of two ( or more ) species @xcite - the @xmath24 model . the basic ingredient of the @xmath24 model is that two species ( atoms , electrons , quarks , nucleons , etc . ) may have either different kinetic energies , or different energy levels ( or chemical potentials ) , i.e there is a mismatch of energy levels of fermionic atoms which participate in pairing . it turns out that this model is an adequate theoretical framework for a number of experimental situations which deal with ultracold fermionic alkali gases , such as for instance @xmath14 and @xmath15 . the present experimental techniques allow realizations of systems with two ( or more ) hyperfine levels . they can be furthermore manipulated in order to maintain various kinds of atomic pairing from bcs - like to strong coupling case as well as to realize the mismatch between the paired `` isotopic '' atomic levels ( chemical potential ) . for further application of this model see below . in the following we apply this model in studying nonuniform superfluidity in ultracold fermionic atomic gases ( @xmath0 ) with oscillating order parameter @xmath25 - which is due to the mismatch effect of two hyperfine states of fermionic atoms . this effect is analogous to the larkin - ovchinikov - fulde - ferrell ( @xmath26 or @xmath27 ) phase in superconductors @xcite , @xcite . based on this effect we propose a novel josephson junction ( @xmath28 ) where the left and right superfluid ( @xmath29 ) gases have uniform order parameters @xmath30 , while the weak link ( @xmath4 ) with the mismatch effect is in the normal state . this gives rise to an oscillating superfluid amplitude inside the weak link @xmath4 and as a result the so - called @xmath2-junction can be realized . we show that if such a junction is a part of the closed ring then _ spontaneous and dissipationless superfluid current _ can flow through the ring depending on the size of the ring . in that case there is _ spontaneous breaking of the time - reversal _ in the system . finally , we discuss possible realizations and generalizations of this novel effect in ultracold fermionic gases . the model with the `` isotope - spin '' pairing ( @xmath24 ) @xcite is a generalization of the bcs pairing mechanism to systems with internal degrees of freedom such as for instance , _ nuclear matter _ - with isospin numbers , _ quark matter _ - with the color and flavor quantum numbers , _ ultracold fermionic gases _ - with quantum numbers of the hyperfine atomic states , _ layered _ and _ multiband superconductors _ - with quantum numbers enumerating layers and bands . the pairing constituents ( electrons , nucleons , quarks , neutral atoms , etc . ) can be either charged or not and in those cases we deal either with real superconductivity or with superfluidity of matter . all these systems posses a natural mismatch in energies ( or masses ) of constituents participating in pairing . this is an important property since in the case when the mismatch parameter @xmath1 is of the order of the bare superconducting gap @xmath31 a nonuniform superfluidity ( superconductivity ) is realized - an analogue of the @xmath26 state in metallic superconductors placed in the zeeman field . the hamiltonian of the @xmath24 model is given by @xcite @xmath32 @xmath33 @xmath34 the @xmath24 order parameter @xmath35 @xmath36 in the following we shall study the bcs - like pairing and josephson effect in ultracold fermionic alkaline gases ( @xmath0 ) in which case @xmath37 and @xmath38 for @xmath39 only - see the explanation below , while the electronic spin states @xmath40 and @xmath41 do not enter ( they are accounted for via @xmath42 ) . for simplicity we omit indices @xmath42 in @xmath43 , i.e. @xmath44 . in the case of two ( @xmath45 ) `` isotopic '' bands with the mismatch in their energies ( or chemical potentials ) - the _ `` isotopic '' band splitting _ ( @xmath1 ) - the quasiparticle spectrum is given by @xmath46note , that the splitting parameter @xmath1 can in principle depend on momenta too , but the results obtained below are qualitatively valid also in this case . it is interesting to mention that the above model and some physical results @xcite were frequently used later on in studying diverse physical problems such as : ( @xmath47 ) _ ultracold _ _ fermionic atomic gases _ - where @xmath48 enumerate hyperfine atomic states of alkali atoms ; ( @xmath49 ) in _ quantum chromodynamics _ - where the problem of pairing in the quark matter , i.e. , the _ color superconductivity _ ( @xmath50 ) , was studied @xcite . in the latter case the @xmath50 pairing of the colored quarks is mathematically more complicated than the standard bcs case , since the former system is characterized by the matrix coupling constant @xmath51 in the internal degrees of freedom , where @xmath52 are quantum numbers of quarks : @xmath53 - _ color _ ; @xmath54 - _ flavor _ ( up , down , strange , ... ) ; @xmath55 - _ spin_. at the same time the hamiltonian @xmath56 ( for instance the nambu - lasinio hamiltonian ) , which is an analogue of the bcs one in the @xmath50 pairing , must be invariant under the symmetry group of strong interactions @xmath57 - see rischke . let us mention some application of this model in _ _ the solid state physics _ _ such as _ _ bi - layered superconductors _ _ in the presence of the zeeman effect ( @xmath58 ) @xmath59 . \label{zeeman}\]]it turns out that the paramagnetic field @xmath60 can compensate the `` isotopic '' mismatch of the chemical potentials ( @xmath61 ) giving rise to some novel effects @xcite . for instance , in case when @xmath62 superconductivity is destroyed by the isotopic - splitting @xmath1 , while the zeeman term can compensate this mismatch and thus inducing superconductivity for @xmath63 - the _ reentrant superconductivity recently , similar effects were found to exist in layered superconductors placed in the zeeman field @xmath60 buzdin - layered . in that case the bi - layer superconductivity was studied in the model ( similar to that in eq.([h0]-[bcs ] ) ) with the _ intralayer _ pairing @xmath9 only . in that case the interplane hoping ( @xmath64 ) in the bi - layer system plays the role of the `` isotopic '' splitting parameter , i.e. @xmath65 , and @xmath66 are the bonding ( @xmath67 ) and antibonding ( @xmath68 ) bands , respectively . it turns out that for @xmath69 the bi - layer superconductivity is realized , i.e , @xmath70 , while for @xmath71 it is destroyed , i.e. , @xmath72 . however , for @xmath73 superconductivity appears again ( reentrant superconductivity ) on the expense of the @xmath2-phase difference between the order parameters on the bilayer ( in the up- and down - plane ) , i.e. @xmath74 . in the language of the @xmath24 model there is a pairing between the antibonding and bonding levels only , i.e. , @xmath75 while @xmath76 . this result is of a potential interest in high - temperature superconductors , especially in bi - layer bisco . one expects that similar effects can be realized also in the _ quark matter _ with the color superconductivity ( @xmath50 ) with two flavor quarks , such as pulsars - magnetars , where the @xmath50 gap is @xmath77 @xmath78 which is of the order of mass difference @xmath79 @xmath78 between up and down quark . having in mind that the magnetic field in magnetars is huge , @xmath80 , then the above compensating effect ( by magnetic field and zeeman effect ) might be operative there . the @xmath24 model renders a number of interesting effects and some of them were discussed in @xcite . in the following we study systems with two levels @xmath45 only and apply it to the ultracold fermionic gases . it is assumed pairing between two species @xmath81 for @xmath39 , only . this two - state model is realistic approximation since in diluted gases , for instance in @xmath15 , the interaction of atoms in different hyperfine states is several order of magnitude larger than in the same states @xcite . recently , the fermionic superfluidity was realized recently by several groups @xcite . however , the definitive experimental prove of the fermionic superfluidity was given recently in the remarkable experiment of the ketterle s group @xcite . they cooled @xmath15 in magnetic trap below the fermi degeneracy . this fermi cloud , consisting of approximately@xmath82 atoms with two lowest hyperfine states @xmath83 and @xmath84 , was loaded into an optical dipole trap . between these states there is a feshbach resonance at @xmath85 @xmath86 . the bec - bcs transition occurs in the region between 780 g and 925 g , where for @xmath87 @xmath88 the system is in the bcs side . since the ( attractive ) coupling between atoms in states @xmath83 and @xmath84 at low t is strong in the @xmath89-channel , this means that the spatial part of the two - body wave function @xmath90 is symmetric under the exchange of atoms , i.e. @xmath91 . as the result the spin part of the wave function is antisymmetric , i.e. , @xmath92 . \label{spin}\]]the pairing is strong between the states @xmath83 and @xmath84 , i.e. , @xmath93 while @xmath94 . the crucial observation in @xcite was the creation of vortices in the rotating systems above some rotation frequency , what can be considered as a definite prove for fermionic superfluidity . since the @xmath24 model strongly reminds on metallic superconductors with the zeeman ( paramagnetic ) effect , this analogy might be useful in studying ultracold fermionic gases . _ first _ , in the case when the `` isotope '' splitting is larger than the critical value @xmath95 , then in the region @xmath96 ( at @xmath97 @xmath98 ) a _ nonmagnetic analogue _ of the @xmath26 phase is realized with the nonuniform order parameter @xmath99 . for @xmath100 the modulation vector @xmath101 is given by @xmath102 where @xmath103 is the fermi velocity . it is known that the structure of the order parameter may contain more wave vectors @xmath104 thus making the so called crystalline structure more favorable @xcite - this ( still unsolved ) problem is not the subject of this paper . concerning the critical value @xmath105 it depends on dimensionality of the system - in 3d one has @xmath106 , in 2d @xmath107 while for 1d @xmath108 . note , that for @xmath109 ( at @xmath97 k ) the system is in uniform superfluid ( superconducting ) state , i.e. @xmath110 . the transition at @xmath105 depends on the type of the nonuniform solution - for instance for the one plane solution we use here ( only for simplicity ) the transition at ( @xmath97 , @xmath111 ) is second order . the phase diagram in the plane ( @xmath112 , @xmath1 ) of the `` isotopic '' @xmath26 phase is shown in @xmath113 . second , it turns out that the @xmath26 state appears at temperatures @xmath114 and in the phase diagram there is a tricritical lifshitz point ( @xmath115 , @xmath116 ) . near this point one can perform the ginzburg - landau ( @xmath117 ) expansion analogously to that obtained in @xcite , @xcite . in order to prove the existence of the @xmath2- junction in the superfluid @xmath0 we shall study the nonuniform superfluidity near the point ( @xmath118 , @xmath119 ) where it is straightforward to derive the @xmath117 expansion by using eqs.([bcs]-[gap ] ) . as the result the ( dimensionless ) density of the free - energy functional @xmath120 ( where @xmath121 is the condensation energy and @xmath122 ) is given by buku - glexpansion , @xcite @xmath123@xmath124here , @xmath125/t^{\ast } $ ] , @xmath126 , @xmath127 , where @xmath128 is the transition temperature from the normal ( @xmath129 ) to the uniform bcs state but in the absence of the @xmath26 state . @xmath130 and @xmath131 are the density of states and superconducting ( superfluid ) coherence length ( at @xmath97 ) , respectively . the parameters @xmath132 , @xmath133 , @xmath134 , @xmath135 , @xmath136 and @xmath137 are of the order of one and their precise values can be found in @xcite . since the term in front of @xmath138is negative it leads to the nonuniform @xmath26 state near the tricritical point ( @xmath118 ; @xmath139 ) . the minimization of @xmath140 gives the transition temperature @xmath141 from the normal to the @xmath142f state ( the dotted line in @xmath113 ) and the wave vector @xmath143 @xmath144@xmath145 in the @xmath26 state the mass - current density ( @xmath146 and @xmath147 ) has much reacher form then the standard g - l expression @xmath148}\psi ^{\ast } \nabla \psi\]]@xmath149 - 2\nabla \psi \nabla ^{2}\psi ^{\ast } \}+c.c . \label{j}\]]in the case of charged superconductors the orbital effect of magnetic field can be accounted for by replacing @xmath150 . in the case of @xmath0 , such as @xmath14 and @xmath15 , the current @xmath151 is the _ mass - current_. we shall demonstrate below that in principle it is possible to have a @xmath2-_josephson contact ( weak link ) _ based on cold fermionic gases which is analogous to the so called @xmath152 contacts ( weak links ) in the solid state physics , where in the ferromagnetic weak link ( @xmath153 ) electrons with different spin projections have different energies ( zeeman - mismatch ) . in this case the superconducting amplitude @xmath154 is induced in @xmath153 by the proximity effect . this amplitude not only decays ( like in the case of normal metal @xmath155 contacts ) but also oscillates between two superconducting banks , i.e. @xmath156 - see @xcite , @xcite . the oscillation in the weak link , especially in the case @xmath8 , is in fact a reminiscence of the @xmath26 state as it was shown in @xcite by using eqs . ( [ f]-[j ] ) . the josephson weak - link with @xmath0 can be in principle realized by using various optical and magnetic trap techniques . if the @xmath0 with only two hyperfine levels are trapped their pairing interaction can be described in some parameter range by eqs.([bcs]-[gap ] ) . we further assume that the left and right part of the weak - link are similar and that the mismatch of the chemical potentials in them is small , @xmath157 . in that case the left ( @xmath158 ) and right ( @xmath159 ) bcssuperfluid - condensates are uniform with the critical temperature @xmath160 and the order parameters @xmath161 and @xmath162 , respectively . furthermore we assume that the weak - link @xmath4 ( @xmath163 ) with the width @xmath164 between the left and right banks is made of @xmath0 with the mismatch @xmath61 but which is in the normal state . in the following we call such a weak - link @xmath28 , where @xmath29 means the bcs superfluid and @xmath4 is the weak link with the mismatch parameter @xmath1 and at temperatures where the system is a _ normal fermi gas_. this means that in @xmath4 the order parameter @xmath165 but the anomalous green s function @xmath166 due to the proximity effect @xcite . at the end we shall argue that the josephson effect can exist even in the case @xmath8 , i.e. far away of the @xmath26 state . however , in this case one should use rather sophisticated microscopic many - body techniques - for instance the eilenberger quasiclassical equations . in order to prove the existence of the @xmath2-josephson junction we shall study the problem when @xmath112 and @xmath1 are just near the tricritical point ( @xmath167 ; @xmath168 ) - see @xmath113 . in such a case the weak link @xmath4 , which is in the normal state but very near to the loff state , is characterized by the induced order parameter @xmath169 which is small near the point ( @xmath170 ; @xmath139 ) and the @xmath117 equation can be linearized . since @xmath171 changes along the @xmath172-axis which is perpendicular to the surface of the weak link ( @xmath4 ) - see @xmath173 , the @xmath117 equation in the weak link @xmath4 reads @xmath174 in order to simplify the analysis we assume that @xmath175 and the solution @xmath176 oscillates with @xmath177 . the general solution in the weak link is@xmath178where @xmath179 and @xmath180 can be obtained from the boundary conditions at @xmath181 and @xmath182 . since at present there is no microscopic theory for the josephson effect in @xmath0 we use the experience from the physics of the @xmath183 weak links . there various boundary conditions do not destroy the oscillations of @xmath184 in the weak link , which is important property for the realization of the @xmath2-contact . therefore , we assume that @xmath171 and @xmath185 are continuous on boundaries at @xmath186 and @xmath187 . since we study the problem near the tricritcal point ( @xmath118 , @xmath119 ) then the standard term ( @xmath188 ) in the current is small and the linearized expression for the current @xmath189 is given by@xmath190 . \label{jx}\]]due to the particle current conservation the josephson current trough the weak link @xmath4 can be calculated at the midpoint @xmath191 , i.e. , @xmath192 is given by@xmath193where@xmath194@xmath195here , @xmath196 is the phase difference on th junction . from eq . ( [ jc ] ) one concludes that @xmath197 can reach negative values whenever @xmath198 . this can be achieved by changing either @xmath112 , @xmath1 or the thickness @xmath187 of the weak link @xmath4 . in such a way one can realize a @xmath2__-junction _ _ - the contact with @xmath199 . to remaind the reader , the @xmath200-contact is characterized by the phase difference @xmath201 in the ground state , while in the standard josephson contacts @xmath202 minimizes the energy of the contact . in the next section we are going to show , that if the @xmath2-junction is placed in a ring with the large radius @xmath203 there is a _ spontaneous superfluid flow _ through the ring , i.e. the _ _ time - reversal symmetry is broken spontaneously__. let us consider a loop made of an ultracold fermionic bcs superfluid with the @xmath2-contact placed in a ring of radius @xmath6 as shown in @xmath205 . if the mass - current flows through the josephson junction and the ring the total energy ( per cross - section @xmath29 of the ring ) of the system @xmath206 is due to the kinetic energy @xmath207 of the _ circulating superfluid current _ and the energy @xmath208 of the josephson contact . in the ring with a small cross - section ( @xmath209 ) the superfluid velocity @xmath210 is practically constant over the cross - section of the ring ( @xmath211 ) and the total energy is given by the simple expression @xmath212here , @xmath213 is the density of superfluid pairs , @xmath214 is the mass of the fermionic atom , and @xmath215 is the phase difference on the contact . the intrinsic phase of the junction is @xmath216 for @xmath217 and @xmath218 for @xmath219 . the relation between the superfluid velocity @xmath220 and the phase @xmath221 can be obtained from the _ quantization condition _ of the superfluid phase @xmath222 along the closed path @xmath223 in the ring - see @xmath205@xmath224the path @xmath225 goes from @xmath187 to @xmath6 - see @xmath205 . as the result one obtains the relation between @xmath226 and @xmath222 @xmath227where @xmath228 . the dimensionless energy of the junction ( contact ) @xmath229 is given by@xmath230the minimization of @xmath231 with respect to @xmath226 gives the condition on the ring radius @xmath6 for the appearance of a spontaneous superflow with @xmath232 @xmath233here , @xmath234 is an effective velocity trough the weak link . in that case the @xmath2-contact ( @xmath218 ) is realized , i.e. there is a _ spontaneous superfluid flow in the ring_. this also means that there is a _ spontaneous breaking of thee time - reversal symmetry _ in the system . from eq.([rc ] ) one concludes that for small critical currents ( velocity @xmath235 ) of the weak link the @xmath2-contact is realized for large @xmath6 . due to the lack of precise experimental data for parameters @xmath103 , @xmath235 , @xmath9 , @xmath236 one can make some qualitative guesses of @xmath237 only . for instance in @xmath0 with @xmath14 , by assuming that @xmath238 @xmath239 , @xmath240 , and that @xmath241 with @xmath242 - the critical depairing velocity , one gets @xmath243 @xmath244 . since @xmath237 scales inversely with the atomic mass it is larger for lower atomic mass . in conclusion , we have shown that in ultracold fermionic atomic gases ( @xmath245 ) with the mismatch in energies ( chemical potentials ) of pairing atoms one can realize @xmath2-josephson junction ( contact ) in @xmath28 weak links . in that case the weak link @xmath4 is made from @xmath0 with the finite mismatch in hyperfine energies ( chemical potentials ) of the pairing atoms , @xmath246 . our proof was given for the case when the energy mismatch @xmath247 is of the order of the fermionic superfluid gap @xmath9 , @xmath248 , in which case @xmath4 is in the normal state but near to the nonuniform @xmath26 state . however , by analogy with the physics of @xmath152 junctions one expects that the @xmath2-josephson junction can be realized also for very large mismatch , i.e. , @xmath8 . in this case the decay - length of the superfluid pairing amplitude induced in the weak link @xmath4 is very short , i.e. @xmath249 . in such a case the problem must be attacked by more sophisticated theoretical methods , for instance by the eilenberger quasiclassical equations in case when @xmath250 . concerning a possible realization of the @xmath2-contact in @xmath0 it seems that traps by optical lattices are more promising way than inhomogeneous magnetic traps . the possibility for the realization of @xmath2-junctions in @xmath5 opens a room for numerous speculations about potential applications . for instance , various contacts of the type @xmath10 as well as trilayer systems such as @xmath251would allow interesting switching properties of these systems . these systems , if they are realizable , would offer a possibility for a new kind of electronics - _ hypertronics_. for instance , the computer memory might be realized in weak links @xmath252 , while the logic processing can be done by manipulating the superfluid state . these are open but very attractive problems in this very promising field . one should admire that the realization and manipulation of josephson junctions in @xmath0 represents a considerable experimental challenge . * acknowledgement*. i express my deep gratitude to i. boovi , a. i. buzdin , p. fulde , w. hofstetter , i. kuli , d. pavuna and d. rischke for support . i am thankful to cristian schunk for critical comments concerning the experimental confirmation of the fermionic superfluidity in @xmath253 . regal , m. greiner , d. s. jin , phys . lett . * 92 * , 040403 ( 2004 ) ; m. w. zwierlein et al . lett . * 92 * , 120403 ( 2004 ) ; j. kinast et al . , lett . * 92 * , 150402 ( 2004 ) ; m. bartenstein et al . * 92 * , 203201 ( 2004 ) ; c. chin et al . , science * 305 * , 1128 ( 2004 ) ; j. kinast , science * 307 * , 1296 ( 2005 )

the bcs - like pairing in ultracold fermionic atomic ( @xmath0 ) gases is studied in the model of `` isotopic - spin '' pairing proposed in 1991 ku - hof - ssc . this model assumes a mismatch ( @xmath1 ) in chemical potentials of pairing fermionic atoms . it is shown that a @xmath2-josephson junction can be realized in @xmath0 systems , where the left and right banks @xmath3 are the @xmath0 superfluids . the weak link @xmath4 consists from the normal @xmath5 with the finite mismatch @xmath1 . if the @xmath2-junction is a part of a closed ring the superfluid mass - current flows spontaneously in the ring , i.e. , the time - reversal symmetry is broken spontaneously . this is realized if the radius of the ring @xmath6 is larger than the critical one @xmath7 . all these effects exist also in the case when @xmath8 , where @xmath9 is the superfluid gap , but with the reduced thickness of the weak link . it is also discussed , that if junctions @xmath10 and trilayers @xmath11 from @xmath0 are realizable this renders a possibility for a novel electronics - _ hypertronics_.

in this study we suggest a new family of spherical mass distribution models that generalizes models by an & evans ( 2006 , hereafter ae ) and models by kuzmin et al . @xcite . the family depends on two structural parameters . it includes plummer s spheres @xcite , hnon s isochrones @xcite and the model by hernquist @xcite as special cases . let us consider the dimensionless potential @xmath0 here @xmath1 and @xmath2 are structural parameters . if @xmath3 , @xmath4 we obtain a model by ae , for @xmath5 , @xmath6 we have a model by kuzmin et al . ( 1969 , 1972 ) . poisson s equation yields the following expression for density @xmath7.\ ] ] it follows from ( [ rho ] ) that our models with @xmath8 are cusped ( as models by ae are ) . the density profiles for different values of parameters are shown in figures [ raspopova - fig1 ] , [ raspopova - fig2 ] . the circular speed is found to be @xmath9 a run of velocity dispersion @xmath10 can be found from an equation of hydrostatic equilibrium @xmath11 the results of calculations for an isotropic velocity distribution ( @xmath12 ) are shown in figure [ raspopova - fig3 ] . central minima will appear in all models with density cusp . it can be obtained from ( [ phi ] ) that @xmath13^{1/p}}{\alpha\phi}.\ ] ] then it is possible to find an augmented density @xmath14 and calculate an isotropic distribution function . stability of such models can be studied using the third antonov law @xcite , namely , if @xmath15 the model is stable against spherical perturbation . the validity of this inequality can be established after some laborious calculations . using the equipotential method by @xcite one can construct axisymmetric generalizations of the suggested model . we considered a potential of such models @xmath16 where @xmath17 is the same function as and @xmath18 is an equation of equipotential surfaces , @xmath19 , @xmath20 being cylindrical coordinates . we considered the equipotentials by @xcite : @xmath21 and by @xcite : @xmath22 here @xmath23 $ ] is a new structure parameter . for spherical systems @xmath24 . we found that for @xmath25 close to @xmath26 the density is positive for @xmath19 , @xmath20 everywhere . so we concluded that such non - spherical model can be used for approximating mass distribution in non - spherical star clusters and non - highly flattened galaxies . an , j. h. , evans , n. w. 2006 , , 131 , 782 binney , j. , tremaine , s. 1987 , galactic dynamics , princeton univ . press , princeton hnon , m. 1959 , ann . dastrophys . , 22 , 126 hernquist , l. 1990 , , 356 , 359 kutuzov , s. a. , ossipkov , l.p . 1981 , , 24 , 17 kuzmin , g.g . , malasidze , g.a . , 1969 , w. struve tartu astrofs . , 38 , 181 ( in russian ) . kuzmin , g. g. , veltmann , .- i . k. 1972 , publ . w. struve tartu astrofs . 40 , 281 ( in russian ) miyamoto , m. , nagai , r. 1975 , publ . japan . , 27 , 533 plummer , h. c. , 1911 , , 71 , 460 satoh , c. 1980 , publ . japan . , 32 , 41

a new two - parametric family of mass distribution for spherical stellar systems is considered . it generalizes families by @xcite and by @xcite . steady velocity dispersions are found for these models by solving an equation of hydrostatic equilibrium . axisymmetric generalizations of the model are discussed .

the contact pattern among individuals in a population is an essential factor for the spread of infectious diseases . in deterministic models , the transmission is usually modelled using a contact rate function , which depends on the contact pattern among individuals and also on the probability of disease transmission . the contact function among individuals with different ages , for instance , may be modelled using a contact matrix @xcite or a continuous function @xcite . however , using network analysis methods , we can investigate more precisely the contact structure among individuals and analyze the effects of this structure on the spread of a disease . the degree distribution @xmath1 is the fraction of vertices in the network with degree @xmath2 . scale - free networks show a power - law degree distribution @xmath3 where @xmath4 is a scaling parameter . many real world networks @xcite are scale - free . in particular , a power - law distribution of the number of sexual partners for females and males was observed in a network of human sexual contacts @xcite . this finding is consistent with the preferential - attachment mechanism ( ` the rich get richer ' ) in sexual - contact networks and , as mentioned by liljeros et al . @xcite , may have epidemiological implications , because epidemics propagate faster in scale - free networks than in single - scale networks . epidemic models such as the susceptible infected ( si ) and susceptible infected susceptible ( sis ) models have been used , for instance , to model the transmission dynamics of sexually transmitted diseases @xcite and vector - borne diseases @xcite , respectively . many studies have been developed about the dissemination of diseases in scale - free networks @xcite and in small - world and randomly mixing networks @xcite . scale - free networks present a high degree of heterogeneity , with many vertices with a low number of contacts and a few vertices with a high number of contacts . in networks of human contacts or animal movements , for example , this heterogeneity may influence the potential risk of spread of acute ( e.g. influenza infections in human and animal networks , or foot - and - mouth disease in animal populations ) and chronic ( e.g. tuberculosis ) diseases . thus , simulating the spread of diseases on these networks may provide insights on how to prevent and control them . in a previous publication @xcite , we found that networks with the same degree distribution may show very different structural properties . for example , networks generated by the barabsi - albert ( ba ) method @xcite are more centralized and efficient than the networks generated by other methods @xcite . in this work , we studied the impact of different structural properties on the dynamics of epidemics in scale - free networks , where each vertex of the network represents an individual or even a set of individuals ( for instance , human communities or animal herds ) . we developed routines to simulate the spread of acute ( short infectious period ) and chronic ( long infectious period ) infectious diseases to investigate the disease prevalence ( proportion of infected vertices ) levels and how fast these levels would be reached in networks with the same degree distribution but different topological structure , using si and sis epidemic models . this paper is organized as follows . in section [ sec : hypothetical ] , we describe the scale - free networks generated . in section [ sec : model ] , we show how the simulations were carried out . the results of the simulations are analyzed in section [ sec : results ] . finally , in section [ sec : conclusions ] , we discuss our findings . we generated scale - free networks following the barabsi - albert ( ba ) algorithm @xcite , using the function barabasi.game(@xmath5 , @xmath6 , directed ) from the r package igraph @xcite , varying the number of vertices ( @xmath5 = @xmath7 , @xmath8 and @xmath9 ) , the number of edges of each vertex ( @xmath6 = 1 , 2 and 3 ) and the parameter that defines if the network is directed or not ( directed = true or false ) . for each combination of @xmath5 and @xmath6 , 10 networks were generated . then , in order to guarantee that all the generated networks would follow the same degree distribution and that the differences on the topological structure would derive from the way the vertices on the networks were assembled , we used the degree distribution from ba networks as input , to build the other networks following the method a ( ma ) @xcite , method b ( mb ) @xcite , molloy - reed ( mr ) @xcite and kalisky @xcite algorithms , all of which were implemented and described in detail in ref . as mentioned above , these different networks have distinct structural properties . in particular , the networks generated by mb are decentralized and with a larger number of components , a smaller giant component size , and a low efficiency when compared to the centralized and efficient ba networks that have all vertices in a single component . the other three models ( ma , mb and kalisky ) generate networks with intermediate characteristics between mb and ba models . the element @xmath10 of the adjacency matrix of the network , @xmath11 , is defined as @xmath12 if there is an edge between vertices @xmath13 and @xmath14 and as @xmath15 , otherwise . we also define the elements of the vector of infected vertices , @xmath16 . if vertex @xmath13 is infected , then @xmath17 , and , if it is not infected , @xmath18 . the result of the multiplication of the vector of infected vertices , @xmath16 , by the adjacency matrix , @xmath11 , is a vector , @xmath19 , whose element @xmath13 corresponds to the number of infected vertices that are connected to the vertex @xmath13 and may transmit the infection @xmath20 using matlab , the spread of the diseases with hypothetical parameters along the vertices of the network was simulated using the following algorithm : 1 . a proportion ( @xmath21 ) of the vertices is randomly chosen to begin the simulation infected . for our simulations , @xmath22 , since we are interested in the equilibrium state and this proportion guarantees that the disease would not disappear due to the lack of infected vertices at the beginning of the simulations . 2 . in the sis ( susceptible infected susceptible ) epidemic model , a susceptible vertex can get infected , returning , after the infectious period , to the susceptible state . for each time step : 1 . we calculate the probability ( @xmath23 ) of a susceptible vertex @xmath13 , that is connected to @xmath24 infected vertices , to get infected , using the following equation : @xmath25 where @xmath26 is the probability of disease spread . [ item2.b ] we determine which susceptible vertices were infected in this time step : if @xmath27uniform(0,1 ) @xmath28 , the susceptible vertex becomes infected . for each vertex infected , we generate the time ( @xmath29 ) that the vertex will be infected following a uniform distribution : @xmath30 uniform ( @xmath31 , @xmath32 ) , where @xmath31 and @xmath32 are , respectively , the minimum and the maximum time of the duration of the disease . 3 . decrease in 1 time step the duration of the disease on the vertices that were already infected , verifying if any of them returned to the susceptible state ; 4 . update the status of all vertices . for the si ( susceptible infected ) epidemic model , we chose @xmath29 in order to guarantee that an infected vertex remains infected until the end of the simulation . varying the values of the parameter @xmath29 , we simulated the behaviour of hypothetical acute and chronic diseases , using different values of @xmath26 , considering that once a vertex gets infected it would remain in this state during an average fixed time ( an approach that can be used when we lack more accurate information about the duration of the disease in a population ) or that there would be a variation in this period , representing more realistically the process of detection and treatment of individuals ( table [ tab : table1 ] shows the diseases simulated and the values of @xmath29 assumed ) . [ cols="^,^,^,^,^ " , ] we adopted a total time of simulation ( @xmath33 ) of 1000 arbitrary time steps . for each spreading model , we carried out 100 simulations for each network , calculating the prevalence of the simulated disease for each time step . then we calculated the average of the prevalence of these simulations on each network . after that we grouped the simulations by network algorithm . finally , we calculated the average prevalence of these network models . on the undirected networks , we observed that the disease spreads independently of the value of @xmath26 used , and that an increase in @xmath26 leads to an increase in the prevalence of the infection ( figure [ fig : figure1 ] ) . also , we observed that the prevalence tends to stabilize approximately in the same level despite the addition of vertices ( figure not shown ) . when we increase the number of edges of each vertex , there is an increase in the prevalence of the infection . a result that stands out is that , when @xmath34 , there is a great difference in the equilibrium level of the prevalence in each network . however , as we increase the value of @xmath6 , the networks tend to show closer values of equilibrium ( figure [ fig : figure3 ] ) . among the undirected networks , the networks generated using the mb algorithm presented the lowest values of prevalence in the spreading simulations ( figure [ fig : figure4 ] ) . on the directed networks , we observed that , despite the simulations of acute diseases , the disease spreads independently of the value of @xmath26 used and , as in the undirected networks , an increase in @xmath26 leads to an increase in the prevalence of the infection ( figure [ fig : figure5 ] ) . also , similarly to what was observed for undirected networks , the prevalence tends to stabilize approximately in the same level despite the addition of vertices ( figure not shown ) . when we increase the number of edges of each vertex , there is an increase in the prevalence of the infection ( figure [ fig : figure7 ] ) . a result that stands out is that , when @xmath35 , the prevalence in the kalisky networks tend to stabilize in a level a little bit higher than the other ones . among the networks , those generated using the ba algorithm presented the lowest values of prevalence in the spreading simulations ( figure [ fig : figure8 ] ) . to compare the numerical results of the simulation with a theoretical approach , it is possible to deduce , for an undirected scale - free network assembled following the ba algorithm , the equilibrium prevalence , given by @xcite @xmath36 . \label{eq:3}\ ] ] this expression applies to the ba undirected network with @xmath37 and a fixed infectious period of one time unit . for instance , for @xmath38 and @xmath39 , we obtain @xmath40 . in figure [ fig : figure9 ] , we observe that the equilibrium prevalence in the simulation reaches the value predicted by equation ( [ eq:3 ] ) . our approach , focusing on different networks with the same degree distribution , allows us to show how the topological features of a network may influence the dynamics on the network . analyzing the results of the spreading simulations , we have , as expected , that the variation in the number of vertices of the hypothetical networks had little influence in the prevalence of the diseases simulated , a result that is consistent with the characteristics of the scale - free complex networks as observed by pastor - satorras and vespignani @xcite . with respect to the effect of the increase of the probability of spreading on the prevalence in undirected networks ( figure [ fig : figure1 ] ) , we observed that the prevalence reaches a satured level . for undirected networks , if the probability of infection is high , there is a saturation of infection on the population for chronic diseases and therefore no new infections can occur . regarding the variation in the number of edges , the increase in the prevalence was also expected since it is known that the addition of edges increases the connectivity on the networks studied , allowing a disease to spread more easily . about the effect of considering the networks directed or undirected , we have that the diseases tend to stay in the undirected networks independently of the spreading model and the value of @xmath26 considered @xcite , while in the directed cases due to the limitations imposed by the direction of the movements , when @xmath34 , the acute diseases tend to disappear on some of the networks . also due to the direction of the links , in directed networks , the disease may not reach or may even disappear from parts of the network , explaining to some degree why the prevalence in directed networks ( figure [ fig : figure3 ] ) is smaller than in undirected networks ( figure [ fig : figure7 ] ) . in the si simulations , we could observe what would be the average maximum level of prevalence of a disease on a network and how fast this level would be reached . in the sis simulations , the oscillations on the stability of the prevalence observed result from the simultaneous recovery of a set of vertices . with a fixed time of infection , the set of vertices that simultaneously recover is greater than in the case of a variable time of infection , since in the latter , due to the variability of the disease duration , the vertices form smaller subsets that will recover in different moments of the simulation . a result that called attention is that , in the directed networks with @xmath34 , when we simulated the chronic diseases using a fixed time , the equilibrium levels achieved were lower than the ones achieved when we used a variable time . examining the results of the simulations on each network model , we have that among the undirected ones , the mb network has the lowest prevalence , with a plausible cause for this being how this network is composed , since there is a large number of vertices that are not connected to the most connected component of the network @xcite . among the directed networks , the ba network has the lowest prevalence observed , what is also probably due to the topology of this network , since it is composed of many vertices with outgoing links only and a few vertices with many incoming and few outgoing links , thus preventing the spread of a disease . using the methodology of networks , it is possible to analyze more clearly the effects that the heterogeneity in the connections between vertices have on the spread of infectious diseases , since we observed different prevalence levels in the networks generated with the same degree distribution but with different topological structures . moreover , considering that the increase in the number of edges led to an increase in the prevalence of the diseases on the networks , we have indications that the intensification of the interaction between vertices may promote the spread of diseases . so , as expected , in cases of sanitary emergency , the prevention of potentially infectious contacts may contribute to control a disease . this work was partially supported by fapesp and cnpq .

the transmission dynamics of some infectious diseases is related to the contact structure between individuals in a network . we used five algorithms to generate contact networks with different topological structure but with the same scale - free degree distribution . we simulated the spread of acute and chronic infectious diseases on these networks , using si ( susceptible infected ) and sis ( susceptible infected susceptible ) epidemic models . in the simulations , our objective was to observe the effects of the topological structure of the networks on the dynamics and prevalence of the simulated diseases . we found that the dynamics of spread of an infectious disease on different networks with the same degree distribution may be considerably different . * adv . studies theor . phys . , vol . 7 , 2013 , no . 16 , 759 - 771 * * hikari ltd , www.m-hikari.com * * http://dx.doi.org/10.12988/astp.2013.3674 * * raul ossada , jos h. h. grisi - filho , * * fernando ferreira and marcos amaku * faculdade de medicina veterinria e zootecnia universidade de so paulo so paulo , sp , 05508 - 270 , brazil copyright @xmath0 2013 raul ossada et al . this is an open access article distributed under the creative commons attribution license , which permits unrestricted use , distribution , and reproduction in any medium , provided the original work is properly cited . * keywords : * scale - free network , power - law degree distribution , dynamics of infectious diseases

non - axisymmetric deformations of rapidly rotating self - gravitating bodies are a rather generic phenomenon in nature and could appear in a variety of astrophysical scenarios like stellar core collapses @xcite , accretion - induced collapse of white dwarfs @xcite , or the merger of two neutron stars @xcite . over the years , a considerable amount of work has been devoted to the search of unstable deformations that , starting from an axisymmetric configuration , can lead to the formation of highly deformed rapidly rotating massive objects @xcite . such deformations would lead to an intense emission of high - frequency gravitational waves ( i.e. in the khz range ) , potentially detectable on earth by next - generation gravitational - wave detectors such as advanced ligo @xcite , advanced virgo and kagra @xcite in the next decade @xcite . from the observational point of view , it is import to get any insight on the possible astrophysical scenarios where such instabilities ( unstable deformation ) are present . it is well known that rotating neutron stars are subject to non - axisymmetric instabilities for non - radial axial modes with azimuthal dependence @xmath4 ( with @xmath5 ) when the instability parameter @xmath6 ( i.e. the ratio between the kinetic rotational energy @xmath7 and the gravitational potential energy @xmath8 ) exceeds a critical value @xmath3 . the instability parameter plays an important role in the study of the so - called dynamical bar - more instability , i.e. the @xmath9 instability which takes place when @xmath1 is larger than a threshold @xcite . previous results for the onset of the classical bar - mode instability have already showed that the critical value @xmath3 for the onset of the instability is not an universal quantity and it is strongly influenced by the rotational profile @xcite , by relativistic effects @xcite , and , in a quantitative way , by the compactness @xcite . however , up to now , significant evidence of their presence when realistic equation of state ( eos ) are consider is still missing . for example in @xcite , using the unified sly eos @xcite , was shown the presence of shear - instability but no sign of the classical bar - mode instability and of its critical behavior have been found . the main aim of the present work is to get more insight on the behavior of the classical bar - mode instability when the matter is described by a stiffer more realistic eos . the investigation in the literature on its dependence on the stiffness of eos usually focused on the values of @xmath2 ( i.e. the adiabatic index of a polytropic eos ) in the range between @xmath10 and @xmath11 @xcite , while the expected value for a real neutron star is more likely to be around @xmath0 at least in large portions of the interior . such a choice for the eos has already been implemented in the past @xcite , even quite recently @xcite , with the aim of maintaining the simplicity of a polytropic eos and yet obtaining properties that resemble a more realistic case . indeed , as it is shown in fig . [ fig : eoss ] , a polytropic eos with @xmath12 and @xmath0 is qualitatively similar to the shen proposal @xcite in the density interval between @xmath13 and @xmath14 . for the sake of completeness , in fig . [ fig : eoss ] we also report the behavior of the @xmath11 polytrope used in @xcite and of the unified sly eos @xcite which describes the high - density cold ( zero temperature ) matter via a skyrme effective potential for the nucleon - nucleon interactions @xcite . the organization of this paper is as follows . in sect . [ sec : setup ] we describe the main properties of the relativistic stellar models we investigated and briefly review the numerical setup used for their evolutions . in sect . [ sec : results ] we present and discuss our results , showing the features of the evolution for models that lie both above and below the threshold for the onset of the bar - mode instability and quantifying the effects of the compactness on the onset of the instability . conclusions are finally drawn in sect . [ sec : conclusions ] . throughout this paper we use a space - like signature @xmath15 , with greek indices running from 0 to 3 , latin indices from 1 to 3 and the standard convention for summation over repeated indices . unless otherwise stated , all quantities are expressed in units in which @xmath16 . in this work we solve the einstein s field equations @xmath17 where @xmath18 is the einstein tensor of the four - dimensional metric @xmath19 and @xmath20 is the stress - energy tensor of an ideal fluid . this can be parametrized as @xmath21 where @xmath22 is the rest - mass density , @xmath23 is the specific internal energy of the matter , @xmath24 is the pressure and @xmath25 is the matter @xmath26-velocity . the evolution equations for the matter follow from the conservation laws for the energy - momentum tensor @xmath27 and the baryon number @xmath28 , closed by an eos of the type @xmath29 . in order to generate the initial data we evolve in this work , we use a @xmath2-type eos of the form @xmath30 where the following relation between @xmath23 and @xmath22 holds : @xmath31 . on the other hand , the evolution is performed using the so - called _ ideal - fluid _ ( @xmath2-law ) eos @xmath32 that allows for increase of the internal energy , by shock heating , if shocks are presents . we have chosen the eos polytropic parameters to be @xmath0 for the adiabatic index and @xmath12 for the polytropic constant . this choice of parameters has the property to closely reproduce the behavior of the shen eos in the interior of a real neutron star ( see fig . [ fig : eoss ] ) . we note that the choice we make here is different from the one of our previous studies @xcite , where we used @xmath11 and @xmath33 , with the explicit intention of determining the difference that such a change implies on the onset of the bar - mode instability . vs. the energy density @xmath34 for two polytropic eoss and two _ realistic _ eos for nuclear matter , namely : ( 1 ) the shen proposal @xcite ; ( 2 ) the unified sly prescription @xcite . [ fig : eoss],title="fig:",scaledwidth=45.0% ] + we solve the above - mentioned set of equations using the usual @xmath35 space - time decomposition , where the space - time is foliated as a tensor product of a three - space and a time coordinate @xmath36 ( which is selected to be the @xmath37 coordinate ) . in this coordinate system the metric can be split as @xmath38 , where @xmath39 has only the spatial components different from zero and can be used to define a riemannian metric @xmath40 on each foliation . the vector @xmath41 , that determines the direction normal to the 3-hypersurfaces of the foliation , is decomposed in terms of the lapse function @xmath42 and the shift vector @xmath43 , such that @xmath44 . we also define the fluid three - velocity @xmath45 as the velocity measured by a local zero - angular momentum observer ( @xmath46 ) , while the lorentz factor is @xmath47 . within this formalism , the conservation of the baryon number suggests the use of the conserved variable @xmath48 with the property that @xmath49 along the time - evolution @xmath36 . , scaledwidth=45.0% ] the initial data of our simulations are computed as stationary equilibrium solutions for axisymmetric and rapidly rotating relativistic stars in polar coordinates @xcite . in generating these equilibrium models we assumed that the metric describing the axisymmetric and stationary relativistic star has the form @xmath50 where @xmath51 , @xmath52 , @xmath53 , and @xmath54 are space - dependent metric functions . similarly , we assume the matter to be characterized by a non - uniform angular velocity distribution of the form @xmath55 \ , , \ ] ] where @xmath56 is the equatorial stellar coordinate radius , and the coefficient @xmath57 is the measure of the degree of the differential rotation , which we set to be @xmath58 , in analogy with other works in the literature . once imported onto the cartesian grid and throughout the evolution , we compute the coordinate angular velocity @xmath59 on the @xmath60 plane as , @xmath61 other characteristic quantities of the system such as the baryon mass @xmath62 , the gravitational mass @xmath63 , the internal energy @xmath64 , the angular momentum @xmath65 , the rotational kinetic energy @xmath7 , the gravitational binding energy @xmath8 and the instability parameter @xmath1 are defined as @xcite : @xmath66 \beta & \equiv t/|w| \ , , \end{aligned}\ ] ] where @xmath67 is the square root of the four - dimensional metric determinant . notice that the definitions of quantities such as @xmath65 , @xmath7 , @xmath8 and @xmath1 are meaningful only in the case of stationary axisymmetric configurations and should therefore be treated with care once the rotational symmetry is lost . all the equilibrium models considered here have been calculated using the relativistic polytropic eos given in eq . ( [ eq : eosp_gamma ] ) , choosing @xmath12 and @xmath0 , in contrast to @xcite , where the values of @xmath33 and @xmath11 have been used . the initial conditions for the evolution have been generated using the nicholas stergioulas rns code @xcite . any model can be uniquely determined ( once the value of the differential rotation parameter has been fixed to @xmath68 ) by two parameters . we decided to denote each of the generated models using the values of the conserved baryonic mass @xmath62 and the @xmath1 parameter at @xmath69 . as a consequence of this choice , in the rest of this paper we will refer to a particular model using the following notation . for example , m1.5b0.270 will denote a model with a conserved baryonic mass @xmath70 and a value of the instability parameter @xmath71 . one of the main features of the generated models is that , due to the high rotation , not all of them have the maximum of the density at the center of the star . for example , if we analyze some of the generated models with a fixed value of the baryonic mass @xmath72 ( see fig . [ fig : densitym2 ] ) , we note that those rotating fastest have the maximum of the density at a distance of their center which is around @xmath73 km . this means that most of the studied models are characterized by a _ toroidal _ configuration ( i.e. the maximum of the density is not on the rotational axis ) . as we will see , there is no correlation between having a toroidal configuration and being unstable against the dynamical bar - mode instability , like in the case of polytropic models with @xmath11 @xcite . axis we report the compactness of each stellar model , while its rotation parameter @xmath1 is on the @xmath74 axis . squares denote models that are not subject to the bar - mode instability while circles represent the ones that are bar - mode ( @xmath75=@xmath76 ) unstable . [ fig : idmodels],scaledwidth=45.0% ] using continuous - black , doted - blue , dash - dash - doted - green , dash - doted - magenta and dashed - green line for the sequences of baryonic mass @xmath77 , @xmath78 , @xmath79 , @xmath78 and @xmath80 , respectively . we also shaded , with the corresponding color , the initial part of the corotation bands . for any value of the frequency of a mode between these two lines there is exactly one radius inside the star which is co - rotating with the mode . [ fig : corotation_band],scaledwidth=45.0% ] an important issue related to the use of polytropic eos in the construction of the initial models is that their properties are fixed in terms of physical scales determined by the value of the polytropic constant @xmath81 that can always be set to @xmath82 by changing the measure units . the assertion that we are generating a model with a giving baryonic mass @xmath62 is therefore related to the actual value chosen for @xmath81 . indeed , in order to claim that the threshold for the instability depends on the stiffness of the eos , we need to eliminate the dependencies on the dimensional scales and then on the chosen value of the polytropic constant @xmath81 . an efficient way to do that is to extrapolate the result for @xmath83 , which corresponds to the newtonian limit , where the general relativistic effects can be neglected . indeed , using the same procedure followed in @xcite , we chose five sequences of constant rest - mass density models , namely with @xmath84 . again in analogy with @xcite , we use a rotational profile with @xmath85 for all models . we restrict the values of the instability parameter @xmath1 to the range between @xmath86 and @xmath87 and we leave the analysis of models with lower values to future work . the positions of all the simulated models in terms of their compactness @xmath88 , i.e. the ratio between the gravitational mass @xmath63 and the equatorial radius @xmath89 , and the rotational parameter @xmath1 are reported in fig . [ fig : idmodels ] . since the models are differentially rotating , fig . [ fig : corotation_band ] shows the corotation bands for the five sequences of models we analyzed . [ cols= " < , < " , ] parameter for different values of the employed resolution . the filled dots mark the time at which the values @xmath90 , used in the fits are evaluated.,scaledwidth=45.0% ] at the time when the distortion parameter @xmath91 reaches 1% of its maximum value , as a function of the inverse of the square of the growth time , for model m1.5b0.270 at different resolutions for the finest grid . the straight line represents the result of the fit ( as a function of @xmath1 ) for the whole sequence of models at fixed value of the conserved baryonic mass @xmath92 and at resolution @xmath93 . the open triangles represent the values of two different models among them . [ fig : tau_resolution],scaledwidth=45.0% ] while the overall dynamics is very similar for all resolutions , there is a consistent shift of the value of the @xmath1 parameter at the beginning of the development of the instability @xmath94 and of the value of the fitted growth time @xmath95 with increasing resolution . more precisely , the analysis of the initial stage of the evolution shows ( see fig . [ fig : beta_evolution_res ] ) that there is indeed a drift ( decrease ) of the value of @xmath1 and , consequently , the fit of @xmath95 is sensible to the value used for a given resolution . this can , at least partially , be explained to be due to the fact that by the time the amplitude of the mode reaches the fit - window , the evolution does not correspond any longer to the original model but it is closer to a new equilibrium model ( through an adiabatic drift ) , characterized by a different value of the rotational parameter @xmath1 . the overall conclusion is that one has to be careful when associating the fitted value for the growth time of the instability to the initial value of the rotational parameter @xmath1 . in fact , as it can be seen in fig . [ fig : beta_evolution_res ] , there is a small shift ( of as much as 2% for the lowest resolution ) of the value of @xmath1 from the start of the simulation up to the time at which the instability is detected ( @xmath96 ) . using this fact , we now report in fig . [ fig : tau_resolution ] the growth - time and the @xmath1 at the beginning of the instability for each resolution and the critical fit of eq . ( [ eq : res_fits ] ) for @xmath92 . that shows we have convergence in the determination of the parameter of the classical bar - mode instability above the threshold , if we use the value @xmath94 to perform the extrapolations . while these results show that the used resolution is enough to explore the dynamical bar - mode instability , we can not draw such conclusion on the parameter of the observed shear - instability at lower @xmath1 . despite that we did observe that such instabilities are present , the resolution used here is not sufficient to draw conclusions about their numerical values . we note that we found candidates of @xmath9 shear instability when examining several models in different resolutions , but the instability dynamics can not be clearly singled out with respect to other instabilities that seem to be present . one example is given in fig . [ fig : m25b200res ] , which shows that the interplay on the growth of various modes shows some dependency on the resolution . such dependency was not observed in the cases where the classical dynamical bar - mode instability is present and it is by far the fastest growing mode . , scaledwidth=45.0% ] this shows that for these values of @xmath1 the used resolution on the finer grid of @xmath93 is not enough to determine the dynamics of the shear instability . this initial analysis show that a resolution of , at least , @xmath97 is needed . moreover , in this case , the use of the numerical discretization error to triggering the fastest growing mode does not seem to be the best strategy to study shear instabilities when a competition between different @xmath75 modes ( like @xmath9 and @xmath98 ) is present . we will leave a detailed study of the low-@xmath99 instabilities to a future work . we have presented a study of the dynamical bar - mode instability in differentially rotating nss in full general relativity for a wide and systematic range of values of the rotational parameter @xmath1 and the conserved baryonic mass @xmath62 , using a polytropic / ideal - fluid eos characterized by a value of the adiabatic index @xmath0 , which allows us to resemble the properties of a realistic eos . in particular , we have evolved a large number of ns models belonging to five different sequences with a constant rest - mass ranging from @xmath100 to @xmath101 , with a fixed degree of differential rotation ( @xmath58 ) and with many different values of @xmath1 in the range @xmath102 $ ] . for all the models with a sufficiently high initial value of @xmath1 we observe the expected exponential growth of the @xmath9 mode which is characteristic of the development of the dynamical bar - mode instability . we compute the growth time @xmath95 for each of these bar - mode unstable models by performing a nonlinear least - square fit using a trial function for the quadrupole moment of the matter distribution . the growth time clearly depends on both the rest - mass and the rotation and in particular we find that the relation between the instability parameter @xmath1 and the inverse square of @xmath95 , for each sequence of constant rest - mass , is linear . this allows us to extrapolate the threshold value @xmath3 for each sequence corresponding to the growth time going to infinity , using the same procedure already employed in @xcite . once the five values of @xmath3 have been computed , we are able to extrapolate the critical value of the instability parameter for the newtonian limit , which is found to be @xmath103 . this value can be directly compared with the one found in @xcite for the `` standard '' @xmath104 case , which is @xmath105 . our results suggest that , even if one can now consider just two values for the adiabatic index , namely the values @xmath106 , considered in the present work , and @xmath104 , considered in @xcite , the use of a stiffer , more realistic eos should be expected to have the effect of reducing the threshold for the onset of the dynamical bar - mode instability . unfortunately , the actual reduction of the threshold @xmath3 is just of the order of @xmath107 , and indeed this reduction does not lead to a significantly higher probability for it to occur in real astrophysical scenarios . we also evolved many models belonging to the same five sequences but having lower values of the instability parameter @xmath1 . we find that many of them show the growth of one or more modes even though their initial value of @xmath1 is below the threshold for the onset of the dynamical bar - mode instability . the modes that show a growth are mainly @xmath9 and the @xmath98 . we compute the frequencies of these growing modes and compare them with the corotation band for their progenitor models , finding that all those frequencies are within this band . we can conclude that such instabilities have to be defined as _ shear instabilities _ , as the ones that were already observed in @xcite . unfortunately , we are not able to measure their growth time , since their dynamics change significantly by changing the resolution of the simulations . in fact , while at a coarse resolution we usually observe only one mode growing exponentially , when improving the resolution other modes develop as well and the interplay between these prevent a clear exponential growth of only one mode which could dominate the evolution . in order to make a quantitative assessment about this phenomenon , either much higher resolution has to be used to see if one of the modes is able to dominate , or seed perturbations have to be introduced with the aim of selecting only a particular mode at a time . we leave this treatment to futures studies . we do have to especially thank n. stergioulas for providing us the rns code that we used to generate the initial stellar configurations . we would also like to thank r. alfieri , s. bernuzzi , n. bucciantini , a. nagar , l. del zanna , for useful discussions and insights in the development of the present work . portions of this research were conducted with high performance computing ( hpc ) resources provided by the european union prace program ( 6@xmath108 call , project `` 3dmagroi '' ) , by the louisiana state university ( allocations hpc_cactus , hpc_numrel and hpc_hyrel ) , by the louisiana optical network initiative ( allocations loni_cactus and loni_numrel ) ; by the national science foundation through xsede resources ( allocations tg - asc120003 , tg - phy100033 and tg - mca02n014 ) , by the infn `` theophys '' cluster and through the allocation of cpu time on the bluegene / q - fermi at cineca for the specific initiative infn - og51 under the agreement between infn and cineca . the work of a. f. has been supported by miur ( italy ) through the infn - suma project . f. l. is directly supported by , and this project heavily used infrastructure developed using support from the national science foundation in the usa ( 1212401 / 1212426 / 1212433 / 1212460 ) .

we present results about the effect of the use of a stiffer equation of state , namely the ideal - fluid @xmath0 ones , on the dynamical bar - mode instability in rapidly rotating polytropic models of neutron stars in full general relativity . we determine the change on the critical value of the instability parameter @xmath1 for the emergence of the instability when the adiabatic index @xmath2 is changed from 2 to 2.75 in order to mimic the behavior of a realistic equation of state . in particular , we show that the threshold for the onset of the bar - mode instability is reduced by this change in the stiffness and give a precise quantification of the change in value of the critical parameter @xmath3 . we also extend the analysis to lower values of @xmath1 and show that low - beta shear instabilities are present also in the case of matter described by a simple polytropic equation of state .

in most history of our universe , gravity has been the dominant force in determining cosmological evolutions . due to the onset of gravitational instability , primordial fluctuations generated in cosmic inflation start growing by attracting dark matter and gases nearby after the matter radiation equilibrium at a temperature approximately @xmath1 @xcite . consequently , structures of the first generation galaxies are seeded . shortly after the forming of proto - galaxies , the hydrogen and helium gases within them begin to condensate and make the first stars . in the formation of any astrophysical body , self - gravity is beyond question one of the key factors in determining the specific destinies of stars , quasars , black holes , and neutron stars . for example , the self - gravity of neutron stars is considered using the affine model @xcite , where the effective relativistic self - gravitational potential is approximated with tolman - oppenheimer - volkoff ( tov ) stellar structure equations . indeed , gravity , though much weaker than the other three fundamental forces , i.e. , strong , weak and electromagnetic forces , with its long range of action and omnipresent character from equivalence principle of general relativity , determines most astronomical evolutions . one persuasive example showing the significance of self - gravity is the boson star . though massive boson stars are prevented from the heisenberg uncertainty principle , they can actually be formed in the early universe when the self - interaction of scalar particles is considered @xcite . various methods are subsequently developed to study different features of boson stars . people deal with a system of self - gravitating bosons either by solving the schrdinger and poisson equations in a newtonian way @xcite , or by fully accounting the relativistic effect using klein - gordon and einstein equations @xcite . when the einstein gravity and the klein - gordon field couple with each other , exotic bosons which possess a small mass may undergo bose - einstein condensation driven by gravity and collapse to form boson stars in the early universe or latter in galaxy center as a candidate for dark matter @xcite . if self - gravity can trigger phenomena at the early universe so dramatic like boson stars or primordial black holes @xcite not mentioned before , the self - gravitational effect in the primordial patch of the early universe should not be easily discarded without a robust proof . an exact evaluation of the self - gravity of relativistic particles at the extremely high energy scale before big bang nucleosynthesis ( bbn ) ( not high enough to reach the boson star critical temperature ) becomes unavoidable . however , most discussions of boson stars are at relatively high energy scale @xmath2 gev @xcite , and at the same time entail the non - zero mass of scalar particles to be greater than around 10 ev to preserve radial stability @xcite . but during the radiation dominated era before bbn , main ingredients are massless photons , and the homogeneity of the early patch is preserved unlike that in boson stars . as a result , it is difficult to fully perceive the self - gravitational behavior of photons in the early universe by simply extending the methods developed for boson stars . although there are several attempts to include the self - gravitational effect of non - relativistic gases at equilibrium , such as monte carlo simulations , analytic mean field methods and low density expansions @xcite , no method so far exists to directly solve the self - gravity of massless particles , i.e. , photons . as a result , we try to develop an appropriate way to deal with the self - gravity of massless bosons in the early universe . the newly developed method will be applied to massive particles as well . before bbn , the universe can be well described as in thermal equilibrium with very high temperature and frequent interactions . remarkable features of the early universe , homogeneity , isotropy and thermal equilibrium together can help us simplify the discussion to a great extent . since the equation of state ( eos ) can serve as a good description of a homogeneous systems , we manage to delve into the problem with statistical physics . as is well known , once including self - interactions between particles , the corresponding partition function shall be changed . such change further leads to modifications in the eos . we try to uncover the self - gravitational effect by examining the corresponding statistical properties of bose and fermi gases first . whether corrections from self - gravity are negligible or bear some new features not considered before shall be tested in detail in this work . in the radiation - dominated epoch , gravities between relativistic particles are commonly neglected in existing models . when the temperature of the universe decreases to approximately @xmath3 , particles are in thermal equilibrium without strong nuclear interactions and the principal ingredients at that time are photons , neutrinos , electrons and their anti - particles . with the energy of the universe dominated by photons , other ingredients , whether massive particles like electrons and positrons or the massless ones , all behave radiation - like due to their frequent collisions with photons and each other at the ultra - high temperature . the matter soup at that time is suggested to follow the equation of state @xmath4 without considering gravitational attractions between particles . from the equivalence principle , massless particles with both high kinetic energies and particle densities may possess a considerable gravitational effect . we deal with self - gravitational interactions between these thermalized particles in a newtonian way . with a statistical approach , we can avoid the theoretical difficulties in solving equations of general relativity as well as numerical errors in actual realization , and at the same time , perceive macro - behaviors without any impediment from micro - complexity . in sec . [ gcp ] , we deduce a new form of grand canonical function for bose and fermi gases in the relativistic and the non - relativistic limits . in sec . [ earlyuni ] , we apply the newly - developed partition function to the early universe before bbn . we also analyze the time dependent term of the eos and set a strict upper bound on the modifications of the state constant from self - gravity . possible influences on relating topics are discussed in sec . [ remark ] . in standard statistical physics , people deal with interactions between gases by including @xmath5 in the grand canonical partition function , where @xmath6 is the potential energy resulting from the interactions between two particles @xmath7 , @xmath8 . the potential energies @xmath6 are set to be zero in general . such approximation works pretty well when interaction is negligible , that is , the density - temperature ratio is extremely low . on the other hand , if the density is relatively high like that in the early universe , the factor from the self - gravitational effect of bosons and fermions becomes appreciable . as uhlenbeck and gropper @xcite proved , the partition function of non - ideal bose or fermi gases should be modified as @xmath9\right ) , \label{eq1}\ ] ] where the plus sign is for fermi gas . the grand canonical partition function becomes @xmath10\right)\mathrm{d } \mathbf{r}_{ij } , \label{eq2}\ ] ] where @xmath11 represents the chemical potential , and @xmath12 is the mass of the particle . for non - relativistic particles , which satisfy the dispersion relation @xmath13 , the first integral in eq . ( [ eq2 ] ) turns @xmath14 since in the non - relativistic case , @xmath15 , the second integral becomes @xmath16\right)\mathrm{d } \mathbf{r}_{ij}=\int e^{-\sum_{i < j}\frac{gm^{2 } } { r_{ij}k_{b}t}}\left(1\pm \sum_{ij } \mathrm{exp}\left[-4\pi mk_{b}tr_{ij}^{2}/h^{2}\right]\right)\mathrm{d}\mathbf{r}_{ij}. \label{eq4}\ ] ] with @xmath17\ll 1 $ ] , the omissions made in eq . ( [ eq4 ] ) are justified . at the same time , we use mean field theory to include only the average effect of @xmath18 . the thermal equilibrium used in mean field theory is guaranteed in the early universe . as @xmath19 we approximate @xmath20 , following which eq . ( [ eq4 ] ) becomes @xmath21\right)\mathrm{d } \mathbf{r_{\ast } } \mathrm{d } \mathbf{r } \nonumber \\ & = & v^{n-1}\left\{\int e^{-\frac{n^{2}gm^{2}}{2r_{\ast } k_{b}t}}\mathrm{d}\mathbf{r}_{\ast } \pm \int \frac{n^{2}}{2}\mathrm{exp}\left[-4\pi mk_{b}tr_{\ast } ^{2}/h^{2}\right]\mathrm{d } \mathbf { r_{\ast } } \right . \nonumber \\ & & \left.\pm \int \frac{n^{2}}{2}\left(e^{-\frac{n^{2}gm^{2}}{2r_{\ast } k_{b}t}}-1\right ) \mathrm{exp}\left[-4\pi mk_{b}tr_{\ast } ^{2}/h^{2}\right]\mathrm{d } \mathbf{r_{\ast } } \right\ } , \label{eq5}\end{aligned}\ ] ] where @xmath22 is the average distance between two particles . we can see that the first term on the right side of eq . ( [ eq5 ] ) represents the gravitational correction of the grand canonical partition function in conventional treatment and the second term corresponds to usual bose - einstein or fermi - dirac statistics , while the last term is a newly - introduced coupling term from the gravity of bosons or fermions . again , the upper sign is for fermi gas . in the early universe , particle number and energy density are relatively high with temperature@xmath23 @xmath3 , i.e. , exp@xmath24\sim 1 $ ] . to integrate the first term of eq . ( [ eq5 ] ) , we expand it into the the first - order taylor series and get @xmath25 this reveals that the upper bound of the integration corresponds to the spacial scale of the system . the second term of eq . ( [ eq5 ] ) is @xmath26\mathrm{d } \mathbf { r_{\ast } } \nonumber \\ & = & 4\pi \int_{0}^{r_{0 } } \frac{n^{2}}{2}\mathrm{exp}\left[-4\pi mk_{b}tr_{\ast } ^{2}/h^{2}\right]r_{\ast } ^{2}\mathrm{d}r_{\ast } \nonumber \\ & = & 4\pi \int_{0}^{\infty } \frac{n^{2}}{2}\mathrm{exp}\left[-4\pi mk_{b}tr_{\ast } ^{2}/h^{2}\right]r_{\ast } ^{2}\mathrm{d}r_{\ast } \nonumber \\ & = & 2^{-5/2}\pi^{3/2}n^{2}\lambda_{\mathrm{nonrela}}^{3 } , \label{eq7}\end{aligned}\ ] ] where @xmath27 is the thermal de broglie wavelength , which should be much smaller than the radius of our interested system . if the scale of the system is equal to or smaller than the thermal de broglie wavelength , a situation which might happen at a much earlier time , such approximation may not be valid , and we should use the error function @xmath28 in the integration instead . again applying the first - order taylor expansion , we get the remaining term @xmath29r_{\ast } \mathrm{d}r_{\ast } = -\frac{n^{4 } \pi gm^{2 } } { 4k_{b}t}\lambda_{\mathrm{nonrela}}^{2}. \label{eq8}\ ] ] finally , we find the new grand canonical partition function @xmath30 , \label{eq9}\ ] ] with which we can obtain the total energy by @xmath31 with @xmath32 , and then the entropy . as suggested above , this partition function is applicable when the scale of the system is much larger than the thermal de broglie wavelength . however , such premises may be wavered at a much earlier stage . the first term of the partition function is the grand canonical function for ideal boltzmann gas . the second term represents the self - gravitational correction of boltzmann gas , and its magnitude corresponds to the portion that gravitational energy @xmath33 contributes to the total energy @xmath34 . in turn it manifests whether gravitational correction is important in dealing with the evolution of the universe . the third term is the correction from identical principle , and its magnitude ( proportional to the cubic square of thermal de broglie wavelength over the volume of the system @xmath35 ) evaluates how strong quantum identity is . as a whole , the self - gravitational effect on fermions is restricted by both the strength of gravity and the intensity of quantum effect . for highly relativistic particles , the rest mass is negligible compared with the kinetic energy , thus the dispersion relation becomes @xmath36 . then the grand canonical partition function is found to be @xmath37\right)\mathrm{d } \mathbf{r}_{ij}. \label{eq10}\ ] ] we can solve it exactly by diagonalizing @xmath38 first . but such treatment is proved to be unnecessary since , on one hand , the self - gravitational effect in the relativistic case acts as a correction term , and on the other hand , for our problem in the homogeneous and isotropic robertson - walker metric background at exceptionally high temperature , relativistic particles are in a thermal equilibrium state evolving in a radiation manner resulting from frequent collisions . therefore , we expect self - gravity alone will not significantly destroy this homogeneity directly shown by cmb observations . all particles in the patch then possess the same characteristic average momentum due to the frequent collision . we then approximate the summation as @xmath39 where @xmath40 is the average momentum of the particles in the system . so the grand canonical partition function becomes @xmath41 to integrate the first term in eq . ( [ eq11 ] ) , we perform the taylor expansion to the first order knowing the magnitude of gravity being small outside the critical sphere with radius @xmath42 @xmath43 the integration from 0 to @xmath44 @xmath45 is neglected since its magnitude is proportional to @xmath46 , much smaller than the first one in our analysis when gravity contributes as a correction term . under the same mean field approximation , eq . ( [ eq12 ] ) becomes @xmath47 before going on to integrate the second term of eq . ( [ eq11 ] ) , we first compare the magnitude of different terms in the denominator @xmath48 . when @xmath49 , i.e. @xmath50 , the corresponding energy density can be approximated as @xmath51 in the planck unit . however , the actual energy density is of magnitude @xmath52 @xcite , meaning that the first term is @xmath53 compared with @xmath54 . besides , since in our interested time period , the temperature varies within one order in magnitude , so long as the number density does not change beyond @xmath55 , i.e. the volume of the universe increases by less than @xmath55 within 1s ( such circumstances are fully guaranteed in a classical radiation - dominated universe ) , @xmath56 can then be safely neglected . but for some earlier stages this treatment may fail . in the same way , @xmath57 is @xmath58 compared with the third term and can be justifiably omitted if the number density varies within @xmath59 . neglecting the first two terms after careful magnitude evaluation , we have @xmath60 therefore only the first integration contributes to our final result , corresponding to a unification of bose and fermi distributions under the relativistic limit . the grand canonical partition function for relativistic particles including the self - gravitational effect becomes @xmath61\mathrm{exp}\left[e^{-\alpha } { \left(\frac{8\pi k_{b}t}{ch}\right)}^{3}v\right ] = \left[1 - 6^{2/3}\frac{3\pi gh}{\lambda_\mathrm{rela}c^3v^{1/3}}\frac{\partial ^{2}}{\partial \alpha ^{2}}\right]\mathrm{exp}\left[e^{-\alpha } \frac{v}{(\lambda_\mathrm{rela})^3}\right ] , \label{eq14}\ ] ] where @xmath62 is the thermal de broglie wavelength for massless particles . in eq . ( [ eq14 ] ) , the first term is merely the ordinary ideal boltzmann gas term , while the second is induced by self - gravity , proportional to @xmath63 , i.e. the gravitational potential energy over the self - energy of particles . by comparing the two terms we can catch a glimpse of how gravity acts on the statistical properties of the relativistic particles . before applying the newly developed method to a specific background , we first review the whole evolution of the early universe . in the very early stages of the evolution , to resolve several problems in the big bang cosmology , like the horizon problem , magnetic - monopole problem , structure formation problem and so on , the universe is suggested to undergo an inflation between time scales @xmath64 and @xmath65 . by the end of the inflation , the background temperature is roughly @xmath66 when strong , weak and electromagnetic forces are unified . as the temperature cools down , the strong force freezes out , weak and electromagnetic interactions are subsequently off stage and before @xmath67 . afterward , neutrinos decouple at @xmath68 . then big bang nucleosynthesis begins at temperature @xmath69 . at the end of bbn , photons decouple from matter , and form the cosmic microwave background radiation we observe today . to use our statistical method in studying the self - gravitational effect , assumptions of homogeneity , isotropy and thermal equilibrium should be justified . we thus choose to study the universe before bbn when three criteria are largely met and the background temperature @xmath70 is of magnitude @xmath3 @xcite , well below the corresponding thermal energy of the rest mass of @xmath71 mesons and strong interactions can be neglected . matter and radiation in the universe at that time are in thermal equilibrium , resulting from the rapid collisions in the exceedingly high temperature despite of the fast universal expansion . by identifying particles whose rest masses are below the corresponding thermal energy of the background temperature @xmath70 , we find the main ingredients to be electrons , neutrinos , their anti - particles and photons . these abundant genres are highly relativistic . for example , the rest energy of the electron corresponds to the thermal energy at temperature @xmath72 , two orders of magnitude smaller than the background temperature . and non - relativistic particles , mainly protons and neutrons , constitute only an extremely small portion of the whole ingredients , approximately @xmath73 as many as that of relativistic ones . below we deal with relativistic and non - relativistic particles separately , though the latter is unable to significantly influence the background evolution . at a temperature higher than @xmath74 , the universe is radiation dominated . consequently , relativistic particles are undoubtedly the main ingredients during our studying period with a temperature approximately @xmath3 . and both massive and massless components act in a radiationlike manner due to frequent collisions . we use the planck unit @xmath75 in all calculations and some of our discussions below . moreover , the volume @xmath76 , which is about four light years in radius in our partition function , corresponds to the biggest thermalized region constrained by cmb anisotropy and grows as the extension of particle horizon of the primordial patch . though the exact size of the early universe at this time is not determined yet . the time span of our study is @xmath77 . we consider the relativistic bosons first and later find that fermions behave similarly as expected . their grand canonical partition function is @xmath78\mathrm{exp}\left[e^{-\alpha } \left(\frac{v}{(\lambda_\mathrm{rela})^3}\right)\right ] , \label{eq25}\ ] ] the chemical potential is set to be zero since the particle number of the photon , which is the main ingredient , does not conserve . here , instead of directly solving the friedman equation with the density and pressure derived in the new grand canonical partition function , we approximate the temperature evolution the same as that in the classical radiation - dominated model neglecting the self - gravitational effect @xmath79 where @xmath80 is a constant to be determined by the initial condition . combining eqs . ( [ eq25 ] ) and ( [ eq26 ] ) , we numerically find that the density of relativistic particles relies on time as @xmath81 where @xmath82 and @xmath83 are separately @xmath84 and @xmath85 . this differs from the conventional result without considering the self - gravitational effect where @xmath86 . that is , once including the self - gravity , energy density decreases much slower . this disparity in energy density can be explained as an increase in gravitational potential caused by the rapid elongation of particle separation in cosmic expansion . this remedies the conventional energy lost with the decrease of particle density and the extension of particle wavelength . the pressure behaves as @xmath87 where @xmath88 . as a result , the state constant no longer stays invariant but instead relies on time @xmath89 where @xmath90 has a numerical value @xmath91 in planck unit which is @xmath92 . knowing that during our studying period @xmath93 in planck unit , or @xmath94 in standard unit , and the volume is @xmath95 in planck unit , or corresponds to four light years in radius justified by cmb observation , we find that the correction term in eq . ( [ eq27++ ] ) is of the magnitude @xmath96 . since the universe is radiation dominated at this time , we find the correction term @xmath97 . differing from the non - relativistic case below , the modification from self - gravity rapidly diminishes as time goes on and volume expands . besides , we can roughly estimate the modification term at much earlier times , such as the end of inflation when @xmath98 and @xmath99 , to be @xmath100 . thus the estimated magnitude of the modification is comparable to the effect in the non - relativistic case . although massive nuclear particles contribute only a negligible portion to the whole energy makeup during the period we study , their abundances are crucial in determining the late time nuclear synthesis . as a result of the rapid expansion , the energy density of the radiation decreases swiftly as @xmath101 in classical treatment , while matter density such as that of protons and neutrons decreases much slower as @xmath102 . additionally , relativistic particles and their anti - particles annihilate with each other as temperature goes down , and the non - relativistic nuclear particle component , therefore , becomes increasingly important . the non - relativistic feature of nuclear particles shall preserve as temperature decreases throughout the thermalization and into nuclear synthesis . since most of the nuclear particles are fermions , we consider only non - relativistic fermion gases without loss of generality . we can safely neglect the chemical potential of bose gases since the photon number is not conserved . but for non - relativistic fermi gases , at extremely high temperature , such a simplification may no longer be appropriate . fortunately , it turns out that the chemical potential fails to show up in density and pressure expressions to the first and second order approximations in our result . this not only significantly simplifies the calculation , but also justifies the usual negligence of chemical potential in the evolution of non - relativistic fermions in the early universe @xcite . for massive fermions , the grand canonical partition function is @xmath103 . \label{eq17}\ ] ] we then obtain the energy density in a conventional way @xmath104 @xmath105 that is , the density is a function of @xmath106 and volume , while @xmath107 and v vary in time . here the background temperature evolution can be well approximated by @xmath108 during the radiation dominated period . in the time span we are studying , @xmath109 varies within one order of magnitude . to simplify the problem correspondingly , we set the temperature to be @xmath3 and solve the density evolution . the mass of general non - relativistic particles in calculation is chosen to be the neutron mass , since masses of other non - relativistic nuclear particles at that time differ only by a coefficient within one order of magnitude . the expression for @xmath110 becomes exceedingly lengthy . facing a huge magnitude range which unavoidably exceeds the calculation limit in the computer , to evaluate the expression without any numerical reduction is impossible . however , after observing the density behavior within a small volume range , an asymptotic behavior is found . we then analyze the density behavior as volume radius increases to four light years . after careful magnitude analysis , we find that as long as the volume of the system exceeds @xmath111 , where @xmath76 is the initial volume of the primordial patch . such a requirement is thus safely guaranteed in the primordial patch before bbn . the first - order density behavior relying on the temperature is found to be @xmath112 where @xmath113 equals @xmath114 in planck unit , or @xmath115 . we use the extreme condition of the early universe between 0.1 s and 1 s to numerically reduce the expression and find that the result goes back to the ideal fermion case , as is shown in fig . [ fig : dis massive ] . relation in planck unit.,scaledwidth=50.0% ] applying the same method , a similar asymptotic behavior of the first order can be found in the pressure - volume relation . we notice that strange negative pressure occurs in the transplanckian scale due to the failure of statistical physics . however , as the radius of the patch grows to four light years , which is the initial value of our calculation , the pressure again becomes almost volume - independent . the reduced pressure is @xmath116 where @xmath117 , as is shown in fig . [ fig : dis pressure ] . then the equation of state in the specific time span becomes @xmath118 where the state constant @xmath119 to the first - order approximation , as can be seen in fig . [ fig : dis massive ] . the recurrence back to classical case implies that the gravitational effect on non - relativistic gases is quite small and can be safely neglected to the first order . we carry on the calculation to the second order and find the density and the pressure behave as @xmath120 @xmath121 where @xmath122 and @xmath123 have numerical values @xmath124 and @xmath125 respectively . the state constant to the second order evolves in time @xmath126 where @xmath127 equals @xmath128 , with temperature @xmath129 . thus we find the total modification induced by the self - gravitational effect in the eos to the second order acts as an additional term of the magnitude @xmath130 and increases over time as @xmath131 . though relatively small , we can see that such a term tends to decrease the pressure between non - relativistic particles as a manifestation of gravitational attraction . besides , the deviation grows as temperature cools down , and finally regains its power in the matter - dominated era . the whole analysis serves as a robust test to the justification of discarding self - gravity when solving the evolution of the non - relativistic nuclear particles during the radiation - dominated period . in general relativity , any kind of dynamic evolution , if inducing local energy fluctuation , can not escape the gravitational back - reaction , i.e. the change of space time geometry itself . from the principle of minimal coupling , such dramatic feature of general relativity indeed does not affect different kinds of interactions significantly enough in most cases . consequently , the theory of general relativity is poorly tested except in a weak - field limit within small regions like solar systems and radio pulsars , no bigger than galaxy scale . however , the friedman equation using robertson - walker metric from general relativity is widely utilized in solving the homogeneous and isotropic expansion of the universe . a test of general relativity in a large scale of cosmic size then proves crucial and necessary in order to justify the foundation of modern cosmology . one of some promising methods is to use particle element abundances @xcite to restrict expansion rate before bbn . the weak interaction before bbn between neutrons and protons should be frequent enough to counter expansion in maintaining equilibrium . accordingly , once the expansion speeds up , the weak interaction between neutrons and protons will be out of equilibrium at higher temperature , causing a larger neutron - to - proton ratio . in turn , the abundances of deuterium , helium-3 , and helium-4 are increased . knowing that helium-4 is hard to destroy during galaxy formation , observations of helium-4 abundance can help test the expansion rate and the cosmological constant as previous work suggested @xcite . modifications of the friedman equation in their work can be parameterized as @xmath132 , \label{modifiedfr}\ ] ] where @xmath133 corresponds to the modified state constant @xmath134 in our case , see eq . ( [ eq27++ ] ) and eq . ( [ w1 ] ) . in existing observation results , such as those by wmap @xcite , peimbert _ et al . _ @xcite , and steigman @xcite , the accuracy of the constraint of @xmath133 is of the magnitude @xmath135 , well beyond the self - gravitational effect in our result . thus , we can exclude self - gravity as the dominant mechanism behind any observable deviation in this method . finally , reexamining the derived formula eq . ( [ eq27++ ] ) for state constant , we compare it with the result in boson stars @xcite , and find that @xmath134 in both cases increases as the volume of the system expands . the merging similarities of the self - gravitational effect between two distinct circumstances are illuminating to understand the fundamental properties of gravity itself . in this paper , we try to evaluate the self - gravitational effect in bose and fermi gases at the early universe and set an exact bound on the magnitude of the modification . we study the bose and fermi gases before bbn synthesis in thermal equilibrium at a temperature in the vicinity of @xmath3 , when particles collide relativistically with each other frequently enough to maintain thermalization in spite of the significant cosmic expansion . we formulate new grand canonical partition functions based on the proof uhlenbeck and gropper made on non - ideal bose and fermi gases . taking advantage of the severe state of the early universe , we reduce the modified density and pressure expressions and derive the equations of state of relativistic and non - relativistic gases , discovering that self - gravity manifests itself in a simple way . in both non - relativistic and relativistic cases , we find the deviation from ideal ones to be small . this justifies the neglecting of the self - gravitational effect when dealing with non - relativistic particles . however , dynamical behavior appears when considering the second order effect . our result serves as a robust proof for neglecting self - gravity when solving the evolutions of relativistic bose and fermi gases as well as the thermalized non - relativistic fermi gases . additionally , we distinguish different effects of self - gravity on the relativistic and the non - relativistic particles . in thermalized relativistic particles , discrepancies between boson and fermion statistics become unapparent , and the self - gravitational effect of magnitude @xmath96 decreases with time . this is naturally expected as matter becomes more and more dominant in the energy content while the energy density of radiation undergoes a more rapid dilution . gravitational attraction consequently diminishes much faster in radiation components than in matter . while in non - relativistic fermi gases , the self - gravitational effect , though still very small of the magnitude @xmath130 , increases over time as the temperature cooling down dramatically and finally becomes significant in the matter - dominated era when both homogeneity and isotropy break down and the eos description is no longer appropriate . as to the evolution of the background metric , we find that self - gravity of any kind will not strongly affect the conventional result using eos of ideal relativistic and non - relativistic boson and fermion statistics . thus the previously not fully justified negligence of self - gravity is theoretically studied from a statistical perspective when dealing with cosmic evolution before bbn . however , our study is restricted to a given temperature range , and the same result may no longer be true at the very early stage of the universe as can be seen in eq . ( [ eq27++ ] ) . at a much earlier epoch , when the volume of the universe is small enough with the basic assumption of our statistical approach of thermal equilibrium still satisfied , self - gravity may no longer be negligible from our result . thus , further study should be carried out at much earlier stages to give a complete evaluation on the effects of self - gravity and then the back - reaction . we thank pisin chen for his comments on the issue at the early stages of this work . y.n . acknowledges enlightening discussions with alexei starobinsky . this work is supported by the national natural science foundation of china ( grant nos . 11021092 , 10975003 , and 11035003 ) , and the national fund for fostering talents of basic science ( grant nos . j0630311 , j0730316 ) . it is also supported by the education foundation for undergraduate research and hui - chun chin and tsung - dao lee chinese undergraduate research endowment ( chun - tsung endowment ) at peking university . p. jetzer , phys . rept . * 220 * , 163 ( 1992 ) . j. ho , s. j. kim , and b. h. lee , [ arxiv : gr - qc/9902040 ] . s. w. hawking , phys . * 26 * , 1344 ( 1971 ) . b. j. carr and s. w. hawking , mon . not . soc . * 168 * , 399 ( 1974 ) . n. bilic and h. nikolic , nucl . b * 590 * , 575 ( 2000 ) [ arxiv : gr - qc/0006065 ] . r. sharma , s. karmakar , and s. mukherjee , [ arxiv : gr - qc/0812.3470 ] . for a review , see h. j. de vega , and n. g. sanchez , phys . b * 490 * , 180 ( 2000 ) [ arxiv : hep - th/9903236 ] . g. e. uhlenbeck and l. gropper , phys . rev . * 41 * , 79 ( 1932 ) . s. weinberg , _ the first three minutes : a modern view of the origin of the universe _ ( basic books , new york , 1993 ) . s. f. daniel , r. r. caldwell , a. cooray , and a. melchiorri , phys . rev . d * 77 * , 103513 ( 2008 ) [ arxiv : astro - ph/0802.1068 ] . d. n. spergel _ et al . _ [ wmap collaboration ] , astrophys . j. suppl . * 170 * , 377 ( 2007 ) [ arxiv : astro - ph/0603449 ] . m. peimbert , v. luridiana , and a. peimbert , astrophys . j. * 666 * , 636 ( 2007 ) [ arxiv : astro - ph/0701580 ] . g. steigman , ann . nucl . part . * 57 * , 463 ( 2007 ) [ arxiv : astro - ph/0712.1100 ] .

we study the self - gravitational effect on the equation of state ( eos ) of bose and fermi gases in thermal equilibrium at the end of reheating , the period after quark - hadron transition and before big bang nucleosynthesis ( bbn ) . after introducing new grand canonical partition functions based on the work of uhlenbeck and gropper , we notice some interesting features of the newly developed eoss with distinct behaviors of relativistic and non - relativistic gases under self - gravity . the usual negligence of the self - gravitational effect when solving the background expansion of the early universe is justified with numerical results , showing the magnitude of the self - gravitational modification of the state constant to be less than @xmath0 . this helps us to clarify the background thermal evolution of the primordial patch . such clarification is crucial in testing gravity theories , evaluating inflation models and determining element abundances in bbn .

the theoretical results of atomic parity non - conservation ( pnc ) when combined with the experimental results is an important probe of physics beyond the standard model of particle physics @xcite . there are two sources of pnc in atoms , nuclear spin - independent ( nsi ) and nuclear spin - dependent ( nsd ) . the nsi - pnc is well studied and experimentally observed in several atoms . the most precise measurement till date is in the case of atomic cs @xcite . the same experiment also indicated a signature of nsd - pnc effects . the most dominant source of which is the nuclear anapole moment ( nam ) , a parity odd nuclear electromagnetic moment arising from parity violating interaction within the nucleus @xcite . however , there are two other contributions to nsd - pnc , these are the nsd electron - nucleus @xmath3 exchange interaction and the combined effect of hyperfine interaction and nsi electron - nucleus @xmath3 exchange interaction . the parameters describing nucleon - nucleon coupling , effect of nam is subsumed into it , extracted from the cs pnc experiment do not concur with the nuclear data @xcite . this certainly calls for the further investigation of the nsd - pnc effects in other atomic systems as well . an example of an alternative experiment is the proposal to measure the pnc in ba@xmath2 ion , suggested by fortson @xcite and is in progress at seattle @xcite . this experiment could lead to an unambiguous observation of nam in the @xmath4 transition , as the nsi - pnc alone does not contribute to this transition . it is important to note that the major difficulty to a clear observation of nam is the large nsi signal , which overwhelms the nsd signature . the ra@xmath2 ion has also been suggested and is considered to be an important candidate for the pnc measurement @xcite . apart from ba@xmath2 and ra@xmath2 ions which are one - valence systems the other promising candidate for pnc , the nam in particular , measurement is the atomic yb . an enhanced effect of pnc has already been reported @xcite in neutral yb , the @xmath5 transition , and for further refinement of the experiment is in progress at berkeley . the @xmath6 transition in yb@xmath2 , has also been suggested to reveal the nam signature and is being investigated at los alamos @xcite . the atomic theory results using reliable and accurate many - body methods are key to estimate the expected value of pnc transition amplitudes and extracting nam . for the theoretical calculations , the relativistic coupled - cluster ( rcc ) theory @xcite can be of great significance , as it is one of the most reliable many - body theory to incorporate electron correlation in atomic calculations . the rcc has been used extensively in atomic structure calculations @xcite of properties like transition energies , hyperfine structure constants , electromagnetic transition amplitudes , intrinsic electric dipole moment and pnc in atoms . apart from atomic physics , it has also been used with great success in nuclear @xcite , molecular @xcite and the condensed matter @xcite physics . in this work , we employ perturbed relativistic coupled - cluster ( prcc ) theory to calculate nsi and nsd - pnc amplitudes of the @xmath06s\;^2s_{1/2}-[4f^{14}]5d\;^2d_{3/2}$ ] transition in the case of @xmath1yb@xmath2 ion . this is timely as there are few theoretical results , sahoo _ et al _ @xcite and dzuba _ et al _ @xcite for nsi - pnc and dzuba _ et al _ @xcite and porsev _ et al _ @xcite for nsd - pnc are the previous works . the nsi - pnc results from ref . @xcite calculated using rcc method differ substantially from ref . @xcite where the correlation - potential - method with sum - over - state approach is employed to calculate nsi and nsd - pnc . the nsd - pnc results reported in ref . @xcite are based on rpa and , in general , is in agreement with the results reported in ref . however , the later is based on the sum - over - state approach , at the level of pnc matrix elements . the prcc method @xcite employed in present work is different from the sum - over - states approach . it accounts for the all singly and doubly excited intermediate states . there are two sets of the cluster amplitudes in the prcc , and the summation over states in the first order time - independent perturbation is incorporated in one set of the cluster amplitudes . the paper is organized as follows . in section . [ method ] , we provide a brief description of the theoretical methods . the unperturbed rcc equations for close - shell and one - valence systems are given to serve as a easy reference . the perturbed rcc is then discussed in detail and prcc equations are derived . the expression for e1pnc using prcc wave function and some leading order diagrams are also discussed . results from the work and uncertainty estimates are presented and discussed in section . [ results ] . in absence of pnc interaction the atomic states are of definite parity , and we consider these as the eigen states of the no - virtual - pair dirac - coulomb hamiltonian @xcite @xmath7 \nonumber \\ & & + \sum_{i < j}\frac{1}{r_{ij } } \lambda_+ , \label{dchamil}\end{aligned}\ ] ] where @xmath8 and @xmath9 are the dirac matrices , @xmath10 is the linear momentum , @xmath11 is the nuclear coulomb potential and the last term is the electron - electron coulomb interactions . the operator @xmath12 projects on the positive energy eigenstates to avoid the negative energy continuum solutions . the hamiltonian @xmath13 satisfies the eigen value equation @xmath14 where @xmath15 is the exact atomic state of the one - valence system and @xmath16 is the corresponding energy . here after , for compact notation , we use @xmath17 to represent @xmath13 . in the present work , we use rcc theory with the single and doubles ( ccsd ) excitation approximation to solve eq . ( [ hdc_eqn ] ) . in rcc , @xmath15 is expressed in terms of the closed - shell and one - valence cluster operators , @xmath18 and @xmath19 respectively , as @xmath20 |\phi_v\rangle , \label{psi_unptrb}\ ] ] where superscript @xmath21 represents the unperturbed rcc operators . the one - valence dirac - fock ( df ) reference state @xmath22 is obtained by adding an electron to the closed - shell reference state , @xmath23 . in the ccsd approximation , @xmath24 and @xmath25 . using the second quantized representation @xmath26 here , @xmath27 and @xmath28 are the cluster amplitudes . the indexes @xmath29 ( @xmath30 ) represent core ( virtual ) states and @xmath31 represent valence states . the operators @xmath32 ( @xmath33 ) and @xmath34 ( @xmath35 ) give single and double replacements after operating on the closed(open)-shell reference states . the diagrammatic representation of these operators are shown in fig . [ ts_fig ] . the open - shell cluster operators are then the solutions of nonlinear equations @xcite [ s0_eqn ] @xmath36 { \bar}{h}{_n}{s } \bar h_n s^{(0)}\ } |\phi_v\rangle & = & e_v^{\rm att}\langle\phi_v^p|s^{(0)}_1|\phi_v\rangle , \label{s01_eqn } \\ \langle \phi_{va}^{pq}|\bar h_n + \{\contraction[0.5ex ] { \bar}{h}{_n}{s}\bar h_n s^{(0)}\ } |\phi_v\rangle & = & e_v^{\rm att}\langle\phi_{va}^{pq}|s^{(0)}_2|\phi_v\rangle , \label{s02_eqn}\end{aligned}\ ] ] where @xmath37 is the similarity transformed hamiltonian , @xmath38 is the normal ordered hamiltonian and @xmath39 is the attachment energy of the valence electron . the operators @xmath18 are the solutions of a similar set of nonlinear coupled equations [ t0_eqn ] @xmath40 the details on the derivation of these equations are given in our previous work @xcite . in presence of pnc interaction atomic states mix with the opposite parity states and the total atomic hamiltonian is @xmath41 here , @xmath42 is the perturbation parameter and @xmath43 represents the any general pnc interaction hamiltonian . it has two components , the nsi and nsd interaction . these are @xmath44 where , @xmath45 is the fermi coupling constant , @xmath46 and @xmath47 are respectively the weak nuclear charge and the weak nuclear moment of the nucleus expressed in terms of neutron and proton numbers , @xmath48 and @xmath49 are the dirac matrices , @xmath50 is the normalized nuclear density and @xmath51 is the nuclear spin . compared to the nsi - pnc , the nsd - pnc require two important considerations because of the nuclear spin operator @xmath52 . first , the cluster operators in the electron space are rank one operators , and second , the atomic states in the one - valence sector are eigenstates of total angular momentum @xmath53 . similar to the unperturbed eigen value equation , eq . ( [ hdc_eqn ] ) , we may write the perturbed eigenvalue equation , satisfied by the total atomic hamiltonian , as @xmath54 where @xmath55 is the perturbed atomic state and @xmath56 is the corresponding energy . to the first - order in @xmath57 , @xmath58 and @xmath59 , where the bar in @xmath60 denotes it s parity is opposite to @xmath61 . from here on , to derive the prcc equations we consider the nsd - pnc interaction hamiltonian . using eq . ( [ hpnc_nsd ] ) , we can rewrite eq . ( [ ht_eqn ] ) as @xmath62 here , @xmath63 is the electronic part of @xmath64 . while writing above equation we have used @xmath65 , as @xmath64 is an odd parity operator it connects opposite parity states only . in the prcc theory , the perturbed wave function is expressed as @xmath66 \left [ 1 + s^{(0 ) } + \lambda \ps \cdot\mathbf{i } \right ] |\phi_v \rangle , \label{psi_ptrb}\ ] ] where @xmath67 and @xmath68 are the closed - shell and one - valence prcc operators , respectively . the superscript @xmath69 is used to indicate the perturbation . the diagrammatic representation of these cluster operators are shown in fig . [ pts_nsd_fig ] . using eq . ( [ psi_ptrb ] ) in eq . ( [ ht_elc_eqn ] ) , we can rewrite the eigenvalue equation as @xmath70 \left [ 1 + s^{(0 ) } \right . \nonumber \\ & & \left . + \lambda\ps\cdot\mathbf{i } \right ] |\phi_v \rangle = e_v e^{t^{(0 ) } } \left [ 1 + \lambda \pt\cdot\mathbf{i } \right ] \left [ 1 + s^{(0 ) } \right . \nonumber \\ & & \left . + \lambda \ps\cdot\mathbf{i } \right ] |\phi_v \rangle.\end{aligned}\ ] ] to derive the prcc equations , we project the above equation on @xmath71 and retain the terms linear in @xmath42 . in addition , for further simplification , we use normal - ordered form of the hamiltonian @xmath72 . after these sequence of operations , the eigenvalue equation is modified to @xmath73 |\phi_v \rangle \nonumber \\ = \left [ \delta e_v \ps + \delta e_v \pt ( 1 + s^{(0 ) } ) \right]|\phi_v \rangle , \label{deltae1v}\end{aligned}\ ] ] where @xmath74 , is the correlation energy of the one - valence system . like @xmath75 introduced earlier , @xmath76 is the similarity transformed nsd - pnc interaction hamiltonian in the electronic space . the prcc equations of @xmath68 can now be derived by projecting eq . ( [ deltae1v ] ) with the excited determinants @xmath77 and @xmath78 as @xmath79{}{h}{_{\rm n}}{s}\bar{h}_{\rm n } \mathbf{s}^{(1 ) } \ } + \ { \contraction{}{h}{_{\rm n}}{s}\bar{h}_{\rm n } \mathbf{t}^{(1 ) } \ } + \ { \contraction[0.5ex]{}{h}{_{\rm n}}{t } \contraction[0.8ex]{}{v}{_{\rm n}t^{(1)}}{s}\bar{h}_{\rm n } \mathbf{t}^{(1)}s^{(0)}\ } + \bar{\mathbf{h}}_{\rm elec}^{\rm nsd } + \ { \contraction[0.5ex]{}{h}{_{\rm elec}^{\rm nsd}}{s } \bar{\mathbf{h}}_{\rm elec}^{\rm nsd}{s}^{(0 ) } \}|\phi_v \rangle & = & e_v^{\rm att } \langle \phi^p_v | \pso|\phi_v \rangle , \\ \langle \phi^{pq}_{vb}|\ { \contraction[0.5ex]{}{h}{_{\rm n}}{s}\bar{h}_{\rm n } \mathbf{s}^{(1)}\}+\ { \contraction[0.5ex]{}{h}{_{\rm n}}{s}\bar{h}_{\rm n } \mathbf{t}^{(1 ) } \ } + \ { \contraction[0.5ex]{}{h}{_{\rm n}}{t } \contraction[0.8ex]{}{v}{_{\rm n}t^{(1)}}{s}\bar{h}_{\rm n } \mathbf{t}^{(1)}s^{(0)}\ } + \bar{\mathbf{h}}_{\rm elec}^{\rm nsd } + \ { \contraction[0.5ex]{}{h}{_{\rm elec}^{\rm nsd}}{s } \bar{\mathbf{h}}_{\rm elec}^{\rm nsd}{s}^{(0 ) } \}|\phi_v \rangle & = & e_v^{\rm att } \langle \phi^{pq}_{vb } | \pst|\phi_v \rangle . \label{ccsptrb1v2}\end{aligned}\ ] ] [ prcc_eqn ] while deriving the equations we have used the relations @xmath80 and @xmath81 . these follows as @xmath67 is an operator of closed - shell sector , it does not contribute to the prcc equation of @xmath82 and @xmath83 . the closed - shell operators @xmath67 are the solutions of the similar set of coupled equations @xcite [ pcceq ] @xmath84 these equations can be derived from the closed - shell perturbed eigenvalue equation . we can also derive a similar set of prcc equations for the nsi - pnc interaction hamiltonian . one major difference is , the cluster operators are rank zero operators . after solving the rcc and prcc equations , we can use the atomic states for the properties calculations . the rcc expressions and the diagrams contributing to the hyperfine structure ( hfs ) constants and the e1 transition amplitudes are derived and discussed in our previous work @xcite . in the present work , we use the same expressions and diagrams to compute hfs constants and e1 transition amplitudes . the pnc induced electric dipole transition amplitude , using prcc wave function , is @xmath85 where @xmath86 is the dipole operator . this expression , unlike the conventional sum - over - sates approach , implicitly account for all the possible intermediate states . from eq . ( [ psi_ptrb ] ) , for the nsd - pnc interaction , the transition amplitude is @xmath87^\dagger \left [ 1 + s^{(0 ) } + \lambda \ps\cdot\mathbf{i } \right]^\dagger \nonumber \\ & & \mathbf{d } e^{t^{(0 ) } } \left [ 1 + \lambda \pt\cdot\mathbf{i } \right ] \left [ 1 + s^{(0 ) } + \lambda \ps\cdot\mathbf{i}\right ] \red \phi_v \rangle.\end{aligned}\ ] ] consider terms linear in @xmath42 and retain only those up to second order in cluster amplitude . define the electronic component as @xmath88 , corresponding to the @xmath89 , it is then given as @xmath90 to calculate @xmath91 , we use diagrammatic analysis to identify the goldstone diagrams from these terms . however , we exclude the structural radiation diagrams , arising from the terms involving two - body cluster operators , for example , @xmath92 . the selected diagrams from the leading order and next to leading order terms are shown in the fig . [ e1pnc_cc_fig ] . for all the calculations we use gaussian type orbitals ( gtos ) or single particle wave functions with @xmath93 potential . as mentioned earlier , to incorporate the relativistic effects we use the dirac - coulomb atomic hamiltonian . for the nuclear potential we consider the finite size fermi density distribution @xmath94 here , @xmath95 . the parameter @xmath96 is the half - charge radius , that is @xmath97 and @xmath98 is the skin thickness . the orbitals are of the form @xmath99 where @xmath100 and @xmath101 are the large and small component radial wave functions , @xmath102 is the relativistic total angular momentum quantum number and @xmath103 are the spinor - spherical harmonics . the radial components are then defined as linear combination of gaussian type functions @xcite @xmath104 the index @xmath105 , where @xmath106 is the number of basis functions and @xmath107 are the coefficients of linear combination . for large component we choose @xmath108 where @xmath109 is an integer and @xmath110 is the normalization constant . the small component are derived from the large components using kinetic balance condition . the @xmath111 follow the general relation @xmath112 the parameters , @xmath113 and @xmath9 , are optimized such that the single particle energies of the core and valence orbitals are in good agreement with the numerical results , obtained from grasp92 @xcite . in table . [ grasp_e_tab ] , we compare the energy of the valence orbitals from the gto with the grasp92 data . .the valence orbital and scf energies of gaussian type orbitals ( gto ) are compared with the grasp92 data . [ cols="^,^,^",options="header " , ] [ nsi_orbital_tab ] cccccccc & + d & @xmath114 & e1pnc & state & d & @xmath115 & e1pnc & state + + & & & nsi - pnc & & & + @xmath116&@xmath117&@xmath118 & @xmath119&@xmath120 & @xmath121&@xmath122 & @xmath123 + @xmath124&@xmath125&@xmath126&@xmath127&@xmath128&@xmath129&@xmath130 & @xmath131 + @xmath132 & @xmath133 & @xmath134&@xmath135&@xmath136&@xmath137&@xmath130 & @xmath138 + @xmath139&@xmath140&@xmath141&@xmath142&@xmath143&@xmath144&@xmath145&@xmath146 + + & & & nsd - pnc & & & + @xmath116&@xmath147&@xmath148 & @xmath119&@xmath120 & @xmath149&@xmath122 & @xmath123 + @xmath124&@xmath150&@xmath126&@xmath127&@xmath128&@xmath151&@xmath130 & @xmath131 + @xmath132&@xmath152&@xmath134&@xmath135&@xmath136&@xmath153&@xmath130&@xmath138 + @xmath139&@xmath154&@xmath141&@xmath142&@xmath143&@xmath155&@xmath156&@xmath146 + [ nsi_orbital_df ] for the nsd - pnc , the dominant contributions from various prcc terms in eq . ( [ e1pnccc1v ] ) are listed in table . [ nsi_comp_tab ] . for hyperfine transitions @xmath157 and @xmath158 , like in nsi - pnc , @xmath159 is the leading order term . it contributes about 231% and -130% , respectively . for transition @xmath160 , however , @xmath161 is the dominant term , contributing about -123% of the total value . the same trend is reported in ref . @xcite , where the contributions are about 252% , -226% and -157% , respectively , for the @xmath157 , @xmath162 and @xmath158 transitions . the next leading order term , @xmath161 , contribute about -114% and 35% , respectively to the @xmath157 and @xmath158 transitions . however , the second leading order term is @xmath159 for the @xmath162 transition , it contributes about 69% . the next two leading order terms are @xmath163 and @xmath164 . the contributions from these terms , in the sequence listed in table . [ nsi_comp_tab ] , are 39% , 11% and -22% , and -24% , -59% and -4% , respectively . non - negligible contributions are also observed from the terms @xmath165 , @xmath166 and @xmath167 . unlike nsi - pnc , in nsd - pnc @xmath168 states also contribute to the e1pnc matrix element . in table . [ nsd_orbital_tab ] , we list dominant contributions from odd parity @xmath169 and @xmath168 states in the prcc calculations for the @xmath157 transition , and specifically the contributions from @xmath159 , @xmath161 and @xmath163 . at df level we present the contributions from the @xmath169 states in table . [ nsi_orbital_df ] . as we see in these tables , both at df and prcc levels , the dominant contribution is from the @xmath170 state . the total contribution from this in the prcc calculations is about 150% , which can be attributed to the 230% and -79% contributions from @xmath159 and @xmath161 , respectively . the large contribution through @xmath159 is due to the strong @xmath43 mixing with @xmath171 . however that is not the case with @xmath172 , which contributes through @xmath161 . at df level @xmath170 contributes only through mixing with @xmath171 , and contribution is about 214% . the state @xmath173 is the third most dominant contributing one , contributing about -40% , through @xmath43 perturbed @xmath172 . the other higher energy orbitals have negligible contribution . the nsd - e1pnc total results are given in table . [ e1pnc_tab ] . for comparison we have listed the df contributions . our df results @xmath174 , @xmath175 and @xmath176 for the three hyperfine transitions listed in table . [ e1pnc_tab ] compare well with the results @xmath177 , @xmath178 and @xmath179 , reported in ref . our total results , however , are on the higher side of the random - phase approximation ( rpa ) based results from ref . @xcite for all hyperfine transitions . the other theoretical nsd - pnc data available for comparison is from dzuba et al @xcite using the correlation - potential - method . our results for transitions @xmath157 and @xmath158 are in good agreement with their results . for transition @xmath162 , however , our result is higher than their value . ccccccccccc & & orbital & & orbital + d & @xmath180 & e1pnc & d & @xmath181 & e1pnc & & d & @xmath182 & e1pnc & + @xmath116 & @xmath183 & @xmath184 & @xmath185 & @xmath186 & @xmath187 & @xmath119 & @xmath120 & @xmath188 & @xmath189 & @xmath123 + @xmath124 & @xmath190 & @xmath191 & @xmath192 & @xmath193 & @xmath136 & @xmath127 & @xmath128 & @xmath194 & @xmath130 & @xmath131 + @xmath132 & @xmath195 & @xmath196 & @xmath197 & @xmath198 & @xmath130 & @xmath135 & @xmath136 & @xmath199 & @xmath130 & @xmath138 + @xmath139 & @xmath200 & @xmath201 & @xmath202 & @xmath203 & @xmath204 & @xmath142 & @xmath143 & @xmath205 & @xmath206 & @xmath146 + @xmath207 & @xmath208 & @xmath209 & @xmath210 & @xmath211 & @xmath212 & @xmath213 & & & & + @xmath214 & @xmath215 & @xmath216 & @xmath217 & @xmath218 & @xmath120 & @xmath219 & & & & + @xmath220 & @xmath221 & @xmath222 & @xmath223 & @xmath224 & @xmath204 & @xmath225 & & & & + + @xmath226 & @xmath227 & @xmath189 & @xmath228 & @xmath229 & @xmath230 & @xmath231 & @xmath130 & @xmath130 & @xmath189 & @xmath232 + @xmath233 & @xmath234 & @xmath189 & @xmath235 & @xmath236 & @xmath196 & @xmath237 & @xmath238 & @xmath239 & @xmath189 & @xmath240 + @xmath241 & @xmath242 & @xmath189 & @xmath243 & @xmath244 & @xmath141 & @xmath245 & @xmath246 & @xmath247 & @xmath189 & @xmath248 + @xmath249 & @xmath250 & @xmath189 & @xmath251 & @xmath252 & @xmath201 & @xmath253 & @xmath254 & @xmath255 & @xmath256 & @xmath257 + @xmath210 & @xmath258 & @xmath189 & @xmath259 & @xmath260 & @xmath261 & @xmath262 & & & & + @xmath263 & @xmath264 & @xmath189 & @xmath265 & @xmath266 & @xmath204 & @xmath267 & & & & + @xmath268 & @xmath269 & @xmath189 & @xmath270 & @xmath271 & @xmath189 & @xmath272 & & & & + [ nsd_orbital_tab ] cccc transition & & other works + & df & prcc & + nsi - pnc & & & + @xmath273 & @xmath274 & @xmath275 & @xmath276 + + nsd - pnc & & & + @xmath277 & @xmath174 & @xmath278 & @xmath279 + @xmath280 & @xmath175 & @xmath281 & @xmath282 + @xmath283 & @xmath176 & @xmath284 & @xmath285 @xmath286 reference@xcite . @xmath287 reference@xcite . @xmath288 reference@xcite . [ e1pnc_tab ] to calculate the uncertainty of our e1pnc results we resort to a analysis based on the sum - over - state approach . in this method , the net uncertainty associated with an intermediate state is @xmath289 where @xmath290 and @xmath291 are the deviations of excitation energy and e1 matrix element form the experimental data . and these are calculated based on results presented in tables . [ ee_tab ] and [ e1_tab ] , respectively . for @xmath292 , uncertainty associated with @xmath43 matrix , we resort to the deviation of @xmath293 from experimental data , where @xmath294 and @xmath295 represent the magnetic dipole hyperfine constants of initial and final states of the e1pnc transition . as discussed earlier , dominant contribution to nsi - pnc is from the @xmath170 state , which contributes through @xmath159 i.e. @xmath43 matrix element @xmath296 , @xmath297 matrix @xmath298 and energy denominator @xmath299 . the uncertainty associated with @xmath43 matrix element , we get using our rcc results for hyperfine constants , is 9.1% . and , the relative uncertainty of @xmath297 matrix element and the energy denominator are calculated as 5.7% and 3.4% , respectively . combining these , the net uncertainty in nsi - pnc result is 18.2% . for nsd - pnc as well @xmath170 is the dominant contributing state . however in this case unlike nsi - pnc , apart from @xmath159 , contribution through @xmath161 is not negligible . the matrix elements involve in this case are @xmath300 and @xmath301 , and the energy denominator is @xmath302 . using the similar analysis we get 4.52% , 11.26% and 2.41% , respectively for @xmath292 , @xmath291 and @xmath303 . combining these , we get 18.19% as the net uncertainty associated with term @xmath161 . the @xmath304 relative uncertainty in nsd - pnc results are then 18.17% . in this work , we present nsi and nsd - pnc transition amplitudes for the @xmath06s^2\;s_{1/2 } - [ 4f^{14}]5d^2\;d_{3/2}$ ] transition in @xmath1yb@xmath2 ion . to estimate the uncertainty of the the pnc results , we also calculate excitation energies , hyperfine structure constants and e1 transition amplitudes for some of the important low lying states , using rcc theory . the e1pnc results are computed using prcc theory , which is formulated based on rcc theory and it incorporates electron correlation effects arising from a class of diagrams to all order in the presence of pnc interaction as a perturbation . our results for excitation energies , hyperfine structure constants and e1 transition amplitudes are in good agreement , in some cases better , with the previous experimental data . our nsi - pnc result lies between the results from two previous studies reported in refs . @xcite and @xcite . the nsd - pnc df results from our work is in excellent agreement with the results reported in ref . @xcite for all hyperfine transitions . the total nsd - pnc result for the @xmath157 hyperfine transition lies between the results of refs . @xcite and @xcite . for the remaining two , our results are slightly on the higher side . the upper bound to the theoretical uncertainty associated with e1pnc results is about 20% . 36 i. b. khriplovich , _ parity nonconservation in atomic phenomena _ ( gordon and breach , new york , 1991 ) . c. s. wood _ et al . _ , science * 275 * , 1759 ( 1997 ) . v. v. flambaum and i. b. khriplovich , zh . fiz * 79 * , 1656 ( 1980 ) [ sov jetp * 52 * 835 ( 1980 ) ] . v. v. flambaum , i. b. khriplovich , and o. p. sushkov , phys . b * 146 * , 367 ( 1984 ) . y. zeldovich , jetp * 6 * , 1184 ( 1958 ) . w. c. haxton and c. e. wieman , annu . nucl . part . sci . * 51 * , 261 ( 2001 ) . n. fortson , phys . * 70 * , 2383 ( 1993 ) . j. a. sherman , t. w. koerber , a. markhotok , w. nagourney , and e. n. fortson , phys . lett . * 94 * , 243001 ( 2005 ) . j. a. sherman , a. andalkar , w. nagourney , and e. n. fortson , phys . a * 78 * , 052514 ( 2008 ) . l. w. wansbeek _ et al . a * 78 * , 050501(r ) ( 2008 ) . o. o. versolato _ et al . a * 82 * , 010501(r ) ( 2010 ) . k. tsigutkin , _ et al . lett . * 103 * , 071601 ( 2009 ) . k. tsigutkin , _ et al . a. * 81 * , 032114 ( 2010 ) . b. p. das , _ proceedings of the workshop on violations of fundamental symmetries in atoms and nuclei , int , seattle _ , 1999 ( unpublished ) . j. torgerson , private communication ( 2010 ) . f. coester , nucl . phys . * 7 * , 421 ( 1958 ) . f. coester and h. kmmel , nucl . phys . * 17 * , 477 ( 1960 ) . e. eliav , u. kaldor , and y. ishikawa , phys . rev . a * 53 * , 3050 ( 1996 ) . r. pal , _ et al . a * 75 * , 042515 ( 2007 ) . b. k. sahoo , l. w. wansbeek , k. jungmann , and r. g. e. timmermans , phys . a * 79 * , 052512 ( 2009 ) . h. s. nataraj , b. k. sahoo , b. p. das , and d. mukherjee , phys . lett . * 101 * , 033002 ( 2008 ) . r. pal , d. jiang , m. s. safronova , and u. i. safronova , phys . a * 79 * , 062505 ( 2009 ) . s. g. porsev , k. beloy , and a. derevianko , phys . d * 82 * , 036008 ( 2010 ) . g. hagen , t. papenbrock , d. j. dean , and m. hjorth - jensen , phys . * 101 * , 092502 ( 2008 ) . t. a. isaev , _ et al . a * 69 * , 030501(r ) ( 2004 ) . r. f. bishop , p. h. y. li , d. j. j. farnell , and c. e. campbell , phys . b * 79 * , 174405 ( 2009 ) . b. k. sahoo and b. p. das , phys . a * 84 * , 010502(r ) ( 2011 ) . v. a. dzuba and v. v. flambaum , phys . a * 83 * , 052513 ( 2011 ) . s. g. porsev , m. s. safronova , and m. g. kozlov phys . a * 86 * , 022504 ( 2012 ) . s. chattopadhyay , b. k. mani , and d. angom phys . a * 86 * , 022522 ( 2012 ) . k. v. p. latha , d. angom , b. p. das , and d. mukherjee , phys . lett . * 103 * , 083001 ( 2009 ) . b. k. mani and d. angom , arxiv:1104.3473v1 . j. sucher , phys . a * 22 * , 348 ( 1980 ) . b. k. mani and d. angom , phys . a * 81 * , 042514 ( 2010 ) . b. k. mani , k. v. p. latha , and d. angom , phys . a * 80 * , 062505 ( 2009 ) . a. k. mohanty and e. clementi , chem * 157 * , 348 ( 1989 ) . r. k. chaudhuri , p. k. panda , and b. p. das , phys . a * 59 * , 1187 ( 1999 ) . f. a. parpia , c. froese fischer , and i. p. grant , comp . . comm . * 94 * , 249 ( 1996 ) . p. pulay , chem . . lett . * 73 * , 393 ( 1980 ) . u. i. safronova and m. s. safronova , phys . a * 79 * , 022512 ( 2009 ) . a. m. martensson - pendrill , d. s. gough , and p. hannaford , phys . a * 49 * , 3351 ( 1994 ) . w. m. itano , phys . rev . a 73 , 022510 ( 2006 ) . r. w. berends and l. maleki , j. opt . b * 9 * , 332 ( 1992 ) . d. engelke and c. tamm , europhys . 33 , 347 ( 1996 ) . m. roberts _ et al . a 60 , 2867 ( 1999 ) . s. olmschenk _ et al . a * 76 * , 052314 ( 2007 ) . s. olmschenk _ et al . a * 80 * , 022502 ( 2009 ) . e. h. pinnington , g. rieger , and j. a. kernahan , phys . rev . a * 56 * , 2421 ( 1997 ) . , http://physics.nist.gov/physrefdata . v. a. dzuba , v. v. flambaum , p. g. silvestrov , and o. p. sushkov , j. phys . b * 20 * , 1399 ( 1987 ) .

we present nuclear spin - independent and dependent parity non - conservation amplitudes for the @xmath06s\;^2s_{1/2}-[4f^{14}]5d\;^2d_{3/2}$ ] transition of @xmath1yb@xmath2 , calculated using perturbed relativistic coupled - cluster theory . as a proxy to estimate theoretical uncertainty of these results we calculate the excitation energies , hyperfine structure constants and e1 transition amplitudes for the important low lying states . the pnc results presented in paper shall be useful in the propose pnc experiments .

one of the active areas of high energy physics research is the study of thermodynamic properties of interacting hadronic matter in the extreme conditions of temperature and/or density for the last few decades . at such high temperature and/or density the hadrons are expected to dissolve into their more fundamental constituents _ viz . _ quarks and gluons , forming a new state of matter called quark gluon plasma ( qgp ) @xcite . the basic theory that describes the strong interaction in terms of quarks and gluons is known as quantum chromodynamics ( qcd ) . it is known that qcd is asymptotically free _ i.e. _ strong interaction becomes weak for processes involving high transferred momenta @xcite . this means that the quarks and gluons appear to be weakly coupled at very short distances . at large separations , the effective coupling becomes progressively stronger resulting in the phenomenon called the confinement of color charges _ i.e. _ at low density and temperature , the quarks and gluons remain confined in color singlet hadrons that constitute hadronic or nuclear matter @xcite . however , when the density or the temperature are high enough , the quarks and gluons start to play dominant roles in determining the thermodynamic properties of the system . at extreme temperature and/or density one expects a transition of hadronic matter into a phase dominated by quarks and gluons , where color is deconfined and the interaction becomes screened @xcite . such a phase , where any length scale of interest is greater than the screening length of the interaction , is known as qgp . at vanishing chemical potential , lattice qcd suggests that this phase transition happens at a temperature around @xmath2 mev @xcite . at non - zero chemical potential , predictions on the order and exact location of the phase transition differ among the various lattice groups @xcite . this is because lattice quantities are not properly defined in such a case so that different groups investigated with different additional approximations . the possibility of creating high temperature qgp by colliding heavy ions in the laboratory and studying this phase of matter has been the goal of experiments at cern sps and at the relativistic heavy ion collider ( rhic ) facility at brookhaven national laboratory ( bnl ) @xcite . alice , atlas and cms collaborations at the large hadron collider ( lhc ) have provided further impetus to these studies @xcite . this experimental search of the qgp needs reliable theoretical estimates of various signals which depend on the pressure , entropy , deconfinement temperature and the equation of state ( eos ) etc . @xcite . in recent years , significant progress has been made to understand the behavior of qgp phase of matter leading to major advancement in the theoretical front addressing some of the subtle issues of the quasiparticles excitation in such an environment @xcite . one of the major developments , in this context , has been the hard thermal loop ( htl ) approximated perturbation theory where these issues are handled in a systematic manner and meaningful results are obtained after performing suitable resummations @xcite . these apart , many calculations have been performed to study the high temperature transport phenomenon of qgp including calculations of various transport coefficients like viscosity , conductivity etc . @xcite . several attempts have also been made to study the eos for interacting qcd matter @xcite . for example , new lattice results for the equation of state of qcd with @xmath3 dynamical flavors were obtained in @xcite . ref.@xcite deals with hybrid model in constructing a deconfining phase boundary between the hadron gas and the qgp and provides a realistic eos for the strongly interacting matter . furthermore , in @xcite the author have studied the thermodynamic properties of weakly interacting unpolarized qgp matter by using a mit bag model within one gluon exchange ( oge ) interaction . such investigations have also been performed in @xcite , where the calculations have been extended to spin polarized matter . being motivated by this series of works , we undertake the present investigation to study the thermodynamic behavior of weakly interacting spin asymmetric qgp matter including correlation corrections by using a mit bag model with non - zero chemical potential . due to asymptotic freedom , one may expect that the quarks and gluons interact weakly . thus , the properties of qcd might be computable perturbatively @xcite . so in our calculations it is assumed qcd coupling constant @xmath4 . it is to be mentioned , that for weak coupling constant @xmath5 , perturbation theory can only be worked out to a finite order . in the strong coupling limit , the perturbation theory fails and one has to resort to lattice results . however , in the present work we assume the coupling to be weak enough for the series expansion in terms of @xmath6 to converge . although the perturbative expansion converge very slowly , the approach makes predictions consistent , when comparison is possible , with lattice results which do not contain such an approximation @xcite . in our model for the computation of the thermodynamic quantities , we require the knowledge of total energy density of spin polarized matter with the inclusion of bubble diagrams @xcite , like what one does for the calculation of the correlation energy for degenerate electron gas @xcite . without such corrections , however , the calculations are known to remain incomplete as the higher order terms beyond the exchange diagrams are affected by infrared divergences due to the exchange of massless gluons @xcite . this indicates the failure of the naive perturbation series . we know that this problem can be cured by summing a class of diagrams that makes the perturbation series convergent and receives logarithmic corrections @xcite . in the present work , we calculate various thermodynamic quantities like energy density , pressure , entropy density with correlation corrections within the one gluon exchange model for non - zero chemical potential and show how these quantities can be expressed in terms of the spin polarization parameter @xmath7 and the temperature @xmath8 . we compare some of our results with the previous calculations , whenever possible . in addition , the present work is extended further to estimate the specific heat , spin susceptibility of qgp with a given fraction of spin - up and spin - down quarks . within our model , we estimate the critical temperature @xmath1 for the transition between the hadronic phase to the quark phase . furthermore , we determine the equation of state of qgp and its dependence on the temperature and the density . the plan of the article is as follows . in sec . ii , we calculate the various thermodynamic quantities with corrections up to @xmath0 . the results are also incorporated in this section . sec . is devoted to the summary and discussions . in this section we calculate the thermodynamic properties , like energy density @xmath9 , pressure @xmath10 , entropy density @xmath11 , free energy @xmath12 , specific heat @xmath13 , spin susceptibility @xmath14 for qgp matter with explicit spin dependent quarks by using a mit bag model with fixed bag pressure @xmath15 . the mit bag model considers free particles confined to a bounded region by the bag pressure . this pressure depends on the quark - quark interaction . for short , the bag constant ( @xmath15 ) is the difference between the energy densities of non - interacting and interacting quarks @xcite . within this model , for the calculation of energy density and other related quantities , we assume the qgp is composed of the light quarks only , i.e. the up and down quarks which interact weakly , and the gluons which are treated as almost free @xcite . we consider the color symmetric forward scattering amplitude of two quarks around the fermi surface by the one gluon exchange interaction . the direct term does not contribute as it involves the trace of single color matrices , which vanishes , while the leading contribution comes from the exchange term @xcite . we are dealing with quasiparticles whose spins are all eigenstates of the spin along a given direction , viz . @xmath16 . the quasiparticle interaction @xmath17 can be decomposed into two parts , spin may be either parallel @xmath18 or antiparallel @xmath19 corresponding to spin nonflip @xmath20 or spin flip @xmath21 scattering amplitudes @xcite , such that [ scat_amp ] f_pp^ss & = & f_pp^nf+f_pp^f , where @xmath22 and @xmath23 are the momentum of the quasiparticles . here , the factor @xmath24 is due to the equal scattering possibilities involving spin - up spin - down and spin - down spin - up quarks , i.e. @xmath25 , where super - scripts denote the spin indices . since spin and momentum have no preferred direction , we take the average over the angles @xmath26 and @xmath27 corresponding to spins @xmath28 and @xmath29 . the angular averaged interaction parameter is given by @xcite [ int01 ] f_pp^ss&= & , where @xmath5 is the coupling constant . in our calculations , it is assumed that the nuclear matter is the initial state with equal neutron and proton densities i.e. @xmath30 , so that @xmath31 , which means the contributions of @xmath32 and @xmath33 quarks are equal . since , we are dealing with ultra - relativistic massless quarks , all of the thermodynamic quantities , as we shall see , are obtained as a function of temperature @xmath34 and fermi momentum @xmath35 . although the fermi momentum is not a suitable experimental quantity , in fact it is better to calculate baryon density as a function of temperature and fermi momentum . since the baryon number of a quark is @xmath36 , we have baryon density @xmath37 and the quark density @xmath38 is given by [ qrk_den ] n_q = _ s= n_q^s & = & n_c n_f _ s = n_p^s(t ) + & = & _ s=. here , @xmath39 and @xmath40 are the color and flavor degeneracy factors , and @xmath41 is the fermi distribution function . using the above eq.([qrk_den ] ) , the fermi momentum as a function of @xmath42 and @xmath8 can be written as [ fermi_mom ] p_f(n_b , t ) & = & \{^2 n_b + } ^1/3 + & & + \{^2 n_b -}^1/3 . here , @xmath43 is the spin polarization parameter with @xmath44 . @xmath45 and @xmath46 correspond to the densities of spin - up and spin - down quarks , respectively . to calculate the total energy density of the qgp , we need to compute the energy densities for both quark and gluon separately , as we treat gluons as non - interacting . the leading contributions to the energy density of quarks are given by three terms viz . kinetic , exchange and correlation energy densities , i.e. e_q & = & e_kin+e_ex+e_corr . the total kinetic energy density for spin - up and spin - down quarks , including the color and flavor degeneracy factors for quarks , is [ kin_eng ] e_kin & = & \{p_f^4[(1+)^4/3+(1-)^4/3 ] + & & + 2 ^ 2t^2p_f^2[(1+)^2/3+(1-)^2/3 ] + ^4t^4 } , where , @xmath47 is the fermi momentum of the unpolarized matter @xmath48 . for spin asymmetric quarks , the exchange energy density consists of two terms @xmath49 , and can be determined by evaluating the following integrals e_ex^nf&=&_s= f_pp^nf n_p^s(t ) n_p^s(t ) , [ exeng_nf ] + e_ex^f&=&n_f n_c^2 f_pp^f n_p^s(t ) n_p^s(t ) . [ exeng_f ] the analytical expression for the total exchange energy density is found to be [ ex_eng ] e_ex&=&\{p_f^4[(1+)^4/3 + ( 1-)^4/3 + 2(1-^2)^2/3 ] + & & + ^2t^2p_f^2[(1+)^2/3+(1-)^2/3 ] + ^4t^4}. it might be noted that the kinetic and the exchange energy density at low temperature have been calculated in ref.@xcite , but here , the relevant quantities have been derived by retaining higher - order terms in @xmath8 . the next higher order correction to the energy density beyond the exchange term is the correlation energy . the detailed calculation of correlation energy for spin polarized matter have been given in @xcite which we quote here : [ cor_eng1 ] e_corr & & _ 0^/2 ^ 2_e d_e \{_l^2 + 2_t^2 } , with @xmath50 and @xmath51 is the gluon @xmath52-momentum . including the leading finite temperature corrections to the gluon self - energy , the relevant @xmath53 and @xmath54 are determined to be @xcite _ l & = & ^-2_e(1-_e _ e ) , [ pil ] + _ t & = & [ 1-^-2_e ( 1-_e _ e ) ] . [ pit ] these are then inserted in eq.([cor_eng1 ] ) and performing the @xmath55 integration , we estimate @xmath56 . the leading @xmath57 order contribution is given by [ cor_eng2 ] e_corr & = & \{p_f^4[(1+)^4/3 + ( 1-)^4/3 + 2(1-^2)^2/3 ] + & & + ^2t^2p_f^2[(1+)^2/3+(1-)^2/3 ] + ^4t^4}. it is to be noted that the exchange energy and the correlation energy for various @xmath7 are always positive for massless quarks @xcite . thus , the interaction between the quarks inside the bag is repulsive and it helps the interacting quarks and the gluons to go out from the bag very easily . therefore a transition to a qgp phase is more favored than in a non - interacting case @xcite . for the energy density of gluons we have [ eng_gluon ] e_g & = & 16 + & = & ^2t^4 , where @xmath58 is the degeneracy of gluons . since our system is ultra - relativistic , there is a simple relation between the pressure and the energy density : & = & e. to obtain the total energy density of the qgp , the bag pressure @xmath59 should be included in addition to the quark and gluon contributions @xcite , [ e_qgp ] e_qgp & = & e_q+e_g+b . the pressure of the system is determined to be [ p_qgp ] p_qgp & = & p_q+p_g - b . once the value of @xmath60 is determined , one can readily calculate the entropy density of the system by evaluating _ qgp & = & ( p_qgp ) . the thermodynamic properties of the system can be obtained by using the helmholtz free energy relation _ qgp & = & e_qgp - ts_qgp . for the numerical estimation of all these quantities , following refs.@xcite , we take @xmath61 , as the coupling constant of qcd and the bag pressure @xmath62 mev @xmath63 for zero hadronic pressure . 0.22 in in fig . [ freng_den ] , we plot the free energy per baryon as a function of baryon density for various order parameter @xmath7 at two different temperatures , @xmath64 mev and @xmath65 mev . this shows that the free energy is larger with higher value of @xmath7 . therefore qgp is more stable when quarks are unpolarized . similarly , in fig . [ freng_temp ] , we show the temperature dependence of the free energy with various @xmath7 at two different densities , @xmath66 @xmath63 and @xmath67 @xmath63 respectively . as expected the free energy decreases , as observed both in fig . [ freng_den ] and fig . [ freng_temp ] , with increasing temperature proving that qgp becomes more stable with unpolarized quarks . 0.2 in in fig . [ entrp_den ] , the variations of entropy per baryon in terms of baryon density have been plotted with different values of order parameter @xmath7 . the entropy per baryon is an important quantity connected to experimental observables of quark - gluon plasma formation in heavy ion collisions @xcite . we find that the entropy decreases with increasing baryon density and at low density , entropy is approximately the same for all the values of the order parameter at a fixed temperature . similarly , fig . [ entrp_temp ] shows the temperature dependence of the entropy per baryon for various @xmath7 . this shows that the entropy per baryon in the qgp is an increasing function of temperature @xcite and that the entropy decreases with increasing @xmath7 . the numerical estimates suggest that the entropy per baryon is continuous along the phase boundary as observed in the right panel of fig . [ entrp_temp ] , while for the left one the temperature reached in these computations is not high enough to make any conclusion . it should be mentioned that when the entropy per baryon varies continuously across the phase boundary , it is a smooth cross over . 0.2 in we have also studied the equation of state . the pressure is illustrated as a function of density in fig . [ eqn_st01 ] with @xmath68 mev ( left panel ) and @xmath69 mev ( right panel ) . the behavior observed for the pressure is driven by the fact that the interaction between the particles is attractive at large distance and repulsive at short distance . in the left panel of fig . [ eqn_st01 ] , the values of pressure and baryon density beyond the point ` a ' correspond to the qgp phase while the segment bc correspond to the hadronic phase . the segment ac denotes the unstable state . in the right panel , we observe that the unstable state shows a tendency to disappear as the temperature increases . we also see , the pressure increases by increasing the baryon density and for a fixed density the pressure of qgp is larger with polarized quark than the unpolarized one . therefore , the transition to the deconfined phase with polarized quark at lower density would be favored in comparison to the one with unpolarized quarks . 0.22 in similarly , in the left panel of fig . [ pres_coup ] , the variations of pressure of interacting qgp matter with baryon density for two different qcd coupling constants , and also for the noninteracting qgp at temperature @xmath64 mev has been shown . it is shown that at the same baryon density the pressure is larger for the interacting cases in comparison with the non - interacting one and if the interaction strength increases also the pressure raises further . thus the interaction makes easier the quarks transition to the deconfined phase at lower density and an increase of the interaction strength pushes the pressure of quarks to be similar to the bag pressure at smaller baryon density . therefore , the interaction of the qcd coupling constant reduces the value of the density to reach a transition @xcite . in the right panel of fig . [ pres_coup ] , the pressure of spin polarized qgp matter as a function of temperature for different @xmath7 has been plotted . we find that the pressure increases by increasing the temperature . it has been observed that at a constant temperature , the pressure of qgp is larger for polarized quarks than for unpolarized ones . these results indicate that the unpolarized state is energetically favorable for the qgp at any temperature and baryon density . the equation of state of spin asymmetric qgp matter becomes stiffer by increasing the baryon density ( temperature ) . it has to be noted that the pattern is qualitatively similar with that in ref.@xcite . 0.2 in the qgp phase diagrams are depicted in fig . [ ph_dia ] ; in the left panel for different @xmath7 at a constant bag pressure and in the right panel at different bag pressures for an unpolarized qgp . in the left panel , we see that the critical temperature is independent of the polarization parameter while critical density is different for unpolarized and polarized qgp . on the other hand , in the right panel it is shown that the critical temperature and density for the phase transition are different for different bag pressures and the critical values increase when increasing the bag pressure . this is because , as mentioned earlier , the interaction between the quarks is repulsive and helps the quarks to escape from the bags . this , in turn , causes the formation of qgp at smaller baryon densities and temperatures @xcite . in the right panel of fig . [ ph_dia ] , it is shown that the critical temperature for the deconfined phase transition lies between @xmath70 mev . the exact location of the phase boundary varies as results are obtained for non - zero chemical potential . when the bag pressure is about @xmath71 mev @xmath63 , the estimated results for critical temperature , i.e. @xmath72 mev , is consistent with the lattice results @xcite . it has to be mentioned that for our case , the critical parameters are determined from the phase boundary by using the condition that the bag pressure @xmath73 is independent of chemical potential @xmath74 and temperature @xmath34 , but in realistic cases @xmath15 may depend on both . the specific heat @xmath75 at constant volume is defined as the quantity of energy needed to increase the temperature of a system by one unit , it reads as @xcite c_v & = & ( ) _ v + & = & 3p_f^2 t + & & + ^2 t^3 . 0.2 in the behavior of specific heat is plotted in fig . [ sp_heat ] at two different densities . we see that the specific heat decreases by increasing the baryon density and polarization parameter . this is because the specific heat is a measure of energy fluctuation with temperature and the fluctuation is smaller when qgp is polarized . we also observe that the specific heat indicates complete continuity in its behavior . the spin susceptibility can be determined by the change in energy of the system as quarks spin are polarized . in the small @xmath7 limit , the energy density behaves like @xcite [ xi_expan ] e_qgp()&= & e_qgp(=0)+_s^2+o(^4 ) , where @xmath76 is the spin stiffness constant . the spin susceptibility @xmath77 is inversely proportional to the spin stiffness ; mathematically @xmath78 @xcite . since only quarks are polarized and the energy density involves three leading terms , the spin susceptibility can be written as @xcite [ chi_inv ] ^-1 & = & ^-1_kin + ^-1_ex + ^-1_corr . with the help of eq.([xi_expan ] ) , each energy contribution to the susceptibility is [ sus_cont ] _ kin^-1 & = & ( 1- ) , + _ ex^-1 & = & -(1- ) , + _ corr^-1 & = & -(1 + ) . using eq.([chi_inv ] ) and eq.([sus_cont ] ) , the sum of all the contribution to the susceptibility is given by [ spin_sus ] & = & _ p^-1 . where @xmath79 is the pauli susceptibility @xcite . the value of @xmath77 can be estimated if @xmath5 , @xmath8 and @xmath47 are exactly known . the numerical estimation of @xmath80 per baryon is given in fig . [ chi ] for two different densities , @xmath66 @xmath63 and @xmath67 @xmath63 , respectively . we see susceptibilities blow up at different temperatures for two different densities . this can be identified as a physical instability of the qgp matter towards an ordered phase @xcite . in this work , we have investigated the thermodynamic properties of the qgp composed of the massless spin - up and spin - down quarks at finite temperature and density using a mit bag model within one gluon exchange ( oge ) interaction . accordingly , we calculated the total free energy , entropy density , pressure as a function of baryon density and temperature for non - zero chemical potential of such a system up to @xmath81 that includes correction due to correlation effects . it was shown that the free energy increases by increasing the polarization parameter . this fact may suggests that the qgp with unpolarized quark is energetically favorable . we found that the entropy is a decreasing function of the density and an increasing function of the temperature . in the present phenomenological model , the equation of state ( eos ) of the qgp with different order parameters has been computed . it was shown that the increase of temperature or density makes the eos stiffer . we obtained a phase boundary for quark - hadron phase transition by making the bag constant independent of @xmath82 and @xmath8 . the values of critical parameters have been estimated for the transition from the hadronic phase to the qgp phase by using the eos . it has been observed that the entropy per baryon is continuous along the phase boundary , indicating a cross over from the hadronic phase to the qgp phase . moreover , we reported how the oge interaction for the massless quarks affects the phase transition of the qgp and causes the system to reach the deconfined phase at smaller baryon densities and temperatures . in addition , the specific heat and the spin susceptibility of qgp have also been calculated and the specific heat decreases by increasing the baryon density in a continuous way . on the other hand , the spin susceptibility blows up at certain temperature for a definite baryon density . this may indicate the physical instability of the ordered phase of qgp . it has to be mentioned that our results depend on the values of the bag pressure while a change in the value of the qcd coupling does not have any dramatic effect in our calculations . moreover , an increase of the bag pressure might improve our results towards the lattice qcd calculations . within the scope of the present model , the value of @xmath1 obtained here , is close to the lattice qcd prediction and the inclusion of multi - gluon exchange processes may strengthen this conclusion . more works in this direction are therefore necessary to examine this issue , especially for multi - flavor systems . leaving aside these questions , our studies might be meaningful for determining the approximate values of temperature , baryon density , specific heat etc . regarding the signals of qgp and their detection for future studies in the collider experiments . 0.2 in * acknowledgments * + i am indebted to late prof . dutt - mazumder who introduced me to this topic and his fruitful discussions motivated me to initiate the present work . i would also like to thank prof . jane alam and prof . pradip k. roy for their critical reading of the manuscript . hwa , _ quark gluon plasma - vol-1 _ , ( world scientific ) ( 1990 ) . hwa , _ quark gluon plasma - vol-2 _ , ( world scientific ) ( 1995 ) . hwa et al . , _ quark gluon plasma - vol-3 _ , ( world scientific ) ( 2004 ) . d. gross and f. wilczek , phys . lett . * 30 * , 1343 ( 1973 ) . d. gross and f. wilczek , phys . d * 8 * , 3633 ( 1973 ) . politzer , phys . lett . * 30 * , 1346 ( 1973 ) . g.t hooft , nucl . b * 138 * , 1 ( 1978 ) . g.t hooft , nucl . b * 190 * , 455 ( 1981 ) . wilson , phys . d * 10 * , 2445 ( 1974 ) . collins and m.j . perry , phys . lett . * 34 * , 1353 ( 1973 ) . m. kisslinger and p. morley , phys . d * 10 * , 2765 ( 1976 ) . shuryak , phys . rept . * 61 * , 71 ( 1980 ) . e.v . shuryak , phys . rept . * 115 * , 151 ( 1984 ) . alford , a. schmitt , k. rajagopal and t. schfer , rev . mod . phys . * 80 * , 1455 ( 2008 ) . a. schmitt , arxiv : nucl - th/0405076 . blaizot , arxiv : 0911.5059 . y. aoki _ et al . _ , jhep * 0906 * , 088 ( 2009 ) . et al . _ , d * 74 * , 054507 ( 2006 ) . z. fodor and s.d . katz , phys . b * 534 * , 87 ( 2002 ) . z. fodor , s.d . katz and k.k . szabo , phys . b * 568 * , 73 ( 2003 ) . m. kitazawa , t. koide , t. kunihiro and y. nemoto , phys . d * 65 * , 091504 ( 2002 ) . h. abuki , t. hatsuda and k. itakura , phys . d * 65 * , 074014 ( 2002 ) . y. nishida and h. abuki , phys . d * 72 * , 096004 ( 2005 ) . na50 collaboration , phys . b * 477 * , 28 ( 2000 ) . star collaboration , nucl . a * 566 * , 277c ( 1994 ) . s. nagamiya , nucl . a * 566 * , 287c ( 1994 ) . alice collaboration , jinst * 3 * , s08002 , ( 2008 ) . atlas collaboration , phys . . lett . * 105 * , 252303 ( 2010 ) . cms collaboration , phys . c * 84 * , 024906 ( 2011 ) . s. plumari , w.m . alberico , v. greco , c. ratti , phys . d * 84 * , 094004 ( 2011 ) . m. modarres and a. mohamadnejad , phys . nucl . lett . * 10 * , 99 ( 2013 ) . m. modarres and h. gholizade , int . e * 17 * , 1335 ( 2008 ) . f.g . gardim and f.m . steffens , nucl . a * 797 * , 50 ( 2007 ) . bordbar , f. kayanikhoo and h. bahri , iranian . a * 37 * , 165 ( 2013 ) . a. rebhan , nucl . a * 702 * , 111 ( 2002 ) . srivastava and c.p . singh , phys . d * 85 * , 114016 ( 2012 ) . blaizot and e. iancu , phys . rept . * 359 * , 355 ( 2002 ) . p. kovtun , d.t . son and a.o . starients , phys . * 94 * , 111601 ( 2005 ) . d * 87 * , 036002 ( 2013 ) . . lett . * 111 * , 172001 ( 2013 ) . s. borsanyi _ et al . _ , jhep * 1011 * , 077 ( 2010 ) . s. borsanyi _ et al . _ , nucl . a * 855 * , 253 ( 2011 ) . k. kajantie , m. laine , k. rummukainen and y. schroder , phys . d * 67 * , 105008 ( 2003 ) . k. pal , s. biswas and a.k . dutt - mazumder , phys . c * 80 * , 024903 ( 2009 ) . m. gell - mann and k.a . brueckner , phys . rev . * 106 * , 364 ( 1957 ) . p. nozires and d. pines , phys . rev . * 111 * , 442 ( 1958 ) . a. chodos _ d * 9 * , 3471 ( 1974 ) . k. pal , s. biswas and a.k . dutt - mazumder , phys . c * 79 * , 015205 ( 2009 ) . t. tatsumi , phys . b * 489 * , 280 ( 2000 ) . k. pal and a.k . dutt - mazumder , phys . c * 80 * , 054911 ( 2009 ) . a. gerhold , arxiv : hep - ph/0503279 . blaizot and e. iancu , phys . rept . * 359 * , 355 ( 2002 ) . a. leonidov , k. redlich , h. satz , e. suhonen and g. weber , phys . d * 50 * , 4657 ( 1994 ) . bannur , int . e * 21 * , 1250090 ( 2012 ) . k. yagi , t. hatsuda and y. miake , quark gluon plasma , ( cambridge university press ) ( 2005 ) . g. boyd , j. engels , f. karsch , e. laermann , c. legeland , m. lutgemeier and b. petersson , phys . lett . * 75 * , 4169 ( 1995 ) . et al . _ , b * 469 * , 419 ( 1996 ) . f. karsch , nucl . a * 698 * , 199 ( 2002 ) . e. laermann and o. philipsen , ann . nucl . part . sci . * 53 * , 163 ( 2003 ) . f. karsch , e. laermann and a. peikert , phys . b * 478 * , 447 ( 2000 ) . landau and e.m . lifschitz , _ course of theoretical physics , _ , vol-5 , ( pergamon press , london , 1958 ) . f. perez , phys . b * 79 * , 045306 ( 2009 ) . shastry , phys . * 38 * , 449 ( 1977 ) . shastry , phys . rev . b * 17 * , 385 ( 1978 ) . k.a . brueckner and k. sawada , phys . rev . * 112 * , 328 ( 1958 ) . r.r.p . singh and d.a . huse , phys . b * 40 * , 7247 ( 1989 ) .

we calculate the free energy , entropy and pressure of the quark gluon plasma ( qgp ) at finite temperature and density with a given fraction of spin - up and spin - down quarks using a mit bag model with corrections up to @xmath0 . the expressions for the specific heat and the spin susceptibility are derived in terms of fermi momentum and temperature . the effects of interaction between the quarks on the properties of the qgp phase are also investigated . within our phenomenological model , we estimate the transition temperature @xmath1 by constructing the phase boundary between the hadronic phase and the qgp phase . finally , we compute the equation of state of the qgp and its dependence on the temperature and the density .

the use of symmetry methods to understand bound states has an ancient origin and has given many of its fruits in atomic and subatomic physics - the work of marcos moshinsky is an excellent example of this @xcite . over the years , however , other approaches to understand more complicated systems have given up symmetry in both classical and quantum mechanics @xcite , @xcite . here we present one of the simplest ways of destroying continuous symmetries with purely quantum - mechanical consequences : a corner . we will show that a sharply bent waveguide supports a finite number of bound states whose nature has no classical counterpart . due to the geometry , the traditional methods of integrability can not be used to obtain the corresponding solutions . this type of systems has been studied in the past , either with mathematical tools @xcite proving the existence of bound states or by employing numerical methods for computing the corresponding energies and eigenfunctions @xcite . the approach in our considerations will be completely analytical , albeit we use approximations to perform computations . we want to show two different approaches to the problem in order to exhibit an apparent complexity of the description given in other studies , mostly numerical @xcite . first we shall deal with a convenient basis of states which shows the binding as a consequence of the presence of corners : a sharp obstacle diffracts waves of an arbitrary wavelength . some limitations of this method , such as the coupling of an infinite number of modes , will be pointed out . finally , the use of conformal coordinates will be introduced as a natural description to find all the bound states for arbitrary bending angles . we would like to emphasize that interesting applications may come in the form of bent wave guides carved in crystals , as suggested in @xcite , using methods connected to the evolution of airy profiles - a recent example can be found in @xcite . the landscape looks promising , considering the many applications of controlling , bending and focusing waves . we show how to obtain a one dimensional schroedinger equation with an effective potential along the longitudinal direction of the guide . one of the challenges for describing our system is to find a suitable set of states . a natural way to introduce box eigenmodes in our system is by using oblique modes , as shown in the figure . . the slope @xmath0 is related to the bending angle @xmath1 . we show an oblique mode with @xmath2 [ fig1 ] let us take units such that @xmath3 with @xmath4 the mass of the particle and write the stationary schroedinger equation in the form ^2_x , y + e = 0 [ 1 ] for the configuration in figure [ fig1 ] with dirichlet conditions , we express the solution in the oblique basis as ( x , y ) = _ n=1^ _ n(x ) ( ( y - a|x|+l ) ) [ 2 ] and @xmath5 for @xmath6 . the matrix elements of the schroedinger operator in this basis can be obtained by using the expansion ( [ 1 ] ) and integrating over the @xmath7 variable . let us denote the vector wave function by @xmath8 . using the following definition for matrix elements _ nm = _ nm n , _ mn = ( 1-(-)^m+n ) [ 7 ] and employing a gauge transformation given by \(x ) = ( - ) ( x ) + ( x ) = ( - ) ( ) [ 8 ] we reduce the schroedinger equation to ( x)=0 [ 9 ] where we see an effective potential of the form v_eff = ( ) ^2 ( a^2 + 1 ) ( x)^2 + ( ) ^2 ^2 . [ 10 ] this matrix couples , in principle , all components of @xmath9 . the origin of @xmath10 and @xmath11 is related to the momentum and the energy of a particle in a box . the @xmath12 dependence of the potential @xmath13 comes completely from the non - commutability of @xmath14 and @xmath10 and is non - differentiable at the corner due to @xmath15 . another important feature is that a change of units @xmath16 in equation ( [ 9 ] ) leads to a scale - free equation , thus proving that the energy @xmath17 must scale as @xmath18 . therefore , the presence of bound states in this problem is independent of the scale . this is in contrast with other bent systems with smooth boundaries @xcite . in previous works by wheeler , bestle and schleich @xcite , the wkb approximation was corrected in the presence of potentials with corners . as a result , the schroedinger equation with a potential could be approximated by a wkb equation with a source term . by neglecting completely the effect of the potential except at the non - differentiable point , the wkb equation with a source could be interpreted as a free schroedinger equation with a @xmath19 potential . in our case , this would cause a binding effect depending on the energy - dependent amplitude of the @xmath19 . we shall apply this idea in order to cope with @xmath13 obtained in the last section , with the additional feature that @xmath13 is a matrix potential . we start with the schroedinger equation + ^-2(x ) ( x ) = 0 [ 1.1 ] where @xmath20 is now a matrix . the corner - corrected wkb formalism given in @xcite can be easily generalized to matrix potentials . a wkb scattering equation ( such as eq ( 13 ) in @xcite ) can be obtained for ( [ 1.1 ] ) . the resulting equation , in turn , can be treated by keeping first order terms in @xmath0 and considering a negligible potential except at @xmath21 . this gives the isolated effect of the corner in the form + ( e- ^2 ) ( x ) = _ sc ( x ) + _ sc - [ ^2 , ] ( e- ^2 ) ^-1 ( x ) . [ 1.12 ] the energies for bound states in this problem are obtained by i ) imposing the appropriate boundary conditions @xmath22 at @xmath23 , ii ) substituting the corresponding solutions in the condition for the jump in the derivative ( due to @xmath19 ) , namely _ 0 + - _ 0- = - [ ^2 , ] ( e- ^2 ) ^-1 _ 0 . [ 1.13 ] the solutions are found to be ( x)= ( ^2 - e ) ^-1/4 ( - |x| ) _ 0 [ 1.14 ] with @xmath24 a normalization factor . the vector @xmath25 is determined by the condition coming from the jump in the derivative ( [ 1.13 ] ) , namely ^-1 _ 0 = 0 [ 1.15 ] this is a linear equation for @xmath25 and the wave function can be obtained by choosing @xmath25 as a member of the null space of the operator in ( [ 1.15 ] ) . the scaled energies @xmath26 are obtained from ( [ 1.15 ] ) by using the determinant ( [ ^2 , ] + 2 ^3/2 ) = 0 [ 1.16 ] the @xmath27 case can be solved easily by inserting in ( [ 1.16 ] ) the diagonal matrix @xmath28 . the resulting energy for the ground state is e_0 = ( ) ^2 ( - ) [ 1.18 ] this energy is below the threshold @xmath29 . the level @xmath30 can also be obtained ; it is above @xmath31 . therefore , in this @xmath27 approximation , the only state which decays exponentially with the distance is the ground state . the wave function can be found straightforwardly once @xmath32 is obtained . we show two cases in figures [ fig2 ] [ fig2bis ] . the solutions thus obtained show that a bound state develops due to the corner alone , producing a @xmath19 potential and coupling the oblique modes - such coupling is essential , since a @xmath33 approximation makes the interaction vanish . yet , we may ask : where are the excited states below threshold due to an increasing value of @xmath0 ? in the following , we analyze this problem using conformal coordinates as a way to obtain a suitable basis . such a basis can not be obtained by means of perturbation theory in @xmath0 using the solutions of this section . . a peak in the relevant region is visible . the decay with distance is exponential , title="fig : " ] . a peak in the relevant region is visible . the decay with distance is exponential , title="fig : " ] . the peak becomes more pronounced as @xmath0 increases . for large values of @xmath0 , the description is no longer valid.,title="fig : " ] . the peak becomes more pronounced as @xmath0 increases . for large values of @xmath0 , the description is no longer valid.,title="fig : " ] our aim now is to show that a non - trivial boundary value problem such as the bent waveguide can be mapped smoothly to a straight guide with an effective potential ( equivalently , a problem with position - dependent mass ) . the simplest way to describe this process is by using complex variables . denote @xmath34 , @xmath35 and consider a transformation such that @xmath36 . we impose the following conditions on @xmath37 : \1 ) the image of the bent waveguide is an infinite straight strip described by @xmath38 . \2 ) the laplacian @xmath39 is transformed into a quadratic form without cross - terms ( the matrix @xmath10 will not appear ) . \3 ) the transformation is smooth except at the points @xmath40 ( upper corner ) and @xmath41 ( lower corner ) in figure [ map2 ] . the condition 1 ) allows the use of a complete set of functions @xmath42 for the dirichlet problem and @xmath43 for the neumann problem , given that the contour lines meet the boundaries orthogonally . the condition 2 ) precludes the appearance of terms proportional to @xmath44 : the integral @xmath45 would produce unbounded matrix elements due to well - known properties of the momentum operator for a particle in a one - dimensional box . requirement 3 ) ensures that any suspicious behavior of the transformation is due to the corners . . we denote the lower corner by @xmath41 and place the upper corner at the origin @xmath40 , see the text . ] the requirements described above are fulfilled by choosing @xmath37 analytic for all the points in the interior of the waveguide . orthogonality is reached by the cauchy - riemann conditions . we introduce the operators _ z = _ x - i _ y , _ z^ * = _ x + i _ y [ 2.1 ] _ = _ u - i _ v , _ ^ * = _ u + i _ v [ 2.2 ] obtaining thus @xmath46 , @xmath47 . using the analytic function @xmath36 , the laplacian operator is transformed as @xmath48 . using the cauchy - riemann conditions we have @xmath49 , the jacobian of the transformation as a function of @xmath50 . our boundary value problem with the schroedinger equation becomes ( _ _ ^ * + ||^2 e ) ( u , v ) = 0 , u , v * r [ 2.5 ] we give an explicit example of our conformal map in the following . consider the schwarz - christoffel ( sc ) transformation @xcite of the upper half - plane onto the semi - infinite strip . by using a composition of a sc map for a trigon with internal angles @xmath51 together with a sc map for a trigon with internal angles @xmath52 and @xmath53 , we arrive at the bijective transformation between one half of the bent waveguide and a semi - infinite straight strip . the expression results in * = f(z ) = _ 0^^2(z/2 ) dx ( 1-x)^- x^- = _ 0^z d((/2))^q + = ( ^2(z/2 ) , , ) [ 2.9 ] where we have used the definition of the incomplete beta function @xcite and we have defined @xmath54 as the complementary angle of @xmath55 in @xmath56 radians . therefore @xmath57 . in passing we note that @xmath58 . the jacobian is given by @xmath59 and one can approximate it in terms of @xmath60 ( or @xmath61 ) for some cases of interest : @xmath62 . it is important to note that as one moves away from the corner in the direction of the arms , the jacobian ( [ 2.14 ] ) becomes a hyperbolic cotangent of a real variable and approches to unity exponentially fast . therefore , interactions vanish for points far from the corners . in connection with the existence of other maps , we stress that any other transformation constructed in this way will have the same behavior near the corners : the fractional power law of the map is completely determined by the angle of the guides and therefore by our physical system . one can try many methods of solution for the transformed boundary value problem given above . here we give an example based on the fact that our careful construction allows certain approximations . let us reduce ( [ 2.5 ] ) to a one - dimensional problem by defining @xmath63 and v_nm(v ) n | ||^2 | m = _ 0^f(c ) du ( ) ( ) ||^2 [ 2.6 ] the set @xmath64 is obviously orthonormal . after multiplying ( [ 2.5 ] ) by our complete set of functions and performing the corresponding integrals over @xmath65 , we find that ( [ 2.5 ] ) is equivalent to _ m ( - _ nm _ v^2 - e v_nm(v ) ) _ m(v ) = _ n _ n(v ) [ 2.8 ] near the origin , our potential ( [ 2.6 ] ) becomes v_nm(v)= _ 0^f(c ) du ( ) ( ) ( u^2 + v^2 ) ^q/(q+1 ) [ 2.15 ] it is justified to approximate the potential by its diagonal entries @xmath66 , since @xmath67 decreases as we move away from the diagonal for fixed @xmath68 . this can be seen by noting that for @xmath69 the integrand can take negative values and oscillates rapidly as @xmath70 increases . we have ( -^2_v -e v_n(v ) ) _ n(v ) = _ n _ n(v ) . [ 2.16 ] we include some plots of @xmath66 and its derivative for different angles , figures [ pot ] , [ slope ] . , @xmath71 increases from @xmath72 to @xmath73 in @xmath74 steps . the transversal mode energy is shown at @xmath75 . ] . near the origin we see an abrupt change ( a discontinuity appears in the limit @xmath76 ) ] the potentials shown in the figure have zero derivative at @xmath77 ( except for @xmath76 ) but , apart from that , their behavior is that of a shallow potential well whose derivative suffers an abrupt change in a small region near the corner . in the plot showing the slopes , we see that there is almost a jump in @xmath78 around @xmath77 . the formalism developed by wheeler , schleich and bestle @xcite in order to deal with kinks should be helpful here again . although the only case in which the derivative is strictly discontinuous corresponds to @xmath76 , the strong variation of @xmath79 occurs in a region which looks effectively like a corner for long wavelengths ( we expect this approximation to work at low energies ) . as the simplest approximation , let us consider the corner - corrected wkb formula for quantization of bound states , namely s = _ 0^v_0 dv , [ 2.17 ] 4s=2(k+)- 4 ( ) , [ 2.18 ] = e ( v_n(0 + ) - v_n(0- ) ) ( 2(_n+e v_n(0 ) ) ) ^-3/2 , [ 2.19 ] remarkably , one can obtain good results by neglecting the classical orbit @xmath80 and keeping the corner effect alone . using ( [ 2.17 ] ) , the approximation implies = ( ) = 1 , k even [ 2.20 ] finding thereby that all even states obey this relation , but the excited states are not supported by such a weak potential . the excitations come from the solutions of ( [ 2.20 ] ) by varying @xmath81 and not @xmath82 . this results in a transversal node structure , instead of longitudinal . with these considerations , the solution to our problem ( [ 2.20 ] ) requires @xmath83 and @xmath84 as the only input . such quantities can be estimated by means of ( [ 2.15 ] ) for @xmath85 . as mentioned before , we replace the jump in the derivative @xmath86 by twice the maximum of @xmath78 around the origin ( see the figure ) . we denote such quantity as @xmath87 . squaring ( [ 2.20 ] ) we obtain a cubic relation for @xmath17 : ( n^2 ^2 ( q/2 ) - e v_n(0 ) ) ^3 = e^2 ( v_max)^2 [ 2.22 ] one can estimate @xmath83 and @xmath87 as v_n(0 ) ( ( q/2 ) ) ^2q/(q+1 ) , v_max q ( ( q/2 ) ) ^(2q+1)/(q+1 ) . [ 2.24 ] with this , the energies in ( [ 2.22 ] ) can be expressed in terms of @xmath81 and @xmath71 using cardano s formula . we present the results graphically in figure [ final ] . as a consequence of the corner . an unlimited number of bound states appear as @xmath71 increases . ] and @xmath88 ( ground state energy @xmath89 ) . the peak of the distribution is located near the upper corner , as expected . this is achieved with only one transversal mode in conformal coordinates ] the agreement of the estimated solutions with previous numerical results @xcite is qualitatively good , considering that the action has been completely neglected in favor of the phase shift . also , the off - diagonal elements of the matrix potential ( [ 2.15 ] ) have been ignored . the trend of the curves as functions of the bending angle is comparable with the numerical results obtained in @xcite for several modes : the energies decrease and more states appear below threshold as @xmath90 . the node structure of our wavefunctions is also compatible with the solutions obtained in @xcite both numerically and experimentally . remarkably , the fractional power behavior of the jacobian alone is responsible for the structure of the spectrum of this system . we include a plot of the resulting wave density for the ground state of an l - shaped guide . in what touches the corrections , we expect the following : the action appears to be important as @xmath71 increases ( deeper wells ) . therefore ( [ 2.22 ] ) receives corrections which increase the energy . as a consequence , we expect a shift of the critical angles towards the value @xmath76 . precisely at this value of @xmath71 we also expect a non - zero lower bound for the energy @xmath91 , as can be seen by identifying such an extreme case with a guide of double thickness . in the first part we have shown that the existence of bound states could be understood as a corner effect and the coupling of transversal modes . we could learn that diffraction was essential to the final result , but the coupling of the modes gave an extra complication . the second part , however , shows that an appropriate set of coordinates may keep the effect of the corner without the coupling of such modes . this makes contact with recent dicussions @xcite in which perturbation theory seems doomed in the presence of unbounded curvature - our example fits perfectly in such class of problems . one can follow either approach to solve the problem , but its apparent complexity - e.g. the many modes used in @xcite for the l - shape - dissapears using the second method . at the end we can conclude that a simple way of breaking a continuous symmetry such as a corner still allows for a simple description as long as the important information is used : diffraction and conformality . the next task is to find the correction to the energy levels due to the classical orbits entering through the action . but we can say that for now our goal has been reached . as a final comment , the problem of bound states in multiple junctions -e.g . x - shaped and @xmath92-shaped geometries - lends itself for this kind of treatment by using neumann boundary conditions . due to the properties of our conformal coordinate system , this can be achieved by replacing @xmath93 by @xmath94 in our discussion . e. sadurn is grateful to the organizers of the symposium _ symmetries in nature _ for their kind hospitality . h. j. stoeckmann , _ quantum chaos : an introduction _ , cambridge university press , 1999 . g. a. luna acosta , a. a. krokhin , m. a. rodrguez and p. h. hernndez tejeda _ phys . b _ * 54 * 11410 ( 1996 ) ; + g. a. luna acosta , k. na , l. e. reichl and a. krokhin , _ phys . e _ * 53 * 3271 ( 1996 ) j. goldstone and r. jaffe _ phys . b _ * 45 * 24 ( 1992 ) . e. n. bulgakov , p. exner , k. n. pichugin and a. f. sadreev , _ phys . b _ * 66 * 155109 ( 2002 ) p. exner and p. seba , _ phys . _ * 30 * 11 ( 1989 ) p. exner , _ phys . a _ * 141 * 5 ( 1989 ) p. exner , p. seba and p. stovicek , _ czech . b _ * 39 * 1181 ( 1989 ) r. schult , d. g. ravenhall and h. w. wyld , _ phys . rev . b _ * 39 * 5476 ( 1989 ) j. p. carini , j. t. londergan , k. mullen and d. p. murdock , _ phys b _ * 48 * 4503 ( 1993 ) . p. polynkin , m. kolesik , , _ science _ * 324 * , 229 - 32 ( 2009 ) ; + p. polynkin , m. kolesik and j. moloney , _ phys . _ * 103 * , 123902 ( 2009 ) ; + j. kasparian , m. rodrguez , , _ science _ * 301 * , 61 - 4 ( 2003 ) . e. kajari , n. l. harshman , , _ appl . . b _ 100 , 43 - 60 ( 2010 ) . j. bjorken and s. orbach _ phys . * 23 * 2243 ( 1981 ) r. schinzinger and p. laura , _ conformal mapping : methods and applications_. dover , new york , 2003 . i. gradshteyn and i. ryzhyk , _ tables of integrals , series and products _ , seventh edition . academic press , amsterdam , 2007 . m. miski - oglu and b. dietz , the numerical solution can be obtained using finite element methods . private communication , 2010 . s. teufel , _ adiabatic perturbation theory in quantum dynamics _ , springer , berlin 2003 .

is it possible to trap a quantum particle in an open geometry ? in this work we deal with the boundary value problem of the stationary schroedinger ( or helmholtz ) equation within a waveguide with straight segments and a rectangular bending . the problem can be reduced to a one dimensional matrix schroedinger equation using two descriptions : oblique modes and conformal coordinates . we use a corner - corrected wkb formalism to find the energies of the one - dimensional problem . it is shown that the presence of bound states is an effect due to the boundary alone , with no classical counterpart for this geometry . the conformal description proves to be simpler , as the coupling of transversal modes is not essential in this case . + * pacs : * 37.10.gh , 42.25.gy , 03.65.ge + * keywords : * bent waveguides , bound states , corners , dirichlet conditions .

let @xmath7 be the tropical semifield , where the tropical sum @xmath8 is taking the maximum @xmath9 , and the tropical product @xmath10 is taking the ordinary sum @xmath11 . let @xmath12 be the sub - semifield of @xmath13 . a tropical curve @xmath5 is a metric graph with possibly unbounded edges . equivalently , in a more formal form , a tropical curve is a compact topological space homeomorphic to a one - dimensional simplicial complex equipped with an integral affine structure over @xmath14 ( see @xcite ) . finite graphs are seen as a discrete version of tropical curves . in relation to the classical algebraic curves , tropical curves and finite graphs have been much studied recently . for example , the riemann - roch formula on finite graphs and tropical curves ( analogous to the classical riemann - roch formula on algebraic curves ) are established in @xcite . the clifford theorem is established in @xcite . in this article , we consider whether the analogy of the following classical theorem holds or not . let @xmath0 be a smooth complex projective curve of genus @xmath15 , and let @xmath1 be the canonical divisor on @xmath0 . let @xmath16 be the canonical ring . then : 1 . @xmath17 is finitely generated as a graded ring over @xmath18 . @xmath17 is generated in degree at most three . our first result is that for a finite graph @xmath3 , the analogous statement ( a ) holds , but that the degrees of generators can not be bounded by a universal constant . for a divisor @xmath19 on @xmath3 , let @xmath20 be the set of rational functions @xmath21 on @xmath3 such that @xmath22 is effective ( see @xcite for details ) . we also refer to [ fgdiv ] for terminology . we show that the direct sum @xmath23 has a graded semi - ring structure over @xmath24 for any finite graph @xmath3 and any divisor @xmath19 on @xmath3 ( lemma [ semiringforfg ] ) . then the following is the first result : [ thm : main:1 ] let @xmath3 be a finite graph and let @xmath25 be the canonical divisor on @xmath3 . we set @xmath26 . then : 1 . @xmath4 _ is _ finitely generated as a graded semi - ring over @xmath24 . 2 . for any integer @xmath27 , there exists a finite graph @xmath28 such that @xmath29 is _ not _ generated in degree at most @xmath30 . for ( a ) , we show that , in fact , the semi - ring @xmath23 is finitely generated as a graded semi - ring over @xmath24 for any divisor @xmath19 on @xmath3 . our next result is that for a tropical curve @xmath5 with integer edge - length , the analogous statement ( a ) does _ not _ hold in general ( hence neither ( b ) ) . we give a sufficient condition for non - finite generation of the canonical semi - ring of tropical curves . for a divisor @xmath19 on @xmath5 , let @xmath31 be the set of rational functions @xmath21 on @xmath5 such that @xmath22 is effective ( see @xcite for details ) . we also refer to [ tcdiv ] for terminology . we show that the direct sum @xmath32 has a graded semi - ring structure over @xmath13 for any tropical curve @xmath5 and any divisor @xmath19 on @xmath5 ( lemma [ semiring ] ) . then the following is the second result : [ thm : main:2 ] let @xmath5 be a @xmath33-tropical curve of genus @xmath15 , and let @xmath34 be the canonical divisor on @xmath5 . assume that there exist an edge @xmath35 of the canonical model of @xmath5 and a positive integer @xmath36 such that @xmath35 is not a bridge and @xmath37 is linearly equivalent to @xmath38 + n(g-1)[q]$ ] , where @xmath39 and @xmath40 are the endpoints of @xmath35 . then the canonical semi - ring @xmath41 is _ not _ finitely generated as a graded semi - ring over @xmath13 . [ cor : main:2 ] 1 . let @xmath5 be a hyperelliptic @xmath33-tropical curve of genus at least @xmath42 . then @xmath6 is _ not _ finitely generated as a graded semi - ring over @xmath13 . 2 . let @xmath43 be a complete graph on vertices at least @xmath44 , and let @xmath5 be the tropical curve associated to @xmath43 , where each edge of @xmath43 is assigned the same positive integer as length . then @xmath6 is _ not _ finitely generated as a graded semi - ring over @xmath13 . for theorem [ thm : main:2 ] , we give , in fact , a sufficient condition for non - finite generation of the graded semi - ring @xmath45 over @xmath13 for any @xmath33-divisor @xmath19 of degree at least @xmath42 on a @xmath33-tropical curve @xmath5 ( theorem [ criterion ] ) . it seems likely that , for _ any _ tropical curve of genus @xmath15 , the canonical semi - ring @xmath46 will not be finitely generated as a graded semi - ring over @xmath13 , which we pose as a question . for the proof of theorem [ thm : main:2 ] , we use the notion of _ extremals _ of @xmath31 introduced by haase , musiker and yu @xcite . then theorem [ thm : main:1](b ) is deduced as a certain discrete version of theorem [ thm : main:2 ] . theorem [ thm : main:1](a ) is shown by using gordan s lemma ( see ( * ? ? ? * , proposition 1 ) ) . in this section , we prove theorem [ thm : main:2 ] and corollary [ cor : main:2 ] . in this section , we first put together necessary definitions and results on the theory of divisors on tropical curves , which will be used later . our basic references are @xcite . in this article , all finite graphs are assumed to be connected and allowed to have loops and multiple edges . for a finite graph @xmath3 , let @xmath47 and @xmath48 denote the set of vertices and the set of edges , respectively . bridge _ is an edge of @xmath3 which makes @xmath3 disconnected . a metric space @xmath5 is called _ a metric graph _ if there exist a finite graph @xmath3 and a function @xmath49 ( called the edge - length function ) such that @xmath5 is obtained by gluing the intervals @xmath50 $ ] for @xmath51 at their endpoints so as to keep the combinatorial data of @xmath3 . the pair @xmath52 is called a _ model _ for @xmath5 . _ in this article , we assume that a metric space @xmath5 is not homeomorphic to the circle @xmath53_. for a point @xmath54 of @xmath5 , we define the _ valence _ @xmath55 of @xmath54 to be the number of connected components in @xmath56 for any sufficiently small neighborhood @xmath57 of @xmath54 . let @xmath58 be a finite subset of @xmath5 which includes all points of valence different from @xmath42 , and @xmath59 be a finite graph whose vertices are the points in @xmath58 and whose edges correspond to the connected components of @xmath60 . if we define a function @xmath61 such that @xmath62 is equal to the length of the corresponding component for each edge @xmath35 , then @xmath63 is a model for @xmath5 . the model @xmath63 is called the _ canonical model _ for @xmath5 if we take the set of all points of valence different from @xmath42 as the finite subset @xmath58 . a metric graph @xmath5 with the canonical model @xmath52 is called a _ @xmath33-metric graph _ if @xmath62 is an integer for each edge @xmath35 of @xmath3 . in this case , the points of @xmath5 with integer distance to the vertices of @xmath3 are called _ @xmath33-points _ , and we denote the set of @xmath33-points by @xmath64 . tropical curves are defined in a similar way as metric graphs . a metric space @xmath5 is called a _ tropical curve _ if there exist a finite graph @xmath3 and a function @xmath65 such that @xmath5 is obtained by gluing the intervals @xmath50 $ ] for @xmath51 at their endpoints so as to keep the combinatorial data of @xmath3 , where the only edges adjacent to a one - valent vertex may have length @xmath66 . the pair @xmath52 is called a _ model _ for @xmath5 . we define the _ canonical model _ of a tropical curve in the same way as that of a metric graph . a tropical curve @xmath5 with the canonical model @xmath52 is called a _ @xmath33-tropical curve _ if @xmath62 is either an integer or equal to @xmath66 for each edge @xmath35 of @xmath3 . in this case , the points of @xmath5 with integer distance to the vertices of @xmath3 are called _ @xmath33-points_. a _ divisor _ on a tropical curve @xmath5 is a finite formal sum of points of @xmath5 , and a _ @xmath33-divisor _ on a @xmath33-tropical curve @xmath5 is a finite formal sum of @xmath33-points of @xmath5 . we denote the set of all divisors on @xmath5 by @xmath67 . if @xmath19 is a divisor on @xmath5 , we write it as @xmath68 , $ ] where @xmath69 is an integer and @xmath70 $ ] is merely a symbol . for a divisor @xmath19 , we define the _ degree _ @xmath71 to be the integer @xmath72 and the _ support _ @xmath73 to be the set of all points of @xmath5 occurring in @xmath19 with a non - zero coefficient . a divisor @xmath19 is called _ effective _ , and we write @xmath74 , if @xmath69 is a non - negative integer for all @xmath75 . on a @xmath33-tropical curve , a divisor @xmath19 is called a _ @xmath33-divisor _ if @xmath73 is a subset of @xmath76 . the _ canonical divisor _ on a tropical curve @xmath5 is defined to be @xmath77 . $ ] a _ rational function _ @xmath21 on a tropical curve @xmath5 is a continuous function @xmath78 that is piecewise linear with finitely many pieces and integer slopes , and may take on values @xmath79 only at the one - valent points . the set of all rational functions on @xmath5 is denoted by @xmath80 . for a rational function @xmath21 and a vertex @xmath54 , we define the _ order _ @xmath81 of @xmath21 at @xmath54 as the sum of outgoing slopes at @xmath54 . the _ principal divisor _ associated to @xmath21 is defined to be @xmath82 .\ ] ] we say that two divisors @xmath19 and @xmath83 are _ linearly equivalent _ , and we write @xmath84 , if there exists a rational function @xmath21 such that @xmath85 . now , we define the most important objects in this article . let @xmath19 be a divisor on a tropical curve @xmath5 . we set @xmath86 for @xmath21 , @xmath87 and @xmath88 , we define the tropical sum @xmath89 and the tropical @xmath13-action @xmath90 as follows : @xmath91 an _ extremal _ of @xmath31 is an element such that @xmath92 implies @xmath93 or @xmath94 for any @xmath95 , @xmath96 . a subset @xmath97 is called a _ subgraph _ if @xmath98 is a compact subset with a finite number of connected components . for a subgraph @xmath98 and a positive real number @xmath99 , we define the rational function _ chip firing move _ @xmath100 as @xmath101 we say that a subgraph @xmath98 can _ fire _ on a divisor @xmath19 if the divisor @xmath102 is effective for a sufficiently small positive real number @xmath99 . here , by a sufficiently small positive real number , we mean that @xmath99 is chosen to be small enough so that the chips " do not pass through each other or pass through points of valence @xmath42 . [ props ] 1 . @xmath31 is a semi - module over @xmath13 . the set of extremals of @xmath31 is finite modulo @xmath13-action . @xmath31 is generated by the extremals . the following lemma is useful for finding extremals : [ extremalcriterion ] a rational function @xmath21 is an extremal of @xmath31 if and only if there are not two proper subgraphs @xmath103 and @xmath104 covering @xmath5 such that each can fire on @xmath22 . let @xmath5 be a tropical curve , and let @xmath105 be the canonical divisor . the direct sum @xmath106 is called the _ canonical semi - ring _ of @xmath5 , and denoted by @xmath6 . for @xmath107 and @xmath108 , we define the tropical product @xmath109 as @xmath110 we show that @xmath6 has indeed a graded semi - ring structure over @xmath13 . [ semiring ] let @xmath5 be a tropical curve . then the canonical semi - ring @xmath6 has naturally a graded semi - ring structure over @xmath13 . for any divisor @xmath19 on @xmath5 , in general , the direct sum @xmath45 has naturally a graded semi - ring structure over @xmath13 . _ proof._we prove only the general case . let @xmath21 and @xmath95 be elements of @xmath111 and @xmath112 , respectively . since the order of a rational function at a point is defined as the sum of outgoing slopes and the tropical product is defined as the ordinary sum , it follows that @xmath113 . therefore @xmath114 . here both @xmath115 and @xmath116 are effective , so @xmath117 is also effective . this means that the tropical product @xmath109 is an element of @xmath118 . together proposition [ props](a ) , we obtain the assertion . 500 since we have @xmath119 , the semi - ring @xmath45 can be seen as a semi - ring over the @xmath120-th part @xmath121 . [ criterion ] let @xmath5 be a @xmath33-tropical curve of genus @xmath15 , and let @xmath19 be a @xmath33-divisor of degree @xmath122 . assume that there exist an edge @xmath35 of the canonical model of @xmath5 and a positive integer @xmath36 such that @xmath35 is not a bridge and @xmath123 is linearly equivalent to @xmath124 + \frac{n d}{2}[q]$ ] , where @xmath39 and @xmath40 are the endpoints of @xmath35 . then @xmath45 is not finitely generated as a graded semi - ring over @xmath13 . _ proof._let @xmath125 be the length of @xmath35 . note that , if @xmath35 is a loop , then @xmath126 . we begin by showing the following lemma . [ mainlemma ] let @xmath19 , @xmath39 , @xmath40 be as in theorem [ criterion ] . if there exists a positive integer @xmath127 such that @xmath128 is linearly equivalent to @xmath129 + \frac{s d}{2}[q]$ ] , then there exists an extremal of @xmath130 which is not generated by elements of @xmath131 over @xmath13 . _ proof._put @xmath132 . since @xmath128 is linearly equivalent to @xmath133 + \frac{n}{2}[q]$ ] , it follows that @xmath134 is linearly equivalent to @xmath135 + l n[q]$ ] . identify the edge @xmath35 with an interval @xmath136 $ ] such that @xmath39 and @xmath40 are identified with @xmath120 and @xmath125 , respectively . let @xmath137 be the point identified with the point @xmath138 of the interval . by definition , @xmath137 is not a @xmath33-point . first , we show two claims . the divisor @xmath139 is linearly equivalent to @xmath140 + ( 2 l n - 1)[r]$ ] . _ proof._let @xmath141 be the rational function which takes on value @xmath120 on @xmath142 , and value @xmath143 at @xmath137 , and is extended linearly to @xmath144 . then the orders of @xmath141 at @xmath39 , @xmath40 , and @xmath137 are @xmath145 moreover , the order of @xmath141 at any point of @xmath146 is equal to @xmath120 by construction . from these values we conclude that @xmath147 + l n [ q ] + { \mathop{\mathrm{div}}}(\tilde{f})$ ] is equal to @xmath140 + ( 2 l n - 1)[r]$ ] . therefore @xmath134 is linearly equivalent to @xmath140 + ( 2l n - 1)[r]$ ] . 500 let @xmath21 be the rational function such that @xmath148 + ( 2l n - 1)[r]$ ] . then @xmath21 is an extremal of @xmath149 . _ proof._since @xmath39 is an endpoint of @xmath35 and @xmath35 is an edge of the canonical model , we have @xmath150 . moreover , by the assumption that @xmath35 is not a bridge , we have @xmath151 . suppose that @xmath103 is a subgraph of @xmath5 that can fire on @xmath148 + ( 2l n - 1)[r]$ ] . then the boundary set @xmath152 of @xmath103 in @xmath5 is contained in @xmath153 . since @xmath151 , we have @xmath154 or @xmath155 . ( here @xmath156 denotes the open interval in @xmath35 connecting @xmath39 to @xmath137 . ) from lemma [ extremalcriterion ] we conclude that @xmath21 is an extremal of @xmath157 . 500 we prove that @xmath21 is not generated by elements of @xmath158 over @xmath13 by contradiction . suppose that @xmath21 is generated by elements of @xmath158 over @xmath13 . then we have @xmath159 where @xmath160 is an element of @xmath161 and the sum @xmath162 is equal to @xmath163 for each @xmath164 . note that there are at least two terms in @xmath165 for each @xmath164 . by lemma [ semiring ] we can take @xmath166 and @xmath167 and @xmath168 such that @xmath169 and @xmath170 . by proposition [ props](b ) we may assume that @xmath171 and @xmath172 are the extremals of @xmath173 and @xmath174 , respectively . then we have @xmath175 where each @xmath171 and @xmath172 is an extremal of @xmath173 and @xmath174 , respectively . since @xmath21 is an extremal , it follows that @xmath21 is equal to @xmath176 after changing indices if necessary . put @xmath177 , @xmath178 , @xmath179 , and @xmath180 . recall that @xmath181 . now , we have @xmath182 + ( 2l n - 1)[r ] = 2s l d + { \mathop{\mathrm{div}}}(f ) = l d + { \mathop{\mathrm{div}}}(g ) + k d + { \mathop{\mathrm{div}}}(h ) .\ ] ] since both @xmath183 and @xmath184 are effective , we may assume that @xmath185 + ( l d-1)[r ] , \\ k d + { \mathop{\mathrm{div}}}(h ) & = k d[r ] , \end{aligned}\ ] ] after changing the role of @xmath95 and @xmath186 if necessary . in this setting , we deduce a contradiction by studying the property of the rational function @xmath186 . since @xmath187 - k d$ ] , all the zeros and poles of @xmath186 lie in @xmath188 . let @xmath189 be the points of @xmath190 in this order , where @xmath191 is @xmath39 and @xmath192 is @xmath40 . moreover , let @xmath193 be the segment which connects @xmath194 to @xmath195 , and @xmath196 be the segment which connects @xmath197 to @xmath198 , where we set @xmath199 and @xmath200 . we denote the each length of @xmath201 and @xmath202 by @xmath203 and @xmath204 , respectively . the sum of outgoing slopes of @xmath186 at @xmath54 as a rational function on @xmath201 and @xmath205 are denoted by @xmath206 and @xmath207 , respectively . , scaledwidth=60.0% ] since @xmath186 is continuous , we have @xmath208 now , by the equality @xmath187 -k d$ ] , we have @xmath209 for @xmath210 , and for @xmath211 , and we have @xmath212 for @xmath213 , and for @xmath214 . from these relations , we deduce that @xmath215 similarly , we deduce that @xmath216 since @xmath19 is a @xmath33-divisor and @xmath137 is not a @xmath33-point , we have @xmath217 . thus @xmath218 it follows that where we use ( [ eq:1 ] ) , ( [ eq:2 ] ) in the second equality , and ( [ eq:3 ] ) in the last equality . by construction , we have @xmath219 hence we have @xmath220 the value @xmath221 is an integer . _ proof._first , we claim that the value @xmath222 is an integer . indeed , since @xmath35 is not a bridge , there exists a path @xmath223 in @xmath142 such that @xmath39 and @xmath40 are the endpoints of @xmath223 . since @xmath186 is a piecewise linear function with integer slopes along @xmath223 and both zeros and poles of @xmath186 on @xmath142 are @xmath33-points , it follows that the difference @xmath224 is an integer . since @xmath5 is a @xmath33-tropical curve and @xmath19 is a @xmath33-divisor , it follows that @xmath203 , @xmath204 , @xmath225 , @xmath226 @xmath227 , @xmath125 are integers . moreover , by definition , @xmath228 is an integer . thus @xmath221 is an integer . 500 since @xmath229 and @xmath230 , and @xmath231 and @xmath232 are relatively prime , respectively , it follows that there exists a positive integer @xmath233 such that @xmath234 . then we have @xmath235 since @xmath236 , it follows that @xmath237 and @xmath238 , but this contradicts @xmath122 . thus @xmath21 is not generated by elements of @xmath239 over @xmath13 . 500 now , we return to the proof of theorem [ criterion ] . we prove that @xmath45 is not finitely generated as a graded semi - ring over @xmath13 by contradiction . suppose that @xmath45 is finitely generated as a graded semi - ring over @xmath13 . since elements of @xmath240 is generated by the extremals over @xmath13 , we may assume that all the generators of @xmath45 is an extremal of @xmath240 for some positive integer @xmath241 . let @xmath233 be the maximal number among such numbers . fix a positive integer @xmath242 such that @xmath243 is bigger than @xmath233 , and put @xmath244 . then we have @xmath245 + \frac{kn d}{2}[q ] = \frac{s d}{2}[p ] + \frac{s d}{2}[q].\ ] ] applying lemma [ mainlemma ] , we get an extremal of @xmath130 which is not generated by the elements of @xmath246 over @xmath13 , but this contradicts the maximality of @xmath233 . thus @xmath45 is not finitely generated as a graded semi - ring over @xmath13 . 500 [ sufficientcondition ] let @xmath5 be a @xmath33-tropical curve of genus @xmath15 . assume that there exist an edge @xmath35 of the canonical model of @xmath5 and a positive integer @xmath36 such that @xmath35 is not a bridge and @xmath37 is linearly equivalent to @xmath38 + n(g-1)[q]$ ] , where @xmath39 and @xmath40 are the endpoints of @xmath35 . then the canonical semi - ring @xmath6 is not finitely generated as a graded semi - ring over @xmath13 . a finite graph is called a _ complete graph _ on @xmath36 vertices if it is a finite graph with @xmath36 vertices in which every pair of distinct vertices is connected by a unique edge . [ corollary ] 1 . let @xmath5 be a hyperelliptic @xmath33-tropical curve of genus at least @xmath42 . then @xmath6 is not finitely generated as a graded semi - ring over @xmath13 . 2 . let @xmath36 be an integer at least @xmath44 , let @xmath43 be a complete graph on @xmath36 vertices , and let @xmath5 be the tropical curve associated to @xmath43 , where each edge of @xmath43 is assigned the same positive integer as length . then @xmath6 is not finitely generated as a graded semi - ring over @xmath13 . _ proof._for ( a ) , let @xmath52 be the canonical model of @xmath5 . by chan s theorem ( * ? ? ? * theorem 3.12 ) , there exists an edge @xmath35 of @xmath3 such that @xmath247 + [ q ] ) = 1 $ ] , where @xmath39 and @xmath40 are the endpoints of @xmath35 . by the riemann - roch formula , it follows that @xmath248 + ( g-1)[q]$ ] is linearly equivalent to @xmath34 . applying theorem [ criterion ] ( with @xmath249 ) , the statement ( a ) follows . for ( b ) , there are two cases , one is that @xmath36 is an odd number , and the other is that @xmath36 is an even number . fix any two vertices @xmath250 and @xmath251 of @xmath43 , and let @xmath35 be the unique edge which connects these vertices . since the genus of @xmath5 is equal to @xmath252 , the degree of the canonical divisor is equal to @xmath253 . if @xmath36 is odd , then @xmath34 is equivalent to @xmath254 + \frac{n(n-3)}{2}[w]$ ] . if @xmath36 is even , then @xmath255 is equivalent to @xmath256 + n(n-3)[w]$ ] . therefore , in both cases , we can apply theorem [ criterion ] , and the statement ( b ) follows . 500 in this section , we prove theorem [ thm : main:1 ] . we denote the valence of a vertex @xmath54 by @xmath55 . a _ divisor _ on a finite graph @xmath3 is a finite formal sum of vertices and we denote the set of all divisors on @xmath3 by @xmath257 . if @xmath19 is a divisor on @xmath3 , we write it as @xmath258 , $ ] where @xmath69 is an integer and @xmath70 $ ] is merely a symbol . the _ degree _ of a divisor and an _ effective _ divisor are defined in the same way as in [ tcdiv ] . the _ canonical divisor _ on a finite graph @xmath3 is defined to be @xmath259 . $ ] a _ rational function _ @xmath21 on a finite graph @xmath3 is a @xmath33-valued function on vertices @xmath47 . the set of all rational functions on @xmath3 is denoted by @xmath260 . for a rational function @xmath21 and a vertex @xmath54 , we define the _ order _ @xmath81 of @xmath21 at @xmath54 as the sum of differences between the value at @xmath54 and at each vertex adjacent to @xmath54 , that is , we define it to be the integer @xmath261 where @xmath262 means that @xmath54 and @xmath263 are the endpoints of @xmath51 . the _ principal divisor _ associated to @xmath21 is defined to be @xmath264 .\ ] ] an _ extremal _ of @xmath269 is an element such that @xmath92 implies @xmath93 or @xmath94 for any @xmath95 , @xmath270 . for a subset @xmath271 of vertices @xmath47 , we define the rational function @xmath272 on @xmath3 as @xmath273 we say that a subset @xmath271 of vertices @xmath47 can _ fire _ on a divisor @xmath19 if the divisor @xmath274 is effective . _ proof._for ( a ) , it is clear that @xmath90 is an element of @xmath269 for @xmath267 and @xmath275 . so it is sufficient to show that @xmath89 is an element of @xmath269 for any @xmath21 , @xmath266 . if we have @xmath276 for a fixed vertex @xmath54 , then it follows that @xmath277 similarly , we have @xmath278 if @xmath279 at @xmath54 . since both @xmath22 and @xmath280 are effective , it follows that @xmath281 is effective . then the statement ( a ) follows . for ( b ) , we identify @xmath20 with the lattice points of a polyhedron in an euclidean space , which is a finite set . since the way of identification will be described in the proof of theorem [ finitea ] , we omit the detail now . then the statement ( b ) follows . [ semiringforfg ] let @xmath3 be a finite graph . then the canonical semi - ring @xmath4 has naturally a graded semi - ring structure over @xmath24 . for any divisor @xmath19 on @xmath3 , in general , the direct sum @xmath287 has naturally a graded semi - ring structure over @xmath24 . let @xmath3 be the finite graph with two vertices and three edges each of which connects the vertices . let @xmath28 be the finite graph obtained by replacing each edge of @xmath3 with a segment which consists of @xmath289 edges . note that @xmath28 has @xmath290 vertices and @xmath291 edges . let @xmath137 be the @xmath292-th vertex counted from @xmath39 on a segment , where @xmath39 is a vertex of valence different from @xmath42 . the canonical divisor @xmath293 is equal to that of @xmath3 by definition , and the divisor @xmath294 is linearly equivalent to @xmath140 + ( 2n - 1)[r]$ ] . it follows that there exists an extremal @xmath21 of @xmath295 such that @xmath296 is equal to @xmath140 + ( 2n - 1)[r]$ ] . suppose that @xmath21 is generated by the elements of @xmath297 over @xmath24 . then we may assume that @xmath21 is equal to @xmath298 , where @xmath95 and @xmath186 are the extremal of @xmath299 and @xmath300 , respectively , and @xmath301 is equal to @xmath36 . it follows that @xmath302 + ( 2l - 1)[r ] , \\ k k_{g_n } + { \mathop{\mathrm{div}}}(h ) & = 2k [ r ] .\end{aligned}\ ] ] let @xmath303 and @xmath251 be the second vertex counted from @xmath39 on each segment different from @xmath35 , where @xmath35 is the segment on which @xmath137 is a vertex . after some calculations , we get @xmath304 then we have @xmath305 and @xmath306 is an integer . _ proof._let the vertices @xmath310 , and let @xmath311 be the graph laplacian , that is , the @xmath312 symmetric matrix such that each entry @xmath313 is equal to the number of edges which connect @xmath195 to @xmath314 if @xmath315 , and the value @xmath316 if @xmath317 . using the laplacian , we can describe @xmath318 as @xmath319 we identify @xmath318 with the lattice points of a polyhedron @xmath320 in @xmath321 by a map @xmath322 which maps @xmath21 to @xmath323 , where @xmath320 is a polyhedron of the form @xmath324 by the fundamental theorem of polyhedra , it follows that @xmath320 is a convex hull of finitely many vectors . _ proof._by the above identification , it is sufficient to show that each lattice point of @xmath0 whose @xmath292-th coordinate is equal to @xmath241 corresponds to a lattice point of @xmath328 . let @xmath329 be a lattice point of @xmath0 . by definition , @xmath330 is an element of @xmath325 . since @xmath331 it follows that @xmath332 hence , @xmath303 is a lattice point of @xmath328 . conversely , let @xmath303 be a lattice point of @xmath328 and let @xmath333 be the vector whose i - th coordinate @xmath334 is equal to @xmath335 . then @xmath251 is an element of @xmath325 in the same way as above . in particular , @xmath336 is a lattice point of @xmath0 . moreover , the sum of the lattice points of @xmath0 corresponds to the product of the elements of @xmath337 for some @xmath241 . this follows from the same reason as the above correspondence , so we omit the detail . by this correspondence and the gordan s lemma ( see ( * ? ? ? * , proposition 1 ) ) , all the elements of @xmath327 for any @xmath241 is generated by the elements of @xmath338 for finitely many @xmath36 over @xmath24 .

for a projective curve @xmath0 and the canonical divisor @xmath1 on @xmath0 , it is classically known that the canonical ring @xmath2 is finitely generated in degree at most three . in this article , we study whether analogous statements hold for finite graphs and tropical curves . for any finite graph @xmath3 , we show that the canonical semi - ring @xmath4 is finitely generated but that the degree of generators are not bounded by a universal constant . for any hyperelliptic tropical curve @xmath5 with integer edge - length , we show that the canonical semi - ring @xmath6 is not finitely generated , and , for tropical curves with integer edge - length in general , we give a sufficient condition for non - finite generation . # 1

the discrete dipole approximation ( dda ) provides a flexible and general method to calculate scattering and absorption of light by objects of arbitrary geometry @xcite . the approximation consists of replacing the continuum target of interest by an array of polarizable points , which acquire oscillating electric dipole moments in response to the electric field due to the incident wave plus all of the other dipoles . broadly speaking , there are two criteria determining the accuracy of the approximation : * the interdipole separation @xmath1 should be small compared to the wavelength of the radiation in the material ( @xmath2 , where @xmath0 is the complex refractive index , and @xmath3 is the wavenumber _ in vacuo _ ) ; * the interdipole separation @xmath1 should be small enough to resolve structural dimensions in the target . with modern workstations , it is now feasible to carry out dda calculations on targets containing up to @xmath4 dipoles @xcite . in addition to calculation of scattering and absorption cross sections , the dda has recently been applied to computation of forces and torques on illuminated particles @xcite if the dipoles are located on a cubic lattice , then in the limit where the interdipole separation @xmath5 , the familiar clausius - mossotti relation ( see , e.g , @xcite ) can be used to determine the choice of dipole polarizabilities required so that the dipole array will approximate a continuum target with dielectric constant @xmath6 . @xcite showed how this estimate for the dipole polarizabilities should be modified to include radiative reaction corrections , and @xcite derived the @xmath7 $ ] corrections required so that an infinite cubic lattice would have the same dispersion relation as a continuum of given dielectric constant . nearly all dda calculations to date have assumed the dipoles to be located on a cubic lattice . if , instead , a rectangular lattice is used , it will still be possible to apply fft techniques to the discrete dipole approximation , in essentially the same way as has been done for a cubic lattice @xcite . however , the ability to use different lattice constants in different directions might be useful in representing certain target geometries . the objective of the present report is to obtain the dispersion relation for propagation of electromagnetic waves on a rectangular lattice of polarizable points . this dispersion relation will be expanded in powers of @xmath8 , where @xmath1 is the characteristic interdipole separation . for a lattice of specified dipole polarizabilities , this will allow us to determine the dispersion relation for electromagnetic waves propagating on the lattice when @xmath9 . alternatively , if we require the lattice to propagate waves with a particular dispersion relation , inversion of the lattice dispersion relation will provide a prescription for assigning dipole polarizabilities when seeking to approximate a continuum material with a rectangular lattice of polarizable points . the problem of wave propagation on an infinite polarizable cubic lattice of point dipoles has been solved previously by ( * ? ? ? * herafter dg93 ) . here we describe the generalization of this problem to rectangular lattices . consider an infinite rectangular lattice with lattice sites at _ = ( n_1 d_1,n_2 d_2,n_3 d_3 ) , where the @xmath10 are integers and the lattice constants @xmath11 are , in general , all different . the density @xmath12 of dipoles is just n = 1d_1d_2d_3 , and it will be convenient to define the characteristic lattice length scale d ( d_1d_2d_3)^1/3 . since the lattice is anisotropic , the polarizability @xmath13 of the dipoles located at lattice sites is a tensor , i.e. , the polarization is ( ) =( ) . thus the polarization vector is not , in general , parallel to the electric field or to the vector potential . from charge conservation we have = 0 , or , since = t , [ eq:4 ] we obtain the transversality condition = 0 . [ eq:5 ] we assume that the dipole moment at the lattice site @xmath14 is _ ( t)= _ 0(0 ) e^i_-it . [ eq:6 ] with the lorentz gauge condition + 1ct = 0 and equation ( [ eq:4 ] ) , the vector potential @xmath15 satisfies the wave equation @xmath16 using these dimensionless quantities , the mode equation ( [ eq:21 ] ) may be written _ l=1 ^ 3 m_il _ lj e_j = e_i , [ eq : mode_equation ] where m_ij _ - d^3^ , [ eq:31 ] and equation ( 25 ) becomes _ _ j=1 ^ 3 _ i _ ij e_j = 0 . [ eq:32 ] it should be noted that the mode equation for a cubic lattice has the same form as equation ( [ eq : mode_equation ] ) . the matrix elements @xmath30 in equation ( [ eq:31 ] ) also have the same form as the matrix elements for a cubic lattice but with the vector @xmath31 replaced by @xmath29 for a rectangular lattice . since the matrix elements for a cubic lattice have been calculated previously , it is convenient to rewrite ( [ eq:31 ] ) as m_ij = m_ij^c + t_ij , [ eq:33 ] where @xmath32 is the matrix element for a cubic lattice , and @xmath33 is t_ij _ - _ . [ eq:34 ] for a given polarizability tensor @xmath34 , the mode equation ( [ eq : mode_equation ] ) allows one to determine the dispersion relation @xmath35 . in the present case , the dispersion relation is known : = m , where @xmath36 is the ( complex ) refractive index and @xmath37 is a unit vector in the direction of propagation , which is fixed . from equation ( [ eq:32 ] ) for a rectangular lattice , the polarization rather than the vector potential is perpendicular to the direction of propagation , i.e. , _ j=1 ^ 3 a_i _ ij e_j = 0 . [ eq:36 ] therefore , the mode equation ( [ eq : mode_equation ] ) can be used to determine dipole polarizabilities @xmath38 such that equation ( [ eq:36 ] ) is satisfied . we seek the dipole polarizabilities in the limit @xmath39 . in this limit , one can write _ ij^(0 ) + _ ij^(1 ) + o(^4 ) , m_ij^c = m_ij^c(0 ) + m_ij^c(1 ) + o(^4 ) , t_ij = t_ij^(0 ) + t_ij^(1 ) + o(^4 ) , = ^(0 ) + ^(1 ) + o(^4 ) , where superscripts 0 and 1 designate the zero frequency limit and the leading order corrections , respectively . the low - frequency expansion of the matrix element @xmath30 for a cubic lattice was given by dg93 and is : m_ij^c(0 ) = m^2 + 23(m^2 - 1)_ij - m^2m^2 - 1a_i a_j , [ eq:41 ] m_ij^c(1 ) = ^2_ij + m^2 ^ 2(1-_ij)c_3a_i a_j- 4 ^ 2i3_ij^3 , [ eq:42 ] where @xmath40 from equation ( [ eq:34 ] ) , one finds t_ij^(0 ) = _ ij r_0 ( i ) , [ eq:43 ] where r_0(i)= _ - _ [ eq : r_0 ] . note that @xmath41 , so that there are only two independent sums which require evaluation . numerical evaluation is discussed in the appendix . @xmath42 \nonumber\\ & - & \nu^2 \delta_{ij } \left [ r_1 + ( m^2 -1 ) r_2(i ) + 8m^2 a_i^2 r_3 ( i , i ) - 4m^2 \sum_{l=1}^3 a_l^2 r_3(i , l ) \right ] \ , , \label{eq:44}\end{aligned}\ ] ] where r_1 = _ - _ , r_2(i ) = _ - _ , r_3(i , j ) = _ - _ . [ eq : r_3 ] in the limit of cubic lattice , @xmath43 , and all the sums @xmath44 in equations ( [ eq : r_0 ] ) - ( [ eq : r_3 ] ) vanish . for a general rectangular lattice , these sums must be evaluated for each set of new lattice constants . care must be used in evaluation of these sums as described in the next section . we note that @xmath45 and @xmath46 may be obtained from @xmath47 : r_1 = _ i=1 ^ 3_j=1 ^ 3 r_3(i , j ) [ eq : r_1fromr_3 ] ; r_2(i ) = _ j=1 ^ 3 r_3(i , j ) [ eq : r_2fromr_3 ] . since @xmath48 , six independent quantities suffice to determine @xmath45 , @xmath46 , and @xmath47 . in the static limit the polarizability tensor @xmath49 is diagonal , and the mode equation and transversality condition become : _ j=1 ^ 3 _ jj^(0 ) e_j^(0 ) = e_i^(0 ) , [ eq:48 ] _ i=1 ^ 3 a_i _ ii^(0 ) e_i^(0 ) = 0 . [ eq:49 ] substituting ( [ eq:41 ] ) and ( [ eq:43 ] ) into ( [ eq:48 ] ) and taking into account condition ( [ eq:49 ] ) one gets _ ii^(0 ) e_i^(0 ) = e_i^(0 ) . [ eq:50 ] for @xmath50 , the static dipole polarizability for a rectangular lattice is _ ii^(0 ) = _ cm 1 + _ cm r_0(i ) , [ eq:51 ] where the clausius - mossotti polarizability is _ cm . it is evident from equation ( [ eq:51 ] ) that the static polarizability for a cubic lattice reduces to the clausius - mossotti result . if @xmath51 in one of the directions , then the polarizability in this direction is undetermined by equation ( [ eq:50 ] ) . however , since the result ( [ eq:51 ] ) is valid for values of @xmath52 arbitrarily close to zero , we take ( [ eq:51 ] ) to be the solution for all directions of the electric field subject to equation ( [ eq:49 ] ) . we now proceed to obtain the leading corrections for small @xmath53 . the first order mode equation and transversality condition are : _ l=1 ^ 3 m_il^(1 ) _ lj^(0 ) e_j^(0 ) + _ j=1 ^ 3 _ l=1 ^ 3 m_il^(0 ) = e_i^(1 ) , [ eq:53 ] _ j=1 ^ 3 a_i = 0 . [ eq : transversality ] from equations ( [ eq:41 ] ) , ( [ eq:43 ] ) , ( [ eq:50 ] ) , and ( [ eq : transversality ] ) one gets @xmath54 & = & \sum_{j=1}^3 \left [ { m^2 + 2 \over 3(m^2 -1 ) } + r_0(i ) \right]\ , \left [ \gamma_{ij}^{(1)}\ , e_j^{(0 ) } + \gamma_{ij}^{(0)}\ , e_j^{(1 ) } \right ] \nonumber \\ & = & \sum_{j=1}^3 { \gamma_{ij}^{(1 ) } \over \gamma_{ii}^{(0 ) } } e_j^{(0 ) } + e_i^{(1 ) } \ , . \ ] ] therefore , equation ( [ eq:53 ] ) becomes [ eq:60 ] _ j=1 ^ 3 e_j^(0)=0 . [ eq : the_equation ] it is convenient to write @xmath55 in the form m_ij^(1 ) = ^2 _ ij l_ii + m^2 ^ 2 a_i a_j k_ij , where , from ( [ eq:42 ] ) and ( [ eq:44 ] ) @xmath56 and k_ij = c_3 + r_1 -2 r_2(i ) -2 r_2(j ) + 8 r_3(i , j ) . we now observe that one can have the square bracket in equation ( [ eq : the_equation ] ) vanish for each @xmath57 if the first - order correction to the polarizability is @xmath58 \ , . \label{eq : gamma_ij^(1)}\end{aligned}\ ] ] equation ( [ eq : gamma_ij^(1 ) ] ) makes it explicit that the frequency - dependent term in the dipole polarizability depends on the direction of propagation of the wave . we observe , however , that this solution is not unique . in fact , the transversality condition ( [ eq : transversality ] ) permits the more general solution _ ij^(1 ) = -^2 _ ii^(0 ) _ jj^(0 ) where @xmath59 could , in principle , be a function of @xmath60 , the direction of propagation , or the polarization . we note , however , that if @xmath61 then eq . ( [ eq : the_equation ] ) is no longer satisfied term by term , but only after summation . since we appear to have freedom in the choice of @xmath59 , numerical experiments using different choices for @xmath59 could be used to see what value of @xmath59 appears to be optimal for scattering calculations using the discrete dipole approximation .. propagation of electromagnetic waves on a cubic lattice has been discussed by dg93 . in their analysis it was assumed that if the dielectric material is isotropic , the polarizability @xmath62 can be taken to be isotropic : @xmath63 and @xmath64 . with this assumption , contraction of eq . ( [ eq:60 ] ) with @xmath52 leads to [ eq : alpha_dg93 ] ^(1)_ij = -^2 ^ 2 _ ij , s_j=1 ^ 3 ( a_je_j^(0))^2 . as in eq.(4.14 ) of @xcite . however , we have seen above that for a cubic lattice ( for which @xmath65 ) , @xmath66 can be made diagonal but not isotropic : if we choose @xmath67 so that @xmath66 is diagonal , we obtain [ eq : alpha_cubic ] ^(1)_ij = -^2 ^ 2 _ ij . which differs from the result of @xcite since @xmath68 , except for the special case of propagation in the ( 1,1,1 ) direction , for which @xmath69 . we now compare discrete dipole approximation ( dda ) calculations of scattering and absorption using ( 1 ) the polarizability prescription ( [ eq : alpha_dg93 ] ) from dg93 and ( 2 ) the present result ( [ eq : alpha_cubic ] ) . we consider spheres where we can use mie theory to easily calculate the exact result for comparison . let @xmath70 and @xmath71 be the cross sections for scattering and absorption , and let @xmath72 and @xmath73 be the dimensionless scattering and absorption efficiency factors for a sphere of radius @xmath74 . figure [ fig : m=1.33 + 0.01i ] shows @xmath75 and @xmath76 calculated using the dda with a target consisting of @xmath77 dipoles , with radius @xmath78 . the dda calculations used a cubic lattice , and were carried out using the public domain code ddscat.6.1 @xcite . calculations were done for values of the scattering parameters @xmath79 satisfying @xmath80 , as this appears to be a good validity criterion for dda calculations . the upper panel shows scattering and absorption efficiencies calculated with mie theory and using the dda with the present prescription for polarizabilities @xmath13 . the middle and lower panels show fractional errors @xmath81 $ ] of the dda scattering and absorption cross sections . because the dda target ( dipoles on a cubic lattice ) is not rotationally symmetric , the scattering problem was solved for 12 different incident directions , in each case for two orthogonal polarizations . the cross sections shown in fig . [ fig : m=1.33 + 0.01i ] are averages over the 12 incident directions and two incident polarizations . in the limit @xmath82 , identical results are obtained for both polarizability prescriptions ; this is expected since they differ only in terms of @xmath83 . note that even at @xmath84 ( i.e. , @xmath85 ) for which the polarizabilities are exactly given by the clausius - mossotti polarizabilities the dda calculation has a nonzero error even as @xmath86 and @xmath87 . this is because ( 1 ) the dipole array only approximates a spherical geometry and ( 2 ) when the `` lattice dispersion relation '' polarizabilities are used , the dda does not accurately model the `` screening '' of the target material close to the surface by the polarization at the target surface . for modest dielectric functions , and sufficient numbers of dipoles , the errors are small for the case shown in figure [ fig : m=1.33 + 0.01i ] the fractional errors in the @xmath86 limit are @xmath88 . as @xmath89 increases , the two prescriptions result in differences in computed values of @xmath75 and @xmath76 , although the differences are small , and comparable in magnitude to the errors in the @xmath86 limit . in figure [ fig : m=1.33 + 0.01i ] our new prescription appears to result in small improvements in accuracy ( relative to dg93 ) for @xmath75 for @xmath90 , and likewise for @xmath76 for @xmath91 . for @xmath92 , however , the dg93 prescription appears to give more accurate results for @xmath76 . note , however , that for both polarizability prescriptions the fractional errors in @xmath75 and @xmath76 for @xmath93 are of the same order as the fractional errors for @xmath87 , so it is not clear that the errors should be attributed to the lattice dispersion relation the errors in @xmath75 and @xmath76 may instead be associated with errors in the computed polarization field near the surfaces , arising from the fact that in the dda the dipoles in the surface layer are imperfectly `` screened '' from the external field by their neighbors . for @xmath80 , the errors in @xmath75 using the new polarizabilities are of order @xmath94 for @xmath95 ( figure [ fig : m=1.33 + 0.01i ] ) , @xmath96 ( figure [ fig : m=1.70 + 0.10i ] ) and @xmath97 ( figure [ fig : m=2.00 + 1.00i ] ) , and of order @xmath98 for @xmath99 ( figure [ fig : m=3.00 + 4.00i ] ) . in all four cases , for wavelengths satisfying @xmath90 , the magnitude of the error in @xmath75 is of the same order as the error at zero frequency , suggesting that it is mainly due to the surface . the errors in @xmath76 tend to be somewhat larger than for @xmath75 , especially for large values of @xmath100 . for @xmath99 ( figure [ fig : m=3.00 + 4.00i ] ) , the new polarizabilities result in a fractional error of @xmath101 for @xmath102 and @xmath77 dipoles . figure [ fig : m=1.70_0.10ib ] shows results calculated for @xmath96 spheres represented by @xmath103 dipoles 8 times as many as in fig . [ fig : m=1.70_0.10i ] . note that the fractional error for @xmath82 is now only 0.5% half as large as for @xmath77 . this can be ascribed to the fact that the fraction of the dipoles in the surface monolayer was reduced by a factor of 2 . note also that the errors for finite @xmath89 are also about a factor of 2 smaller than for @xmath77 . these results appear to confirm the suspicion that the errors arise primarily from the surface layer . @xcite recently proposed a method for assigning dipole polarizabilities that takes into account the effects of target geometry , resulting in dipole polarizabilties that are location - dependent even when the target being approximated has homogeneous composition . @xcite have extended this treatment by adding finite - wavelength corrections to the `` surface - corrected '' polarizability prescription , and used dda calculations for spheres and ellipsoids to show that accurate results for @xmath75 and @xmath76 can be obtained for modest numbers of dipoles even when the target material has a large refractive index @xmath104 . the current calculations does _ not _ include such `` surface corrections '' , the prescription ( [ eq : alpha_cubic ] ) depends only on local properties so it comes as no surprise that there are systematic errors even in the limit @xmath86 . @xcite show that the error can be greatly reduced if the polarizabililites are assigned in a way that allows for both target geometry ( i.e. , surface corrections ) and the finite wavelength corrections obtained from the lattice dispersion relation analysis . using these `` surface - corrected lattice dispersion relation polarizabilities '' , @xcite show that for the case @xmath99 , the error in @xmath76 can be reduced to of order @xmath105 for @xmath106 using as few as @xmath107 dipoles , whereas for @xmath106 , the error is seen from figure [ fig : m=3.00 + 4.00i ] to be @xmath108 even with @xmath77 dipoles . such surface corrections will obviously be of value for dda calculations , but at the moment have been applied only to sperical , spheroidal , or ellipsoidal targets @xcite . the principal results of this paper are as follows : 1 . we have derived the exact mode equation ( [ eq : mode_equation ] ) for propagation of electromagnetic waves through a rectangular lattice of polarizable points . if the electric polarizability tensor @xmath109 for the lattice points is specified , then we can solve the mode equation to obtain the dispersion relation for waves propagating on the lattice . alternatively , if we wish the lattice to have the same dispersion relation as a continuum medium of complex refractive index @xmath0 , then we can solve the mode equation for the appropriate polarizability tensor . 3 . in the case of a cubic lattice , our derived `` lattice dispersion relation '' polarizability tensor ( [ eq : alpha_cubic ] ) is diagonal but anisotropic , and differs from the isotropic polarizabilities obtained by dg93 . the new polarizability prescription has been tested in dda scattering calculations for spheres . the resulting cross sections have accuracies comparable to those obtained with the dg93 prescription . we thank robert lupton for making available the sm software package . this work was supported in part by nsf grants ast-9988126 and ast-0216105 . the coefficients @xmath110 are obtained numerically from the sums ( [ eq : r_0 ] ) . note that eq . ( [ eq : r_0 ] ) for @xmath113 involves the difference between two sums which separately diverge at infinity . to handle this divergence , one notes that in nature one is always interested in finite wavelengths and nonzero frequencies ; when dealing with finite wavelengths , the lattice sums converge because of the oscillations introduced at large distances by retardation effects . to obtain the proper behavior when going to the zero frequency limit , we modify the summations to suppress the contribution from large distances . the manner in which this is achieved is not critical . we compute r_0(i)=_0 [ eq : evalr_0 ] the exponential cutoff allows one to limit the summations to @xmath114 and @xmath115 . the summation ( [ eq : evalr_0 ] ) is carried out for various small values of @xmath109 and it is found , as anticipated , that for @xmath116 the numerical result for @xmath113 is insensitive to the choice of @xmath109 . each of the sums in ( [ eq : evalr_0 ] ) is over a lattice ; @xmath117 defines the distance from the origin for the cubic lattice , and @xmath118 is the distance from the origin for the reciprocal lattice . each sum diverges at infinity , where we can replace the discrete summation by an integration ; the reason why the divergences cancel is that the two lattices have the same density of lattice points , so that at large distances , both approach the same integration . collinge , m.j . , & draine , b.t . 2004 , `` discrete - dipole approximation with polarizabilities that account for both finite wavelength and target geometry '' , j. opt . am . , a , submitted ( http://xxx.arxiv.org/abs/astro-ph/0311304v1 )

we discuss the propagation of electromagnetic waves on a rectangular lattice of polarizable point dipoles . for wavelengths long compared to the lattice spacing , we obtain the dispersion relation in terms of the lattice spacing and the dipole polarizabilities . we also obtain the dipole polarizabilities required for the lattice to have the same dispersion relation as a continuum medium of given refractive index @xmath0 ; our result differs from previous work by draine & goodman ( 1993 ) . our new prescription can be used to assign dipole polarizabilities when the discrete dipole approximation is used to study scattering by finite targets . results are shown for selected cases .

entanglement is considered to be the key resource of the rapidly developing fields of quantum information processing and quantum computation , for an introduction see e.g. @xcite . among the early proposals concerning the applications of entangled states are those for quantum key distribution @xcite , quantum dense coding @xcite , and quantum teleportation @xcite . despite the resulting great interest in entangled quantum states , their complete characterization is still an unsolved problem . it is known that entanglement can be fully identified by applying all positive but not completely positive ( pncp ) maps to a given state @xcite . the problem of this approach , however , consists in the fact that the general form of the pncp maps is essentially unknown . the presently best studied pncp map is the partial transposition ( pt ) @xcite . it is known that pt gives a complete characterization of entanglement in hilbert spaces of dimension @xmath0 and @xmath1 @xcite . bipartite entanglement can also be completely characterized via pt in infinite - dimensional hilbert spaces , as long as only gaussian states are considered , whose moments up to second order fully describe their properties @xcite . by using higher - order moments , a complete characterization has been given for those entangled states which exhibit negativities after application of the pt map @xcite . this approach gives no insight into bound entangled states remaining positive after pt . to overcome this limitation , to the matrices of moments other kinds of pcnp maps have been applied @xcite , including kossakowsky , choi and breuer maps @xcite . the identification of bound entanglement in this way , however , turned out to be a cumbersome problem . an equivalent approach of identifying entanglement is based on special types of hermitian operators , called entanglement witnesses . the witnesses were introduced as a class of linear operators , whose mean values are non - negative for separable states but can become negative for entangled states @xcite . presently only some classes of entanglement witnesses are available . once a witness is known , an optimization can be performed @xcite . also nonlinear witnesses have been studied @xcite , which may increase the number of entangled states to be identified by a single witness , in comparison with a given linear witness . however , if one is able to construct the general form of the linear witnesses , the problem of identifying entanglement is completely solved . in the present contribution we show that any entanglement witness can be expressed in terms of completely positive hermitian operators , whose general form is well known . on this basis we derive entanglement inequalities , which are formulated solely in terms of general hermitian operators . we also provide an approach for optimizing such inequalities , by introducing a separability eigenvalue problem . our method is a powerful tool for analyzing experimental data , to verify any kind of entanglement . one may also identify general bound entangled states whose density operator has a positive partial transposition ( ppt ) . the paper is structured as follows . in sec . [ sec : ii ] we derive the most general form of entanglement conditions in terms of hermitian operators . this leads us to an optimization problem the separability eigenvalue problem which is studied in sec . [ sec : iii ] . in sec . [ sec : iv ] this problem is solved for a class of projection operators and a general numerical implementation of entanglement tests for arbitrary quantum states is given . a method for the identification of bound entangled states is considered in sec . [ sec : v ] . finally , in sec . [ sec : vi ] we give a brief summary and some conclusions . let us consider two systems @xmath2 and @xmath3 , represented by arbitrary hilbert spaces @xmath4 and @xmath5 with orthonormal bases being @xmath6 and @xmath7 respectively , with @xmath8 and @xmath9 being arbitrary sets . note that the hilbert spaces are not necessarily finite or separable . even spaces with an uncountable bases are under study . an entanglement witness is a bounded hermitian operator @xmath10 , which has positive expectation values for separable states and it has negative eigenvalues @xcite . for our purposes a generalization of the class of entanglement witnesses is useful . one can think of bounded hermitian operators , which have positive expectation values for separable states : @xmath11 all operators fulfilling the conditions ( [ cond1 ] ) and ( [ cond2 ] ) shall define the set @xmath12 , operators in this set are called partial positive operators . all operators @xmath13 fulfilling the conditions @xmath14 @xmath15 together with eq . ( [ cond2 ] ) , with @xmath13 in place of @xmath10 , shall denote the set @xmath16 of positive semi - definite operators . so all entanglement witnesses are elements of the difference of these sets : @xmath17 . it was shown by the horodeckis @xcite , that for any entangled state @xmath18 there exists an entanglement witness @xmath19 , so that the expectation value becomes negative : @xmath20 . for this inseparability theorem only linear entanglement witnesses were used , which are sufficient to identify all entangled states . for this reason we restrict our considerations to linear witnesses , which are elements of the set @xmath12 . let us consider the important example of witnesses based on the pt map . recently it has been shown @xcite , that for any state @xmath18 with a negative pt ( npt ) there exists an operator @xmath21 , such that @xmath22 these operators have been studied in detail as functions of the annihilation and creation operators of two harmonic oscillators : @xmath23 . all the resulting @xmath24 are examples of elements of @xmath12 , in particular they include all entanglement witnesses for npt states . now we will turn to the construction of entanglement witnesses in their most general form . as outlined above , the problem of finding all entanglement witnesses via pncp maps is very difficult . here we will introduce a different but equivalent approach to entanglement witnesses , which requires the class of @xmath16 operators only . a hermitian operator @xmath13 is positive , if and only if it can be written as @xmath25 in the first step we will now generate any entanglement witness out of a difference of positive operators . for any entanglement witness @xmath10 exists a real number @xmath26 and a positive hermitian operator @xmath13 so that @xmath10 can be written as @xmath27 . proof . : : the bounded operator in spectral decomposition is @xmath28 , with @xmath29 being a projection - valued measure and @xmath30 the bounded set of eigenvalues . let the supremum of all eigenvalues be @xmath31 . for all separable quantum states , @xmath10 must be a positive map , which implies @xmath26 . @xmath32 which is the demanded form . to formulate a new entanglement theorem for positive hermitian operators , we need the definition of optimal entanglement witnesses as given by lewenstein _ et al . _ an entanglement witness @xmath33 is finer than @xmath34 , if and only if the entanglement of any state detected by @xmath35 is also detected by @xmath33 ( beside other entangled states , which are not detected by @xmath34 ) . an entanglement witness @xmath36 is optimal , if and only if no witness is finer than @xmath36 . therefore a state is separable , if and only if for all optimal entanglement witnesses the expectation value is positive . to find these witnesses , we need the function @xmath37 , which maps a general hermitian operator @xmath38 to its maximal expectation value for a separable state : @xmath39 it is obvious , that a hermitian operator @xmath27 is a general element of @xmath12 , if and only if @xmath40 . and it is optimal , if and only if @xmath41 . [ theo1 ] a state @xmath18 is entangled , if and only if there exists @xmath42 : @xmath43 . proof . : : let @xmath36 be an optimal witness , which detects the entanglement of @xmath18 : @xmath44 the other way around , a state @xmath45 is separable , if and only if for all @xmath13 : @xmath46 . this is a kind of distance criterion . our entanglement theorem [ theo1 ] does no longer require the explicit form of any entanglement witness . entanglement is completely verified by hermitian operators @xmath47 , which are given by eq . ( [ nonneg ] ) . the needed functions @xmath48 are readily obtained from eq . ( [ distfct ] ) . let us now consider a bounded hermitian operator @xmath38 , which can always be expressed in terms of a positive operator @xmath13 and a real number @xmath49 , @xmath50 it is obvious , that all bounded hermitian operators can be written in the form of @xmath38 . this can be used to further simplify the theorem [ theo1 ] . [ theo2 ] a state @xmath18 is entangled , if and only if there exists a hermitian operators @xmath38 : @xmath51 . proof . : : note that @xmath52 . the function @xmath53 is @xmath54 from theorem [ theo1 ] follows : @xmath55 from theorem [ theo2 ] entanglement can be numerically tested . the set @xmath16 and the set of bounded hermitian operators have the same cardinality . now the construction of positive hermitian operators , eq . ( [ nonneg ] ) , becomes superfluous and the number of tests does not increase . let us consider a simple implication of theorem [ theo2 ] . the entanglement test for a bounded hermitian operator @xmath56 reads as @xmath57 the entanglement condition for the state @xmath58 with the operator @xmath59 is @xmath60 | a , b \rangle \ } & < { \rm tr}(\hat{\varrho}[-\hat{a}])\\ \leftrightarrow \quad \inf\ { \langle a , b | \hat{a } | a , b \rangle \ } & > { \rm tr}(\hat{\varrho}\hat{a}).\label{eq:2ndcond}\end{aligned}\ ] ] equation ( [ eq:2ndcond ] ) is a second entanglement condition for the operator @xmath56 and it is equivalent to the original condition ( [ eq : cond2 ] ) for the operator @xmath59 . entanglement witnesses of the form @xmath61 had been considered before , see @xcite . here we gave the proof that any entanglement witness can be given in this form , which is a much stronger statement . this has been done for arbitrary dimensional hilbert spaces . let us consider the calculation of the function @xmath53 . the hermitian operator @xmath38 has the following projections : @xmath62= \langle a | \hat{a } | a\rangle , \\ \hat{a}_b&={\rm tr}_b [ \hat{a } \left ( \hat{1}_a\otimes |b\rangle\langle b| \right ) ] = \langle b | \hat{a } | b\rangle.\end{aligned}\ ] ] now the extrema of @xmath63 can be obtained . from eq . ( [ distfct ] ) , the extremum of the function @xmath64 is calculated under the constraints @xmath65 and @xmath66 . the functional derivatives are denoted by @xmath67 and @xmath68 . this leads to two lagrange multipliers @xmath69 and the conditions @xmath70 this can be written as : @xmath71 multiplying the first equation with @xmath72 and the second with @xmath73 , the lagrange multipliers are obtained as @xmath74 . this leads to the equations @xmath75 with the constraints @xmath76 , are called separability eigenvalue equations . the separability eigenvalue problem can be solved by computers or , in simple cases , by hand . the closed set @xmath77 is bounded in @xmath78 . the smooth function @xmath79 defined on this set is bounded by @xmath80 . therefore a solution of the separability eigenvalue equation exists . according to the calculation ( [ beginlem2 ] ) ( [ endlem2 ] ) we can formulate the following the function @xmath53 can be written as @xmath81 with @xmath82 being eigenvalues of the separability eigenvalue equations . : : see the calculations ( [ beginlem2 ] ) ( [ endlem2 ] ) and eq . ( [ distfct ] ) . we also obtain @xmath83 , with @xmath82 separability eigenvalue . in the following we want to consider some properties of the solution of the separability eigenvalue equations . let @xmath84 , @xmath85 be a solution of the separability eigenvalue problem of the bounded hermitian operator @xmath56 the vector @xmath86 has a schmidt decomposition , see @xcite , with the term @xmath87 , @xmath88 2 . if @xmath89 , @xmath90 ( or @xmath91 ) is another solution then @xmath92 ( or @xmath93 ) . if @xmath89 , @xmath94 is another solution then @xmath85 and @xmath94 are linearly independent . proof . : : 1 . the general form of the vector is @xmath95 . @xmath96 thus , @xmath97 . 2 . the first part of separability eigenvalue equations read as @xmath98 these are eigenvalue equations for the hermitian operator @xmath99 acting on @xmath4 for different eigenvalues . thus , @xmath92 . 3 . assume linear dependence , @xmath100 . since @xmath101 , we obtain @xmath102 . this would imply @xmath103 , which is a contradiction to @xmath89 . these properties of the separability eigenvalue equations can be easily seen for the solution of the example given in sec . [ ssec : a ] . in the following we want to study two aspects of the results obtained so far . in sec . [ ssec : a ] , an analytical solution of the separability eigenvalue problem will be derived for a special class of projection operators . in sec . [ ssec : b ] , a general entanglement test for arbitrary quantum states is under study . for example , let us solve the separability eigenvalue equations for the special class of operators of the form @xmath104 . the normalized vector @xmath105 can be expanded as @xmath106 in the same way the vectors @xmath107 and @xmath108 can be written as @xmath109 and @xmath110 . the separability eigenvalue equations can be written for each component as @xmath111 inserting eqs . ( [ exse1 ] ) and ( [ exse2 ] ) into each other and using eq . ( [ beginlem2 ] ) in the form @xmath112 for @xmath113 ( @xmath114 being a trivial case ) we can separate eq . ( [ exse1 ] ) from eq . ( [ exse2 ] ) , @xmath115 with the interpretation @xmath116 we get @xmath117 the positive and compact operators @xmath118 and @xmath119 can be given in spectral decomposition as @xmath120 thus the non - trivial separability eigenvalues are @xmath121 . using @xmath122 and @xmath123 , the state reads as @xmath124 and @xmath125 are orthonormal in each hilbert space . this is the schmidt decomposition of @xmath105 , cf . e.g. @xcite . by the above calculations we get the solutions @xmath126 for @xmath127 the state @xmath128 is factorized , and @xmath129 does not detect any entanglement . in all other cases , @xmath129 is useful to identify entanglement . now we can write the special condition for entanglement , by use of @xmath129 in theorem [ theo2 ] , as @xmath130 let us consider the example of two harmonic oscillators in a mixture of a superposition of coherent states , @xmath131 , with vacuum , @xmath132 , @xmath133 where @xmath134 and @xmath135^{-1/2}$ ] . for this state , higher - order moments are needed to verify npt - entanglement even for @xmath136 , see @xcite . based on the bell state @xmath137 , with @xmath138 , the condition ( [ purecond ] ) reads as @xmath139 it identifies entanglement of @xmath140 for certain values of @xmath141 , @xmath142 and @xmath143 . for other choices of @xmath105 , even the simple condition ( [ purecond ] ) may identify entanglement for more general parameters of the mixed state under study . let us now deal with the general form of theorem [ theo2 ] . we may explicitly construct all optimal entanglement witnesses as @xmath144 more generally , any element of @xmath12 can be written as @xmath145 ( @xmath146 ) . note that , if @xmath147 is a positive real number and @xmath10 an entanglement witness , then @xmath148 is also an entanglement witness , which is as fine as @xmath10 . in the following we consider an implementation of our method for a finite dimensional hilbert space @xmath149 . a numerical implementation could be a set of hermitian operators @xmath150 with the properties : @xmath151 this is a kind of a spherical grid , with a maximal distance @xmath152 for any hermitian operator @xmath56 to at least one element @xmath153 , for arbitrarily small @xmath154 . for example let us consider the following construction . we use the most general form of a hermitian operator @xmath38 together with the norm @xmath155 , @xmath156 with @xmath157 . we obtain each element of this set @xmath150 for instance by varying each @xmath158 in the following way . the absolute value @xmath159 can be increased from @xmath160 to @xmath161 with steps @xmath162 . the argument @xmath163 can be increased from @xmath160 to @xmath164 with steps @xmath165 . each operator has the norm @xmath166 . we neglect operators with @xmath167 and renormalize the other operators as @xmath168 . using the definition of the norm , we obtain that for each @xmath56 with @xmath169 exist one element @xmath153 with @xmath170 . in this way we may construct the test operators @xmath153 for a desired precision @xmath154 . now one can solve with an appropriate algorithm the separability eigenvalue equations for each @xmath171 . this gives the values of @xmath172 and according to equation ( [ optwitcr ] ) the optimal witnesses @xmath173 let us consider a grid of only 6 hermitian operators . figure 1 indicates , to which extend these optimal witnesses identify entanglement . the gray area represents those entangled states , which are not identified . identify a manyfold of entangled states , except those in the gray areas . the set of separable states is approximated by the @xmath174.,width=245 ] the test for entanglement is connected with an error depending on the chosen value of @xmath154 . if @xmath154 becomes smaller , the gray area in fig . 1 is decreasing . for any given entangled state @xmath18 there exists an @xmath152 , so that its entanglement can be identified by at least one of the chosen hermitian operators @xmath171 . in practice , the possible values of @xmath154 can be related to the available experimental precision . the construction outlined above shows clearly that we need only a finite number of entanglement tests for the desired precission . as we have mentioned above , for npt states all witnesses can be given as @xmath175 . let us define the function @xmath176 . the following proposition shows that the solution of the separability eigenvalue problem for @xmath175 becomes superfluous . if a general bounded hermitian operator @xmath56 has the solution @xmath177 of the separability eigenvalue equations , then @xmath178 has the solution @xmath179 . it follows @xmath180 . : : since for the hermitian operators @xmath181 , we find for all @xmath182 @xmath183\\ & = { \rm tr } [ \hat{a}^{\rm pt } \left(|a\rangle\langle a| \otimes |b\rangle\langle b| \right)^{\rm pt}]\\ & = { \rm tr } [ \hat{a}^{\rm pt } \left(|a\rangle\langle a| \otimes |b^\ast\rangle\langle b^\ast| \right)]\\ & = g_{\rm pt}(a , b^\ast ) . \end{aligned}\ ] ] thus , the optimization will lead to the solutions @xmath179 . we find for the operators @xmath184 presented in sec . [ ssec : a ] the function @xmath185 , with @xmath186 being the schmidt coefficients . since for this example @xmath114 is a separability eigenvalue as well , see eq . ( [ eq : zeroeval ] ) , we get @xmath187 to obtain the known characterization for npt entangled states , @xmath188 , we use the entanglement condition as presented in eq . ( [ eq:2ndcond ] ) , which simplifies to @xmath189^{\rm pt})<0 $ ] . . from the npt side , a typical witness is @xmath190 . , width=283 ] note that an arbitrary operator @xmath178 is a hermitian operator as well . since we can shift any operator @xmath191 without changing the entanglement witness , @xmath192 any entanglement test can be performed with an operator of the form @xmath175 . all entangled states which remain non - negative under pt are bound entangled ones , see @xcite . the following characterization of ppt bound entangled states can be given : @xmath193 the first condition refers to ppt . the second condition identifies entanglement . the difference between @xmath175 as a witness for npt entanglement and for entanglement in general is equal to @xmath194 the minimal separability eigenvalue , see fig . 2 . in the present paper we have proven the general form of entanglement witnesses . on this basis we have derived necessary and sufficient conditions for bipartite entanglement . optimal entanglement inequalities have been given in the most general form . they have been formulated with arbitrary hermitian operators , which are easy to handle because of their well known structure and they are useful for applications in experiments . separability eigenvalue equations have been formulated . they serve for the optimization of the entanglement conditions for all chosen hermitian operators . some properties of the solution of these equations have been analyzed . the separability eigenvalue problem resembles the ordinary eigenvalue problem of hermitian operators , with the additional restriction that the solution is a factorizable vector . we have analytically solved the separability eigenvalue equations for a special class of projection operators . using these solutions , we could demonstrate entanglement of a mixed state given in terms of continuous variables . a general entanglement test of the proposed form can be implemented numerically , with its error being related to the available experimental precision . the method under study can also be used to identify any bound entangled state with a positive partial transposition . this requires to test the given states for the negativity of its partial transposition . it turned out that the separability eigenvalues remain unchanged under partial transposition . eventually , bound entanglement can be demonstrated by a combination of a general entanglement test and a test for the negativity of the partial transposition of the state under study .

necessary and sufficient conditions for bipartite entanglement are derived , which apply to arbitrary hilbert spaces . motivated by the concept of witnesses , optimized entanglement inequalities are formulated solely in terms of arbitrary hermitian operators , which makes them useful for applications in experiments . the needed optimization procedure is based on a separability eigenvalue problem , whose analytical solutions are derived for a special class of projection operators . for general hermitian operators , a numerical implementation of entanglement tests is proposed . it is also shown how to identify bound entangled states with positive partial transposition .

a shift of finite type ( sft ) is a dynamical system defined by finitely many local transition rules . these systems have been studied for their own sake @xcite , and they have also served as important tools for understanding other dynamical systems @xcite . each sft can be described as the set of bi - infinite sequences on a finite alphabet that avoid a finite list of words over the alphabet . thus , there are only countably many sfts up to the naming of letters in an alphabet . for the sake of simplicity , we state our results in terms of sfts in the , even though we prove more general results in terms of sequences of directed graphs in the subsequent sections . let @xmath0 be a nonempty sft ( for definitions , see section [ sftpresentations ] ) . let @xmath1 be the set of words of length @xmath2 that appear in @xmath0 . for @xmath4 in @xmath8 $ ] , let @xmath9 be the probability measure on the power set of @xmath1 given by choosing each word in @xmath1 independently with probability @xmath4 . the case @xmath10 puts uniform measure on the subsets of @xmath1 . for notation , let @xmath11 be the power set of @xmath1 . to each subset @xmath3 of @xmath1 , we associate the sft @xmath5 consisting of all points @xmath12 in @xmath0 such that each word of length @xmath2 in @xmath12 is contained in @xmath3 . with this association , we view @xmath9 as a probability measure on the sfts @xmath5 that can be built out of the subsets of @xmath1 . briefly , if @xmath0 has entropy @xmath13 and @xmath2 is large , then a typical random sft @xmath5 is built from about @xmath14 words , an @xmath4 fraction of all the words in @xmath1 , but not all of these words will occur in any point in @xmath5 . our main results can be stated as follows . let @xmath15 denote the artin mazur zeta function of @xmath0 ( see definition [ zetadefn ] ) . the first theorem deals with the likelihood that a randomly chosen sft is empty . [ emptythmintro ] let @xmath0 be a nonempty sft with entropy @xmath16 . let @xmath17 be the event that @xmath5 is empty . then for @xmath4 in @xmath8 $ ] , @xmath18$.}\ ] ] thus , when @xmath4 is in @xmath19 , there is an asymptotically positive probability of emptiness . the next theorem gives more information about what happens when @xmath4 lies in @xmath20 . [ subcritthmintro ] let @xmath0 be a nonempty sft with entropy @xmath16 . let @xmath21 be the event that @xmath5 has zero entropy , and let @xmath22 be the random variable on @xmath11 which is the number of irreducible components of @xmath5 . then for @xmath23 , @xmath24 ; the sequence @xmath25 converges in distribution to the random variable @xmath26 such that @xmath27 and for @xmath28 , @xmath29 where @xmath30 is an enumeration of the periodic orbits in @xmath0 ; the random variable @xmath26 has exponentially decreasing tail and therefore finite moments of all orders . our next result describes the entropy of the typical random sft when @xmath4 lies in @xmath31 $ ] . [ entropythmintro ] let @xmath0 be an sft with positive entropy @xmath16 . then for @xmath32 and @xmath33 , @xmath34 and the convergence to this limit is exponential in @xmath2 . finally , we have a result concerning the likelihood that a random sft will have a unique irreducible component of positive entropy when @xmath4 is near @xmath7 . [ msftthmintro ] let @xmath0 be an irreducible sft with positive entropy@xmath35 . let @xmath36 be the event that @xmath5 has a unique irreducible component @xmath37 of positive entropy and @xmath37 has the same period as @xmath0 . then there exists @xmath38 such that for @xmath39 , @xmath40 furthermore , the convergence to this limit is exponential in @xmath2 . there have been studies of other objects called random subshifts of finite type in the literature @xcite , but the objects studied here are rather different in nature . the present work is more closely related to perturbations of sfts , which have already appeared in works by lind @xcite in dimension @xmath7 and by pavlov @xcite in higher dimensions . in those works , the main results establish good uniform bounds for the entropy of an sft obtained by removing any single word of length @xmath2 from a sufficiently mixing sft as @xmath2 tends to infinity . random sfts may also be interpreted as dynamical systems with holes @xcite , in which case the words of length @xmath2 in @xmath0 that are forbidden in the random sft @xmath5 are viewed as ( random ) holes in the original system @xmath0 . the question of whether an sft defined by a set of forbidden words is empty has been studied in formal language theory and automata theory , and in that context it amounts to asking whether the set of forbidden words is _ unavoidable _ @xcite . also , the random sfts considered here can be viewed as specific instances of random matrices ( see @xcite ) or random graphs ( see @xcite ) , and the concept of directed percolation on finite graphs has appeared in the physics literature in the context of directed networks @xcite . to the best of our knowledge , the specific considerations that arise for our random sfts seem not to have appeared in any of this wider literature . the paper is organized as follows . section [ preliminaries ] contains the necessary background and notation , as well as some preliminary lemmas . the reader familiar with sfts and directed graphs may prefer to skip sections [ sftpresentations ] and [ seqsgraphs ] , referring back as necessary . in section [ emptiness ] we discuss the likelihood that a random sft is empty , and , in particular , we prove theorem [ emptythmintro ] . the remainder of the main results are split into two sections according to two cases : @xmath41 and @xmath42 $ ] . the case @xmath41 is treated in section [ subcriticalphase ] , and the case @xmath43 $ ] is addressed in section [ supercriticalphase ] . section [ remarks ] discusses some corollaries of the main results . for a detailed treatment of sfts and their presentations , see @xcite . in this section we describe three ways to present an sft : with a finite list of forbidden words over a finite alphabet , with a finite , directed graph , or with a square , nonnegative integer matrix . let @xmath44 be a finite set , which we will call the _ alphabet_. an element @xmath45 is called a _ word _ of length @xmath2 . let @xmath46 , endowed with the product topology induced by the discrete topology on @xmath44 . then @xmath47 is a compact metrizable space , which is called the _ full shift _ on @xmath48 . let @xmath49 be the left shift , that is , for @xmath50 in @xmath47 , let @xmath51 . with this definition @xmath52 is a homeomorphism of @xmath47 . a subset @xmath0 of @xmath47 is called shift - invariant if @xmath53 . a closed , shift - invariant subset of @xmath47 is called a _ subshift_. for any subshift @xmath0 , the _ language _ @xmath54 of @xmath0 is the collection of all finite words ( blocks ) that appear in some sequence @xmath12 in @xmath0 . note that @xmath55 , where @xmath1 is the set of all words of length @xmath2 that appear in some sequence @xmath12 in @xmath0 . [ by convention we set @xmath56 , where @xmath57 denotes the empty word . ] given a set @xmath58 of words on @xmath48 , we may define a subshift @xmath59 as the set of sequences @xmath12 in @xmath47 such that no word in @xmath60 appears in @xmath12 . one may check that this procedure indeed defines a subshift . if @xmath0 is a subshift and there exists a _ finite _ set of words @xmath61 such that @xmath62 , then @xmath0 is called a _ subshift of finite type _ ( sft ) . the natural notion of isomorphism for sfts is called conjugacy . two sfts @xmath0 and @xmath63 are _ conjugate _ , written @xmath64 , if there exists a homeomorphism @xmath65 such that @xmath66 . an sft @xmath0 is _ irreducible _ if for every two nonempty open sets @xmath67 and @xmath68 and every @xmath69 in @xmath70 , there exists @xmath71 such that @xmath72 . an sft @xmath0 is _ mixing _ if for every two nonempty open sets @xmath67 and @xmath68 in @xmath0 , there exists @xmath73 in @xmath70 such that for all @xmath74 , we have @xmath72 . mixing and irreducibility are conjugacy - invariant . we now define the higher block presentations of an sft . [ higherblockdef ] let @xmath0 be an sft . the @xmath2-_block presentation _ of @xmath0 , denoted @xmath75}$ ] , is defined as follows . the alphabet for @xmath75}$ ] is @xmath1 . we define the code @xmath76 by the equation @xmath77\ ] ] for all @xmath12 in @xmath0 . then @xmath75 } = \phi_n(x)$ ] . for all @xmath78 , we have that @xmath75 } \cong x$ ] , where the conjugacy is given by @xmath79 . the _ entropy _ of an sft @xmath0 is defined as @xmath80 . alternatively , one may define sfts in terms of finite directed graphs . a directed graph @xmath81 consists of a set of vertices @xmath68 and a set of edges @xmath82 such that for each edge @xmath83 , there is a unique initial vertex , , and a unique terminal vertex , @xmath84 . we view the edge @xmath85 as going from @xmath86 to @xmath87 . we allow self - loops , but for the sake of convenience we assume ( without loss of generality for our considerations ) that there are no multiple edges . in this paper we make the standing convention that `` graph '' means directed graph . we will collect our standing assumptions in standing assumptions [ stassump ] . [ edgeshiftdefn ] given a directed graph @xmath88 , we define the _ edge shift _ @xmath89 to be the set of all bi - infinite ( oriented ) walks on @xmath88 , _ that is _ , @xmath90 . any edge shift is an sft ( trivially ) . let us show that any sft is conjugate to an edge shift . if @xmath91 is an sft and @xmath92 is a finite set of forbidden words , then @xmath93 , where @xmath81 is defined as follows . let @xmath94 . then let @xmath95 and @xmath96 . further , for any edge @xmath97 , we let @xmath98 $ ] and @xmath99 $ ] . the same construction works with @xmath2 in place of @xmath73 for any @xmath74 . if @xmath88 is a graph such that @xmath93 , we say that @xmath89 is an edge presentation of @xmath0 , or sometimes just a presentation of @xmath0 . the _ adjacency matrix _ @xmath100 of a directed graph @xmath88 may be defined as follows . fix an enumeration of the vertices in @xmath88 . then let @xmath101 be the number of distinct edges @xmath85 in @xmath88 such that @xmath102 and @xmath103 . a square , nonnegative integral matrix @xmath100 is _ irreducible _ if for each pair @xmath104 and each @xmath69 , there exists @xmath105 such that @xmath106 . a matrix @xmath100 is _ nondegenerate _ if it has no zero row and no zero column . if @xmath100 is nondegenerate , then the edge shift @xmath107 is irreducible if and only if @xmath100 is irreducible also , if @xmath100 is nondegenerate , then the edge shift @xmath89 is mixing if and only if there exists @xmath73 such that for all @xmath74 and all pairs @xmath104 , it holds that @xmath108 . a matrix is _ primitive _ if it satisfies the latter property . a path in @xmath88 is a finite sequence @xmath109 of edges such that @xmath110 for @xmath111 . if @xmath112 is a path in @xmath88 , we say that @xmath113 goes from vertex @xmath114 to vertex @xmath115 . we denote by @xmath116 the set of paths of length @xmath117 in @xmath88 . by convention , we set @xmath118 . for a path @xmath113 in @xmath88 , let @xmath119 and @xmath120 be the set of vertices and the set of edges traversed by @xmath113 , respectively . let @xmath0 be an sft . an _ irreducible component _ @xmath63 of @xmath0 is a nonempty , maximal sft contained in @xmath0 such that @xmath63 is irreducible . let @xmath88 be a graph . an _ irreducible component _ @xmath37 of @xmath88 is a nonempty , maximal subgraph of @xmath88 such that the adjacency matrix of @xmath37 is irreducible . the reader should be advised that in some papers the definition of irreducible component includes trivial components ( a single vertex with no edges adjacent to it ) , but the definition given here does not include trivial components . [ nblockgraphdefn ] let @xmath88 be a finite , directed graph . for @xmath78 , define @xmath121 } = ( v^{[n]},e^{[n]})$ ] , the @xmath2-_block graph _ of @xmath88 , as follows . let @xmath122 } = b_{n-1}(g)$ ] and @xmath123 } = b_{n}(g)$ ] , such that if @xmath124}$ ] , then @xmath125 $ ] and @xmath126 $ ] . note that @xmath127 } = g$ ] . if @xmath128 for some graph @xmath88 , then it follows immediately from the definitions that @xmath75 } = x_{g^{[n]}}$ ] . [ pthpowerdefn ] let @xmath81 be a graph . for @xmath129 in @xmath70 , we define the @xmath129th power graph , @xmath130 , as follows . let @xmath131 and @xmath132 . if @xmath133 is an edge in @xmath134 , then we let @xmath135 and @xmath136 . [ transposedefn ] let @xmath81 be a graph . define the transpose graph , @xmath137 , as follows . let @xmath138 and @xmath139 , where an edge @xmath85 in @xmath140 goes from @xmath87 to @xmath86 . in other words , the transpose graph is just the graph formed by reversing the direction of all the edges in @xmath88 . given a square , nonnegative , integral matrix @xmath100 , one may also define an sft @xmath141 as follows . let @xmath88 be a directed graph whose adjacency matrix is exactly @xmath100 ( such a graph always exists ) . then let @xmath141 be the edge shift defined by @xmath88 . recall the following basic facts ( which may be found in @xcite ) . for an sft @xmath0 , we have @xmath142 . if @xmath0 is a nonempty sft and @xmath143 for a square , nonnegative integral matrix @xmath100 , then @xmath16 , where @xmath144 is the spectral radius of @xmath100 . by the perron frobenius theorem , if @xmath100 is nonnegative and irreducible , then there exists a strictly positive ( column ) vector @xmath145 such that @xmath146 , and there exists a strictly positive ( row ) vector @xmath147 such that @xmath148 . furthermore , @xmath145 and @xmath147 are each unique up to a positive scalar . [ nonzerospec ] for any nonnegative integer matrix @xmath100 , let @xmath149 be the spectral radius of @xmath100 , and let @xmath150 be the characteristic polynomial of @xmath100 . then let @xmath151 be the _ nonzero spectrum _ of the matrix @xmath100 , which is defined as the multiset of nonzero roots of @xmath150 listed according to their multiplicity . if @xmath100 is the adjacency matrix of the graph @xmath88 , we define @xmath152 and @xmath153 . if @xmath154 for two nonnegative integral matrices @xmath100 and @xmath155 , then @xmath156 . also , if @xmath100 is primitive , then @xmath157 . finally , if @xmath100 is irreducible , then there exists a unique @xmath52-invariant borel probability measure @xmath158 on @xmath159 of maximal entropy . let us describe some basic properties of @xmath160 . we associate a word @xmath161 in @xmath0 to the cylinder set @xmath162 = b\}$ ] . in this way we interpret the measure of words in @xmath163 as the measure of the corresponding cylinder set . let @xmath145 be a positive right eigenvector of @xmath100 and @xmath147 a positive left eigenvector of @xmath100 , and suppose they are normalized so that @xmath164 . our standing assumption that there are no multiple edges means that @xmath165 for all @xmath104 . then for a vertex @xmath166 in @xmath68 , we have @xmath167 , and for @xmath168 , we have that @xmath169 now we define two objects , the period and the zeta function , which contain combinatorial information about the cycles in a graph @xmath88 ( alternatively , one may refer to the periodic points in an sft @xmath0 ) . [ perioddefn ] for an sft @xmath0 , let @xmath170 be the greatest common divisor of the sizes of all periodic orbits in @xmath0 . for a graph @xmath88 , let @xmath171 be the greatest common divisor of the lengths of all cycles in @xmath88 . [ zetadefn ] let @xmath0 be an sft and @xmath172 . then the artin mazur zeta function of @xmath0 ( see @xcite ) is , by definition , @xmath173 for a graph @xmath88 , let @xmath174 . for a graph @xmath88 , note that @xmath175 is the number of cycles of ( not necessarily least ) period @xmath129 in @xmath88 , and @xmath176 also , @xmath177 has radius of convergence @xmath178 and @xmath179 . in this work we consider sequences of graphs @xmath180 that grow in some way . a particular example of such a sequence is the sequence of @xmath2-block graphs of an sft @xmath0 . indeed , by taking @xmath180 to be such a sequence in theorems [ emptythm ] , [ subcritthm ] , [ entropythm ] and [ msftthm ] , we obtain the theorems stated in the . generalizing to the graph setting also allows one to consider sequences of graphs presenting sfts which are conjugate to a fixed sft @xmath0 , where the sequences need not be the @xmath2-block sequence for @xmath0 . to indicate the generality of the arguments further , though , we formulate and prove the results for sequences of graphs that do not necessarily present conjugate sfts . before we move on to these results , we need to define several notions regarding the manner of growth of the sequence @xmath180 . let @xmath88 be a finite , directed graph with adjacency matrix @xmath100 . we will have use for the following notation . [ perdefn ] let @xmath181 for @xmath113 in @xmath182 , let @xmath183 be the set of all paths @xmath184 in @xmath182 such that there exists a natural number @xmath185 such that @xmath186 , where @xmath187 is the permutation of @xmath188 defined in cycle notation by @xmath189 . for each vertex @xmath166 in @xmath88 , let @xmath190 and @xmath191 . then let @xmath192 in order to measure the separation of periodic orbits in @xmath88 , we make the following definition . [ defnofz ] let @xmath193 where @xmath119 is the set of vertices traversed by the path @xmath113 . as a measure of the size of @xmath88 , we consider the following quantity . if @xmath100 has spectral radius @xmath194 , then let @xmath195 to measure a range for uniqueness of paths in @xmath88 , we make the following definitions . [ udefn ] let @xmath196 we use the transition length as a type of diameter of @xmath88 . let @xmath197 here we briefly recall the notion of the weighted cheeger constant of an irreducible , directed graph @xmath88 . the weighted cheeger constant was defined and studied in @xcite . let @xmath160 be the measure of maximal entropy of @xmath89 , and let @xmath198 $ ] be given by @xmath199 . for any vertex @xmath145 in @xmath68 , let @xmath200 . then for any subset of vertices @xmath201 , let @xmath202 , and for any two subsets @xmath203 , let @xmath204 in general , @xmath205 is not symmetric in @xmath206 and @xmath207 since @xmath88 is directed . let @xmath208 be the set of edges @xmath85 in @xmath88 such that @xmath209 and @xmath210 . let @xmath211 . [ cheegerdefn ] the weighted cheeger constant of @xmath88 is defined as @xmath212 and the unweighted cheeger constant of @xmath88 is defined as @xmath213 [ uniformexpseqdef ] we say that @xmath88 is a directed @xmath113-expander graph if @xmath214 . also , a sequence of directed graphs @xmath180 is a _ uniform expander sequence _ , if there exists a @xmath215 such that @xmath216 is a directed @xmath113-expander for each @xmath2 . we will also have use for the following quantity related to the spectral gap of @xmath88 . [ spgapdefn ] let @xmath217 . we make the following standing assumptions , even though some of the statements we make may hold when these restrictions are relaxed . in particular , theorems [ emptythm ] and [ subcritthm ] do not require that @xmath218 is irreducible , nor do they require that @xmath194 ( see remark [ relaxrmk ] ) . [ stassump ] recall that `` graph '' means directedgraph . let @xmath180 be a sequence of graphs with an associated sequence of adjacency matrices @xmath219 . unless otherwise stated , we will make the following assumptions : * for each @xmath2 , each entry of @xmath218 is contained in @xmath220 ; * each @xmath218 is irreducible ; * for each @xmath2 , @xmath221 ; * @xmath222 ; * @xmath223 . note that @xmath224 , which depends only on @xmath225 and @xmath129 . therefore , the standing assumptions imply that @xmath226 does not depend on @xmath2 , and , therefore , @xmath227 and @xmath228 do not depend on @xmath2 . additional conditions that we place on sequences of graphs will come from the following list . [ different theorems will require different assumptions , but the sequence of @xmath2-block graphs of an irreducible graph with spectral radius greater than @xmath7 will satisfy conditions ( c1)(c8 ) below by proposition [ highergraphseqlemma ] . ] [ conditions ] we define the following conditions on a sequence of graphs @xmath180 with a sequence of adjacency matrices @xmath219 : there exists @xmath229 such that @xmath230 for all @xmath2 ( bounded degree ) ; @xmath231 tends to infinity as @xmath2 tends to infinity ( separation of periodic points ) ; there exists @xmath232 such that @xmath233 for all @xmath2 ( fast separation of periodic points ) ; there exists @xmath232 such that @xmath234 for all @xmath2 ( local uniqueness of paths ) ; there exists @xmath235 such that @xmath236 for all @xmath2 ( small diameter ) ; there exists @xmath237 such that @xmath238 for all @xmath2 ( bounded distortion of vertices ) and @xmath239 for all @xmath2 ( bounded distortion of edges ) ; there exists @xmath237 such that @xmath240 and @xmath241 for all @xmath2 , where @xmath242 is a positive left eigenvector of @xmath218 and @xmath243 is a positive right eigenvector of @xmath218 ( bounded distortion of weights ) ; @xmath180 is a uniform expander sequence , and @xmath244 is a uniform expander sequence ( forward / backward expansion ) . now we establish some lemmas , which will be used in the subsequent sections . [ c7impliesc1c6 ] let @xmath180 be a sequence of graphs satisfying the standing assumptions [ stassump ] . then implies and for both @xmath180 and @xmath244 . first note that if ( c7 ) holds for @xmath180 , then it also holds for @xmath244 since a positive left eigenvector for @xmath245 is given by @xmath246 and a positive right eigenvector for @xmath245 is given by @xmath247 . therefore , we only need to show that ( c7 ) for @xmath180 implies ( c1 ) and ( c6 ) for @xmath180 [ since the same argument will apply to @xmath244 ] . let @xmath242 and @xmath243 be positive left and right eigenvectors for @xmath218 , respectively , and assume that @xmath249 . recall with this normalization , if @xmath166 is a vertex in @xmath250 , then @xmath251 . then condition ( c7 ) implies that there exists @xmath252 such that for all @xmath2 , @xmath253 similarly , ( c7 ) implies that there exists @xmath254 such that for all @xmath2 , we have that @xmath255 [ recall that @xmath256 . thus , ( c7 ) implies ( c6 ) . note that for @xmath85 in @xmath257 , we have that @xmath258 then condition ( c7 ) implies that there exists a uniform constant @xmath252 such that @xmath259 for all @xmath2 and all @xmath85 in @xmath257 . we also have that @xmath260 since @xmath216 is irreducible ( by standing assumptions [ stassump ] ) , we know that @xmath261 , and , therefore , we have that for any @xmath2 , and any @xmath166 in @xmath250 , @xmath262 which implies that @xmath263 is uniformly bounded in @xmath2 . a similar argument shows that @xmath264 is uniformly bounded in @xmath2 , which shows that @xmath265 is uniformly bounded in @xmath2 and gives ( c1 ) . recall that for a graph @xmath88 , the quantities @xmath266 and @xmath267 were defined in definitions [ spgapdefn ] and [ cheegerdefn ] , respectively . [ mixexplemma ] let @xmath88 be a graph with primitive adjacency matrix @xmath100 . then it holds that @xmath268 . this lemma is a consequence of @xcite , theorems 4.3 and 5.1 , as we now explain . since @xmath100 is primitive , there exists a strictly positive vector @xmath145 and @xmath269 such that @xmath146 . let @xmath270 be the stochastic matrix defined by @xmath271 . then @xmath270 is the transition probability matrix corresponding to the random walk defined by the measure of maximal entropy @xmath160 on @xmath89 . we have that @xmath272 . given such a transition probability matrix , chung defines a laplacian @xmath273 and proves ( @xcite , theorem 4.3 ) , that the smallest nonzero eigenvalue of @xmath273 , denoted @xmath274 , satisfies the following inequality : @xmath275 we remark that the left - hand side of the inequality in @xcite , theorem 4.3 , is equal to the left - hand side of ( [ chung4.3 ] ) since @xmath100 is primitive ( not just irreducible ) . note that the left - hand side of ( [ chung4.3 ] ) equals @xmath266 , as defined in definition [ spgapdefn ] . after defining the weighted cheeger constant ( as in definition [ cheegerdefn ] ) , chung proves ( @xcite , theorem 5.1 ) , that @xmath276 combining the inequalities in ( [ chung4.3 ] ) and ( [ chung5.1 ] ) , we obtain the desired inequality . recall that the @xmath129th power graph was defined in definition [ pthpowerdefn ] . [ irrexplemma ] let @xmath88 be a graph with an irreducible adjacency matrix . let @xmath277 . let @xmath278 be an irreducible component of @xmath134 , the @xmath129th power graph of @xmath88 . let @xmath279 ( which does not depend on the choice of irreducible component in @xmath134 ) . then there exists @xmath280 , depending only on @xmath281 and @xmath129 , such that @xmath282 . let @xmath88 , @xmath129 and @xmath281 be as in the statement of the lemma . if @xmath283 , then lemma [ mixexplemma ] immediately gives the result . now we assume @xmath284 . the fact that @xmath88 is irreducible and @xmath285 implies that there is a partition of the vertices into @xmath129 nonempty subsets , @xmath286 , such that for each edge @xmath85 with @xmath287 , it holds that @xmath288 , where the superscripts are taken modulo @xmath129 . let @xmath289 ( definition [ edgeshiftdefn ] ) , and for each @xmath290 , let @xmath291 . for any set @xmath292 with @xmath293 and @xmath290 , define @xmath294 recall that we denote by @xmath160 the measure of maximal entropy on @xmath0 , and we may write @xmath267 as follows : @xmath295 we also use the following notation : @xmath296 let us establish a useful inequality . for @xmath297 and @xmath298 , note that each point @xmath12 in @xmath299 also lies in @xmath300 for @xmath301 . thus , @xmath302 to complete the proof , we will find @xmath215 in terms of @xmath281 and @xmath129 so that for @xmath292 with @xmath293 , we have that @xmath303 the bound @xmath113 will be the minimum of four bounds , each coming from a particular type of set @xmath292 . consider the following conditions on the set @xmath206 , which we will use to break our proof into cases : there exists @xmath304 such that @xmath305 ; @xmath306 for each @xmath307 , or @xmath308 for each @xmath307 ; @xmath309 for each @xmath307 . now we consider cases . _ case _ : ( i ) holds , _ that is _ , there exists @xmath304 such that @xmath310 . assume first that @xmath311 , which implies that @xmath312 . choose @xmath313 such that @xmath314 , and finally choose @xmath315 such that @xmath316 ( @xmath317 ) . then by inequality ( [ csestimate ] ) and the shift - invariance of @xmath160 , we have that @xmath318 now assume @xmath319 . choose @xmath313 such that @xmath320 , and finally choose @xmath298 such that @xmath321 ( @xmath322 ) . then by ( [ csestimate ] ) and the shift - invariance of @xmath160 , @xmath323 let @xmath324 , and note that if condition ( i ) holds , then the inequality in ( [ sboundeqn ] ) holds with @xmath325 in place of @xmath113 . _ case _ : ( i ) does not hold , but ( ii ) holds , _ that is _ , @xmath326 for all @xmath307 , or @xmath327 for all @xmath307 . assume first that @xmath328 for all @xmath307 . since @xmath329 and @xmath330 for all @xmath307 , there exists @xmath313 such that @xmath331 . then by ( [ csestimate ] ) and the definition of @xmath332 in ( [ r_idefn ] ) , @xmath333 let @xmath334 be the irreducible component of @xmath134 with vertex set @xmath335 . then @xmath334 has primitive adjacency matrix , and @xmath336 . lemma [ mixexplemma ] gives that @xmath337 . since @xmath338 and @xmath339 , we have that @xmath340 , and , thus , @xmath341 let @xmath342 . we have shown that for @xmath206 such that @xmath343 for each @xmath307 , the inequality in ( [ sboundeqn ] ) holds with @xmath344 in place of @xmath113 . for @xmath206 such that @xmath345 for each @xmath307 , choose @xmath313 such that @xmath346 . then an analogous argument gives that the inequality in ( [ sboundeqn ] ) holds with @xmath344 in place of @xmath113 . _ case _ : ( iii ) holds , _ that is _ , @xmath347 for all @xmath307 . a simple calculation yields that @xmath348 and @xmath349 for each @xmath307 . using ( [ csestimate ] ) , we see that for each @xmath313 , @xmath350\\[-8pt ] & \geq & \frac{1}{3p } \cdot\frac { \mu(c_s^j \cap\sigma^{-p } { \hspace*{1pt}\overline}{c}{}^j_s ) } { \mu(c_s^j ) } \nonumber\end{aligned}\ ] ] and @xmath351\\[-8pt ] & \geq&\frac{1}{3p } \cdot \frac { \mu(c_s^j \cap\sigma^{-p } { \hspace*{1pt}\overline}{c}{}^j_s ) } { \mu(\overline { c}_s^j ) } .\nonumber\end{aligned}\ ] ] then since @xmath334 has a primitive adjacency matrix , lemma [ mixexplemma ] and inequalities ( [ caseiiiestimatei ] ) and ( [ caseiiiestimateii ] ) give that the inequality in ( [ sboundeqn ] ) holds with @xmath352 in place of @xmath113 . _ case _ : each of ( i ) , ( ii ) and ( iii ) does not hold , _ that is _ , we assume that @xmath206 is such that @xmath353 for each @xmath307 , there exists @xmath354 and @xmath355 such that @xmath356 and @xmath357 , and there exists @xmath358 such that either @xmath359 or @xmath360 . suppose first that @xmath359 . choose @xmath313 such that @xmath361 , and choose @xmath315 such that @xmath362 ( @xmath322 ) . calculation gives that @xmath363 . then by ( [ csestimate ] ) and the shift - invariance of @xmath160 , @xmath364 now assume @xmath360 . choose @xmath313 such that @xmath365 and choose @xmath315 such that @xmath366 ( @xmath322 ) . calculation reveals that @xmath367 . then by ( [ csestimate ] ) and the shift - invariance of @xmath160 , @xmath368 let @xmath369 . we have shown that for @xmath206 in this case , the inequality in ( [ sboundeqn ] ) holds with @xmath370 in place of @xmath113 . now let @xmath371 , which depends only on @xmath281 and @xmath129 . we have shown that @xmath372 . recall that the transpose graph @xmath140 of a graph @xmath88 was defined in definition [ transposedefn ] . [ unifexplemma ] let @xmath180 be a sequence of graphs satisfying the standing assumptions [ stassump ] and such that both @xmath180 and @xmath244 have bounded degrees and bounded distortion of edges and vertices [ conditions and in definition [ conditions ] ] . then @xmath180 and @xmath244 are both uniform expander sequences [ condition in definition [ conditions ] ] . we check that conditions ( c1 ) and ( c6 ) for @xmath180 together imply that @xmath180 is a uniform expander sequence , and then the same argument will apply to @xmath244 since ( c1 ) and ( c6 ) also hold for @xmath244 . recall the following notation . let @xmath373 $ ] be given by @xmath199 , where @xmath160 is the measure of maximal entropy on @xmath374 . also , @xmath375 denotes the weighted cheeger constant of @xmath216 ( definition [ cheegerdefn ] ) . by the standing assumptions [ stassump ] , @xmath376 for each @xmath2 . therefore , @xmath227 does not depend on @xmath2 , and we let @xmath377 . let @xmath378 be an irreducible component of the @xmath129th power graph of @xmath216 , and let @xmath379 . since @xmath380 only depends on the nonzero spectrum of @xmath216 , which is constant in @xmath2 by the standing assumptions [ stassump ] , we have the @xmath380 is constant in @xmath2 . let @xmath381 . by lemma [ irrexplemma ] , there exists @xmath382 , depending only @xmath380 and @xmath227 , such that @xmath383 . since we have that @xmath384 and @xmath385 for all @xmath2 , we may choose @xmath386 , and we obtain that @xmath387 for all @xmath2 . now we relate @xmath375 to @xmath388 ( definition [ cheegerdefn ] ) using properties ( c1 ) and ( c6 ) . for notation , let @xmath389 . since @xmath180 satisfies conditions ( c1 ) and ( c6 ) , there exists @xmath390 such that for every @xmath2 and every subset @xmath391 , @xmath392 and @xmath393 , which implies that for every @xmath206 such that @xmath394 , @xmath395 now assume @xmath396 . if @xmath397 , then @xmath398 , @xmath399 . if @xmath398 , @xmath400 , then we have@xmath398 , @xmath401 . combining these estimates gives that for all @xmath206 such that @xmath402 , we obtain that @xmath403 which shows that @xmath180 is a uniform @xmath404-expander sequence . [ implicationslemma ] let @xmath180 be a sequence of graphs satisfying the standing assumptions [ stassump ] and bounded distortion of weights [ condition in definition [ conditions ] ] . then : there exists @xmath237 such that for all @xmath2 , @xmath117 and @xmath405 , @xmath406 there exists a @xmath237 such that for all @xmath2 , @xmath117 , @xmath407 , and @xmath405 , @xmath408 where @xmath409 ; there exists @xmath237 such that for all @xmath2 , @xmath117 , and @xmath410 , it holds that @xmath411 , where @xmath412 ; there exists @xmath237 such that for all @xmath2 , @xmath413 , and @xmath414 , it holds that @xmath415 , where @xmath416 and @xmath417 was defined in definition [ udefn ] . for notation , let @xmath389 and @xmath418 . _ proof of _ ( 1 ) . we have that @xmath419 then condition ( c7 ) implies that there exists @xmath420 such that for each @xmath2 and @xmath166 in @xmath250 , @xmath421 by the definition of @xmath422 , there exists @xmath423 such that @xmath424 such that for each @xmath2 and @xmath166 in @xmath250 , @xmath425 then ( c7 ) implies that there exists @xmath426 such that for any @xmath2 and any three vertices @xmath166 , @xmath427 and @xmath428 in @xmath250 , @xmath429 finally , we conclude that there exists @xmath430 such that for each @xmath2 , @xmath117 and @xmath113 in @xmath431 , we have that @xmath432 the statement in ( 1 ) follows . _ proof of _ ( 2 ) . the statement in ( 2 ) follows from the statement in ( 1 ) and the fact that @xmath433 . _ proof of _ ( 3 ) . note that from ( 1 ) we have that there exists @xmath252 such that @xmath434 since @xmath225 does not depend on @xmath2 by our standing assumptions [ stassump ] , we have that @xmath435 does not depend on @xmath2 . clearly , @xmath436 is bounded as @xmath437 tends to infinity . therefore , there exists @xmath438 such that @xmath439 as desired . _ proof of _ ( 4 ) . by ( 2 ) , we have that there exists @xmath420 such that for all @xmath2 , @xmath440 , and @xmath166 in @xmath250 , @xmath441 by ( 2 ) , there exists @xmath423 such that for all @xmath2 , @xmath442 , and @xmath166 in @xmath250 , @xmath443 by definition of the uniqueness parameter @xmath67 , each path in @xmath444 can be continued in at most one way to form a path in @xmath445 . therefore , with @xmath446 , we have that for all @xmath2 , @xmath447 , and @xmath166 in @xmath250 , @xmath448 [ highergraphseqlemma ] let @xmath449 be a graph with irreducible adjacency matrix @xmath450 having entries in @xmath220 and spectral radius @xmath451 . let @xmath452}$ ] for @xmath453 . then the sequence @xmath180 satisfies the standing assumptions [ stassump ] and conditions . moreover , @xmath454 for all @xmath2 ; there exists @xmath235 such that @xmath455 for all @xmath2 ; @xmath456 for all @xmath2 ; @xmath457 for all @xmath2 ; @xmath458 for all @xmath2 . one may easily check from the definitions that each @xmath218 has entries in @xmath220 , each @xmath218 is irreducible , and @xmath376 . we show below that @xmath459 tends to infinity as @xmath2 tends to infinity , which gives that @xmath180 satisfies the standing assumptions [ stassump ] . the set of in - degrees that appear in @xmath216 is constant in @xmath2 , and so is the set of out - degrees that appear in @xmath216 . therefore , @xmath460 , which implies condition ( c1 ) . by definition , since @xmath462}$ ] , we have that @xmath463 . by the standard perron frobenius theory , there exist constants @xmath464 and @xmath465 such that @xmath466 . it follows that there exists a constant @xmath235 such that @xmath467 , and , in particular , @xmath459 tends to infinity . recall the higher - block coding map @xmath468 ( see definition [ higherblockdef ] ) . if @xmath12 is a point in @xmath469 , then let @xmath470 be the set of vertices in @xmath216 traversed by @xmath471 . let us show that @xmath472 . recall fine and wilf s theorem @xcite , which can be stated as follows . let @xmath12 be a periodic sequence with period @xmath129 , and @xmath473 be a periodic sequence with period @xmath474 . if @xmath475 = y[i+1,i+n]$ ] for @xmath476 and @xmath307 in @xmath477 , then @xmath478 . it follows from this theorem that if @xmath12 and @xmath473 lie in distinct periodic orbits of @xmath469 and have periods less than or equal to @xmath479 , then @xmath480 . thus , @xmath472 , and , in particular , @xmath180 satisfies conditions ( c2 ) and ( c3 ) . note that the map @xmath79 gives a bijection between @xmath431 and @xmath481 for all @xmath482 . using this map , we check that @xmath457 as follows . for any two paths @xmath483 , there is at most one path of length @xmath484 in @xmath449 of the form @xmath485 ( since every edge in such a path is specified by either @xmath113 or @xmath184 ) . this fact implies that @xmath486 . now if @xmath113 is in @xmath487 and @xmath488 are given , then there is at most one path @xmath184 in @xmath489 such that @xmath490=b$ ] and @xmath491 = c[t , t+n-2]$ ] ; indeed , if @xmath184 is such a path , then @xmath490 $ ] is determined by @xmath113 , and @xmath492 $ ] is determined by the periodicity condition @xmath491=c[t , t+n-2]$ ] . this fact implies that @xmath493 , and , thus , we have that @xmath457 , which , in particular , gives condition ( c4 ) . let us check that @xmath458 , which will imply that @xmath180 satisfies condition ( c5 ) . the statement that @xmath494 is equivalent to the statement that for any two paths @xmath495 , there exists a path @xmath496 in @xmath449 of length less than or equal to @xmath497 such that @xmath498 is a path in @xmath449 . in this formulation , the statement is clearly true , since , by the definition of @xmath497 , there is a path @xmath496 from @xmath499 to @xmath500 of length less than or equal to @xmath497 , and then the concatenation @xmath498 gives a path in @xmath449 . let @xmath501 be a positive left ( row ) eigenvector for @xmath450 ( corresponding to the eigenvalue @xmath144 ) , and let @xmath502 be a positive right ( column ) eigenvector for @xmath450 ( corresponding to the eigenvalue @xmath503 ) . let @xmath504 . then let @xmath505 and @xmath506 . then @xmath242 is a positive left eigenvector for @xmath218 and @xmath243 is a positive right eigenvector for @xmath218 . it follows that @xmath180 conditions ( c6 ) and ( c7 ) . in fact , to satisfy ( c7 ) , we may choose @xmath507 , where @xmath508 and @xmath509 . condition ( c8 ) follows from the fact that @xmath180 satisfies condition ( c7 ) ( by applying lemmas [ c7impliesc1c6 ] and [ unifexplemma ] in succession ) . let @xmath510 be the probability space consisting of the set @xmath511 and the probability measure @xmath9 , where @xmath9 is the product of the bernoulli measures on each coordinate with parameter @xmath512 $ ] . there is a natural partial order on @xmath510 , given by the relation @xmath513 if and only if @xmath514 for @xmath515 . we say that a random variable @xmath516 on @xmath510 is _ monotone increasing _ if @xmath517 whenever @xmath513 . an event @xmath100 is monotone increasing if its characteristic function is monotone increasing . monotone decreasing is defined analogously . monotone random variables and events have been studied extensively @xcite ; however , we require only a small portion of that theory . in particular , we will make use of the following proposition , a proof of which may be found in @xcite . [ fkg ] if @xmath0 and @xmath63 are monotone increasing random variables on @xmath511 , then @xmath518 . it follows easily from the fkg inequality that if @xmath519 is a finite intersection of monotone decreasing events , then @xmath520 ( use induction and note that if @xmath521 is the characteristic function of the monotone decreasing event @xmath522 , then @xmath523 is monotone increasing ) . in fact , we only use this corollary , but we nonetheless refer to it as the fkg inequality . for a finite , directed graph @xmath88 , we consider the discrete probability space on the set @xmath524 , where @xmath9 is the product of the bernoulli(@xmath4 ) measures on each coordinate . the set @xmath525 corresponds to the power set of @xmath82 in the usual way : @xmath3 in @xmath525 corresponds to the set @xmath522 in @xmath526 such that @xmath85 is in @xmath522 if and only if @xmath527 . furthermore , @xmath525 corresponds to the space of subgraphs of @xmath88 : for @xmath3 in @xmath525 , define the subgraph @xmath528 to have vertex set @xmath68 and edge set @xmath529 , where an edge @xmath85 in @xmath82 is included in @xmath530 if and only if @xmath531 . in the percolation literature , the edges @xmath85 such that @xmath531 are often called `` open , '' and the remaining edges are called `` closed . '' since we are interested in studying edge shifts defined by graphs , we will refer to an edge @xmath85 as `` allowed '' when @xmath531 and `` forbidden '' when @xmath532 . finally , each @xmath3 in @xmath525 can be associated to the sft @xmath5 defined as the set of all bi - infinite , directed walks on @xmath88 that traverse only _ allowed _ edges ( with respect to @xmath3 ) . the probability measure @xmath9 corresponds to allowing each edge of @xmath88 with probability @xmath4 , independently of all other edges . for the sake of notation , we suppress the dependence of @xmath9 on the graph @xmath88 . in this work we consider the following conjugacy invariants of sfts . let @xmath533 be the property containing only the empty shift . let @xmath534 be the property containing all sfts with zero entropy . by convention , we let @xmath535 . for any sft @xmath0 , let @xmath536 be the topological entropy , and let @xmath537 be the number of irreducible components of @xmath0 . if @xmath0 is nonempty , let @xmath538 be defined by the equation @xmath539 . if @xmath0 is empty , let @xmath540 . if @xmath541 is a property of sfts and @xmath88 is a finite directed graph , then let @xmath542 be the set of @xmath3 in @xmath525 such that @xmath5 has property @xmath541 . if @xmath543 is a function from sfts to the real numbers and @xmath88 is a finite directed graph , then let @xmath544 be the function @xmath545 . recall that @xmath546 , @xmath177 and @xmath547 were defined in definitions [ nonzerospec ] , [ zetadefn ] and [ defnofz ] , respectively . [ emptythm ] let @xmath180 be a sequence of graphs such that @xmath376 for all @xmath2 and either @xmath548 or @xmath549 and @xmath231 tends to infinity as @xmath2 tends to infinity . . then @xmath551$.}\ ] ] [ emptyrmk ] theorem [ emptythmintro ] can be obtained as a corollary of theorem [ emptythm ] by taking @xmath180 to be the sequence of @xmath2-block graphs of @xmath0 . indeed , if the sft @xmath0 in theorem [ emptythmintro ] has zero entropy , then @xmath552 , and the conclusion of theorem [ emptythmintro ] follows from case ( i ) in theorem [ emptythm ] . if the sft @xmath0 in theorem [ emptythmintro ] has positive entropy , then @xmath194 and @xmath231 tends to infinity by the exact same argument in the proof of proposition [ highergraphseqlemma](iii ) , and , therefore , the conclusion of theorem [ emptythmintro ] follows from case ( ii ) in theorem [ emptythm ] . in this section we provide a proof of theorem [ emptythm ] . before proceeding with the proof , we state a fact that will be useful in the investigations that follow . recall that for a path @xmath113 , we denote by @xmath119 the set of vertices traversed by @xmath113 . [ perlemma ] suppose @xmath88 is a directed graph . suppose @xmath113 is in @xmath553 such that @xmath554 . then there exists a path @xmath184 in @xmath553 such that @xmath555 and @xmath556 . let @xmath145 be in @xmath119 . then there exists a return path to @xmath145 following @xmath113 , and we may choose a shortest return path @xmath184 to @xmath145 using only vertices in @xmath119 . then @xmath184 is in @xmath553 and @xmath555 , as desired . proof of theorem [ emptythm ] recall that an sft is nonempty if and only if it contains a periodic point ( see @xcite ) . first , assume that case ( i ) holds , which means that @xmath552 . in this case , each @xmath374 contains finitely many orbits . further , the number of periodic orbits of each period in @xmath374 is constant , and the probability of each periodic orbit being allowed in @xmath5 is constant . therefore , the conclusion follows immediately , since the sequence @xmath557 is constant . now assume that case ( ii ) holds . for the moment , consider a fixed natural number @xmath2 . let @xmath558 be an enumeration of the periodic orbits of @xmath374 such that if @xmath559 , then @xmath560 . let @xmath561 . let @xmath562 be the vertices in @xmath216 traversed in the orbit @xmath563 and let @xmath564 be the edges in @xmath216 traversed in the orbit @xmath563 . now for each @xmath313 , let @xmath565 be the event that @xmath563 is allowed , which is the event that all of the edges in @xmath564 are allowed . let @xmath566 be the event that @xmath563 is forbidden , which is @xmath567 , the complement of @xmath565 . notice that @xmath565 is a monotone increasing event ( if @xmath3 is in @xmath565 and @xmath568 , then @xmath569 is in @xmath565 ) , and @xmath566 is a monotone decreasing event . the fact that an sft is nonempty if and only if it contains a periodic point implies that @xmath570 . combining the definition of @xmath231 and lemma [ perlemma ] , we obtain that if @xmath571 , then @xmath572 . it follows that @xmath573 for each @xmath307 such that @xmath574 . furthermore , the definition of @xmath231 implies that the events @xmath575 such that @xmath574 are all jointly independent . these observations give that @xmath576 using lemma [ perlemma ] , we see that there is great redundancy in the intersection @xmath519 . eliminating some of this redundancy , we obtain the following : @xmath577 then using lemma [ perlemma ] again and the fact that @xmath578 , we see that the intersection on the right in ( [ infinitecaptofinite ] ) is actually a finite intersection . applying the fkg inequality , we obtain that @xmath579 combining the inequalities in ( [ emptyupperbd ] ) , ( [ emptyupperbd2 ] ) , ( [ emptylowerbd1 ] ) and ( [ emptylowerbd2 ] ) gives that for each @xmath2 , @xmath580 by the standing assumptions that @xmath581 , we have that @xmath582 is independent of @xmath2 . since @xmath231 and @xmath583 tend to infinity as @xmath2 tends to infinity , equation ( [ emptycombinedbds ] ) gives that @xmath584 then theorem [ emptythm ] follows from the well - known product formula for @xmath585 ( see @xcite ) , which may be stated as @xmath586 along with the fact that @xmath587 converges for @xmath588 and diverges to @xmath589 for @xmath590 . in this section we study random sfts in the subcritical phase : @xmath591 . the main result of this section is theorem [ subcritthm ] . let us fix some notation for this section . we consider a sequence of graphs @xmath180 such that @xmath581 and @xmath231 tends to infinity as @xmath2 tends to infinity , with @xmath592 and @xmath550 . since @xmath376 , there exist shift - commuting bijections @xmath593 . in other words , there exist bijections @xmath79 from the set of cyclic paths in @xmath449 to the set of cyclic paths in @xmath216 such that if @xmath113 is in @xmath594 , then @xmath595 is @xmath596 . if @xmath113 is in @xmath553 , then we refer to @xmath183 ( recall definition [ perdefn ] ) as a cycle . using the fixed bijections @xmath79 , we may refer to a cycle @xmath597 as being in @xmath216 for any @xmath2 . we fix an enumeration of the cycles in @xmath449 , @xmath598 , and then since the bijections @xmath79 are fixed , this choice simultaneously gives enumerations of all the cycles in each @xmath216 . for any @xmath599 in @xmath70 , let @xmath600 . let us begin with a lemma . [ finiteirrcomplemma ] let @xmath180 be a sequence of graphs such that @xmath376 and @xmath231 tends to infinity as @xmath2 tends to infinity , with @xmath592 and @xmath601 . given a nonempty , finite set @xmath206 in @xmath70 , let @xmath602 be the event that the set of allowed cycles is @xmath603 . then @xmath604$.}\ ] ] the proof of lemma [ finiteirrcomplemma ] is an easy adaptation of the proof of theorem [ emptythm ] , and we omit it for the sake of brevity . recall that @xmath537 denotes the number of irreducible components in the sft @xmath0 , and for any graph @xmath88 , the random variable @xmath605 is defined by the equation @xmath606 . [ subcritthm ] let @xmath180 be a sequence of graphs such that @xmath376 and either @xmath607 or @xmath549 and @xmath231 tends to infinity as @xmath2 tends to infinity . let @xmath550 . then for @xmath608 , @xmath609 ; the sequence @xmath610 converges in distribution to the random variable @xmath26 such that @xmath611 and for @xmath28 , @xmath612 where @xmath30 is an enumeration of the cycles in @xmath449 ; the random variable @xmath26 has exponentially decreasing tail and therefore finite moments of all orders . one obtains theorem [ subcritthmintro ] as a consequence of theorem [ subcritthm ] by taking @xmath180 to be the sequence of @xmath2-block graphs of a nonempty sft @xmath0 . indeed , if the sft @xmath0 in theorem [ subcritthmintro ] has zero entropy , then @xmath552 , and the conclusions of theorem [ subcritthmintro ] follow from the case ( i ) in theorem [ subcritthm ] . if the sft @xmath0 in theorem [ subcritthmintro ] has positive entropy , then @xmath194 and @xmath231 tends to infinity by the exact same argument in the proof of proposition [ highergraphseqlemma](iii ) , and , therefore , the conclusions of theorem [ subcritthmintro ] follow from case ( ii ) in theorem [ subcritthm ] . proof of theorem [ subcritthm ] let @xmath180 be as above . let @xmath613 . first , assume that case ( i ) holds , which means that @xmath552 . conclusion ( 1 ) follows immediately , since for each @xmath2 , we have that @xmath614 [ the random sft @xmath5 satisfies @xmath615 . also , the fact that @xmath552 is equivalent to the fact that @xmath449 ( and therefore @xmath216 ) contains only finitely many cycles . then conclusions ( 2 ) and ( 3 ) also follow immediately , since the sequence @xmath616 is constant . now assume that case ( ii ) holds . recall that we have an enumeration @xmath617 of the cycles in @xmath449 , which we refer to as an enumeration of the cycles in @xmath216 , for any @xmath2 , using the bijections @xmath79 . also recall that for any nonempty , finite set @xmath618 , we denote by @xmath602 the event in @xmath619 consisting of all @xmath3 such that the set of cycles in @xmath620 is exactly @xmath621 . _ proof of _ ( 1 ) . recall that an sft has zero entropy if and only if it has at most finitely many periodic points @xcite . then we have that @xmath622 also note that by the definition of @xmath602 , the union in ( [ zeroentunion ] ) is a disjoint union . thus , we have that @xmath623 be distinct , nonempty , finite subsets of @xmath70 . then by theorem [ emptythm ] and lemma [ finiteirrcomplemma ] , we have that @xmath624 since @xmath625 and @xmath626 were arbitrary , we conclude that @xmath627 using the facts that @xmath628 and @xmath629 ( which implies that the relevant infinite products and series converge uniformly ) , one may easily check that @xmath630 thus , we have shown that @xmath631 . since @xmath632 , we conclude that @xmath633 . _ proof of _ ( 2 ) . since @xmath616 takes values in @xmath634 , the sequence @xmath610 converges in distribution to @xmath26 if and only if @xmath635 converges to @xmath636 for each @xmath117 in @xmath634 . note that @xmath637 if and only if @xmath3 is in @xmath638 , which implies that @xmath639 . thus , for @xmath629 , theorem [ emptythm ] implies that@xmath640 converges to @xmath641 as @xmath2 tends to infinity . now let @xmath117 be in @xmath70 . recall that @xmath642 is an enumeration of the cycles in @xmath449 , and we have fixed bijections between these cycles and the cycles in each @xmath216 . by theorem [ subcritthm](1 ) , we have that @xmath643 , and , therefore , @xmath644 , where @xmath645 tends to @xmath646 as @xmath2 tends to infinity . thus , we need only focus on events of the form @xmath647 for some @xmath117 . now if @xmath3 is in @xmath648 , then @xmath649 is the number of periodic orbits in @xmath650 . thus , @xmath651 for any @xmath2 in @xmath70 , let @xmath652 . for @xmath117 in @xmath70 and @xmath2 in @xmath70 , let @xmath653 we have that @xmath654 , and , therefore , @xmath655 by theorem [ subcritthm](1 ) . also , using lemma [ finiteirrcomplemma ] , we have that @xmath656 , where @xmath657 and for @xmath117 in @xmath70 , @xmath658 further , we have that @xmath659 . it follows from these facts that @xmath660 . thus , we have shown that for @xmath117 in @xmath70 , @xmath661 as desired . _ proof of _ ( 3 ) . for @xmath117 in @xmath70 , let @xmath662 we show that there for any real number @xmath663 , there exists @xmath664 such that @xmath665 for all @xmath666 . let @xmath667 . since @xmath629 , we have that @xmath668 now choose @xmath664 such that @xmath669 in the following sums , we will use that any set @xmath618 with @xmath670 can be written as @xmath671 , where @xmath672 . note that in this case @xmath673 . then for @xmath666 we have @xmath674 since @xmath629 , we have that @xmath675 , and we conclude that @xmath665 for all @xmath666 . we recognize the distribution of @xmath26 as the sum of countably many independent bernoulli trials , where the probability of success of trial @xmath676 is given by @xmath677 for some enumeration @xmath678 of the cycles in @xmath449 ( or any @xmath216 ) . we record some facts about this distribution in the following corollary . with the same hypotheses as in theorem [ subcritthm ] , the characteristic function of @xmath26 is given by @xmath679 where the product is over all periodic orbits in @xmath0 . it follows that the moment generating function of @xmath26 is given by @xmath680 in theorems [ emptythm ] and [ subcritthm ] , we assert the existence of various limits to certain values . beyond the bounds given in our proofs , we do not know at which rates these sequences converge to their limits . in this section we study random sfts in the supercritical phase . the main results are theorems [ entropythm ] and [ msftthm ] . on a first reading , the reader may prefer to skip section [ itsection ] and refer back to it as necessary . our proof of theorem [ entropythm ] relies , in part , on showing that with large probability the number of allowed words of length @xmath117 in a random sft is close to @xmath681 , for a particular choice of @xmath117 . in our proof , we choose @xmath117 to be polynomial in @xmath389 for two reasons . first , we need @xmath117 to dominate @xmath422 , so that the @xmath117th root of the number of words of length @xmath117 gives a good upper bound on the perron eigenvalue of the random sft . second , @xmath117 should be subexponential in @xmath422 , essentially because most paths in @xmath216 with length subexponential in @xmath422 are self - avoiding , and we need good bounds on the probability of paths of length @xmath117 that exhibit `` too - soon - recurrence . '' for context , we recall a result of ornstein and weiss @xcite . in fact , their result is quite general , but we only recall it in a very specific case . let @xmath0 be an irreducible sft with measure of maximal entropy @xmath160 . for @xmath12 in @xmath0 , let @xmath682 be the first return time ( greater than 0 ) of @xmath12 to the cylinder set @xmath683 $ ] under @xmath52 . then the result of ornstein and weiss implies that for @xmath160-a.e . @xmath12 in @xmath0 , @xmath684 . it follows from this result that for @xmath117 polynomial in @xmath2 , the @xmath160-measure of the set of words of length @xmath117 with a repeated @xmath2-word tends to @xmath646 . in the following lemmas , we give some quantitative bounds on the @xmath160-measure of the set of paths of length @xmath117 in @xmath216 with @xmath685 repeated edges , where the important point for our purposes is that the bounds improve exponentially as @xmath313 decreases . to get these bounds , we employ some of the language and tools of information theory . after getting a handle on the @xmath160-measure of paths in @xmath216 with certain self - intersection properties , our assumption that @xmath180 satisfies condition ( c7 ) in definition [ conditions ] implies that the @xmath160-measure on paths is the same as the counting measure up to uniform constants . in keeping with the convention of information theory , @xmath686 denotes the base @xmath687 logarithm of @xmath12 . a binary @xmath2-_code _ on an alphabet @xmath48 is a mapping @xmath688 , where @xmath689 is the set of all finite words on the alphabet @xmath220 . we may refer to such mappings simply as codes . a code is _ faithful _ if it is injective . the function that assigns to each @xmath147 in @xmath690 the length of the word @xmath691 is called the _ length function _ of the code , and it will be denoted by @xmath692 when the code is understood . a code is a _ prefix code _ if whenever @xmath691 is a prefix of @xmath693 . a _ shannon code _ with respect to a measure @xmath694 on @xmath695 is a code such that @xmath696 . we note that for a measure @xmath694 on @xmath695 , there is a prefix shannon code on @xmath695 with respect to @xmath694 @xcite . we will also require the following two lemmas from information theory . [ maininfotheorylemma ] let @xmath48 be an alphabet . let @xmath697 be a prefix - code on @xmath695 , and let @xmath160 be a shift - invariant borel probability measure on @xmath698 . then @xmath699 let @xmath700 . then for any @xmath147 in @xmath155 , we have that @xmath701 . the kraft inequality for prefix codes ( @xcite , page 73 ) states that since @xmath692 is a prefix code , @xmath702 . hence , @xmath703 [ eliaslemma ] there is a prefix code @xmath704 such that @xmath705 , where @xmath706 is the length of @xmath707 . a prefix code satisfying the conclusion of lemma [ eliaslemma ] is called an _ elias code_. recall that if @xmath113 is a path in the graph @xmath81 , then we denote by @xmath120 the set of edges traversed by @xmath113 . let @xmath180 be a sequence of graphs satisfying our standing assumptions [ stassump ] . [ ndef ] for each @xmath2 , @xmath117 , and @xmath708 , let @xmath709 [ ddef ] for each @xmath2 , @xmath117 , and @xmath710 , let @xmath711 [ qdef ] for each @xmath2 , @xmath117 , and @xmath708 , let @xmath712 [ sdef ] for each @xmath2 , @xmath117 , and @xmath710 , let @xmath713 for any of the sets defined in definitions [ ndef][sdef ] , we use a `` hat '' to denote the set with `` @xmath714 '' replaced by `` @xmath715 '' in the definition . for example , @xmath716 the `` hat '' notation will only appear in the proof of theorem [ entropythm ] . the following four lemmas find bounds on @xmath717 , @xmath718 , @xmath719 and @xmath720 . the following lemma bounds the @xmath160-measure ( and therefore the cardinality ) of the set of paths of length @xmath117 in @xmath216 that traverse at most @xmath721 edges . the proof relies on a general principle in information theory ( made precise by lemma [ maininfotheorylemma ] ) : a set of words that can be encoded `` too efficiently '' must have small measure . in order to use this principle , we find an efficient encoding of the paths of length @xmath117 in @xmath216 that traverse at most @xmath313 edges . the basic observation behind the coding is trivial : a path of length @xmath117 that only traverses @xmath722 edges must have @xmath685 repeated edges . therefore , instead of encoding each of the @xmath685 repeated edges explicitly , we simply encode some combinatorial data that specifies when `` repeats '' happen and when the corresponding edges are first traversed . [ manyrepeatslemma ] let @xmath180 be a sequence of graphs satisfying the standing assumptions [ stassump ] and such that @xmath180 has local uniqueness of paths and bounded distortion of weights [ conditions and in definition [ conditions ] ] . then there exists a polynomial @xmath723 and @xmath73 such that for each @xmath74 , @xmath413 and @xmath708 , @xmath724 and @xmath725 consider @xmath180 , @xmath2 , @xmath117 and @xmath313 as in the hypotheses . let @xmath726 and @xmath418 . a path @xmath113 in @xmath727 from vertex @xmath599 to vertex @xmath728 contributes @xmath729 to @xmath730 . the condition ( c7 ) gives a uniform constant @xmath731 such that @xmath732 is bounded below by @xmath733 . therefore , the bound on @xmath717 follows from the bound on @xmath734 , since @xmath735 [ as in lemma [ implicationslemma](1 ) ] . we now proceed to show the bound on @xmath730 . let @xmath736 . consider @xmath113 in @xmath727 . then there exists @xmath737 such that @xmath738 for some @xmath739 , for each @xmath740 , where @xmath741 . now we define a set @xmath742 by induction . let @xmath743 and @xmath744 . assuming by induction that @xmath745 and @xmath746 have been defined and that @xmath747 , we define @xmath748 and @xmath749 as follows : * if @xmath750 , let @xmath751 ; * otherwise , if @xmath752 , then let @xmath753 let @xmath754 . this induction procedure terminates when @xmath755 for some @xmath756 , and we denote this terminal @xmath313 by @xmath757 . let @xmath758 . note that for each @xmath759 , we have that @xmath760 it follows that @xmath761 . having defined the set @xmath762 , we now decompose the integer interval @xmath763 into subintervals . first , let @xmath764 let @xmath765 be the maximal disjoint subintervals ( with singletons allowed ) of @xmath188 such that @xmath766 and @xmath767 . note that @xmath768 be the maximal disjoint subintervals of @xmath188 such that : * @xmath769 for each @xmath770 ; * @xmath771 ; * and for each @xmath772 , we have that @xmath773 is nonempty and @xmath774 . in summary , we have that @xmath775 , and only @xmath776 may be empty . for any @xmath777 , let @xmath778 . by lemma [ implicationslemma](3 ) , there exists a uniform constant @xmath464 such that @xmath779 for notation , if @xmath780 is a subset of @xmath781 , then @xmath782 is @xmath113 restricted to @xmath780 . since @xmath160 is a @xmath7-step markov on @xmath374 , we have that @xmath783\\[-10pt ] & & { } \times\prod_{\ell= 2}^{n+1 } \mu ( b_{i_{\ell } } | a_{s_1,t_1 } \cap b_{i_1 { \cdots}j_{\ell-1 } } ) \nonumber\\[-2pt ] \label{decompofmu2 } & = & \mu ( b_{i_1 } | a_{s_1,t_1 } ) \prod_{\ell= 1}^{n } \mu ( b_{j_{\ell } } b_{i_1 { \cdots}j_{\ell-1}}).\end{aligned}\ ] ] given @xmath113 , we may form @xmath784 , @xmath785 , @xmath773 and @xmath786 as above , and then we encode @xmath113 as follows : encode @xmath787 and @xmath788 using an elias code ; encode @xmath789 using a prefix shannon code with respect to @xmath790 ; assuming @xmath791 has been encoded , we encode @xmath792 by encoding @xmath784 and @xmath785 for each @xmath307 in @xmath762 such that @xmath793 , using an elias code ( and note that this information completely determines @xmath792 by definition of @xmath67 and construction of @xmath625 ) ; assuming @xmath794 has been encoded , we encode @xmath795 using a prefix shannon code with respect to latexmath:[$\mu ( \cdot now we analyze the performance of the code . since the code is a concatenation of prefix codes , it is a prefix code . since @xmath67 tends to infinity as @xmath2 tends to infinity [ by ( c4 ) ] and @xmath442 , there exists @xmath73 such that for @xmath74 and @xmath797 , the length of the codeword in the elias encoding of @xmath599 is less than or equal to @xmath798 . then we have , neglecting bits needed to round up , @xmath799 combining ( [ decompofmu1 ] ) , ( [ decompofmu2 ] ) and ( [ lengthofcode ] ) , we have that @xmath800 now by lemma [ implicationslemma](2 ) and ( 3 ) , there exist uniform constants @xmath465 and @xmath801 such that @xmath802 by construction , @xmath803 , @xmath804 and @xmath805 . then by lemma [ maininfotheorylemma ] , there exists a uniform constant @xmath806 such that @xmath807 letting @xmath808 , for some uniform constant @xmath809 , we obtain that @xmath810 which completes the proof . the following lemma bounds the @xmath811-measure ( and therefore the cardinality ) of the set of pairs paths of length @xmath117 in @xmath216 that share at least one edge and together traverse at most @xmath812 edges . the general strategy of encoding pairs of paths using combinatorial data and appealing to information theory is similar to that of lemma [ manyrepeatslemma ] . lemma [ masterdoublelemma ] involves the additional hypothesis that there exists a uniform bound @xmath813 such that for any pair of paths @xmath814 in @xmath216 , there exists a path @xmath815 in @xmath216 with @xmath816 . using this hypothesis , one observes that pairs of paths can essentially be concatenated in @xmath216 and then treated as single paths as in lemma [ manyrepeatslemma ] . [ masterdoublelemma ] let @xmath180 be a sequence of graphs satisfying the standing assumptions [ stassump ] and such that @xmath180 has local uniqueness of paths , small diameter and bounded distortion of weights [ conditions , and in definition [ conditions ] ] . then there exists a polynomial @xmath817 and @xmath818 such that for @xmath819 , @xmath820 and @xmath710 , @xmath821 and @xmath822 consider @xmath180 , @xmath2 , @xmath117 and @xmath313 as in the hypotheses . let @xmath726 , @xmath823 and @xmath824 . note that the bound on @xmath718 follows from the bound on @xmath825 , since condition ( c7 ) implies that there exists a uniform constant @xmath731 such that @xmath826 [ as in lemma [ implicationslemma](1 ) ] . we now proceed to show the bound on @xmath825 . by the definition of @xmath813 , for every pair @xmath827 , there exists a path @xmath828 in @xmath216 such that @xmath829 and @xmath498 is in @xmath830 . we choose a single such @xmath828 for each pair @xmath831 , and we choose a ( possibly empty ) path @xmath832 such that @xmath833 is in @xmath834 ( whose existence is guaranteed by the fact that @xmath216 is irreducible ) . if @xmath835 , then @xmath836 is in @xmath837 . using condition ( c5 ) , we have that @xmath838 for a uniform constant @xmath37 . then we have that there exist uniform constants @xmath464 , @xmath465 and @xmath801 such that for each @xmath2 , each @xmath117 and each pair @xmath831 in @xmath839 , @xmath840 thus , lemma [ manyrepeatslemma ] implies that there exists a polynomial @xmath723 and @xmath73 such that for @xmath74 , @xmath841 with @xmath842 and @xmath843 for a uniform constant @xmath844 , we have @xmath845 which completes the proof . the following two lemmas ( lemmas [ doubleperrepeatslemma ] and [ manydoubleperrepeatslemma ] ) give bounds on the measure ( and therefore the cardinality ) of the set of pairs of periodic paths in @xmath216 with certain overlap properties . the general ideas are similar to those in lemmas [ manyrepeatslemma ] and [ masterdoublelemma ] , but in order to get precise bounds on the relevant sets , we exploit the fact that these sets consist of pairs of periodic paths . in other words , when we encode paths using their pattern of `` repeats , '' we also take into account their assumed periodicity . [ doubleperrepeatslemma ] let @xmath180 be a sequence of graphs satisfying the standing assumptions [ stassump ] and bounded distortion of weights [ condition in definition [ conditions ] ] . then there exists a polynomial @xmath846 and @xmath847 such that for each @xmath848 and @xmath849 , @xmath850 and @xmath851 consider @xmath180 , @xmath2 and @xmath117 as in the hypotheses . let @xmath389 and @xmath418 . note that the bound on @xmath719 follows from the bound on @xmath852 , since condition ( c7 ) implies that there exists a uniform constant @xmath731 such that @xmath853 [ as in lemma [ implicationslemma](1 ) ] . we now proceed to show the bound on @xmath852 . let @xmath113 be in @xmath854 . let @xmath85 be in @xmath855 . for @xmath856 , let @xmath857 be the set of paths @xmath184 of length @xmath117 in @xmath216 such that @xmath858 . then lemma [ implicationslemma ] [ parts ( 1 ) and ( 4 ) ] implies that there exist uniform constants @xmath464 and @xmath465 such that @xmath859 let @xmath37 be the set of paths @xmath184 of length @xmath117 in @xmath216 such that @xmath860 . then @xmath861 , and by shift - invariance of @xmath160 , @xmath862 since @xmath863 was arbitrary , it follows from inequality ( [ aboveinequality1 ] ) that @xmath864 since @xmath865 was arbitrary , we conclude that there exists a uniform constant @xmath801 such that @xmath866 where the last inequality follows from lemma [ implicationslemma](4 ) . this inequality completes the proof . [ manydoubleperrepeatslemma ] let @xmath180 be a sequence of graphs satisfying the standing assumptions [ stassump ] and such that @xmath180 has local uniqueness of paths , small diameter and bounded distortion of weights [ conditions , and in definition [ conditions ] ] . then there exists a polynomial @xmath867 and @xmath868 such that for @xmath869 , @xmath413 and @xmath710 , @xmath870 and @xmath871 consider @xmath180 , @xmath2 , @xmath117 and @xmath313 as in the hypotheses . let @xmath872 , @xmath418 and @xmath824 . note that the bound on @xmath720 follows from the bound on @xmath873 , since condition ( c7 ) implies that there exists a uniform constant @xmath731 such that @xmath874 [ as in lemma [ implicationslemma](1 ) ] . we now proceed to show the bound on @xmath873 . let @xmath85 be in @xmath257 and let @xmath875 be the set of paths @xmath113 of length @xmath117 in @xmath216 such that @xmath876 . then it follows from lemma [ implicationslemma](4 ) that there exists a uniform constant @xmath464 such that @xmath877 to each pair @xmath831 in @xmath878 , let us associate a particular path of length @xmath879 in @xmath216 , which we construct as follows . let @xmath831 be in @xmath878 . by definition of @xmath878 , there is at least one edge @xmath85 in @xmath880 . let @xmath187 be the cyclic permutation of @xmath188 of order @xmath117 given by @xmath881 . let @xmath187 act on periodic paths of length @xmath117 in @xmath216 by permuting the indices : @xmath882 . then let @xmath883 be in @xmath884 . similarly , let @xmath885 be in @xmath886 . now choose a path @xmath828 in @xmath216 such that @xmath887 and @xmath888 is a path in @xmath216 ( the existence of such a path @xmath828 is guaranteed by the definition of @xmath813 ) . by irreducibility of @xmath216 , we also choose a ( possibly empty ) path @xmath832 in @xmath216 such that @xmath889 is in @xmath890 . we associate the path @xmath889 to the pair @xmath831 , and note that there exist uniform constants @xmath465 , @xmath801 and @xmath844 [ by lemma [ implicationslemma](1 ) and condition ( c5 ) ] such that @xmath891 now we use the same construction as in the proof of lemma [ manyrepeatslemma ] with only slight modification . we encode the words @xmath889 as follows : construct @xmath762 , @xmath625 and the partition of @xmath892 as in the proof of lemma [ manyrepeatslemma ] , with the additional condition that @xmath893 . ( in other words , we ignore any `` repeats '' introduced by @xmath496 . ) encode @xmath883 as in the proof of lemma [ manyrepeatslemma ] . to encode the path @xmath828 , we first encode the fact that @xmath894 ( by encoding @xmath117 and @xmath895 using an elias code ) , and then encode @xmath828 using a prefix shannon code with respect to @xmath896 . encode @xmath885 as in the proof of lemma [ manyrepeatslemma ] . encode @xmath832 using a prefix shannon code with respect to @xmath897 . for large @xmath2 , encoding the fact that @xmath898 adds less than @xmath899 to @xmath900 . on the other hand , we have that there is a uniform constant @xmath430 such that @xmath901 , by lemma [ implicationslemma](4 ) . thus , there exists @xmath868 and a uniform constant @xmath902 such that for @xmath903 , we have @xmath904\\[-8pt ] & & \qquad\leq(|{\mathcal{i}}|+1)\bigl(4 \log ( 2k+r)\bigr ) + n k_6 - ( m+u+|j|)\log\lambda\nonumber\end{aligned}\ ] ] with @xmath905 , @xmath804 and @xmath906 . then by lemma [ maininfotheorylemma ] , there is a polynomial @xmath907 such that for @xmath869 , @xmath908 note that the number of pairs @xmath831 associated to the path @xmath889 is at most @xmath909 , and , hence , @xmath910 now let @xmath911 , and the proof is complete . recall that if @xmath88 is a graph , then @xmath912 is the random variable such that @xmath913 is the spectral radius of the adjacency matrix of @xmath528 . [ entropythm ] let @xmath180 be a sequence of graphs that satisfies the standing assumptions [ stassump ] and such that @xmath180 has local uniqueness of paths , small diameter and bounded distortion of weights [ conditions , and in definition [ conditions ] ] . then for @xmath32 and @xmath33 , @xmath914 and the convergence to the limit is exponential in @xmath459 . if we assume that @xmath0 is irreducible in the statement of theorem [ entropythmintro ] , then theorem [ entropythmintro ] is a direct corollary of theorem [ entropythm ] , obtained by choosing @xmath180 to be the sequence of @xmath2-block graphs of an irreducible sft with positive entropy ( and using the fact that such a sequence satisfies the hypotheses of theorem [ entropythm ] by proposition [ highergraphseqlemma ] ) . in the case when @xmath0 is reducible , @xmath0 has a finite number of irreducible components of positive entropy , @xmath915 , and there exists @xmath307 such that @xmath916 . for all large @xmath2 , we have that @xmath917 for @xmath918 , which means that the entropies of the random subshifts appearing inside each of these components are mutually independent . applying theorem [ entropythm ] to each of these components , we obtain theorem [ entropythmintro ] for reducible @xmath0 . proof of theorem [ entropythm ] let @xmath4 be in @xmath31 $ ] . let @xmath389 and @xmath418 . let @xmath113 be a path in @xmath919 . let @xmath920 be the random variable defined by @xmath921 now let @xmath922 for each @xmath2 and @xmath117 , we have that @xmath923 . indeed , @xmath924 is the average number of loops of length @xmath117 based at a vertex in @xmath216 . thus , there is at least one vertex @xmath145 with at least @xmath924 loops of length @xmath117 based at @xmath145 , and it follows that @xmath925 since these loops may be concatenated freely . also , it follows from subadditivity that @xmath926 , which implies that @xmath927 for all @xmath2 and @xmath117 . fix @xmath928 , and let @xmath929 , where @xmath307 is chosen such that @xmath930 and @xmath931 divides @xmath117 . recall that if @xmath180 is the sequence of @xmath2-block graphs of a fixed graph @xmath88 , then by proposition [ highergraphseqlemma ] we have that @xmath422 and @xmath2 differ by at most a uniform constant , and , thus , @xmath932 . we will show below that as @xmath2 tends to infinity , @xmath933 tends to @xmath934 ; @xmath935 tends to @xmath934 ; there exists @xmath420 and @xmath936 such that @xmath937 ; there exists @xmath938 and @xmath939 such that @xmath940 . recall definitions [ ndef][sdef ] , as well as the modification of these definitions using `` hats . '' notice that @xmath941 also , @xmath942 regarding variances , we have @xmath943 for the remainder of this proof , we use the following notation : if @xmath944 and @xmath945 are two sequences , then @xmath946 means that the limit of the ratio of @xmath947 and @xmath948 tends to @xmath7 as @xmath2 tends to infinity . _ proof of _ ( i ) . by lemma [ implicationslemma](1 ) , there exists a uniform constant @xmath420 such that @xmath949 taking @xmath117th roots , letting @xmath2 tend to infinity , and using that @xmath950 tends to @xmath646 , we obtain that @xmath951 . by lemmas [ implicationslemma](1 ) and [ manyrepeatslemma ] , we have that there exists @xmath73 , a polynomial @xmath723 , and a uniform constant @xmath423 such that for @xmath74 , @xmath952 by condition ( c4 ) and the fact that @xmath953 , we have that : * @xmath422 tends to infinity as @xmath2 tends to infinity by the standing assumptions [ stassump ] ; * @xmath954 , which tends to zero as @xmath2 tends to infinity ; * @xmath955 , which tends to infinity as @xmath2 tends to infinity . thus , taking @xmath117th roots and letting @xmath2 tend to infinity , we have that @xmath956 which concludes the proof of ( i ) . _ proof of _ ( ii ) . let @xmath957 . note that since @xmath129 divides @xmath117 , there exists a uniform constant @xmath958 such that @xmath959 for large enough @xmath117 . we choose @xmath2 large enough so that this inequality is satisfied . then we have that @xmath960 taking @xmath117th roots , letting @xmath2 tend to infinity , and using that @xmath950 tends to @xmath646 , we get that @xmath961 . recall that @xmath962 . therefore , it follows from ( i ) that @xmath963 . thus , we have shown ( ii ) . _ proof of _ ( iii ) . for @xmath964 , lemma [ masterdoublelemma ] implies that there is @xmath818 and a polynomial @xmath965 such that @xmath966 and @xmath967 for @xmath968 . now using that @xmath969 [ see ( [ inequalityin(i ) ] ) ] , we obtain that there exists a uniform constant @xmath809 such that @xmath970 using the facts that @xmath971 and @xmath953 , we have that @xmath972 is asymptotically bounded above by @xmath973 . since @xmath974 , it holds that @xmath975 tends to @xmath7 . thus , we obtain that for any @xmath976 , there exists @xmath977 and @xmath847 such that for @xmath848 , it holds that @xmath978 , which proves ( iii ) . _ proof of _ ( iv ) . for @xmath964 , lemma [ manydoubleperrepeatslemma ] together with ( c4 ) implies that there is @xmath868 and a polynomial @xmath979 such that @xmath980 for @xmath869 . also , lemma [ doubleperrepeatslemma ] implies that there is @xmath981 and a polynomial @xmath982 such that @xmath983 for @xmath984 . now using that @xmath985 , we obtain that there exists @xmath986 such that , with @xmath987 , @xmath988 using the facts that @xmath971 and @xmath953 , we have that @xmath989 is asymptotically bounded above by @xmath973 . since @xmath974 , it holds that @xmath990 tends to @xmath7 . thus , we obtain that for any @xmath991 , there exists @xmath992 and @xmath993 such that for @xmath994 , @xmath995 which proves ( iv ) . _ proof of theorem _ [ entropythm ] _ using _ ( i)(iv ) . recall that @xmath996 . let @xmath33 . since @xmath997 , we may assume without loss of generality that . then @xmath998\\[-8pt ] & & \qquad = { \mathbb{p}_{\alpha } } ( \beta_n \geq{\alpha}\lambda+ \varepsilon ) + { \mathbb{p}_{\alpha}}(\beta_n \leq{\alpha}\lambda- \varepsilon)\nonumber \\ \label{entropyineq2 } & & \qquad = { \mathbb{p}_{\alpha}}\bigl ( \beta_n^k \geq({\alpha}\lambda+ \varepsilon)^k \bigr ) + { \mathbb{p}_{\alpha}}\bigl ( \beta _ n^k \leq({\alpha}\lambda- \varepsilon)^k\bigr ) \\ \label{entropyineq3 } & & \qquad \leq { \mathbb{p}_{\alpha}}\bigl ( \phi_{n , k } \geq({\alpha}\lambda+ \varepsilon)^k\bigr ) + { \mathbb{p}_{\alpha}}\bigl(\psi _ { n , k } \leq({\alpha}\lambda- \varepsilon)^k \bigr).\end{aligned}\ ] ] we will bound each of the two terms in ( [ entropyineq3 ] ) . notice that @xmath999 let @xmath1000 . then by chebyshev s inequality , @xmath1001 the denominator in the right - hand side of ( [ chebbound2 ] ) might be @xmath646 for finitely many @xmath2 , but by properties ( i ) and ( iii ) , there exists @xmath1002 such that for large enough @xmath2 , @xmath1003 similarly , we let @xmath1004 , and then chebyshev s inequality gives that @xmath1005 again , the denominator in the right - hand side might be @xmath646 for finitely many @xmath2 , but by properties ( ii ) and ( iv ) , there exists @xmath1006 such that for large enough @xmath2 , @xmath1007 in conclusion , we obtain that there exists @xmath1008 such that for large enough @xmath2 , @xmath1009 [ msftthm ] let @xmath180 be a sequence of graphs that satisfies the standing assumptions [ stassump ] , with @xmath957 , and such that : * @xmath180 has bounded degrees [ condition in definition [ conditions ] ] , * @xmath180 has fast separation of periodic points [ condition in definition [ conditions ] ] , * and @xmath180 has uniform forward and backward expansion [ condition in definition [ conditions ] ] . let @xmath1010 be the event in @xmath619 that @xmath1011 contains a unique irreducible component @xmath37 of positive entropy . also , let @xmath1012 be the event ( contained in @xmath1013 ) that the induced edge shift on @xmath37 has period @xmath129 . then there exists @xmath38 such that for @xmath1014 , @xmath1015 and the convergence to these limits is exponential in @xmath459 . theorem [ msftthmintro ] is a corollary of theorem [ msftthm ] : if @xmath0 is an irreducible sft of positive entropy , then the sequence of @xmath2-block graphs for @xmath0 satisfies the hypotheses of theorem [ msftthm ] by proposition [ highergraphseqlemma ] . in fact , if @xmath0 is a reducible sft , we may apply theorem [ msftthmintro ] to each irreducible component independently , which allows us to conclude the following . let @xmath0 be a reducible sft with irreducible components @xmath1016 such that @xmath1017 for each @xmath307 . let @xmath1018 be the event in @xmath11 that @xmath5 has exactly @xmath1019 irreducible components with periods @xmath1020 . then there exists @xmath38 such that for @xmath1021 $ ] , we have that @xmath1022 , with exponential ( in @xmath2 ) convergence to the limit . [ followerdefs ] let @xmath88 be a directed graph . for each vertex @xmath145 in @xmath88 , and for each @xmath3 in @xmath1023 , let @xmath1024 be the union of @xmath1025 and the set of vertices @xmath166 in @xmath88 such that there is an allowed path from @xmath145 to @xmath166 in @xmath528 . similarly , for each vertex @xmath145 in @xmath88 and each @xmath3 in @xmath1023 , let @xmath1026 be the union of @xmath1027 and the set of vertices @xmath166 in @xmath88 such that there is an allowed path from @xmath166 to @xmath145 in @xmath528 . also , let @xmath1028 , which is the vertex set of the irreducible component containing @xmath145 in @xmath528 . the proof of the following proposition is an adaptation of the proof of lemma 2.2 in @xcite . [ posentimpliesbig ] let @xmath180 be a sequence of graphs satisfying the standing assumptions [ stassump ] and such that @xmath180 has bounded degrees and uniform forward and backward expansion [ conditions and in definition [ conditions ] ] . let @xmath1029 be a sequence of integers such that @xmath1030 , for some @xmath1031 , for all large @xmath2 . let @xmath1032 be the event in @xmath619 consisting of all @xmath3 such that there exists a vertex @xmath145 in @xmath216 with @xmath1033 . then there exists @xmath1034 such that for @xmath1035 , @xmath1036 and the convergence of this limit is exponential in @xmath459 . furthermore , the same statement holds with @xmath1037 replaced by @xmath1038 . let @xmath389 . let @xmath280 be such that both @xmath180 and @xmath244 are @xmath113-expander sequences [ where the existence of such a @xmath113 is guaranteed by condition ( c8 ) ] . we use the notation in definition [ followerdefs ] . for any @xmath145 in @xmath250 and any @xmath3 in @xmath1039 , the set @xmath1024 has the property that all edges in @xmath1040 are forbidden ( by @xmath3 ) . then the fact that @xmath216 is a @xmath113-expander implies that for a particular subset @xmath206 of @xmath250 , the probability that @xmath1041 for some @xmath145 is bounded above by @xmath1042 that could appear as @xmath1024 for some @xmath145 is bounded above by @xmath1043 , where @xmath85 is the base of the natural logarithm ( @xcite , lemma 2.2 ) ( see also @xcite , lemma 2.1 , or @xcite , page 396 , exercise 11 ) . then for @xmath4 such that @xmath1044 , we have that for any @xmath1045 , @xmath1046 thus , there is a @xmath38 ( depending only on @xmath1047 , @xmath113 , @xmath144 and @xmath1048 ) such that if @xmath1035 , then the right - hand side of the inequality in ( [ rhsgoestozero ] ) tends to zero exponentially in @xmath459 as @xmath2 tends to infinity . in particular , we may take @xmath1049 since @xmath244 is also a uniform @xmath113-expander , the same estimates hold with @xmath1050 in place of @xmath1032 . proof of theorem [ msftthm ] let @xmath180 be as in the statement of theorem [ msftthm ] . let @xmath726 , @xmath1051 , and @xmath1052 . we use the notation in definition [ followerdefs ] . consider the following events : @xmath1053 recall that condition ( c3 ) gives @xmath1054 such that @xmath1055 . note that proposition [ posentimpliesbig ] gives @xmath1034 such that for @xmath1014 , there exists @xmath1056 and @xmath1057 such that for large @xmath2 , @xmath1058 fix such an @xmath4 , and note that for all large enough @xmath2 , we have the following estimate : @xmath1059 . consider @xmath3 in @xmath1060 . suppose that there exists @xmath1061 and @xmath1062 in @xmath250 such that @xmath1063 and @xmath1064 . then by definition of @xmath1060 , we must have that @xmath1065 and @xmath1066 . it follows that there is a path from @xmath1061 to @xmath1062 in @xmath620 , and there is a path from @xmath1062 to @xmath1061 in @xmath620 . thus , @xmath1067 . we have shown that for @xmath3 in @xmath1060 , there is at most one irreducible component of cardinality greater than @xmath1068 . note that this argument implies that for @xmath3 in @xmath1060 , all allowed periodic orbits @xmath597 such that @xmath1069 must lie in the same irreducible component . by definition of @xmath1070 , if @xmath1071 is an irreducible component of @xmath620 with positive entropy , then @xmath1072 ( since it must contain at least two periodic orbits with overlapping vertex sets ) . we deduce that for @xmath3 in @xmath1060 , there is at most one irreducible component of @xmath620 with positive entropy . we now show that there exists an irreducible component of positive entropy with probability tending exponentially to @xmath7 . let @xmath1073 , where @xmath307 is chosen ( for each @xmath2 ) such that @xmath1074 and @xmath129 divides @xmath1075 . then let @xmath1076 . consider the following sequences of random variables : @xmath1077 note that by the definition of @xmath1070 and lemma [ perlemma ] , we have that @xmath1078 for any periodic path @xmath113 with period less than or equal to @xmath1070 . furthermore , any two such paths are disjoint . therefore , the random variables @xmath1079 are jointly independent , and the random variables @xmath1080 are also jointly independent . thus , @xmath1081 as @xmath2 tends to infinity , @xmath1070 tends to infinity since @xmath1055 and @xmath422 tends to infinity . then by the standing assumptions [ stassump ] [ in particular , we use that @xmath376 ] and the fact that @xmath129 divides @xmath1075 and @xmath1083 , we have that each of the sequences @xmath1084 and @xmath1085 tends to a finite , nonzero limit as @xmath2 tends to infinity ( and in fact the limit is @xmath129 ) . for two sequences @xmath947 and @xmath948 of positive real numbers , let @xmath946 denote the statement that their ratio tends to a finite , nonzero limit as @xmath2 tends to infinity . then we have that @xmath1086 and @xmath1087 . note that @xmath1088 and @xmath1089 . a simple application of chebyshev s inequality implies that @xmath1090 and @xmath1091 we have shown that the probability that there is no periodic orbit of period @xmath1075 tends to @xmath646 exponentially in @xmath422 as @xmath2 tends to infinity , and the probability that there exists no periodic orbit of period @xmath1083 tends to @xmath646 exponentially in @xmath422 as @xmath2 tends to infinity . in summary , we have shown that the following events occur with probability tending to @xmath7 exponentially in @xmath422 as @xmath2 tends to infinity : * there exists a periodic point of period @xmath1092 ; * there exists a periodic point of period @xmath1093 ; * any two periodic points of period greater than @xmath1094 lie in the same irreducible component ( of necessarily positive entropy ) ; * there is at most one irreducible component of positive entropy . we conclude that with probability tending to @xmath7 exponentially in @xmath422 as @xmath2 tends to infinity , there exists a unique irreducible component of positive entropy , and the induced edge shift on that component has period @xmath129 . [ relaxrmk ] the proofs of theorems [ emptythm ] and [ subcritthm ] do not require all of the standing assumptions [ stassump ] . in fact , these proofs only use that @xmath376 for each @xmath2 and that @xmath231 tends to infinity as @xmath2 tends to infinity . theorem [ emptythm ] states that at the critical threshold @xmath1095 , the probability of emptiness tends to zero . using the fact that entropy is a monotone increasing random variable ( as defined in section [ probframework ] ) , one may deduce from theorem [ entropythm ] that for @xmath1096 , the probability that the random sft has zero entropy tends to @xmath7 . it might be interesting to know more about the behavior of typical random sfts at the critical threshold . we have considered only random @xmath477-sfts , but one may also consider random @xmath1097-sfts for any @xmath496 in @xmath70 by adapting the construction of @xmath11 and @xmath9 in the obvious way . it appears that most of the proofs presented above may not be immediately adapted for @xmath1098 , but there is one exception , which we state below . let @xmath0 be a nonempty @xmath1097-sft . for @xmath1099 , there are various zeta functions for @xmath0 ( for a definition distinct from ours , see @xcite ) ; we consider @xmath1100 where @xmath1101 is the number of periodic points @xmath12 in @xmath0 such that the number of points in the orbit of @xmath12 divides @xmath129 . the function @xmath1102 has radius of convergence @xmath1103 , where @xmath1104 . for example , for a full @xmath1097 shift on @xmath1047 symbols , , regardless of @xmath496 . using exactly the same proof as presented in section [ emptiness ] , we obtain that @xmath1105$.}\ ] ] for @xmath1106 , this bound implies that the limiting probability of emptiness is @xmath646 . in this context , we note that there is no algorithm , which , given a @xmath1097-sft @xmath0 defined by a finite list of finite forbidden configurations , will decide whether @xmath0 is empty @xcite . nonetheless , we may be able to compute the limiting probability of emptiness . for example , if @xmath0 is a full shift on @xmath1047 symbols , then for @xmath1107 , we have that the limiting probability of emptiness is @xmath646 . one may also consider more general random subshifts . recall that a set @xmath1108 is a subshift if it is closed and shift - invariant . for a nonempty subshift @xmath0 and a natural number @xmath2 , we may consider the ( finite ) set of subshifts obtained by forbidding words of length @xmath2 from @xmath0 . after defining a probability measure @xmath9 on this space as in section [ preliminaries ] , we obtain random subshifts of @xmath0 . now we may investigate the asymptotic probability of properties of these random subshifts . recall that any subshift @xmath0 can be written as @xmath1109 , where @xmath1110 is a sequence of sfts ( called the markov approximations of @xmath0 ) and @xmath1111 . a subshift @xmath0 is called _ almost sofic _ @xcite if there exists a sequence @xmath1110 of irreducible sfts such that and @xmath1112 . using this inner and outer approximation by sfts , the conclusion of theorem [ entropythmintro ] still holds if the system @xmath0 is only assumed to be an almost sofic subshift . theorem [ msftthm ] asserts the existence of a constant , but we are left with several questions about this constant . fix a sequence @xmath180 satisfying the hypotheses of theorem [ msftthm ] . let @xmath1113 . what is @xmath1114 ? what is @xmath1114 in the case that @xmath180 is the sequence of @xmath2-block graphs of a mixing sft of positive entropy ( or even a full shift ) ? the author expresses great thanks to his advisor , mike boyle , for his guidance and encouragement in general , and also for many helpful conversations and comments regarding this work in particular . further , the author thanks the anonymous referee for reading the paper carefully and making helpful suggestions .

let @xmath0 be an irreducible shift of finite type ( sft ) of positive entropy , and let @xmath1 be its set of words of length @xmath2 . define a random subset @xmath3 of @xmath1 by independently choosing each word from @xmath1 with some probability @xmath4 . let @xmath5 be the ( random ) sft built from the set @xmath3 . for each @xmath6 and @xmath2 tending to infinity , we compute the limit of the likelihood that @xmath5 is empty , as well as the limiting distribution of entropy for @xmath5 . for @xmath4 near @xmath7 and @xmath2 tending to infinity , we show that the likelihood that @xmath5 contains a unique irreducible component of positive entropy converges exponentially to @xmath7 . these results are obtained by studying certain sequences of random directed graphs . this version of `` random sft '' differs significantly from a previous notion by the same name , which has appeared in the context of random dynamical systems and bundled dynamical systems . .

it is well known that an essential characteristic of compounds forming liquid crystals is the rod - like shape of their constituent molecules , with an high length to breadth ratio . therefore the molecules are supposed to be cylindrically symmetrical . for example , the ordering matrix which is often used to describe the partial alignment in a mesophase , contains only one independent element and this can be determined by some techniques [ 1 ] . the fact that the molecular cylindrical symmetry is assumed is appealing to a statistical mechanician , because the pairwise anisotropic intermolecular potential required in any calculation is simple for such particles [ 2 ] . however the molecules , in fact , are lath - like and thus do not possess the high symmetry . the ordering matrix has two principal components and therefore these components are required to describe the orientational order of a uniaxial mesophase composed of lath - like molecules . in this sense , the deviation of this ordering matrix from cylindrical symmetry was found to be significant [ 3 ] . the importance of deviations from cylindrical symmetry may be inferred from unambiguous determinations of the ordering matrix for rod - like molecules , such as @xmath0 [ 4 ] . moreover it is found that these matrices are comparable to those estimated for a pure mesophase [ 3 ] . there are some studies in which the consequences of deviations from molecular cylindrical symmetry are investigated . it is shown that a system consisting of particles with a lower symmetry than @xmath1 is capable of existing either as a uniaxial or a biaxial liquid crystal [ 5 ] . the possible existence of a biaxial phase is studied in detail for a system of hard rectangular plates using a lattice model [ 6 ] , the landau approach [ 7 ] and the molecular field approximation [ 8 ] . the deviations of the ordering matrix from cylindrical symmetry is clearly determined by the molecular symmetry and the element of the ordering matrix for the long axis will also be influenced by the form of the pseudo - intermolecular potential . the calculations of the ordering matrix for an ensemble of hard rectangular particles is performed in [ 9 ] . it must be emphasized that although these calculations are of some interest , they may not be particularly realistic because some experiments indicate that dispersion forces may make a dominant contribution to the anisotropic intermolecular potential [ 9 ] . considering the cases above , luckhurst et al . developed a theory [ 10 ] for non - cylindrically symmetric molecules interacting via a completely general intermolecular potential within molecular field approximation . for a decade , nonextensive statistics has an increasing interest and recently tsallis thermostatistics ( tt ) has been applied to the nematic - isotropic transition [ 11 - 13 ] as a nonextensive statistics . in [ 11 ] , the maier - saupe mean field theory has been generalized within tt and applied to a nematic liquid crystal , para - azoxyanisole . in the other study , [ 12 ] , the the effects of the nonextensivity on the dimerization process has been studied , and finally the mean field theory of anisotropic potentail of rank @xmath2 has been generalized within tt and the effect of the nonextensivity on the order parameters has been illustrated in [ 13 ] . up to now , the mean field theories for uniaxial nematogens formed by cylindrically symmetric molecules have been studied by using tt . in this manner , we aim , in this study , to enlarge the applications of tt to the liquid crystal systems and to handle luckhurst et al.s theory which considers the molecules to be non - cylindrically symmetric . in doing so , we first give some essential properties of luckhurst et al.s theory . then we mention on tt and its axioms . finally , we apply tt to the luckhurst et al.s theory and some possible concluding remarks are made . we must emphasize that we would like to give only the possible contributions of the nonextensivity to the theory . so we must keep in mind that since one relies on the generalized theory or not , more extensional studies related with it must be performed in the nematic - isotropic transition . however , we believe that this study is sufficient to give motivation for further applications of tt to the liquid crystals . the intermolecular potential for particles of general shape is given by [ 10 ] @xmath3 in a product basis of wigner rotation matrix [ 14 ] , where @xmath4 is the distance between molecules @xmath5 and @xmath6 . the orientation of molecule @xmath7 in a coordinate system containing the intermolecular vector as the @xmath8 axis is denoted by @xmath9 . this potential energy is invariant under rotation of the coordinate system about @xmath10 axis . therefore the summation in eq.(1 ) can be restricted as follows [ 10 ] : @xmath11 in what follows , the redundant subscripts on the coefficient @xmath12 will be suppressed . because it is convenient to define the molecular orientation in terms of a common coordinate system , the potential energy @xmath13 could be transformed to a laboratory frame . the choice of this coordinate system is determined according to the symmetry of the liquid crystal phase , so for a uniaxial mesophase , the laboratory @xmath10 axis can be taken to be parallel to the symmetry axis of the mesophase . the transformation of @xmath14 is carried out by performing the rotation from the intermolecular vector to the molecule in two steps using the relationship @xmath15 where the subscript @xmath16 is the rotation from the laboratory to the intermolecular frame , @xmath17 denotes that from the laboratory to the molecule coordinate system . then the intermolecular potential can be written as @xmath18 if the distribution function for the intermolecular vector is independent of orientation , then one could use the orthogonality of the rotation matrices to evaluate the ensemble average : @xmath19 the partially averaged potential may then be written as @xmath20 then now it is the time to average over the orientations adopted by particle @xmath6 . however one needs @xmath21 . this average is taken to be independent of the orientation of molecule @xmath5 within the molecular field approximation and it is to be identified with the normal ensemble average . since we only consider a uniaxial mesophase , @xmath22 vanishes when @xmath23 and @xmath24 is even . in this manner , the potential is reduced to the following form : @xmath25 where the summation is over @xmath26 ( @xmath24 even ) . the scalar contribution could be ignored because we would only like to study the orientational properties of the mesophase . the average of the orientational pseudo - potential for molecule @xmath5 is given by @xmath27 the expansion coefficients in eq.(8 ) are defined by @xmath28 where @xmath29 is the scalar potential of the ensemble [ 10,15 ] . the pseudo - potential is conveniently written as @xmath30 where @xmath31 with @xmath32 and @xmath33 is the euler angles and define the orientation of the director in the molecular coordinate system . luckhurst et al . made some comments in [ 10 ] on this form of the pseudo - potential and the symmetry properties of the coefficients @xmath34 . because luckhurst et al . investigated the influence of deviations from molecular cylindrical symmetry on various orientational order parameters for a uniaxial nematic mesophase , they tried to minimize the number of the variables in the calculation without any loss of the essential physics . as a first approximation , they considered only those terms with @xmath24 equal to @xmath6 in the expansion of the pseudo - potential . the neglect of terms higher than quadratic could be a good approximation , since similar assumptions for cylindrically symmetric molecules had let to results in reasonable accord with experiment [ 10,16 ] . as mentioned in [ 10 ] , the number of expansion coefficients are reduced by appealing to some specific model for the interactions or by imposing symmetry restrictions on the molecules . if one follows straley [ 9 ] and takes the molecules to be hard rectangular parallelopipeds , then the coefficients @xmath35 are independent of the sign of either @xmath36 or @xmath37 and zero if either of these subscripts is odd , that is @xmath38 /3 , \\ c_{220 } & = & \left ( l^{2}-bw\right ) \left ( b - w\right ) /\sqrt{6 } , \\ c_{222 } & = & -l\left ( w - b\right ) ^{2}/2,\end{aligned}\ ] ] where @xmath24 is the length , @xmath39 the breadth and @xmath40 the width of the molecule . it must be noted that this parametrization is only approximate [ 9 ] and also anisotropic repulsive forces are probably not dominant in determining the behaviour of real nematics . therefore assuming the molecules to interact via dispersion forces , the expressions for @xmath35 could be derived . the similar restriction can be imposed on the expansion coefficients @xmath41 , without appealing to specific forms of the intermolecular interactions , by using more formal arguments based on the molecular symmetry and its influence on the pair potential . for example , if each molecule has a centre of symmetry , then [ 10 ] @xmath42 and consequently @xmath43 in addition , if the molecules also possess a plane of symmetry orthogonal to their @xmath10 axes , then @xmath44 and @xmath45 however eqs.(16,18 ) can only be consistent when both @xmath36 and @xmath37 have even values . moreover the coefficients are independent of the sign of @xmath36 and @xmath37 . then the pseudo - potential may be written as @xmath46 \left [ d_{0,j}^{(l)}(\beta \gamma ) + d_{0,-j}^{(l)}(\beta \gamma ) \right ] /(1+\delta _ { 0j})(1+\delta _ { 0p}).\ ] ] according to the restrictions above . if limiting the summation to those terms with @xmath24 equal to @xmath6 , one has @xmath47 d_{0,0}^{(2)}(\beta ) \nonumber \\ & & + 2\left [ c_{200}\overline{d_{0,0}^{(2)}}+2c_{222}\overline{d_{02}^{(2)}}% \cos ( 2\gamma ) \right ] d_{0,2}^{(2)}(\beta ) \cos ( 2\gamma ) , \end{aligned}\ ] ] where @xmath48 is a reduced wigner rotation matrix [ 14 ] . the second rank order parameters @xmath49 and @xmath50 in this expression are related to the principal elements of the ordering matrix by @xmath51 and @xmath52 the order parameter @xmath53 indicates the deviation of the ordering matrix from cylindrical symmetry . for simplicity , it is convenient to write the pseudo - potential as @xmath54 , \ ] ] where @xmath55 /kt\ ] ] and @xmath56 /kt.\ ] ] the partition function for this single particle potential is then given by @xmath57 \sin \beta d\beta \,d\gamma .\ ] ] the orientational molar potential energy and the molar entropy are given as follows @xmath58 , \\ s & = & -r\left [ a\,\overline{d_{0,0}^{(2)}}+b\,\overline{d_{0,2}^{(2)}\cos ( 2\gamma ) } \right ] + r\ln ( z).\end{aligned}\ ] ] respectively . then the orientational molar free energy is straightforward : @xmath59 -rt\ln ( z).\ ] ] the order parameters are calculated from the equations below : @xmath60 \sin \beta d\beta \,d\gamma } { % z}\ ] ] and @xmath61 \sin \beta d\beta \,d\gamma } { % z}.\ ] ] it is worthwhile to note that @xmath49 in eq.(30 ) is simply @xmath62 , ( @xmath63 is the second legendre polynomial ) . the free energy @xmath64 will be a minimum provided the consistency equations for these order parameters are satisfied . with the aid of eqs.(24,25 ) , one can write the following equations : @xmath65 , \ ] ] and @xmath66 where @xmath67 @xmath68 in eq.(34 ) is the ratio @xmath69 and it is a measure of the deviation from cylindrical symmetry . if @xmath68 is taken to be zero , the theory reduced to the maier - saupe mean field theory . now we would like to give the description of tsallis thermostatistics with its axioms below . tsallis thermostatistics ( tt ) has been extensively used to investigate the concepts of statistical mechanics and thermodynamics . in this context , it has been applied to various phenomena [ 17 ] and tt seems to be appropriate for the study of nonextensive systems . as mentioned above , we have choosen the nematic - isotropic transition as an application area of tt in earlier studies [ 11 - 13 ] and the present study would be the extensional study to investigate the nematic liquid crystals . in all of these earlier studies , we consider that the uniaxial nematic liquid crystals are formed by cylindrically symmetric molecules . however , we wonder in this study that if the mesophase could be formed by non - cylindrically symmetric molecules , how does the nonextensivity affect the nematic - isotropic transition ? therefore our starting point is luckhurst et al.s theory [ 10 ] which is a molecular field theory and assumes the molecules of the nematic liquid crystals to be non - cylindrical and we examine the effects of the nonextensivity on the nematic - isotropic transition by considering this molecular field theory . the understanding of the basic principles and properties of tt has gain fundamental importance . in this manner , tsallis et al . have studied the role of the constraint within tt . tsallis proposed an entropy definition in 1988 [ 18 ] : @xmath70 where @xmath71 is a constant , @xmath72 is the probabilty of the system in the @xmath73 microstate , @xmath40 is the total number of configurations and @xmath74 is called the entropic index whose meaning will be given below . in the limit @xmath75 , the entropy defined in eq.(35 ) reduced to the boltzmann - gibbs ( bg ) entropy . in other words , tt contains boltzmann - gibbs statistics as a special case with @xmath76 . as well known , bg statistics is a powerful one to study a variety of the physical systems . however , it fails for the systems which @xmath77 have long range interactions , @xmath78 have long range memory effects and @xmath79 evolve in multi - fractal space - time . the systems which has these properties is called nonextensive and if a system does obey these restrictions , bg statistics seems to be inappropriate and one might need a nonextensive formalism to study on the physical system . it will be useful to write the pseudo - additivity entropy rule , @xmath80 which reflects the character of the entropic index @xmath74 and also of the nonextensivity . in this equation , @xmath81 and @xmath39 are two independent systems . since all cases @xmath82 ( which is called nonnegativity property ) , @xmath83 , @xmath76 and @xmath84 , correspond to superadditivity , additivity and subadditivity respectively . because of some unfamiliar consequences of first two energy constraints discussed in [ 13,19 ] , tsallis et al . present the third internal energy constraint as follows : @xmath85 this choice is commonly considered to study physical systems and is denoted as the tsallis - mendes - plastino ( tmp ) choice . the optimization of tsallis entropy given by eq.(35 ) according to the third choice of the energy constraint results in @xmath86 ^{\frac{1}{1-q}}}{% z_{q}^{(3)}}\ ] ] with @xmath87 ^{\frac{1}{1-q}}.\ ] ] therefore the @xmath74-expectation value of any observable is defined as @xmath88 where @xmath81 represents any observable quantity which commutes with hamiltonian . as can be expected , when @xmath76 , the @xmath74-expectation value of the observable reduces to the conventional one . at this stage , the important point is to solve eq.(40 ) which is an implicit one , and tsallis et al . suggest two different approaches ; @xmath89 and @xmath90{0xb4}}% % beginexpansion { \acute{}}% % endexpansion $ ] transformation . the nonextensivity , in fact , appears in systems where long range interactions and/or fractality exist , and such properties have been invoked in recent models of manganites [ 20,21 ] as well as in the interpretation of experimental results . these properties are also appear in [ 22 ] where the role of the competition between different phases to the physical properties of these materials is emphasized . the formation of micro - clusters of competing phases , with fractal shapes , randomly distributed in the material is considered in [ 23,24 ] , and the role of long range interactions to phase segregation [ 25,26 ] . if we look at the recent studies in which tt is employed , we see also that the entropic index is frequently established from the dynamics of the system under consideration analytically . some of such studies are : boghosian et al.s study [ 27 ] for the incompressible navier - stokes equation in boltzmann lattice models , baldovin et al.s studies [ 28,29 ] at the edge of chaos for the logistic map universality class , oliveira et al. studies [ 20,21 ] on manganites . according to these axioms of tt given above , the order parameters are calculated by using eq.(40 ) instead of eqs.(30 ) and ( 31 ) . the exponential form of any function , say x , is given by @xmath91 ^{\frac{1}{1-q}}\ ] ] in tsallis thermostatistics . substituting this equation in the calculation , eq.(40 ) is employed and then the self - consistent equations are solved to obtain the order parameters . the generalized free energy then follows within tt : @xmath92 -rt\ln _ { q}(z_{q}),\ ] ] where @xmath93 . now it is the time to give the results of application of tt to the luckhurst et al.s theory presented above . the first stage in our calculation is the identification of the transition from the uniaxial nematic phase to the isotropic phase . this may be accomplished for a given molecular symmetry by determining the value of @xmath94 at which the orientational free energy vanishes , provided there is no volume change at the transition . the results of such calculations are listed tables ( 1 - 3 ) . table 1 shows the transition temperatures , the critical values of order parameters @xmath49 and @xmath95 at the transition temperature and the entropy changes at the transition concerning @xmath96 for some values of @xmath74 . we can see immediately that increasing the entropic index increases @xmath97 and so decreases the nematic - isotropic transition temperature for constant @xmath98 . similar behaviour is also seen from tables ( 2 ) and ( 3 ) concerning @xmath99 and @xmath100 respectively . also we see from tables ( 1 - 3 ) that increasing the entropic index increases the order parameter @xmath101 at the transition for @xmath102 and @xmath100 . however there is a somewhat suprising result that as @xmath74 is increased , the secondary order parameter @xmath53 also increases at the transition for @xmath103 while this is not the case for @xmath96 , that is , the secondary order parameter decreases with increasing @xmath74 for @xmath104 . we also observe from the tables ( 1 - 3 ) that the entropy changes at the transition increases with increasing @xmath74 . in fig.(1 ) , we illustrate the dependence of the order parameter @xmath105 and on reduced temperature for some @xmath74 values , with @xmath106 only . the results are plotted as a function of the reduced variable @xmath107 which is identical to the reduced temperature @xmath108 provided the coefficients @xmath35 are themselves independent of temperature . we observe that as @xmath74 increased , the transition becomes more markedly first - order . it is well known from [ 10 ] that the effect of introducing deviations from cylindrical symmetry is to lower the curves for @xmath109 and @xmath100 respectively , as well as changing their slopes . it is clear from fig.(1 ) that as @xmath74 is decreased , the similar behaviour is also seen , that is , the slope of the curve for @xmath106 decreases with increasing the entropic index @xmath74 . the similar behaviour could also expected for other @xmath68 values @xmath110 . the secondary order parameter @xmath111 is plotted as a function of @xmath49 , that is @xmath62 , in fig.(2 ) with @xmath99 for various values of the entropic index . these results exhibit an unusual behaviour , for the order parameter @xmath53 is observed to increase with increasing @xmath62 pass through a maximum and then decrease . it is interesting to note that as @xmath74 is decreased , the maximum value of the secondary order parameter assumes a higher value than those concerning lower value of @xmath74 . another important point is that maier - saupe theory gives a universal value of the @xmath49 at the transition temperature @xmath112 . however both luckhurst et al.s theory and the generalized form of the maier - saupe theory , including the present study , assume different values of this order parameter with changing @xmath68 and @xmath74 parameters . this fact is consistent with experiment in that experimental values of @xmath49 for various nematics assume in a range of @xmath113 [ 30 ] . while @xmath68 denotes the ratio @xmath69 , i.e. it is responsible from the deviations from the cylindrical symmetry , @xmath74 is a measure of the nonextensivity of the system . but it is interesting that they exhibit some similar behaviour about the critical value of the order parameter @xmath49 at the transition . [ cols="^,^,^,^,^",options="header " , ] table 1 . @xmath114 c = 0.2 + q @xmath115_k@xmath115@xmath115@xmath115s_c / nk @xmath1160.99@xmath1154.335@xmath1150.332@xmath1150.0334@xmath1150.256@xmath1160.995@xmath1154.36@xmath1150.336@xmath1150.0335@xmath1150.265@xmath1161@xmath1154.386@xmath1150.341@xmath1150.0336@xmath1150.275@xmath1161.005@xmath1154.412@xmath1150.348@xmath1150.0337@xmath1150.301@xmath1161.01@xmath1154.438@xmath1150.354@xmath1150.0338@xmath1150.314@xmath117 table 2 . @xmath114 c = 0.3 + q @xmath115_k@xmath115@xmath115@xmath115s_c / nk @xmath1160.99@xmath1154.104@xmath1150.2@xmath1150.0399@xmath1150.101@xmath1160.995@xmath1154.132@xmath1150.203@xmath1150.0404@xmath1150.105@xmath1161@xmath1154.160@xmath1150.207@xmath1150.041@xmath1150.112@xmath1161.005@xmath1154.1894@xmath1150.212@xmath1150.0417@xmath1150.117@xmath1161.01@xmath1154.218@xmath1150.218@xmath1150.0424@xmath1150.126@xmath117 table 3 . many molecular theories assume that an essential characteristic of compounds forming liquid crystals is the rod - like shape of their constituent molecules that means an high length to breadth ratio . this assumption leads to the approximation that the molecules might be assumed to have cylindrical symmetry . however the molecules , in fact , are lath - like and thus do not possess the high symmetry . in this sense , a molecular theory was developed [ 10 ] for an ensemble of such particles based upon a general expansion of the pairwise intermolecular potential together with the molecular field approximation . tsallis thermostatistics has been commonly employed in studying on the physical systems as a nonextensive statistics and many studies are investigating the nonextensive effects in different systems and phenomena . thus we would like to handle , in this study , this molecular field theory [ 10 ] and to investigate the effects of the nonextensivity for uniaxial nematics formed by non - cylindrically symmetric molecules . with this aim , we investigate the dependence of the long range order parameter on reduced temperature and report the variation of the critical values of the order parameters and entropy change at the transition temperature with the entropic index . [ 1 ] a. saupe , z. naturf . a. 19 ( 1964 ) 161 ; p. diehl and c. l. khetrapal , nmr basic principles prog . 1 ( 1969 ) 1 . [ 2 ] j. a. pople , proc . r. soc . a 221 ( 1954 ) 498 . [ 3 ] r. alben , j. r. mccoll , c. s. shih , solid st . commun . 11 ( 1972 ) 1081 . [ 4 ] niederberger , w. , diehl , p. , lunazzi , l. , molec . phys . 26 ( 1973 ) 571 . [ 5 ] m. j. freiser , phys . ( 1970 ) 1041 ; liquid crystals 3 , 1972 , edited by g. h. brown and m. m. labes , part 1 , 281 . [ 6 ] c. s. shih , r. alben , j. chem . 57 ( 1972 ) 3057 . [ 7 ] r. alben , phys . rev 30 ( 1973 ) 778 . [ 8 ] j. p. straley , phys . a 10 ( 1974 ) 1881 . [ 9 ] e. sackman , p. krebs , j. u. rega , j. voss , h. mhwald , molec . crystals and liq . crystals 24 ( 1973 ) 283 . [ 10 ] g. r. luckhurst , c. zannoni , p. l. nordio , u. segre , mol . phys . 30 ( 1975 ) 1345 . [ 11 ] o. kayacan , f. bykkili , d. demirhan , physica a 301 ( 2001 ) 255 . [ 12 ] o. kayacan , physica a 328 ( 2003 ) 205 . [ 13 ] o. kayacan , chem . phys . in press . [ 14 ] m. e. rose , elementary theory of angular momentum , john wiley & sons , 1957 . [ 15 ] t. d. shultz , liquid crystals , 3 , 1972 , edited by g. h. brown and m. m. labes , part 1 , 263 . [ 16 ] a. saupe , angew . int . edition 7 ( 1968 ) 97 . [ 17 ] see http://tsallis.cat.cbpf.br for an updated bibliography . [ 18 ] c. tsallis , j. stat . 52 ( 1988 ) 479 . [ 19 ] c. tsallis , r.s . mendes and a.r . plastino , physica a ( 1998 ) 534 . [ 20 ] m. s. reis , j. p. arajo , v. s. amaral , e. k. lenzi , and i. s. oliveira , phys . b 66 ( 2002 ) 134417 . [ 21 ] m.s . reis , j.c.c . freitas , m.t.d . orlando , e.k . lenzi , i.s . oliveira , europhys . 58 ( 2002 ) 42 . [ 22 ] e. dagotto , t. hotta , a. moreo , phys . 344 ( 2001 ) 1 . [ 23 ] m. mayr , a. moreo , j.a . verges , j. arispe , a. feiguin , e. dagotto , phys . 86 ( 2001 ) 135 . [ 24 ] a.l . malvezzi , s. yunoki , e. dagotto , phys . b 59 ( 1999 ) 7033 . [ 25 ] a. moreo , s. yunoki , e. dagotto , science 283 ( 1999 ) 2034 . [ 26 ] j. lorenzana , c. castellani , c.d . castro , phys . b 64 ( 2001 ) 235127 . [ 27 ] b.m . boghosian , p.j . love , p.v . coveney , i.v . karlin , s. succi , j. yepez , galilean - invariant lattice - boltzmann models with h - theorem , preprint ( 2002 ) , [ e - print : cond - mat/0211093 ] . [ 28 ] f. baldovin , a. robledo , phys . e 66 ( 2002 ) 045104 ( r ) . [ 29 ] f. baldovin , a. robledo , nonextensive pesin identity : exact renormalization group analytical results for the dynamics at the edge of chaos of the logistic map , preprint ( 2003 ) , [ e - print : cond - mat/0304410 ] . [ 30 ] a. beguin , j. c. dubois , p. le barny , j. billard , f. bonamy , j. m. busisine , p. cuvelier , sources of thermodynamic data on mesogens , mol . cryst . 115 ( 1984 ) 1 . table 1 . the order parameters and entropy change at the nematic - isotropic transition with @xmath96 for some @xmath74 values .

many molecular theories of nematic liquid crystals consider the constituent molecules as cylindrically symmetric . in many cases , this approximation may be useful . however the molecules of real nematics have lower symmetry . therefore a theory was developed ( mol . phys . 30 ( 1975 ) 1345 ) for an ensemble of such particles based upon a general expansion of the pairwise intermolecular potential together with the molecular field approximation . in this study , we would like to handle this molecular field theory by using tsallis thermostatistics which has been commonly used for a decade to study the physical systems . with this aim , we would like to investigate the dependence of the order parameters on temperature and would like to report the variation of the critical values of the order parameters at the transition temperature with the entropic index . pacs number(s ) : 05.20.-y , 05.70.-a , 61.30.cz , 61.30 . gd . keywords : tsallis thermostatistics , maier - saupe theory , nematic liquid crystals , non - cylindrically symmetrical molecules .

the enrichment of the intergalactic medium ( igm ) with heavy elements has , over the past decade , become a key tool in understanding star and galaxy formation by providing a fossil record of metal formation and galactic feedback . absorption line spectroscopy has revealed , among other findings , that the low - density ( @xmath10 ) intergalactic medium ( igm ) , as probed by the ly@xmath11 forest and through , , , , , , and other transitions , is at least partly enriched at all redshifts and densities probed . in particular , recent studies indicate that : * when smoothed over large ( @xmath12 kpc ) scales , the abundance of carbon decreases as gas overdensity @xmath13 does and has a scatter of @xmath14dex at fixed density . there is carbon in at least some gas at all densities down to at least the mean cosmic density , with the median carbon metallicity obeying [ c / h ] @xmath15 at @xmath16 ( schaye et al . 2003 , hereafter * ? ? ? * on smaller ( @xmath17 kpc ) scales the distribution of metals is less well known , but observations suggest that the metals may be concentrated in small , high - metallicity patches @xcite . * there is no evidence for metallicity evolution from redshift @xmath18 to @xmath19 @xcite and metals exist at some level at @xmath20 @xcite . in connection with this observed widespread distribution of metals , a general picture has emerged that galactic winds driven largely from young and/or starburst galaxies have enriched the igm . the same feedback may account for the dearth of low - luminosity galaxies relative to the halo mass function ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , and also for the mass - metallicity relationship of galaxies ( e.g. , * ? ? ? * ; * ? ? ? . however , a detailed understanding of the various feedback processes is lacking and there are still open questions and controversies concerning the time and relative importance of the various enrichment processes , and concerning the implications for galaxy formation . both theoretical modeling and observations of intergalactic ( ig ) enrichment are now advancing to the point where comparison between the two can provide crucial insight into these issues , but this comparison is not without problems . two key difficulties concern the ionization correction required to convert observed ionic abundances into elemental abundances . first , while the oft - studied ions and are observationally convenient , they are poor probes of hot ( @xmath21k ) gas , because the ion factions /c and /si both fall dramatically with temperature . thus , the hot remnants of fast outflows might be largely invisible in these ions . second , the dominant uncertainty in both the absolute and relative abundance inferences stems from uncertainty in the spectral shape of the ultraviolet ionizing background radiation ( uvb ) . analysis of oxygen , as probed by , has the potential to shed light on both problems : this ionization state becomes prevalent in some of the very phases in which and become rare , and its abundance depends on the uvb shape differently than those of other ions , helping break the degeneracy between abundances and uvb shape . the challenge posed by is that at @xmath22 it is strongly contaminated by both ly@xmath11 and ly@xmath23 lines , making its identification and quantification difficult . previous studies of high-@xmath24 oxygen enrichment using line fitting @xcite or pixel statistics @xcite have reliably detected oxygen in the igm , and quantified its abundance in relatively dense gas , but have not assessed the oxygen abundance with a very large data sample , at very low - densities , or in a unified treatment with other available ions . here we extend to our application of the pixel optical depth " technique ( e.g. , aguirre , schaye & theuns , hereafter * ? ? ? * ) to a large set of high quality vlt / uves and keck / hires spectra . the results , when combined with previous studies of and @xcite and of and ( aguirre et . 2004 ; hereafter * ? ? ? * ) , give a comprehensive observational assessment of ig enrichment by carbon , silicon and oxygen , with significantly reduced uncertainties due to the uvb shape , as well as new data on the importance of hot , collisionally ionized gas . we have organized this paper as follows . in [ sec - data ] and [ sec - overview ] we briefly describe our sample of qso spectra . the analysis method is described briefly in [ sec - overview ] and then in greater depth in the remainder of [ sec - meth ] , with heavy reference to papers i , ii and iii . the basic results are given in [ sec - resrel ] and discussed in [ sec - discuss ] . finally , we conclude in [ sec - conc ] . all abundances are given by number relative to hydrogen , and solar abundance are taken to be @xmath25 , @xmath26 , and @xmath27 @xcite . we analyze 17 of the 19 high - quality ( @xmath28 velocity resolution , s / n @xmath29 ) absorption spectra of quasars used in papers ii and iii . the two highest - redshift spectra used in those previous studies were excluded here because the severe contamination of the region by lines makes detection of nearly impossible and also introduces very large continuum fitting errors in the region . fourteen spectra were taken with the uv - visual echelle spectrograph ( uves , * ? ? ? * ) on the very large telescope and three were taken with the high resolution echelle spectrograph ( hires , * ? ? ? * ) on the keck telescope . for convenience , the observed qsos are listed in table [ tbl : sample ] . llcccclll q1101 - 264 & 2.145 & 1.878 & 2.103 & 3050.00 & uves & 1 & 1.6 + q0122 - 380 & 2.190 & 1.920 & 2.147 & 3062.00 & uves & 2 & 0.6 + j2233 - 606 & 2.238 & 1.963 & 2.195 & 3055.00 & uves & 3 & 1.1 + he1122 - 1648 & 2.400 & 2.112 & 2.355 & 3055.00 & uves & 1 & 1.4 + q0109 - 3518 & 2.406 & 2.117 & 2.361 & 3050.00 & uves & 2 & 1.5 + he2217 - 2818 & 2.406 & 2.117 & 2.361 & 3050.00 & uves & 3 & 1.6 + q0329 - 385 & 2.423 & 2.133 & 2.377 & 3062.00 & uves & 2 & 1.2 + he1347 - 2457 & 2.534 & 2.234 & 2.487 & 3050.00 & uves & 1,2 & 2.5 + pks0329 - 255 & 2.685 & 2.373 & 2.636 & 3150.00 & uves & 2 & 1.5 + q0002 - 422 & 2.76 & 2.441 & 2.710 & 3055.00 & uves & 2 & 1.6 + he2347 - 4342 & 2.90 & 2.569 & 2.848 & 3428.00 & uves & 2 & 1.5 + q1107 + 485 & 3.00 & 2.661 & 2.947 & 3644.36 & hires & 4 & 2.3 + q0420 - 388 & 3.123 & 2.774 & 3.068 & 3760.00 & uves & 2 & 1.8 + q1425 + 604 & 3.20 & 2.844 & 3.144 & 3736.20 & hires & 4 & 2.1 + q2126 - 158 & 3.268 & 2.906 & 3.211 & 3400.00 & uves & 2 & 2.0 + q1422 + 230 & 3.62 & 3.225 & 3.552 & 3645.24 & hires & 4 & 3.1 + q0055 - 269 & 3.655 & 3.257 & 3.586 & 3423.00 & uves & 1 & 4.0 regions within @xmath30 from the quasars , where @xmath31 is the hubble parameter at redshift @xmath24 extrapolated from its present value ( @xmath32 ) assuming @xmath33 , were excluded to avoid proximity effects . regions thought to be contaminated by absorption features that are not present in our simulated spectra ( e.g. , damped ly@xmath11 systems ) were also excluded from the analysis . lyman continuum contamination increases significantly towards lower wavelengths , whereas ( as described below ) our correction for this contamination assumes that it is non - evolving . to mitigate this effect , only the red portion [ @xmath34 of the qso spectra used in papers ii and iii is analyzed in this paper . as in @xcite , this was found to result in smaller errors than using the full region . further details concerning the sample and data reduction are given in paper ii ( 2 ) . the pixel optical depth method we use for measuring is similar to that described in papers i , ii , and iii . section [ sec - overview ] contains a brief outline of the method ; [ sec - confit ] and [ sec - recovery ] describe continuum fitting and contamination corrections , which have been changed slightly from the methods described in papers ii and iii ; [ sec - ovitest ] describes tests of the recovery , and [ ioncorr ] discusses the ionization balance of the relevant species , and describes how ionization corrections are performed . the basic method for analysis of each qso spectrum is as follows : 1 . optical depths due to ly@xmath11 ( @xmath35 ) absorption are recovered for all pixels in the ly@xmath11 forest region , using higher - order lyman lines to estimate optical depths for saturated pixels . the pixel optical depth at the corresponding wavelengths of the metal lines ( @xmath36 ) , ( @xmath37 ) , and ( @xmath38 ) are recovered , making several corrections to reduce contamination and noise . the recovered optical depth in one transition is compared with that of another , by binning the pixels in terms of the optical depth of , , or , and plotting the median ( or some other percentile of ) optical depth of . a correlation then indicates a detection of absorption . an example is shown in fig . [ fig : oviscat ] . as was done in papers ii and iii , an identical analysis is applied to synthetic spectra generated using a cosmological , hydrodynamical simulation , kindly provided by tom theuns . for each observed quasar we generate 50 corresponding simulated spectra with the same noise properties , wavelength coverage , instrumental broadening , and pixel size as the observed spectra . for each uvb model ( of which several are used ; see below ) the carbon distribution as measured in paper ii , and the value of [ si / c ] from paper iii , are imposed on the fiducial spectra . an oxygen abundance is assigned by assuming a _ constant , uniform _ value of [ o / c ] . ionization balances are calculated using cloudy(version 94 ; see ferland et al . 1998 and ferland 2000 for details ) . a direct comparison of the results from these simulated and observed spectra allows for inferences about the distribution of oxygen , carbon , and silicon . the same simulation was used in papers i ( 3 ) , ii , and iii , to which the reader is referred for details . this study employs the identical uvb models used in papers ii ( 4.2 ) and iii , excluding model `` qgs3.2 '' . all models are from haardt & madau ( 2001 , hereafter hm01 ) . these have been renormalized ( by a redshift - dependent factor ) such that the simulated spectra match the observed evolution of the mean ly@xmath11 absorption ( paper ii ) . the fiducial model , `` qg '' , includes contributions from both galaxies ( with a 10% escape fraction for ionizing photons ) and quasars ; `` q '' includes only quasars ; `` qgs '' is an artificially softened version of qg : its flux has been reduced by a factor of ten above 4 ryd . the uvb used in the simulation only affects the igm temperature , and was chosen to match the measurements by @xcite . a major source of error in optical depths is continuum fitting in the absorption region , where contamination by ly@xmath11 and ly@xmath23 lines is heavy . to make this fitting as accurate as possible , and to furnish an estimate of the continuum fitting error , we have applied the following procedure to the region analyzed for absorption ( in the case of observed spectra , this was done after the spectra had been continuum fitted by eye as described in paper ii 2 ) : 1 . the spectral region is divided into 20 ( rest - frame ) segments . 2 . in segments with large unabsorbed regions , an automatic continuum fitting algorithm is applied ( see 5.1 of paper ii ) , in which pixels @xmath39 below the continuum are iteratively removed . 3 . in segments without large unabsorbed regions , we identify small unabsorbed regions or regions absorbed only in ly@xmath23 ; the latter are identified by superimposing the region of the spectrum corresponding to ly@xmath11 absorption . the continuum level of the segment is fit by minimizing the deviation of identified unabsorbed regions from unit flux , and deviation of the ly@xmath23 regions from the corresponding scaled ly@xmath11 features . a spline is interpolated between the fits to all segments and the spectrum is rescaled by this spline . this procedure was applied to all observed spectra as well as to to one simulated spectrum per observed spectrum , where a 10 - 20% error in the continuum was introduced on scales of 1 , 4 and 16 segments . the median absolute errors remaining after blindly fitting the continua of the simulated spectra are given in table [ tbl : sample ] as an estimate of continuum fitting errors in the corresponding observed spectra . because the procedure is not fully automatic , we were unable to apply it to all of the simulated spectra . the region redwards of ly@xmath11 was fit in both simulated and observed spectra using the procedure described in 5.1 step i of paper ii . continuum fitting errors are much smaller for this region ( @xmath400.01@xmath41 ) . after continuum fitting the spectra , ( ly@xmath11 ) optical depths @xmath9 are derived for each pixel between the quasar s ly@xmath11 and ly@xmath23 emission wavelengths , save for regions close to the quasar to avoid proximity effects ( see 2 ) . if ly@xmath11 is saturated ( i.e. , @xmath42 , where @xmath43 and @xmath44 are the flux and noise arrays , see paper i , 4.1 ; paper ii , 5.1 , step 2 ) , higher - order lyman lines are used to estimate @xmath9 . corresponding , , and optical depths ( @xmath45 , @xmath46 , @xmath47 ) are subsequently derived for each pixel . we exclude regions of the quasar spectrum that are contaminated by absorption features that are not included in our simulated spectra , such as ly@xmath11 lines with damping wings . for @xmath46 and @xmath47 , corrections are made for self - contamination and contamination by other metal lines , as described in paper i , 4.2 . as shown in fig . [ fig : oviscat ] , when plotting each percentile in @xmath45 absorption against absorption in some other ion , the correlation disappears below some optical depth @xmath48 ( corresponding to a value @xmath49 in the other ion ) that is determined by noise , continuum fitting errors , and contamination by other lines . these effects may then be corrected for by subtracting @xmath48 from the binned optical depths , thus converting most points below @xmath49 into upper limits . for each realization and for each percentile , we compute @xmath48 as the given percentile of optical depth for the set of pixels with optical depth @xmath50 . we use values @xmath51 when binning in or , and of @xmath52 when binning in , as we never see a correlation extending below these values .. for ovi the correlations are generally less strong than for civ and we fix @xmath49 `` by hand . '' ] the error on @xmath48 for an individual realization is computed by dividing the spectrum into 5 segments , then bootstrap - resampling the spectrum by choosing these chunks with replacement , and finally computing the variance of @xmath48 as computed from 100 such resampled spectra . when the realizations are combined , @xmath53 is instead computed as the median among the realizations , and the error on this value is computed by bootstrap - resampling the realizations . for further details see paper ii , 5.1 , step 4 and paper iii , 3.4 . as noted above , ly@xmath11 and higher lyman transitions heavily contaminate the absorption regime . this can add substantial error in the recovered @xmath45 . two corrections are made to minimize this contamination . first , after recovering @xmath9 and @xmath45 , an initial correction is made for contamination by higher - order lines by subtracting @xmath54 where _ f _ is the transition oscillator strength , @xmath55 is the redshifted lyman @xmath11 wavelength corresponding to ly@xmath56 absorption observed at wavelength @xmath57 and _ i _ corresponds to first five higher order lyman lines ( i.e. , ly@xmath23 @xmath58 through ly@xmath59 @xmath60 ) . a second correction is made by taking the minimum of the doublet , @xmath61 where ` a ' and ` b ' denote the stronger and weaker doublet components , respectively . these corrections are described in detail in paper i , 4.2 . another potential contamination issue is that due to strong ly@xmath11 or ly@xmath23 absorption , some higher percentiles in absorption can become dominated by saturated pixels , so that the particular value of the percentile is determined by the contaminating lines rather than by the distribution . to remove these unreliable percentiles from consideration , the average noise @xmath62 is calculated for saturated pixels and is converted into a maximum optical depth ( @xmath63 ) . after optical depths are binned , those percentile bins with @xmath64 are excluded from the analysis . because there is substantial processing of the recovered optical depths , it is important to test how efficiently the true optical depths are recovered by our procedures . to do so , simulations were produced just as described in [ sec - overview ] , but with ( effectively ) perfect resolution , no noise , and only @xmath65 absorption ( and hence no contamination or self - contamination ) . in fig . [ fig : true_rec ] the recovered pixel optical depths are plotted against these `` true '' optical depths for a set of 60 simulated spectra , for two representative qsos ( see paper i for more such tests ) . of particular note is the efficacy of subtracting the `` flat level '' @xmath48 as described in the preceding section . ideally , @xmath48 would be determined using pixels with negligible absorption ; this is possible in the present case ( as the true optical depths are known ) and corresponding results are shown in the left panels . in a realistic case , a proxy for must be used ; in the right panels of fig . [ fig : true_rec ] , is employed and @xmath48 is computed using all pixels with @xmath66 . in both cases , the subtracted @xmath48 values are shown as horizontal dashes on the right axes of fig . [ fig : true_rec ] . overall , we find that using @xmath9 to calculate @xmath67 is effective at recovering @xmath68 for the 31st and median percentiles ; for higher percentiles the recovery is accurate only at higher @xmath45 , but the large ` scatter ' indicates that this is random , rather than systematic error . in papers i and ii it was shown from simulations that there exists a tight correlation between @xmath9 and the absorbing gas density and temperature , which could be used to predict an ionization correction ( i.e. , the ratios of /o and /h ) as a function of density ; for details see 6 in paper i and 5.1 in paper ii . as noted in papers ii and iii , this works well for and less well for , due to their mild and strong ionization correction dependence on @xmath9 , respectively . in the upper panel of figure [ fig : ionpred ] we provide a contour plot of the logarithm of the predicted fraction of ions versus temperature and density . the middle and lower panels show @xmath69 for [ o / c]=0 , and @xmath70 for [ o / h]=0 , respectively . for photoionized gas ( @xmath71 ) the fraction is highest for @xmath72 , and only weakly dependent on the temperature . however , for higher densities ( and @xmath73k ) , the fraction falls quickly , resulting in a very large ionization correction for saturated pixels . at the same time , the fraction at high density increases with @xmath74 for @xmath75k , so that collisionally ionized gas might be detected relatively easily . therefore at high @xmath9 , collisionally ionized gas can easily swamp photoionized . this can be seen most clearly in the bottom panel , which shows that for fixed [ o / h ] and density , @xmath8 increases quickly at @xmath75k , particularly if the density is high . because our @xmath76 relation is dominated by photoionized gas ( with @xmath77 ) , the effects of collisionally ionized gas are potentially important and are discussed at length in [ sec - collisional ] below . the strategy employed here is to employ our fiducial ionization corrections as in papers ii and iii , but to recognize that at high - density , the results may be significantly affected by collisionally ionized gas . it is important to note that the importance of collisionally ionized gas may be underestimated by our simulation , because we did not include a mechanism for generating galactic winds , which could shock - heat the gas surrounding galaxies . our simulation does , however , include heating by gravitational accretion shocks . once the ionization correction has been determined , and corrected optical depths and recovered optical depths @xmath9 have been obtained , the oxygen abundance can be calculated : @xmath78 = \log \left ( { \tau_{\rm ovi } \over \tau_{\rm hi } } { ( f\lambda)_{\rm hi } \over ( f\lambda)_{\rm ovi } } { n_{\rm o } \over n_{ovi}}{n_{\rm hi } \over n_{h}}\right ) - ( { \rm o / h})_\odot , \label{eq : metallicity}\ ] ] where @xmath79 and @xmath80 are the oscillator strength and rest wavelength of transition @xmath56 , respectively ( @xmath81 , @xmath82 , @xmath83 , @xmath84 ) , and we use the solar abundance @xmath25 ( number density relative to hydrogen ; anders & grevesse 1989 ) . an example of the results from this analysis applied to the observed spectrum of q1422 + 230 is shown in figure [ fig:1422ioncorr ] . in figure [ fig : trueinvsumm_o6 ] we show a test in which we have generated simulated spectra using the `` qg '' ionizing background , recovered optical depths , and applied the just - described ionization correction to recover the oxygen abundance . the true metallicity is given by the carbon distribution of paper ii ( for the qg background ) , with a fixed [ o / c]=@xmath85 ( i.e. , [ o / h]=-3.15 + 0.65@xmath13 ) , and is shown on the plot as a dashed line . for @xmath86 , the abundance recovery is promising : it overestimates by less than 0.3 dex , and the dependence of @xmath13 is reproduced . however , the overestimation appears to increase for @xmath87 , reaching approximately 1 dex for the highest overdensity bin . the difference in the high- versus low-@xmath13 gas can probably be attributed to the collisionally ionized gas residing in and around dense regions , due to gravitational accretion shocks . we should thus keep in mind that we expect to overestimate the oxygen abundance associated with strong absorbers . our basic result will be an estimate of [ o / c ] for the low - density igm as a whole , computed using four different but consistent methods , which we describe in turn . to extract as much information as possible from our data we have , as in papers ii and iii , combined the data points obtained from our entire sample . figure [ fig : o_c.fwd.zbin ] shows @xmath88 versus @xmath89 , in bins of @xmath24 . to generate these points , we begin with @xmath45 values binned in @xmath47 for each qso , as in fig . [ fig : oviscat ] for q1422 + 230 . we then subtract from each the `` flat level '' @xmath48 for that qso to adjust for noise , contamination , etc . ( see [ sec - overview ] ) , then divide by the central value of the @xmath47 bin . these points , gathered from all qsos , are rebinned by determining , for each @xmath47 bin in fig . [ fig : o_c.fwd.zbin ] , the best constant - level @xmath90-fit to all of the points in the specified redshift bin . the errors represent 1- and 2-@xmath44 confidence intervals ( @xmath91 and @xmath92 ) on this fit . the plotted lines indicate corresponding optical depths from synthetic spectra drawn from the simulation , using several uvb models , the corresponding [ c / h ] distributions as determined in paper ii , and a constant [ o / c ] value determined as follows . for each background , we generate simulated @xmath93 points in the same way as we did for the observations , but averaging over 50 simulated realizations as described in [ sec - overview ] . we then calculate a @xmath90 between all valid observed original ( not rebinned ) points and the corresponding simulated points . because we use 50 simulated realizations , the simulation errors are almost always negligible compared to the observed errors , but they are still taken into account by calculating the total @xmath90 using the formula : @xmath94^{-1 } , \label{eq : chi2}\ ] ] where @xmath95 and @xmath44 is the error in this quantity . we then add a constant offset to the simulated points ( which corresponds to scaling [ o / c ] ) such that @xmath90 is minimized . in each panel the lines connect the scaled , rebinned simulation points . the first evident result is that @xmath96 and appears to be at most weakly dependent on @xmath97 , from @xmath98 to 0 . this is unlike @xmath46 , which increases by @xmath992 dex in this @xmath47 regime ( paper ii ) . at the lowest densities the data exhibits a decline . comparing these panels suggest there is little dependence on redshift in this interval this can be seen more clearly in figure [ fig : o_c.fwd.dbin ] , which show @xmath100 versus @xmath24 in bins of @xmath47 : there is no evidence , in either the simulated or observed points , for evolution in @xmath93 for 1.5@xmath1013.5 . the observed trends in @xmath102 are reproduced well by the simulations . therefore , because @xmath93 scales with [ o / c ] , the offset in @xmath93 obtained by minimizing the @xmath90 ( eq . [ eq : chi2 ] ) against the observations can be used to reliably compute the best fit [ o / c ] . as an example , for our fiducial uvb model qg , the simulated spectra were generated with [ o / c]=0.65 and best fit by an offset of @xmath103dex ( implying a best - fit [ o / c]=0.66 ) , with @xmath104=42.7/82 . as we found in paper ii and for q1422 + 230 above , the reduced @xmath90 is somewhat low . this is due to a slight overestimate of the errors at low-@xmath47 ( paper ii ) and to the fact that the data points are not completely independent ( because single absorbers contribute to multiple data points ) . the fitted [ o / c ] values and corresponding @xmath105 , are listed in table [ tbl : allfits ] , with errors computed by bootstrap - resampling the quasars used in the @xmath90 minimization . for our fiducial model , qg , the best fit [ o / c]@xmath106 . the quasar - only background q ( which is probably too hard ; see paper ii ) gives a much lower value of [ o / c]@xmath107 . the softer qgs backgrounds gives implausibly high values of [ o / c]@xmath108 . coupling this with results from paper ii suggesting that the qgs background is unrealistically soft , strongly disfavors this uvb model . lcccccccccc qg & @xmath109 & 42.7/82 & @xmath110 & @xmath111 & 65.7/115 & @xmath112 & -3.47@xmath113 & 0.08@xmath114 & 0.65@xmath115 & 114.1/184 + q & @xmath116 & 59.9/82 & @xmath117 & @xmath118 & 65.6/115 & @xmath119 & -2.91@xmath120 & -0.06@xmath121 & 0.17@xmath122 & 113.8/184 + qgs & @xmath123 & 26.7/82 & @xmath124 & @xmath125 & 73.8/115 & @xmath126 & -4.14@xmath127 & 0.54@xmath128 & 1.31@xmath120 & 114.2/184 + we may also subdivide our sample by redshift to test the dependence of [ o / c ] on these . first , computing [ o / c ] using only spectra that have a median absorption redshift @xmath129 ( see table [ tbl : sample ] ) yields [ o / c]=@xmath130 , versus [ o / c]=@xmath131 using the spectra with @xmath132 ; these are consistent to about 1@xmath44 . using the q and qgs uvbs , the [ o / c ] values inferred from the redshift subsamples are marginally consistent with each other and with the full sample . while the @xmath93 ratios give the most direct constraints on [ o / c ] , it is also useful to examine @xmath8 , since comparing the simulated to the observed @xmath8 ratios gives an additional ( but related ) estimate of [ o / c ] ( recall that our simulation reproduces the observed @xmath133 ) .. figure [ o_h.fwd.zbin ] shows @xmath70 versus @xmath9 for our combined sample , in bins of redshift . lines again connect the simulation points ( with an overall scaling to best match the observations ) , which reproduce the observed trends in @xmath24 and @xmath9 . the scalings correspond to best - fit [ o / c ] of 0.69 @xmath4 0.06 , 0.19 @xmath4 0.06 , and 2.25 @xmath4 0.06 , for qg , q , and qgs , respectively . for qg and qgs , the inferred [ o / c ] are consistent with the results found using @xmath93 ; for our hardest uvb model , q , the [ o / c ] values are more discrepant , but still within 1.5@xmath44 . however , while the simulations reproduce the overall trends present in the data , there are some possible discrepancies . although the @xmath134 points are upper limits , the simulations also appear to fall marginally above the data in this regime . at @xmath135 , on the other hand , the simulations slightly but significantly underpredict @xmath45 . this can be seen more clearly in figure [ o_h.fwd.dbin ] , where @xmath70 versus @xmath24 in bins of @xmath9 is shown , exhibiting a clear discrepancy for points at high-@xmath24 and high @xmath9 . indeed , if we consider subsamples above and below @xmath136 , we find that for @xmath1371 we obtain [ o / c ] values consistent those obtained from @xmath93 ( 0.62 @xmath4 0.07 , 0.13 @xmath4 0.07 and 2.18 @xmath4 0.07 for qg , q , and qgs respectively ) , but for @xmath138 we obtain [ o / c ] of @xmath139 , @xmath140 , and @xmath141 for qg , q , and qgs respectively . this discrepancy in [ o / c ] for between low- and high-@xmath9 subsamples is significant at @xmath142 for the qg and qgs models , and at @xmath143 for the q model . given the tight ( but @xmath24-dependent ) relation between @xmath9 and gas density ( see paper ii , fig . 2 , and the upper axis in the top two panels of fig . [ o_h.fwd.zbin ] ) , it is useful also to divide our sample into high- and low - density subsamples . we have done this by recomputing [ o / c ] from @xmath8 using only bins with @xmath9 corresponding to @xmath144 or @xmath145 . for @xmath14610 , we obtain [ o / c ] of 0.56 @xmath4 0.08 , 0.06 @xmath4 0.08 , and 2.11 @xmath4 0.09 ( all consistent at 1@xmath44 with the full sample ) , whereas for for @xmath147 we obtain [ o / c ] of 1.02 @xmath4 0.09 , 0.51 @xmath4 0.09 , and 2.60 @xmath4 0.09 for qg , q , and qgs , respectively ( all discrepant at @xmath148 ) . this high-@xmath13 and ( less significantly ) high-@xmath9 difference might have several causes . first , it might correspond to a genuine change in [ o / c ] with gas density . in this case , however , such an effect would also be expected in the trends of @xmath93 vs.@xmath47 and no such effect is evident . thus , we consider it more likely that there exists a significant portion of hot gas not present in the simulations that contains , but lacks and . as suggested in [ ioncorr ] and fig . [ fig : ionpred ] , and further discussed below in [ sec - collisional ] , this might indicate the presence of a significant amount of collisionally ionized ( @xmath149k ) gas in the igm . a third check on our measured [ o / c ] is provided by the @xmath150 ratio . [ o_s.fwd.zbin ] shows @xmath150 versus @xmath46 for our full sample , for two cuts in @xmath24 . for each uvb the constant si / c ratio derived in paper iii and the c distribution measured in paper ii are imposed on the simulations and the [ o / c ] ratio is varied to minimize the @xmath90 difference between the observed and simulated data points . though the detection of @xmath150 is weak , the simulations appear to adequately represent the observations in the redshift range where @xmath46 is best detected , @xmath151 . from these fits we infer , for this redshift interval , [ o / c ] of @xmath152 , @xmath153 , and @xmath154 for uvb models qg , q , and qgs , respectively ; all are consistent with the results obtained from @xmath93 , though all are somewhat higher . because probes higher density gas than ( see paper iii ) , this again suggests that the simulations underpredict the amount of in and near dense gas . as a final method , we can apply the `` inversion method '' developed in paper ii , to convert @xmath8 vs. @xmath9 into [ o / h ] vs. @xmath13 by applying a density - dependent ionization correction . then , using the measured distribution of carbon from paper ii , an independent measurement of [ o / c ] can be obtained . in figure [ o_h.inv.zbin ] we show the derived [ o / h ] versus @xmath13 for our preferred uvb model qg , with data from all @xmath24 combined . the data points from the individual quasar spectra ( an example is shown in fig . [ fig:1422ioncorr ] ) have been binned in density bins of 0.25 dex . the solid line shows the least - squares fit to the data points with @xmath144 and the dotted curves indicate the 1 @xmath44 confidence limits , with the resulting fit given in the upper left corner . the errors on the fits were determined by bootstrap resampling the qsos . the dashed line is the value of [ o / h ] given by the derived [ o / c]=@xmath155 result using @xmath93 , and assuming a [ c / h ] distribution from paper ii . for @xmath156 , the [ o / h ] derived from the ionization correction agrees very well with that determined from both @xmath93 , and @xmath8 . this strengthens the result from [ sec - resrel - oc ] and [ sec - ovihi ] that [ o / c ] is indeed constant for -0.5@xmath157 1 . however , for @xmath158 the ionization correction results in substantially more [ o / h ] than that predicted using the previously - determined [ c / h ] distribution and a fixed [ o / c ] ratio . this is similar to the breakdown between the simulations and observed @xmath8 seen in fig . [ o_h.fwd.zbin ] , and the erroneously high [ o / h ] recovered for high-@xmath13 ( see fig . [ fig : trueinvsumm_o6 ] ) when applying the ionization correction to the simulated spectra . once again , this suggests the presence of collisionally ionized . the primary source of systematic uncertainty in , e.g. [ o / c ] or [ o / h ] , is the complex modeling that must be performed to extract these values from the pixel correlations . the greatest combination of importance and uncertainty is clearly the uncertainty in the shape of the uvb ; but this is discussed at length below in sec . [ sec - nuc ] so we here focus on other aspects of the modeling . the good agreement between the four methods we have employed indicates that the method is sound ; but it is also clear that there are real differences between the universe and our simulation . in particular , it is clear that the real universe has an extra component of at high density , almost certainly due to collisionally ionized gas that is not captured by those simulations . nonetheless , if we exclude those high-@xmath9 regions , the small discrepancies between [ o / c ] as measured using the different methods indicate that such effects probably do not contribute more than @xmath159 dex uncertainty to our basic results . another source of error that may be inaccurately assessed by our bootstrap resampling technique is that from continuum fitting . as shown in table [ tbl : sample ] , our estimated rms error in the absorption region is @xmath160% . to test the effect of this error on our results , we have imposed an additional error on each observed spectrum on scales of 20 , 80 , and 320 , for a total added rms error of 2% , then recomputed our results . we find that our best fit [ o / c ] from @xmath45 versus @xmath47 , and from @xmath45 versus @xmath9 , are both within @xmath161 dex . further , both the individual binned points and linear fits of the [ o / h ] values computed from the full sample ( fig . [ o_h.inv.zbin ] ) are all affected by this continuum error to a lesser degree than the quoted random errors . thus , we conclude that the continuum fitting error is not a significant systematic affecting our results . a final possible source of systematic error is that the and recombination rates used in the version of cloudy we have employed are too high by @xmath162% compared with recent experimental values , for the temperatures relevant to low - density photoionized gas . ( savin , _ private communication _ ) . for the range of densities @xmath163 we cover , this would imply a ( density - dependent ) correction of @xmath164dex . this might change our overall fits by an amount comparable to the statistical errors , but is still much smaller than uncertainties stemming from the uvb shape , and can not account for the 0.5dex of excess in @xmath45 at high @xmath165 shown in fig . [ o_h.inv.zbin ] . combining these possible sources , we estimate probable systematic errors of @xmath166dex in our basic [ o / c ] and [ o / h ] values ; this uncertainty may be somewhat greater in subsets of the data , particularly at high - density . it is important to emphasize that each quoted result is sensitive to , and applies to , only a certain range of gas densities . at the upper end , our results nominally concern gas of up to @xmath167 , though ( as noted at length above ) the high-@xmath9 range of our data is likely to be affected by collisional ionization . that range is , however , not dominant : we have checked that if all pixels with @xmath168 are excluded from the analysis , the results given for [ o / c ] in columns 2 and 4 of table [ tbl : allfits ] change only within the quoted errors . the lower end of the gas density range probed is most straightforward in results from @xmath8 ( figs . [ o_h.fwd.zbin ] and [ o_h.inv.zbin ] , which formally show ovi detections at 1@xmath44 using all qsos at z=@xmath169 for @xmath170 , or @xmath171 and confident detections at @xmath172 . in paper iii , the si abundance results were sensitive to @xmath173 @xmath174 , and in paper ii , c abundances were measured in much lower density gas . thus our results provide indirect constraints on [ o / c ] and [ o / si ] down to @xmath175 . however , there are important caveats . first , the quoted results pertain to the the full density range probed and thus are not necessarily very sensitive to the lowest densities . second , because the pixel method only works if the element on the x - axis is more easily detectable than the element on the y - axis , direct measurements of [ o / c ] and [ o / si ] from @xmath93 and @xmath150 are dominated by much higher - density that gives @xmath176 ; thus while these results are consistent with our indirect constraints , they do not address the ( somewhat implausible ) possibility that at low densities o , si , and c come from completely different gas phases . on the other hand , at _ high _ densities it is quite possible that ovi and siiv emission are dominated by different phases , so the indirectly inferred [ si / o ] values are probably both reliable and well - measured only in the moderate density range @xmath177 . although mass or volume filling - factors corresponding to these results are not well constrained ( see , e.g. , schaye & aguirre 2005 ) , the forewarned reader can convert the density range @xmath5 correspond to into a volume using paper ii . the best fitting metallicities and corresponding @xmath90/d.o.f . from papers ii , iii and this work are shown in table [ tbl : allfits ] for each uvb model . two interesting results stand out . first , for all uvb models carbon is underabundant relative to both silicon and oxygen ( being only marginally consistent with solar for uvb model q ) . second , all abundance ratios are sensitive to the uvb shape . a harder uvb results in a lower inferred [ o / c ] but a higher inferred [ si / c ] , making the [ si / o ] ratio particularly sensitive to the spectral hardness of the uvb . the extreme sensitivity of the inferred [ si / o ] ratio to the spectral shape of the uv background makes it possible to constrain feasible uvb models by making only weak assumptions about the [ si / o ] ratio . since si and o are both @xmath11 elements , they are expected to trace each other relatively well . for example , using the nucleosynthetic yields of and and a @xcite initial mass function from @xmath178 , the [ si / o ] ratio of the ejecta of a population of age @xmath179 yr is predicted to be about 0.12 and -0.03 for stars of solar and 1 percent solar metallicity , respectively . this agrees well with observations of metal - poor stars , which find [ si / o]@xmath3 @xcite . tallying the results of this work with those of papers ii and iii yields [ si / o]@xmath180 , @xmath181 , and @xmath182 for uvb models qg , q , and qgs , respectively . thus , our preferred model , qg , is nicely consistent with the expectations , but models q and qg lead to inferred [ si / o ] ratios that are highly inconsistent with both nucleosynthetic yields and observations of metal - poor stars . ( assuming [ si / c ] @xmath183 and [ si / o ] @xmath184 in the q background , for example , raises the @xmath90 of the fits in figures [ fig : o_c.fwd.zbin ] and [ o_h.fwd.zbin ] by 65 and 92 , respectively ; requiring this for qgs likewise raises the @xmath90 by 138 and 81 . ) we conclude that the uvb has a spectral shape similar to that of model qg . while our result using the qg uvb are broadly consistent with the abundance ratios in metal - poor stars and in yield calculations , the [ o / c ] and [ si / c ] may be somewhat high , by @xmath185dex . this is comparable to our systematic errors , but nevertheless interesting if taken seriously . for example , the models of @xcite that include the contributions of hypernovae ( defined as supernovae with kinetic energy @xmath186 that of normal core - collapse sne ) produce [ o / c ] @xmath99 0.6 , and [ si / c ] @xmath99 0.65 , in agreement with our results . in paper ii we combined the median [ c / h]@xmath187 with the width @xmath188)(\delta , z)$ ] of the lognormal probability distribution of [ c / h ] for @xmath189 to determine the mean c abundance versus @xmath13 . this was then integrated over the mass - weighted probability distribution @xmath13 ( obtained from our hydrodynamical simulation ) to compute the contribution by gas in this density range to the overall mean cosmic [ c / h ] . assuming that [ o / c ] is _ constant _ over this density range over which we have reliably _ measured _ oxygen abundances . ] we obtain , for our fiducial uvb model qg , [ o / h]@xmath190 , corresponding to @xmath191 + 2.1}\left ( { \omega_b \over 0.045}\right ) .\ ] ] extrapolating our [ c / h ] and [ o / c ] results even further to the full density range of the simulation would yield values @xmath192dex higher but with more uncertainty , as we have argued that our results are unreliable at the highest densities . note that these results are relatively insensitive to the uvb ( unlike those for @xmath193 in paper iii ) because for a harder uvb , the inferred [ c / h ] increases , while [ o / c ] decreases . for our quasar - only model q , these effects almost entirely cancel , yielding [ o / h]@xmath194 , and an @xmath195 value 30@xmath41 lower than for model qg . note also that these estimates include the oxygen that resides in gas that is observable in and , but they do not include oxygen in intergalactic gas that is very hot ( @xmath196 k ) or very cold ( @xmath197 k ) and shielded from ionizing radiation . if , following @xcite , we take @xmath198 , we then infer an intergalactic metal reservoir of @xmath199 this can be compared to their estimate of the total @xmath200 metal budget " of @xmath201 , indicating that @xmath202 of metals produced prior to @xmath203 reside in the component of the igm that is studied here . previous studies have explored oxygen abundances in the igm using both line - fitting ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and pixel optical depths . previous pixel studies did not attempt to convert their detections into oxygen abundances ; but we can compare to their recovered optical depths . both @xcite and claim detection of down to @xmath204 . using our combined data set binned in density ( see fig . [ o_h.inv.zbin ] ) , we obtain 1-@xmath44 detections down to about the mean density ( @xmath205 at @xmath206 ) . while is in principle an excellent tracer of metal in very low - density gas , in practice we find that for the higher redshifts where low densities are more easily probed , contamination is severe . thus we are not in practice able to constrain metals in underdense gas as claimed in previous studies , in spite of a large sample and improved techniques of removing contaminants . on the high-@xmath9 side , both and @xcite exclude pixels saturated in , so can not probe @xmath207 . ( this accounts , for example , for our detection of in q1422 + 230 , while @xcite had no detection in the same qso ) . the study of @xcite did probe high @xmath9 , where their results are broadly consistent with ours . the studies of @xcite , @xcite , @xcite , and @xcite did perform ionization corrections and we can compare our abundance determinations ( relatively directly ) to theirs . @xcite assumed to abundance of oxygen relative to carbon to be solar and inferred metallicities for various uvb models and a number of absorbers in @xmath208 systems at @xmath19 . they found that for relatively hard uv backgrounds , comparable to our model q , the ionization models yielded densities in agreement with theoretical predictions for self - gravitating clouds with the observed column densities ( @xmath209 ; @xcite ) and metallicities of @xmath210 solar , in excellent agreement with our measurement of [ o / c]@xmath211[c / h]@xmath212 for model q at @xmath213 and @xmath214 . we note that if @xcite would have allowed oxygen to be overabundant relative to carbon , they would have found that softer uvb models are required to obtain density estimates that agree with theoretical expectations for gravitationally confined clouds . @xcite employed the faint object spectrograph on _ hst _ to search for in qso spectra from redshift 1.6@xmath215 2.9 . the / ratio found by the survey favored a uvb background similar to our q , which they use to derive a metallicity of @xmath216[o / h ] @xmath217 . while somewhat higher than our value , much of the difference may be attributable to their use of the 78th percentile in the / ratio . assuming that [ o / h ] has a scatter at fixed density similar to that in [ c / h ] ( see @xcite ) , this would correspond to a _ median _ of @xmath218dex less , or @xmath219[o / h ] @xmath220 , in fairly good agreement with our numbers . @xcite divide their sample into metal - poor absorbers " with @xmath221@xmath222@xmath223 , which they take to be predominantly photoionized , and metal - rich absorbers " with @xmath221@xmath222@xmath224 , for which they assume a hotter phase . for the metal - poor systems they use a hard " uvb and assume [ o / c]@xmath225 to derive a range of @xmath226 [ o / h ] @xmath227 , and for the metal - rich phase they infer a median [ o / h ] @xmath228 to @xmath229 , depending upon the assumptions regarding the ionization balance . combining their samples they estimate cosmic density @xmath230 , corresponding to @xmath231 [ o / h ] @xmath232 if divided by the cosmic gas density @xmath233 . while precise comparison is difficult , these numbers are consistent with our corresponding estimates of [ o / h]@xmath194 , or @xmath234 using the q uvb . @xcite _ assume _ [ o / c]=0.5 and 0.0 for uvb backgrounds comparable to our qg and q , for @xmath235 , so ( as for the above studies ) comparing derived [ o / c ] values is less useful than several other points of comparison . first , in both backgrounds , their dependence of [ o / h ] upon @xmath13 is similar to that found for [ c / h ] vs. @xmath13 in paper ii , consistent with a constant [ o / c ] value . second , @xcite find that there is a clear jump in in the median [ o / h ] at @xmath23610 , while a corresponding jump is not seen in [ c / h ] , similar to our results in [ sec - ioncorr ] . they also interpret this as possibly indicating that stronger absorbers are physically more complex or multiphased . third , @xcite compute an overall contribution @xmath237 , using their hard " background ; this would correspond to a cosmic average contribution of [ o / h ] = @xmath238 ; these numbers are @xmath239dex lower than our values , but this should be regarded as good agreement given the number of assumptions made in each computation . finally , the studies of @xcite@xcite are similar to ours in employing simulated spectra to attempt to match observed absorption and thus constrain [ o / c ] and [ o / h ] . @xcite generated simulated spectra from a constant - metallicity simulation and compared ionic ratios to extant data , inferring [ c / h ] @xmath240 and evidence for overabundance of si and o relative to carbon . @xcite used q1422 + 230 and a quasar - only haardt & madau uvb model like ` q ' . the found that their data is consistent with [ o / c ] @xmath3 and [ c / h]@xmath241 ; these are quite consistent with our results using the q uvb . as an alternative interpretation , they note that a softer uvb would give high [ o / c ] ( more characteristic of type ii supernova yields ) , but also lower [ c / h ] , again consistent with our findings ; however they interpret this softness as due to patchy reionization , whereas we favor its explanation by the contribution of galaxies to the uvb . as discussed above , the difference between the inferred [ o / h ] for @xmath242 and @xmath243 is probably due to collisionally ionized gas , for reasons sketched in [ ioncorr ] . [ fig : ionpred ] shows that for @xmath244k , collisional ionization dominates and the optical depth ratios become independent of the density . for @xmath245 k , on the other hand , the / and the / ratios both drop rapidly with increasing density . consequently , these ratios can be many orders of magnitude higher in hot , dense gas than in warm , dense gas . the lower the density , the smaller the differences become . in fact , at high densities the fraction in warm gas is too small for to be observable . therefore , any at the redshift of very strong absorption is likely to arise in a different phase than the associated and possibly even the associated . the phase must either have a much lower density or a much higher temperature . because the fraction of hot gas in our simulation is small at all densities , we effectively assume the gas to be photo - ionized when we compare with synthetic spectra and when we correct for ionization ( as in fig . [ o_h.inv.zbin ] ) . in the latter case , we also implicitly assume that and absorption arise in the same gas phase . hence , our results from [ sec - ioncorr ] suggest the existence of a detectable amount of enriched hot ( @xmath246k ) gas associated with strong absorption . a possible physical explanation is that such systems coincide with outer regions of high @xmath24 galactic halos , where the effects of galactic winds may dominate the heating process . if the temperature exceeds 10@xmath247k in these regions , then this gas would contain significant , while lacking and . the latter two would then arise in a cooler gas phase , which has to be fairly dense in order to account for the strong absorption . the high density of the cooler phase implies that it would not produce significant photo - ionized . these results and inferences are consistent with other observational studies of . @xcite and @xcite both find that the majority of the detected absorption systems had temperatures determined from the line widths too low for collisional ionization , but can not rule out higher temperatures for some absorbers . indeed , the study of @xcite find their detected high - column density lines associated with strong absorbers are broad enough to be consistent with collisional ionization . as noted above , @xcite find a jump in [ o / h ] at @xmath23610 interpretable as a transition to a regime in which collisionally ionized gas affects the abundance inferences . while collisionally ionized gas complicates oxygen abundance inferences , the flip - side is that then provides an important probe of hot , enriched ig gas that is difficult to detect in ( e.g. , paper ii ) . indeed , hydrodynamical simulations by @xcite and @xcite predict that a significant portion of is collisionally ionized . hence , it would thus be very interesting to compare such simulations , employing as a metallicity tracer , with the observations analyzed here . we have studied the relative abundance of oxygen in the igm by analyzing , , , and pixel optical depths derived from a set of high - quality vlt and keck spectra of 17 qsos at @xmath248 , and we have compared them to realistic , synthetic spectra drawn from a hydrodynamical simulation to which metals have been added . our fiducial model employs the ionizing background model ( `` qg '' ) taken from haardt & madau ( 2001 ) for quasars and galaxies ( rescaled to reproduce the observed mean ly@xmath11 absorption ) . the simulation assumes a silicon abundance as calculated in paper iii , [ si / c]@xmath249 , and a carbon abundance as derived in paper ii : at a given overdensity @xmath13 and redshift @xmath24 , [ c / h ] has a lognormal probability distribution centered on @xmath250 and of width 0.70 dex . the main conclusions from this analysis are as follows : * for 1.9@xmath2153.6 , @xmath251 , and @xmath25210 ( when smoothed on the scale of the absorption , @xmath253 kpc ) , the fiducial simulation utilized in papers i , ii , and iii consistently agrees with the observed @xmath1 , @xmath6 , and ( to a lesser degree ) @xmath7 . fitting @xmath1 yields a constant [ o / c ] = 0.66 @xmath4 0.06 , with estimated systematic errors within @xmath254 dex . converting the observed @xmath6 into [ o / c]@xmath255 using the ionization correction method of paper ii further supports these results . * the relative abundances [ o / c ] and ( especially ) [ o / si ] are sensitive to the uvb shape . we find that our fiducial ( haardt & madau 2001 , quasars and galaxies ) spectrum gives reasonable results for both , but that significantly softer or harder uvbs , such as the haardt & madau 2001 quasar - only uvb , give results that are highly inconsistent with both theoretical yields and observed abundance ratios in other low - metallicity environments , and should not be considered tenable . * our results , both from applying the ionization correction and from comparing the simulations to the observations , suggest no evolution in [ o / h ] over the redshift range 1.9@xmath2563.6 , but a strong dependence on @xmath13 . both results are consistent with those found in paper ii for [ c / h ] . * for @xmath257 and @xmath13 @xmath258 10 the value of [ o / c ] ( derived by comparison to the simulations ) is inconsistent with that found at lower densities , and @xmath218dex higher than that predicted using the carbon distribution of paper ii and a density - independent [ o / c ] value . this might in principle suggest a density - dependent [ o / c ] ratio , but we favor the interpretation that a fraction of the high-@xmath13 absorbing gas is collisionally ionized , and that this leads to an erroneously large ionization correction in this regime . this interpretation is supported by our simulated spectra as well as by the observation that lines associated with strong absorbers tend to be broader than those associated with weak systems @xcite . we are grateful to wallace sargent , michael rauch and tae - sun kim for providing the keck / hires and vlt / uves data used here and in papers i - iii . we are also extremely grateful to daniel savin for his assistance in understanding and assessing the systematic uncertainties in recombination rates . thanks also to rob wiersma for computing the expected si / o ratio from nucleosynthetic yields taken from the literature . we thank the anonymous referee for providing comprehensive and helpful feedback that improved the manuscript . aa and cdh gratefully acknowledge support from nsf grant ast-0507117 and js from marie curie excellence grant mext - ct-2004 - 014112 .

we have studied the abundance of oxygen in the igm by analyzing , , , and pixel optical depths derived from a set of high - quality vlt and keck spectra of 17 qsos at @xmath0 . comparing ratios @xmath1 to those in realistic , synthetic spectra drawn from a hydrodynamical simulation and comparing to existing constraints on [ si / c ] places strong constraints on the ultraviolet background ( uvb ) model using weak priors on allowed values of [ si / o ] : for example , a quasar - only background yields [ si / o ] @xmath2 , highly inconsistent with the [ si / o ] @xmath3 expected from nucleosynthetic yields and with observations of metal - poor stars . assuming a fiducial quasar+galaxy uvb consistent with these constraints yields a primary result that [ o / c ] = 0.66 @xmath4 0.06 @xmath4 0.2 ; this result is sensitive to gas with overdensity @xmath5 . consistent results are obtained by similarly comparing @xmath6 and @xmath7 to simulation values , and also by directly ionization - correcting @xmath8 as function of @xmath9 into [ o / h ] as a function of density . subdividing the sample reveals no evidence for evolution , but low- and high-@xmath9 samples are inconsistent , suggesting either density - dependence of [ o / c ] or more likely prevalence of collisionally - ionized gas at high density .

dynamic traffic assignment ( dta ) is the descriptive modeling of time - varying flows on traffic networks consistent with traffic flow theory and travel choice principles . dta models describe and predict departure rates , departure times and route choices of travelers over a given time horizon . it seeks to describe the dynamic evolution of traffic in networks in a fashion consistent with the fundamental notions of traffic flow and travel demand ; see @xcite for some review on dta models and recent developments . _ dynamic user equilibrium _ ( due ) of the open - loop type , which is one type of dta , remains a major modern perspective on traffic modeling that enjoys wide scholarly support . it captures two aspects of driving behavior quite well : departure time choice and route choice @xcite . within the due model , travel cost for the same trip purpose is identical for all utilized route - and - departure - time choices . the relevant notion of travel cost is _ effective travel delay _ , which is a weighted sum of actual travel time and arrival penalties . in the last two decades there have been many efforts to develop a theoretically sound formulation of dynamic network user equilibrium that is also a canonical form acceptable to scholars and practitioners alike . there are two essential components within the due models : ( 1 ) the mathematical expression of nash - like equilibrium conditions ; and ( 2 ) a network performance model , which is , in effect , an embedded _ dynamic network loading _ ( dnl ) problem . the dnl model captures the relationships among link entry flow , link exit flow , link delay and path delay for any given set of path departure rates . the dnl gives rise to the notion of _ path delay operator _ , which is viewed as a mapping from the set of feasible path departure rates to the set of path travel times or , more generally , path travel costs . properties of the path delay operator are of fundamental importance to due models . in particular , continuity of the delay operators plays a key role in the existence and computation of due models . the existence of dues is typically established by converting the problem to an equivalent mathematical form and applying some version of brouwer s fixed - point existence theorem ; examples include @xcite and @xcite . all of these existence theories rely on the continuity of the path delay operator . on the computational side of analytical due models , every established algorithm requires the continuity of the delay operator to ensure convergence ; an incomplete list of such algorithms include the fixed - point algorithm @xcite , the route - swapping algorithm @xcite , the descent method @xcite , the projection method @xcite , and the proximal point method @xcite it has been difficult historically to show continuity of the delay operator for general network topologies and traffic flow models . over the past decade , only a few continuity results were established for some specific traffic flow models . @xcite use the link delay model @xcite to show the continuity of the path delay operator . their result relies on the _ a priori _ boundedness of the path departure rates , and is later improved by a continuity result that is free of such an assumption @xcite . in @xcite , continuity of the delay operator is shown for networks whose dynamics are described by the lwr - lax model @xcite , which is a variation of the lwr model that does not capture vehicle spillback . their result also relies on the _ a priori _ boundedness of path departure rates . @xcite consider vickrey s point queue model @xcite and show the continuity of the corresponding path delay operator for general networks without invoking the boundedness on the path departure rates . all of these aforementioned results are established for network performance models that do not capture vehicle spillback . to the best of our knowledge , there has not been any rigorous proof of the continuity result for dnl models that allow queue spillback to be explicitly captured . on the contrary , some existing studies even show that the path travel times may depend discontinuously on the path departure rates , when physical queue models are used . for example , @xcite uses the cell transmission model and signal control to show that the path travel time may depend on the path departure rates in a discontinuous way . such a finding suggests that the continuity of the delay operator could very well fail when spillback is present . this has been the major hurdle in showing the continuity or identifying relevant conditions under which the continuity is guaranteed . this paper bridges this gap by articulating these conditions and providing accompanying proof of continuity . this paper presents , for the first time , a rigorous continuity result for the path delay operator based on the lwr network model , which explicitly captures physical queues and vehicle spillback . in showing the desired continuity , we propose a systematic approach for analyzing the well - posedness of two specific junction models : a merge and a diverge model , both originally presented by @xcite . the underpinning analytical framework employs the _ wave - front tracking _ methodology @xcite and the technique of _ generalized tangent vectors _ @xcite . a major portion of our proof involves the analysis of the interactions between kinematic waves and the junctions , which is frequently invoked for the study of well - posedness of junction models ; see @xcite for more details . such analysis is further complicated by the fact that vehicle turning ratios at a diverge junction are determined endogenously by drivers route choices within the dnl procedure . as a result , special tools are developed in this paper to handle this unique situation . as we shall later see , a crucial step of the process above is to estimate and bound from below the minimum network supply , which is defined in terms of local vehicle densities in the same way as in @xcite . in fact , if the supply somewhere tends to zero ( that is , when traffic approaches the jam density ) , the well - posedness of the diverge junction may fail , as we demonstrate in section [ subsubsecexample ] . this has also been confirmed by the earlier study of @xcite , where a wave of jam density is triggered by a signal red light and causes spillback at the upstream junction , leading to a jump in the path travel times . remarkably , in this paper we are able to show that ( 1 ) if the supply is bounded away from zero ( that is , traffic is bounded away from the jam density ) , then the diverge junction model is well posed ; and ( 2 ) the desired boundedness of the supply is a natural consequence of the dynamic network loading procedure that involves only the merge and diverge junction models we study here . this is a highly non - trivial result because it not only plays a role in the continuity proof , but also implies that gridlock can never occur in the network loading procedure in any finite time horizon . the final continuity result is presented in section [ subseccontinuity ] , following a number of preliminary results set out in previous sections . although our continuity result is established only for networks consisting of simple merge and diverge nodes , it can be extended to networks with more complex topologies using the procedure of decomposing any junction into several simple merge and diverge nodes @xcite . moreover , the analytical framework employed by this paper can be invoked to treat other and more general junction topologies and/or merging and diverging rules , and the techniques employed to analyze wave interactions will remain valid . the main contributions made in this paper include : * formulation of the lwr - based dynamic network loading ( dnl ) model with spillback as a system of _ partial differential algebraic equations _ ( pdaes ) ; * a continuity result for the path delay operator based on the aforementioned dnl model ; * a novel method for estimating the network supply , which shows that gridlock can never occur within a finite time horizon . the rest of this paper is organized as follows . section [ seclwrdnl ] recaps some essential knowledge and notions regarding the lwr network model and the dnl procedure . section [ secpdae ] articulates the mathematical contents of the dnl model by formulating it as a pdae system . section [ subsecwellposed ] introduces the merge and diverge models and establishes their well - posedness . section [ seccontinuityproof ] provides a final proof of continuity and some discussions . finally , we offer some concluding remarks in section [ secconclusion ] . throughout this paper , the time interval of analysis is a single commuting period expressed as @xmath0 $ ] for some @xmath1 . we let @xmath2 be the set of all paths employed by travelers . for each path @xmath3 we define the path departure rate which is a function of departure time @xmath4 $ ] : @xmath5~\rightarrow~\mathbb{r}_+\ ] ] where @xmath6 denotes the set of non - negative real numbers . each path departure rate @xmath7 is interpreted as a time - varying path flow measured at the entrance of the first arc of the relevant path , and the unit for @xmath7 is _ vehicles per unit time_. we next define @xmath8 to be a vector of departure rates . therefore , @xmath9 can be viewed as a vector - valued function of @xmath10 , the departure time . instead of @xmath11 to denote path flow vectors . ] the single most crucial ingredient t is the path delay operator , which maps a given vector of departure rates @xmath12 to a vector of path travel times . more specifically , we let @xmath13,\quad \forall p\in \mathcal{p}\ ] ] be the path travel time of a driver departing at time @xmath10 and following path @xmath14 , given the departure rates @xmath12 associated with all the paths in the network . we then define the path delay operator @xmath15 by letting @xmath16 , which is a vector consisting of time - dependent path travel times @xmath17 . we recap the network extension of the lwr model @xcite , which captures the formation , propagation , and dissipation of spatial queues and vehicle spillback . discussion provided below relies on general assumptions on the fundamental diagram and the junction model , and involves no _ ad hoc _ treatment of flow propagation , flow conservation , link delay , or other critical model features . we consider a road link expressed as a spatial interval @xmath18\subset \mathbb{r}$ ] . the _ partial differential equation _ ( pde ) representation of the lwr model is the following scalar conservation law @xmath19\times[a,\,b]\ ] ] with appropriate initial and boundary conditions , which will be discussed in detail later . here , @xmath20 denotes vehicle density at a given point in the space - time domain . the fundamental diagram @xmath21\rightarrow [ 0,\ , c]$ ] expresses vehicle flow at @xmath22 as a function of @xmath20 , where @xmath23 denotes the jam density , and @xmath24 denotes the flow capacity . throughout this paper , we impose the following mild assumption on @xmath25 : * ( f ) . * the fundamental diagram @xmath25 is continuous , concave and vanishes at @xmath26 and @xmath27 . an essential component of the network extension of the lwr model is the specification of boundary conditions at a road junction . derivation of the boundary conditions should not only obey physical realism , such as that enforced by entropy conditions @xcite , but also reflect certain behavioral and operational considerations , such as vehicle turning preferences @xcite , driving priorities @xcite , and signal controls @xcite . articulation of a junction model is facilitated by the notion of _ riemann problem _ , which is an initial value problem at the junction with constant initial condition on each incident link . there exist a number of junction models that yield different solutions of the same riemann problem . in one line of research , an entropy condition is defined based on a minimization / maximization problem @xcite . in another line of research , the boundary conditions are defined using link _ demand _ and _ supply _ @xcite , which represent the link s sending and receiving capacities . models following this approach include @xcite and @xcite . the solution of a riemann problem is given by the _ riemann solver _ ( rs ) , to be defined below . we consider a general road junction @xmath28 with @xmath29 incoming roads and @xmath30 outgoing roads , as shown in figure [ chapdnl : figlwrjunction ] . incoming links and @xmath30 outgoing links . ] we denote by @xmath31 the incoming links and by @xmath32 the outgoing links . in addition , for every @xmath33 , the dynamic on @xmath34 is governed by the lwr model @xmath35\times [ a_i,\,b_i]\ ] ] where link @xmath34 is expressed as the spatial interval @xmath36 , and we always use the subscript ` @xmath37 ' to indicate dependence on the link @xmath34 . the initial condition for this conservation law is @xmath38\ ] ] notice that the above @xmath39 initial value problems are coupled together via the boundary conditions to be specified at the junction . such a system of coupling equations is commonly analyzed using the riemann problem . [ chapdnl : rpdef]*(riemann problem ) * the riemann problem at the junction @xmath28 is defined to be an initial value problem on a network consisting of the single junction @xmath28 with @xmath29 incoming links and @xmath30 outgoing links , all extending to infinity , such that the initial densities are constants on each link : @xmath40,\qquad \qquad i\in\{1,\,\ldots,\,m\ } \\ \rho_j(0,\,x)~\equiv~\hat\rho_j\qquad & x\in [ a_j,\,+\infty ) , \qquad\qquad j\in\{m+1,\,\ldots,\,m+n\ } \end{cases}\ ] ] where @xmath41 $ ] are constants , @xmath42 . a riemann solver for the junction @xmath28 is a mapping that , given any @xmath39-tuple of riemann initial conditions @xmath43 , provides a unique @xmath39-tuple of boundary conditions @xmath44 such that one can solve the initial - boundary value problem for each link , and the resulting solutions constitute a weak entropy solution of the riemann problem at the junction . a precise definition of the riemann solver is given as follows . [ chapdnl : lwrrsdef ] * ( riemann solver ) * a riemann solver for the junction @xmath28 with @xmath29 incoming links and @xmath30 outgoing links is a mapping @xmath45 & ~\longrightarrow~ \prod_{k=1}^{m+n}\left[0,\,\rho^{jam}_k\right ] \\ \big(\hat\rho_{1},\,\ldots,\ , \hat\rho_{m+n}\big)&~\mapsto~\big(\overline\rho_1,\,\ldots,\,\overline\rho_{m+n}\big)\end{aligned}\ ] ] which relates riemann initial data @xmath46 to boundary conditions @xmath47 , such that the following hold . 1 . the solution of the riemann problem restricted on each link @xmath48 is given by the solution of the initial - boundary value problem with initial condition @xmath49 and boundary condition @xmath50 , @xmath42 . the rankine - hugoniot condition ( flow conservation ) holds : @xmath51 3 . the _ consistency condition _ holds : @xmath52\big]~=~rs[\hat\rho]\ ] ] three conditions must be satisfied by the riemann solver ( rs ) . item @xmath53 above requires that the boundary condition on each link must be properly given so that the initial value problems not only have well - defined solutions , but these solutions must also be compatible and form a sensible solution at the junction . simply stipulates the conservation of flow across the junction . is a desirable property and is sometimes referred to as the _ invariance property _ @xcite . for the same riemann problem , there exist many riemann solvers that satisfy conditions ( i)-(iii ) above . despite their varying forms , most existing riemann solvers rely on a flow maximization problem at the relevant junction subject to constraints related to turning ratio , right - of - way , or signal controls ; see @xcite and @xcite . for each link @xmath34 , we let @xmath54 be the critical density at which the flow is maximized . the demand @xmath55 for incoming links @xmath34 and the supply @xmath56 for outgoing links @xmath57 are defined in terms of the density near the exit and entrance of the link , respectively @xcite : @xmath58 in prose , the demand represents the maximum flow at which cars can be discharged from the incoming link ; and the supply represents the maximum flow at which cars can enter the outgoing link . notice that the demand and supply are both expressed as functions of density , and they are always greater than or equal to the fundamental diagram @xmath59 or @xmath60 ; see figure [ figds ] for an illustration . in our subsequent presentation , without causing confusion we will use notations @xmath55 and @xmath61 interchangeably where the former indicates the demand as a time - varying quantity , and the latter emphasizes demand as a function of density . the same applies to the supply . the aim of this section is to formulate the lwr - based dynamic network loading ( dnl ) problem as a system of partial differential algebraic equations ( pdaes ) . the proposed pdae system uses vehicle density and queues as the primary unknown variables , and computes link dynamics , flow propagation , and path delay for any given vector of departure rates . the pdae system captures vehicle spillback explicitly , and accommodates a wide range of junction types and riemann solvers . we consider a network @xmath62 expressed as a directed graph with @xmath63 being the set of links and @xmath64 being the set of nodes . let @xmath2 be the set of paths employed by travelers , and @xmath65 be the set of origin - destination pairs . each path @xmath3 is expressed as an ordered set of links it traverses : @xmath66 where @xmath67 is the number of links in this path . there are several crucial components of a complete network loading model , each of which is elaborated in a subsection below . throughout the rest of this paper , for each node @xmath68 , we denote by @xmath69 the set of incoming links and @xmath70 the set of outgoing links . for each @xmath71 , the density dynamic is governed by the scalar conservation law @xmath72~=~0\qquad ( t,\,x)\in[0,\,t]\times [ a_i,\,b_i]\ ] ] subject to initial condition and boundary conditions to be determined in section [ subsecpdaebdycond ] . the fundamental diagram @xmath73 satisfies condition * ( f ) * stated at the beginning of section [ secpreliminaries ] . in order to explicitly incorporate drivers route choices , for every @xmath3 such that @xmath74 we introduce the function @xmath75 , @xmath76\times [ a_i,\,b_i]$ ] , which represents , in every unit of flow @xmath77 , the fraction associated with path @xmath14 . we call these variables _ path disaggregation variables _ ( pdv ) . for each car moving along the link @xmath34 , its surrounding traffic can be distinguished by path ( e.g. 20% following path @xmath78 , 30% following path @xmath79 , 50% following path @xmath80 ) . as this car moves , such a composition will not change since its surrounding traffic all move at the same speed under the first - in - first - out ( fifo ) principle ( i.e. no overtaking is allowed ) . in mathematical terms , this means that the path disaggregation variables , @xmath81 , are constants along the trajectories of cars @xmath82 in the space - time diagram , where @xmath83 is the trajectory of a moving car on link @xmath34 . that is , @xmath84 which , according to the chain rule , becomes @xmath85 which further leads to another set of partial differential equations on link @xmath34 : @xmath86 here , @xmath87 is the solution of . the following obvious identity holds @xmath88 where @xmath89 means path @xmath14 contains ( or traverses ) link @xmath34 " , and the summation appearing in is with respect to all such @xmath14 . by convention , if @xmath90 , then @xmath91 for all @xmath89 . for reason that will become clear later , we introduce the concept of an _ ordinary node_. an ordinary node is neither the origin nor the destination of any trip . we use the notation @xmath92 to represent the set of ordinary nodes in the network . as mentioned earlier , the partial differential equations on links incident to @xmath28 are all coupled together through a given junction model , i.e. , a riemann solver . a common prerequisite for applying the riemann solver is the determination of the flow distribution ( turning ratio ) matrix @xcite , which relies on knowledge of the pdvs @xmath93 for all @xmath94 . we define the time - dependent flow distribution matrix associated with @xmath28 : @xmath95^{|\mathcal{i}^j| + |\mathcal{o}^j|}\ ] ] where by convention , we use subscript @xmath37 to indicate incoming links , and @xmath96 to indicate outgoing links . each element @xmath97 represents the turning ratios of cars discharged from @xmath34 that enter downstream link @xmath57 . then , for all @xmath14 that traverses @xmath28 , the following holds . @xmath98 it can be easily verified that @xmath99 $ ] and @xmath100 according to . we are now ready to express the boundary conditions for the ordinary junction @xmath101 . let @xmath102 \rightarrow \prod_{k=1}^{|\mathcal{i}^j|+|\mathcal{o}^j|}[0,\,\rho^{jam}_k]\ ] ] be a given riemann solver . notice that the dependence of the riemann solver on @xmath103 has been indicated with a superscript . the boundary conditions for pdes read @xmath104\qquad \forall i_k \in \mathcal{i}^j \\ \label{pdaersdef2 } \rho_{l}(t,\,a_l)&~=~rs^{a^j}_l\left[\big(\rho_{i}(t,\ , b_i-)\big)_{i_i\in\mathcal{i}^j}~,~\big(\rho_{j}(t,\,a_j+)\big)_{i_j\in\mathcal{o}^j}\right]\qquad \forall i_l\in \mathcal{o}^j\end{aligned}\ ] ] where @xmath105 $ ] denotes the @xmath106-th component of the mapping , @xmath107 . intuitively , - mean that , given the current traffic states @xmath108 and @xmath109 adjacent to the junction @xmath28 , the riemann solver @xmath110 specifies , for each incident link @xmath48 or @xmath111 , the corresponding boundary conditions @xmath112 or @xmath113 . in prose , at each time instance the riemann solver inspects the traffic conditions near the junction , and proposes the discharging ( receiving ) flows of its incoming ( outgoing ) links . such a process is based on the flow distribution matrix @xmath114 and often reflects traffic control measures at junctions . furthermore , the riemann solver operates with knowledge of every link incident to the junction , thus the boundary condition of any relevant link is determined jointly by all the links connected to the same junction . therefore , the lwr equations on all the links are coupled together through this mechanism . for this reason the lwr - based dnl models are highly challenging , both theoretically and computationally . the upstream boundary conditions associated with pdes are : @xmath115 where the numerator @xmath116 expresses the exit flow on link @xmath34 associated with path @xmath14 , which , by flow conservation , is equal to the entering flow of link @xmath57 associated with the same path @xmath14 ; the denominator represents the total entering flow of link @xmath57 . unlike the density - based pde , the pdv - based pde does not have any downstream boundary condition due to the fact that the traveling speeds of the pdvs are the same as the car speeds ( they can be interpreted as lagrangian labels that travel with the cars ) ; thus information regarding the pdvs does not propagate backwards or spills over to upstream links . we consider a node @xmath68 that is either the origin or the destination of some path @xmath14 . one immediate observation is that the flow conservation constraint no longer holds at such a node since vehicles either are ` generated ' ( if @xmath117 is an origin ) or ` vanish ' ( if @xmath117 is a destination ) . a simple and effective way to circumvent this issue is to introduce a _ virtual link_. a virtual link is an imaginary road with certain length and fundamental diagram , and serves as a buffer between an ordinary node and an origin / destination ; see figure [ chapdnl : figvirtuallink ] for an illustration . by introducing virtual links to the original network , we obtain an augmented network @xmath118 in which all road junctions are ordinary , and hence fall within the scope of the previous section . . right : a virtual link connecting a destination ( t ) to an ordinary junction @xmath28 . ] let us denote by @xmath119 the set of origins in the augmented network @xmath118 . for any @xmath120 , we denote by @xmath121 the set of paths that originate from @xmath122 , and by @xmath123 the virtual link incident to this origin . for each @xmath124 we denote by @xmath7 the departure rate ( path flow ) along @xmath14 . it is expected that a buffer ( point ) queue may form at @xmath122 in case the receiving capacity of the downstream @xmath123 is insufficient to accommodate all the departure rates @xmath125 . for this buffer queue , denoted @xmath126 , we employ a vickrey - type dynamic @xcite ; that is , @xmath127 where @xmath128 denotes the supply of the virtual link @xmath123 . the only difference between and vickrey s model is the time - varying downstream receiving capacity provided by the virtual link . it remains to determine the dynamics for the path disaggregation variables ( pdv ) . more precisely , we need to determine @xmath129 for the virtual link @xmath123 where @xmath124 , and @xmath130 is the upstream boundary of @xmath123 . this will be achieved using the vickrey - type dynamic and the fifo principle . specifically , we define the queue exit time function @xmath131 where @xmath10 denotes the time at which drivers depart and join the point queue , if any ; @xmath131 expresses the time at which the same group of drivers exit the point queue . clearly , fifo dictates that @xmath132 where the two integrands on the left and right hand sides of the equation are flow entering the queue and flow leaving the queue , respectively . we may determine the path disaggregation variables as : @xmath133 notice that , if @xmath134 , then the flow leaving the point queue at time @xmath131 is also zero ; thus there is no need to determine the path disaggregation variables . therefore , the identity is well defined and meaningful . with all preceding discussions , we may finally express the path travel times , which are the outputs of a complete dnl model . the path travel time consists of link travel times plus possible queuing time at the origin . mathematically , the link exit time function @xmath135 for any @xmath34 is defined , in a way similar to , as @xmath136 for a path expressed as @xmath137 , the time to traverse it is calculated as @xmath138 where @xmath139 means the composition of two functions . this is due to the assumption that cars leaving the previous link ( or queue ) immediately enter the next link without any delay . we are now ready to present a generic pdae system for the dynamic network loading procedure . let us begin by summarizing some key notations . * the original network with link set @xmath63 and node set @xmath64 ; * the set of virtual links ; * the augmented network including virtual links ; * the set of origins in @xmath140 ; * the set of paths originating from @xmath120 ; * the set of ordinary junctions in @xmath140 ; * the set of incoming links of a junction @xmath101 ; * the set of outgoing links of a junction @xmath101 ; * the flow distribution matrix associated with junction @xmath28 ; * the riemann solver for junction @xmath28 , which depends on @xmath103 . we also list some key variables of the pdae system below . * the path departure rate along @xmath3 ; * the vehicle density on link @xmath141 ; * the proportion of flow on link @xmath34 associated with path @xmath14 ( path disaggregation variable ) ; * the point queue at the origin @xmath120 ; * the point queue exit time function at origin @xmath120 . given any vector of path departure rates @xmath142 , the proposed pdae system for calculating path travel times is summarized as follows . @xmath143=0\qquad & ( t,\,x)\in[0,\,t]\times [ a_i,\,b_i ] \\ \label{chapdnl : lwrpdaeeqn3 } \partial_t\mu_i^p\big(t,\,x\big)+v_i\big(\rho_i(t,\,x)\big)\cdot \partial_x\mu_i^p\big(t,\,x\big)=0\qquad & ( t,\,x)\in[0,\,t]\times [ a_i,\,b_i ] \\ \label{chapdnl : lwrpdaeeqn65 } \mu_s^p\big(\lambda_s(t),\,a_s\big)~=~ { h_p(t)\over \sum_{q\in\mathcal{p}^s}h_q(t ) } \qquad & \forall s\in\mathcal{s},\ , p\in\mathcal{p}_s \\ \label{chapdnl : lwrpdaeeqn66 } \mu_{j}^{p}(t,\ , a_j)~=~{f_{i}\big(\rho_{i}(t,\ , b_i)\big)\cdot \mu_{i}^p(t,\ , b_i ) \over f_{j}\big(\rho_{j}(t,\,a_j)\big ) } \qquad & \forall p \supset \{i_i,\,i_j\ } \\ \label{chapdnl : lwrpdaeeqn4 } a^j(t)=\big\{\alpha^j_{ij}(t)\big\},\quad \alpha^j_{ij}(t)=\sum_{p\ni i_i,\,i_j } \mu_{i}^p(t,\,b_i ) \qquad & \forall i_i\in\mathcal{i}^j,\ , i_j\in\mathcal{o}^j \\ \label{chapdnl : lwrpdaeeqn5 } \rho_{k}(t,\,b_k)~=~rs^{a^j}_k\left[\big(\rho_{i}(t,\ , b_i-)\big)_{i_i\in\mathcal{i}^j}~,~\big(\rho_{j}(t,\,a_j+)\big)_{i_j\in\mathcal{o}^j}\right]\qquad & \forall i_k \in \mathcal{i}^j \\ \label{chapdnl : lwrpdaeeqn6 } \rho_{l}(t,\,a_l)~=~rs^{a^j}_l\left[\big(\rho_{i}(t,\ , b_i-)\big)_{i_i\in\mathcal{i}^j}~,~\big(\rho_{j}(t,\,a_j+)\big)_{i_j\in\mathcal{o}^j}\right]\qquad & \forall i_l\in \mathcal{o}^j \\ \label{chapdnl : lwrpdaeeqn7 } d_p(t,\,h)~=~\lambda_s \circ \lambda_1 \circ \lambda_2 \ldots \circ\lambda_{m(p ) } ( t ) \qquad & \forall p\in\mathcal{p},\quad \forall t\in[0,\,t]\end{aligned}\ ] ] eqn describes the ( potential ) queuing process at each origin . eqns and express the queue exit time function for a point queue , and the link exit time function for a link , respectively . eqns - express the link dynamics in terms of car density and pdv ; eqns - specifies the upstream boundary conditions for the pdv as these variables can only propagate forward in space . eqns - determine the boundary conditions at junctions . finally , eqn determines the path travel times . the above pdae system involves partial differential operators @xmath144 and @xmath145 . solving such a system requires solution techniques from the theory of numerical _ partial differential equations _ ( pde ) such as finite difference methods @xcite and finite element methods @xcite . in mathematical modeling , the term _ well - posedness _ refers to the property of having a unique solution , and the behavior of that solution hardly changes when there is a slight change in the initial / boundary conditions . examples of well - posed problems include the initial value problem for scalar conservation laws @xcite , and the initial value problem for the hamilton - jacobi equations @xcite . in the context of traffic network modeling , well - posedness is a desirable property of network performance models capable of supporting analyses and computations of dta models . it is also closely related to the continuity of the path delay operator , which is the main focus of this paper . this section investigates the well - posedness ( i.e. continuous dependence on the initial / boundary conditions ) of two specific junction models . these two junctions are depicted in figure [ figtwojunc ] , and the corresponding merge and diverge rules are proposed initially by @xcite in a discrete - time setting with fixed vehicle turning ratios and driving priority parameters . we first consider the diverge junction shown on the left part of figure [ figtwojunc ] , with one incoming link @xmath146 and two outgoing links @xmath147 and @xmath148 . the demand @xmath149 and the supplies , @xmath150 and @xmath151 , are defined by and respectively . the riemann solver for this junction relies on the following two conditions . * cars leaving @xmath146 advance to @xmath147 and @xmath148 according to some turning ratio which is determined by the pdv @xmath152 in the dnl model . * subject to ( a1 ) , the flow through the junction is maximized . in the original diverge model @xcite , the vehicle turning ratios , denoted @xmath153 and @xmath154 with obvious meaning of notations , are constants known _ a priori_. this is not the case in a dynamic network loading model since they are determined endogenously by drivers route choices , as expressed mathematically by eqn . the diverge junction model , described by ( a1 ) and ( a2 ) , can be more explicitly written as : @xmath155 where @xmath156 denotes the exit flow of link @xmath146 , and @xmath157 denotes the entering flow on link @xmath57 , @xmath158 . we now turn to the merge junction in figure [ figtwojunc ] , with two incoming links @xmath159 and @xmath160 and one outgoing link @xmath161 . in view of this merge junction , assumption ( a1 ) becomes irrelevant as there is only one outgoing link ; and assumption ( a2 ) can not determine a unique solution . to address this issue , we consider a _ right - of - way _ parameter @xmath162 and the following priority rule : + ( r1 ) the actual link exit flows satisfy @xmath163 . + ) are the optimal solution sets of . left : rule ( r1 ) is compatible with ( a2 ) ; and there exists a unique point @xmath164 satisfying both ( a2 ) and ( r1 ) . right : ( r1 ) is incompatible with ( a2 ) ; in this case , the model selects point @xmath164 within the set @xmath165 that is closest to the ray ( @xmath166 ) from the origin with slope @xmath167 . ] notice that ( r1 ) may be incompatible with assumption ( a2 ) ; and we refer the reader to figure [ figmerge ] for an illustration . whenever there is a conflict between ( r1 ) and ( a2 ) , we always respect ( a2 ) and relax ( r1 ) so that the solution is chosen to be the point that is closest to the line @xmath168 among all the points yielding the maximum flow . clearly , such a point is unique . mathematically , we let @xmath165 be the set of points @xmath169 that solves the following maximization problem : @xmath170 moreover , we define the ray @xmath171 . then the solution of the merge model is defined to be the projection of @xmath166 onto @xmath165 ; that is , @xmath172\ ] ] where @xmath173 $ ] denotes the euclidean distance between the point @xmath174 and the ray @xmath166 : @xmath175~=~\min_{(x,\,y)\in r}\left\| ( f^{out}_{4},\,f^{out}_{5 } ) - ( x,\,y)\right\|_2\ ] ] in this section , we investigate the well - posedness of the diverge junction model . unlike previous studies @xcite , a major challenge in this case is to incorporate drivers route choices , expressed by the _ path disaggregation variable _ @xmath176 , into the model and the analysis . in effect , we need to establish the continuous dependence of the model on the initial / boundary conditions in terms of both @xmath177 and @xmath176 . as we shall see in section [ subsubsecexample ] below , such a continuity does not hold in general . following this , section [ subsubsecsufficient ] provides sufficient conditions for the continuity to hold . these sufficient conditions are crucial for the desired continuity of the delay operator . it has been shown by @xcite that the diverge model ( section [ subsecdiverge ] ) with constant turning ratios , @xmath153 and @xmath154 , is well - posed . however , the assumption of fixed and exogenous turning ratios does not hold in dnl models ; see . as a result the well - posedness is no longer true , which is demonstrated by the following counterexample . we consider the diverge junction ( figure [ figtwojunc ] ) and assume the same fundamental diagram @xmath25 for all the links for simplicity . we consider a series of constant initial data parameterized by @xmath178 on the three links @xmath179 , and @xmath148 : @xmath180,\qquad\qquad \rho_3(0,\,x)~\equiv~\hat\rho_3^{\,\varepsilon}\in ( 0,\,\rho^c)\ ] ] where @xmath181 and @xmath23 are the critical density and the jam density , respectively . @xmath182 , @xmath183 , and @xmath184 satisfy : @xmath185 where @xmath186 is a parameter . this configuration of initial data implies that link @xmath146 and link @xmath147 are both in the congested phase , while link 3 is in the uncongested ( free - flow ) phase , see figure [ figcounterex ] for an illustration . two paths exist in this example : @xmath187 and @xmath188 . the initial conditions for @xmath189 and @xmath190 , which arise from travelers route choices , are as follows . @xmath191 notice that these initial conditions are part of the initial value problem at the junction . the solution of this initial value problem depends on the value of @xmath178 ; in particular , we have the following two cases . * when @xmath192 , we claim that the initial conditions @xmath193 and @xmath194 satisfying - constitute a constant solution at the junction . to see this , we follow the junction model and the definitions of demand and supply - to get @xmath195 where @xmath24 denotes the flow capacity . thus @xmath196 , @xmath197 and @xmath194 form a constant solution at the junction . * when @xmath198 , the turning ratios satisfy @xmath199 , @xmath200 . effectively , link @xmath146 is only connected to link @xmath148 . we easily deduce that @xmath201 as a result , the solution on link @xmath146 is given by a backward - propagating rarefaction wave ( or expansion wave , fan wave ) with @xmath196 and @xmath181 on the two sides . on link @xmath148 , a forward - propagating rarefaction wave with @xmath181 and @xmath202 ( that is , the limit of @xmath184 as @xmath203 ) on the two sides is created . link @xmath147 remains in a completely jam state with full density @xmath23 . see figure [ figsolution ] for an illustration of the solutions . , @xmath147 and @xmath148 for the two cases : @xmath192 and @xmath198 . the dashed lines represent the kinematic waves ( characteristics ) , and darker color indicates higher density . notice that jumps in the boundary conditions across these two cases exist on @xmath146 and @xmath148 . ] the two sets of solutions corresponding to @xmath192 and @xmath198 are shown in figure [ figsolution ] . we first notice that in these two cases , the boundary flows are very different , despite the infinitesimal difference in the pdv , namely from @xmath192 to @xmath198 . in particular , both @xmath156 and @xmath204 jump from @xmath205 and @xmath206 ( when @xmath192 ) to @xmath24 ( when @xmath198 ) . this is a clear indication of the discontinuous dependence of the diverge junction model on its initial conditions . let us zoom in on the mechanism that triggers such a discontinuity . according to , the expression for @xmath156 when @xmath192 is @xmath207 as long as @xmath178 is positive , the fraction @xmath208 . however , when @xmath198 we have an expression of ` @xmath209 ' , which should be equal to @xmath210 since link @xmath147 has effectively no influence on the junction , and we have @xmath211 . the above argument amounts to the following statement : @xmath212 which explains the jump in the solutions when @xmath178 tends to zero . the fact that @xmath213 tends to @xmath23 as @xmath214 plays a key role in this example . as we shall see later in theorem [ divwpthm ] , bounding @xmath213 away from the jam density is essential for the well - posedness . as we have previously demonstrated , the diverge model with time - varying vehicle turning ratios may not depend continuously on its initial conditions , which are defined in terms of the two - tuple @xmath215 . in this section , we propose additional conditions that guarantee the continuous dependence with respect to the initial data at the diverge junction . our analysis relies on the method of wave - front tracking @xcite and the technique of _ generalized tangent vectors _ @xcite . in order to be self - contained while keeping our presentation concise , we move some general background on these subjects and essential mathematical details to [ secappemb ] . * ( well - posedness of the diverge model)*[divwpthm ] consider the diverge junction ( figure [ figtwojunc ] ) and assume that 1 . there exists some @xmath216 such that the supplies @xmath217 , @xmath218 for all @xmath10 ; 2 . the path disaggregation variable @xmath219 has bounded total variation in @xmath10 . then the solution of the diverge junction depends continuously on the initial and boundary conditions in terms of @xmath177 and @xmath176 . the proof is long and technical , and thus will be presented in [ appsecthm1 ] following necessary mathematical preliminaries in [ secappemb ] . for the merge junction depicted in figure [ figtwojunc ] , the path disaggregation variables @xmath176 become irrelevant since there is only one downstream link . according to theorem 5.3 of @xcite , the solution at the merge junction , in terms of density @xmath177 , depends continuously on the initial and boundary conditions . moreover , according to , the propagation speed of @xmath176 is the same as the vehicle speed @xmath220 , we thus conclude that @xmath176 also depends continuously on the initial and boundary conditions . this shows the well - posedness of the merge junction . the proof of the continuity of the path delay operator is outlined as follows . we first show that the first hypothesis of theorem [ divwpthm ] holds for networks that consists of only the merge and diverge junctions discussed in section [ secmergediverge ] ; this will be done in section [ secsupplymin ] . then , we propose some mild conditions on the fundamental diagram and the departure rates in section [ secpdv ] to ensure that the second hypothesis of theorem [ divwpthm ] holds . it then follows that both the merge and diverge models are well posed . finally , in section [ sec : dep ] we show the well - posedness of the model in terms of @xmath177 and @xmath176 at each origin , where a point queue may be present . put altogether , these results lead to the desired continuity for the delay operator . this section provides a lower bound on the supply , which is a function of density , on any link in the network during the entire time horizon @xmath0 $ ] . our finding is that the jam density can never be reached within any finite time horizon @xmath221 , and the supply anywhere in the network is bounded away from zero . as a consequence , grid lock will never occur in the dynamic network . we denote by @xmath222 the set of destinations in the augmented network with virtual links ( see section [ subsecvirtuallink ] ) ; that is , every destination @xmath223 is incident to a virtual link that connects @xmath224 to the rest of the network ; see figure [ figflowintod ] . for each @xmath223 , we introduce its supply , denoted @xmath225 , to be the maximum rate at which cars can be discharged from the virtual link connected to @xmath224 . effectively , there exists a bottleneck between the virtual link and the destination ; and the supply of the destination is equal to the flow capacity of this bottleneck ; see figure [ figflowintod ] . notice that in some literature such a bottleneck is completely ignored and the destination is simply treated as a sink with infinite receiving capacity . this is of course a special case of ours once we set the supply @xmath225 to be infinity . however , such a supply may be finite and even quite limited under some circumstances due to , for example , ramp metering , limited parking spaces , or the fact that the destination is an aggregated subnetwork that is congested . . ] we introduce below a few more concepts and notations . @xmath226 : : : the minimum link length in the network , including virtual links . @xmath227 : : : the minimum link flow capacity in the network , including virtual links . @xmath228 : : : the maximum backward wave speed in the network . @xmath229 : : , where @xmath230 is the priority parameter for the merge junction @xmath28 ( see section [ subsecmergemodel ] ) , @xmath231 denotes the set of merge junctions in the network . @xmath232 : : : the minimum supply among all the destination nodes . @xmath233}\inf_{x\in[a_i,\,b_i]}s_i\left(\rho_i(t,\,x)\right)$ ] : : : the minimum supply at any location in the network during time interval @xmath234 , where @xmath235 [ thmdelta ] consider a network consisting of only merge and diverge junctions as shown in figure [ figtwojunc ] . given any vector of path departure rates , the dynamic network loading procedure described by the pdae system yields the following property in the solution : @xmath236 we postpone the proof until [ subsecappthmdelta ] for a more compact presentation . theorem [ thmdelta ] guarantees that the supply values anywhere in the network during the period @xmath237 $ ] are uniformly bounded below by some constant that depends only on @xmath106 , although such a constant may decay exponentially as @xmath106 increases . given that any dynamic network loading problem is conceived in a finite time horizon , we immediately obtain a lower bound on the supplies , as shown in the corollary below . [ cordelta ] under the same setup as theorem [ thmdelta ] , we have @xmath238}s_i\big(\rho_i(t,\,x)\big)~\geq~\bar p^{\lfloor { t\over l/\lambda}\rfloor } \left\{\delta^{\mathcal{d}}~,~\bar p c^{min}\right\}\qquad\forall t\in[0,\,t]\ ] ] and @xmath238}\inf_{t\in[0,\,t]}s_i\big(\rho_i(t,\,x)\big)~\geq~\bar p^{\lfloor { t\lambda\over l}\rfloor}\min\left\{\delta^{\mathcal{d}}~,~\bar p c^{min}\right\}\ ] ] where the operator @xmath239 rounds its argument to the nearest integer from below . both inequalities are immediately consequences of . corollary [ cordelta ] shows that , for any network consisting of the merge and diverge junctions , the supply values are uniformly bounded from below at any point in the spatial - temporal domain . in particular , a jam density can never occur . moreover , such a network is free of complete gridlock . is formed somewhere in the network and the static queues do not dissipate in finite time . ] in this section we establish some properties of the path disaggregation variable @xmath176 , which serve to validate the second hypothesis of theorem [ divwpthm ] . [ lemma : ben ] assume that there exists @xmath240 and @xmath241 so that the following hold : 1 . for all links @xmath34 , the fundamental diagrams @xmath59 are uniformly linear near the zero density ; more precisely , @xmath242 is constant for @xmath243 $ ] for all @xmath141 . each @xmath59 has non - vanishing derivative , i.e. @xmath244 $ ] . the departure rates @xmath245 are uniformly bounded , have bounded total variation ( tv ) , and bounded away from zero when they are non - zero ; i.e. @xmath246 and @xmath247 $ ] for all @xmath3 and almost every @xmath4 $ ] . then the path disaggregation variables @xmath248 are either zero , or uniformly bounded away from zero . moreover , they have bounded variation . the proof is moved to [ subsecapplemma : ben ] . notice that assumptions 1 and 2 of lemma [ lemma : ben ] are satisfied by the newell - daganzo ( triangular ) fundamental diagrams , where both the free - flow and congested branches of the fd are linear . moreover , given an arbitrary fundamental diagram , one can always make minimum modifications at @xmath26 and @xmath249 to comply with conditions 1 and 2 . assumption 3 of lemma [ lemma : ben ] is satisfied by any departure rate that is a result of a finite number of cars entering the network . and , again , any departure rate can be adjusted to satisfy this condition with minor modifications . in this section we discuss the continuous dependence of the queues @xmath126 and the solutions @xmath250 and @xmath251 with respect to the path departure rates @xmath7 , @xmath3 . the analysis below relies on knowledge of the wave - front tracking algorithm and generalized tangent vector , for which an introduction is provided in [ secappemb ] . following @xcite , we introduce a generalized tangent vector @xmath252 for the triplet @xmath253 where @xmath254 is a scalar shift of the queue @xmath255 , i.e. the shifted queue is @xmath256 , while the tangent vectors of @xmath250 and @xmath251 are defined in the same way as in [ secgtv ] . the tangent vector norm of @xmath257 is simply its absolute value @xmath258 . [ lemmabenqueue ] assume that the departure rates @xmath259 are piecewise constant , and let @xmath260 be a tangent vector defined via shifting the jumps in @xmath261 . then the tangent vector @xmath252 is well defined , and its norm is equal to that of @xmath260 and bounded for all times . the proof is postponed until [ subsecapplemmabenqueue ] . at the end of this paper , we are able to prove the continuity result for the delay operator based on a series of preliminary results presented so far . * ( continuity of the delay operator ) * consider a network consisting of merge and diverge junctions described in section [ secmergediverge ] , under the same assumptions stated in lemma [ lemma : ben ] , the path delay operator , as a result of the dynamic network loading model - , is continuous . we have shown that at each node ( origin , diverge node , or merge node ) , the solution depends continuously on the initial and boundary values . in addition , between any two distinct nodes , the propagation speeds of either @xmath177-waves or @xmath176-waves are uniformly bounded . thus such well - posedness continues to hold on the network level . consequently , the vehicle travel speed @xmath262 for any @xmath34 depends also continuously on the departure rates . we thus conclude that the path travel times depend continuously on the departure rates . the assumption that the network consists of only merge and diverge nodes is not restrictive since junctions with general topology can be decomposed into a set of elementary junctions of the merge and diverge type @xcite . in addition , junctions that are also origins / destinations can be treated in a similar way by introducing virtual links . here , we would like to comment on the assumptions made in the proof of the continuity . in a recent paper @xcite , a counterexample of uniqueness and continuous dependence of solutions is provided under certain conditions . more precisely , the authors construct the counterexample by assuming that the path disaggregation variables @xmath251 have infinite total variation . this shows the necessity of the second assumption of theorem [ divwpthm ] . moreover , again in @xcite , the authors prove that for density oscillating near zero the solution may not be unique ( even in the case of bounded total variation ) , which shows the necessity of the third assumption in lemma [ lemma : ben ] . @xcite provides an example of discontinuous dependence of the path travel times on the path departure rates using the cell transmission model representation of a signal - controlled network . in particular , the author showed that when a queue generated by the red signal spills back into the upstream junction , the experienced path travel time jumps from one value to another . this , however , does not contradict our result presented here for the following reason : the jam density caused by the red signal in @xcite does not exist in our network , which has only merge and diverge junctions ( without any signal controls ) . indeed , as shown in corollary [ cordelta ] , the supply functions at any location in the network are uniformly bounded below by a positive constant , and thus the jam density never occurs in a finite time horizon . the reader is reminded of the example presented in section [ subsubsecexample ] , where the ill - posedness of the diverge model is caused precisely by the presence of a jam density . the counterexample from @xcite is constructed essentially in the same way as our example , by using signal controls that create the jam density . we offer some further insights here on the extension of the continuity result to second - order traffic flow models . to the best of our knowledge , second - order models on traffic networks are mainly based on the aw - rascle - zhang model @xcite and the phase - transition model @xcite . solutions of the aw - rascle - zhang system may present the vacuum state , which , in general , may prevent continuous dependence @xcite . on the other hand , the phase transition model on networks behaves in a way similar to the lwr scalar conservation law model . however , continuous dependence may be violated for density close to maximal ; see @xcite . therefore , extensions of the continuity result to second - order models appear possible only for the phase - transition type models with appropriate assumptions on the maximal density achievable on the network , but this is beyond the scope of this paper . this paper presents , for the first time , a rigorous continuity result for the path delay operator based on the lwr network model with spillback explicitly captured . this continuity result is crucial to many dynamic traffic assignment models in terms of solution existence and computation . similar continuity results have been established in a number of studies , all of which are concerned with non - physical queue models . as we show in section [ subsubsecexample ] , the well - posedness of a diverge model may not hold when spillback occurs . this observation , along with others made in previous studies @xcite , have been the major source of difficulty in showing continuity of the delay operator . in this paper , we bridge this gap through rigorous mathematical analysis involving the wave front tracking method and the generalized tangent vectors . in particular , by virtue of the finite propagation speeds of @xmath177-waves and @xmath176-waves , the continuity of the delay operator boils down to the well - posedness of nodal models , including models for the origins , diverge nodes , and merge nodes . minor assumptions are made on the fundamental diagram and the path departure rates in order to provide an upper bound on the total variations of the density @xmath177 and the path disaggregation variables @xmath176 , which subsequently leads to the desired well - posedness of the nodal models and eventually the continuity of the operator . a crucial step of the above process is to estimate and bound from below the minimum network supply , which is defined in terms of local vehicle densities . in fact , if the supply of some link tends to zero , the well - posedness of the diverge junction may fail as we demonstrate in section [ subsubsecexample ] . this has also been confirmed by an earlier study @xcite , where a wave of jam density is triggered by a signal red light and causes spillback at the upstream junction , leading to a jump in the path travel times . remarkably , in this paper we are able to show that ( 1 ) if the supply is bounded away from zero , then the diverge junction model is well posed ; and ( 2 ) the desired boundedness of the supply is a natural consequence of the dynamic network loading procedure that involves only the simple merge and diverge junction models . this is a highly non trivial result because it not only plays a role in the continuity proof , but also implies that gridlock can never occur in the network loading procedure in a finite time horizon . however , we note that in numerical computations gridlock may very well occur due to finite approximations and numerical errors , while our no - gridlock result is conceived in a continuous - time and analytical ( i.e. , non - numerical ) framework . it is true that , despite the correctness of our result regarding the lower bound on the supply , zero supply ( or gridlock ) can indeed occur in real - life traffic networks as a result of signal controls . in fact , this is how @xcite constructs the counter example of discontinuity . nevertheless , signal control is an undesirable feature of the dynamic network loading model for far more obvious reasons : the on - and - off signal control creates a lot of jump continuities in the travel time functions , making the existence of dynamic user equilibria ( due ) almost impossible . given that the main purpose of this paper is to address fundamental modeling problems pertinent to dues , it is quite reasonable for us to avoid junction models that explicitly involve the on - and - off signal controls . nevertheless , there exist a number of ways in which the on - and - off signal control can be approximated using continuum ( homogenization ) approaches . some examples are given in @xcite and @xcite . regarding the exponential decay of the minimum supply , we point out that , as long as the jam density ( gridlock ) does not occur , our result is still applicable . from a practical point of view most networks reach congestion during day time but are almost empty during night time . in other words , the asymptotic gridlock condition is never reached in reality . therefore , we think that our result is a reasonable approximation of reality for due problems . we would also like to comment on the continuity result for other types of junction models . prove that lipschitz continuous dependence on the initial conditions may fail in the presence of junctions with at least two incoming and two outgoing roads . notably , the counterexample that they use to disprove the lipschitz continuity is valid even when the turning ratio matrix @xmath103 is constant and not dependent on the path disaggregation variables @xmath251 . it is interesting to note that , in the same book , continuity results for general networks are provided , but only for riemann solvers at junctions that allow re - directing traveling entities a feature not allowed in the dynamic network loading model but instead is intended for the modeling of data networks . the findings made in this paper has the following important impacts on dta modeling and computation . ( 1 ) the analytical framework proposed for proving or disproving the well - posedness of junction models and the continuity property can be applied to other junction types and riemann solvers , leading to a set of continuity / discontinuity results of the delay operators for a variety of networks . ( 2 ) the established continuity result not only guarantees the existence and computation of continuous - time dues based on the lwr model , but also sheds light on discrete - time due models and provides useful insights on the numerical computations of dues . in particular , the various findings made in this paper provide valuable guidance on numerical procedures for the dnl problem based on the cell transmission model @xcite , the link transmission model @xcite , or the kinematic wave model @xcite , such that the resulting delay operator is continuous on finite - dimensional spaces . as a result , the existence and computation of discrete - time dues will also benefit from this research . the wave - front tracking ( wft ) method was originally proposed by @xcite as an approximation scheme for the following initial value problem @xmath263 where the initial condition @xmath264 is assumed to have _ bounded variation _ ( bv ) @xcite . the wft approximates the initial condition @xmath264 using _ piecewise constant _ ( pwc ) functions , and approximates @xmath25 using _ piecewise affine _ ( pwa ) functions . it is an event - based algorithm that resolves a series of wave interactions , each expressed as a riemann problem ( rp ) . the wft method is primarily used for showing existence of weak solutions of conservation laws by successive refinement of the initial data and the fundamental diagram ; see @xcite and @xcite . @xcite extend the wft to treat the network case and show the existence of the weak solution on a network . we provide a brief description of this procedure below . fix a riemann solver ( rs ) for each road junction in the network . consider a family of piecewise constant approximations @xmath265 of the initial condition @xmath266 on each link and a family of piecewise affine approximation of the fundamental diagrams @xmath267 , where @xmath178 is a parameter such that @xmath268 and @xmath269 as @xmath214 . a wft approximate solution on the network is constructed as follows . 1 . within each link , solve a riemann problem at each discontinuity of the pwc initial data . at each junction , solve a riemann problem with the given rs . 2 . construct the solution by gluing together the solutions of individual rps up to the first time when two traveling waves interact , or when a wave interacts with a junction . 3 . for each interaction , solve a new rp and prolong the solution up to the next time of any interaction . 4 . repeat the processes 2 - 3 . to show that the procedure described above indeed produces a well - defined approximate solution on the network , one needs to ensure that the following three quantities are bounded : ( 1 ) the total number of waves ; ( 2 ) the total number of interactions ( including wave - wave and wave - junction interactions ) ; and ( 3 ) the _ total variation _ ( tv ) of the piecewise constant solution at any point in time . these quantities are known to be bounded in the single conservation law case ; in fact , they all decrease in time @xcite . however , for the network case , one needs to proceed carefully in estimating these quantities as they may increase as a result of a wave interacting with a junction , which may produce new waves in all other links incident to the same junction . the reader is referred to @xcite for more elaborated discussion on these interactions . for a sequence of approximate wft solutions @xmath270 , @xmath192 , if one can show that the total variation of @xmath270 is uniformly bounded , then as @xmath214 a weak entropy solution on the network is obtained . the generalized tangent vector is a technique proposed by @xcite to show the well - posedness of conservation laws , and is used later by @xcite to show the well - posedness of junction models in connection with the lwr model . its mathematical contents are briefly recapped here . given a piecewise constant function @xmath271\to \mathbb{r}$ ] , a _ tangent vector _ is defined in terms of the shifts of the discontinuities of @xmath272 . more precisely , let us indicate by @xmath273 the discontinuities of @xmath274 where @xmath275 , and by @xmath276 the values of @xmath274 on @xmath277 . a tangent vector of @xmath272 is a vector @xmath278 such that for each @xmath241 , one may define the corresponding perturbation of @xmath272 , denoted by @xmath279 and given by @xmath280 for @xmath281 , where we set @xmath282 . the norm of the tangent vector @xmath283 is defined as @xmath284 in other words , the norm of the tangent vector is the sum of each @xmath285 multiplied by the magnitude of the shifted jumps . the procedure of showing the well - posedness of a junction model using the tangent vectors is as follows . given piecewise constant initial / boundary conditions on each link incident to the junction , one considers their tangent vectors . by showing that the norms of their tangent vectors are uniformly bounded in time among all approximate wave - front tracking solutions , it is guaranteed that the @xmath286 distance of any two solutions is bounded , up to a multiplicative constant , by the @xmath286 distance of their respective initial / boundary conditions . more precisely , we have the following theorem if the norm of a tangent vector at any time is bounded by the product of its initial norm and a positive constant , among all approximate wave - front tracking solutions , then the junction model is well posed and has a unique solution . see @xcite . throughout this paper , we employ the notation @xmath287 to represent a wave interacting with the junction from road @xmath34 , where @xmath288 is the density value in front of the wave ( in the same direction as the traveling wave ) and @xmath250 is the density behind the wave . after the interaction , a new wave may be created on some road @xmath57 ( it is possible that @xmath289 ) , and we use @xmath290 and @xmath291 to denote the density at @xmath28 on road @xmath57 before and after the interaction , respectively . [ lemma : tangent - vectors - diverge ] consider the diverge model in section [ subsecdiverge ] . if a wave @xmath287 on @xmath34 interacts with @xmath28 then the shift @xmath292 produced on any @xmath57 , as a result of the shift @xmath293 on @xmath34 , satisfies @xmath294 where @xmath295 are the jumps in flow across these two waves . to fix the ideas , we assume that @xmath34 is the incoming road @xmath146 ( the other cases can be treated in a similar way ) . let @xmath296 be the interacting wave , then it must be that @xmath297 . if no wave is reflected on road @xmath146 after the interaction ( i.e. no new backward wave @xmath298 is created on @xmath146 as a result of the interaction ) , then we can conclude the lemma by following the same estimate as shown in @xcite . if , on the other hand , a wave @xmath299 is reflected then @xmath300 , @xmath296 is a rarefaction wave , and @xmath301 . in other words , part of the change in the flow caused by the interaction passes through the junction , and part of it is reflected back onto @xmath146 . we may then apply the same argument as in @xcite to the part that passes through the junction ( @xmath302 ) and conclude the lemma . if a @xmath177-wave from @xmath34 interacts with the junction , producing a @xmath177-wave on link @xmath57 , we call @xmath34 and @xmath57 the source and recipient of this interaction , respectively . such an event is denoted @xmath303 . we observe that @xmath304 is precisely the @xmath286-distance of two initial conditions on @xmath34 with and without the shift @xmath293 at the discontinuity @xmath287 . similarly , @xmath305 is the @xmath286-distance of the two solutions on @xmath57 as a result of having or not having the initial shift @xmath293 on @xmath34 , respectively . furthermore , if @xmath293 is the only shift in the initial condition on @xmath34 , then @xmath306 is the norm of the tangent vector for @xmath34 ( see definition ) before the interaction , and @xmath307 is the norm of the tangent vector for @xmath57 after the interaction . from , we have @xmath308 therefore , to bound the norm of the tangent vectors one has to check that the multiplication factors @xmath309 remain uniformly bounded , regardless of the number of interactions that may occur at this junction . let us consider the diverge junction , and assume that a wave interacts with @xmath28 from road @xmath34 and produces a wave on road @xmath57 , where @xmath310 . as in the proof of lemma [ lemma : tangent - vectors - diverge ] , we can restrict to the flow passing through the junction ( indeed the reflected part has the same flow variation and smaller shift since the reflected wave is a slow big shock ) . according to , @xmath311 and @xmath312 . consequently , we have the following matrix of multiplication factors @xmath313 : @xmath314 we denote this matrix by @xmath315 . according to @xcite , in order to estimate the tangent vector norm , it suffices to keep tract of just one single shift and show the corresponding tangent vector norm is bounded regardless of the number of wave interactions . to this end , we consider the only meaningful sequence of wave interaction , which is of the form @xmath303 , @xmath316 , . we observe that , if @xmath153 and @xmath154 are nonzero constants , then @xmath317 for any @xmath318 . this means that no matter how many interactions occur , the multiplication factor is always an element of the matrix @xmath319 and is uniformly bounded . thus the diverge model with fixed turning ratios is well posed . however , in the dnl model where @xmath153 and @xmath154 are time - varying , this is no longer true as we have shown in the counterexample in section [ subsubsecexample ] . the well - posedness requires some additional conditions to hold , and the corresponding proof needs more elaborated arguments that take into account the @xmath176-waves . these will be done in theorem [ divwpthm ] . this proof is completed in several steps . + * [ step 1 ] . * we invoke the wave - front tracking framework and the generalized tangent vector to show the desired continuous dependence . as discussed in [ secgtv ] , it boils down to showing that the increase in the tangent vector norm , as a result of arbitrary number of wave interactions ( including both @xmath177-wave and @xmath176-wave ) , remains bounded . + * [ step 2 ] . * first of all , the end of [ secgtv ] shows that , for constant values of @xmath176 ( that is , with fixed vehicle turning ratios ) , the tangent vector norm is uniformly bounded regardless of the interactions between the @xmath177-waves and the junction . next , one needs to consider the case where @xmath176 changes from one value to another , i.e. when a @xmath176-wave interacts with the junction . in general , we consider two consecutive interaction times of the @xmath176-waves : @xmath320 and @xmath321 , and let @xmath322 be the multiplication factor for the tangent vector norm of @xmath177 relative to the times @xmath320 and @xmath321 . more precisely , we have @xmath323 , where @xmath324 is the tangent vector norm of @xmath177 at time @xmath10 . clearly , @xmath322 can only take value in the following matrix where @xmath325 and @xmath326 are given by the constant @xmath176 value during @xmath327 ; see . @xmath328,\qquad k=0,\,1,\,2,\,\ldots\ ] ] similar arguments can be carried out to all other intervals @xmath329 $ ] , @xmath330 $ ] and so forth . in the rest of the proof , our goal is to estimate the multiplication factor for the tangent norm of @xmath177 under the wft framework , assuming arbitrary interaction patterns of the @xmath177-waves and @xmath176-waves . + * [ step 3 ] . * we define @xmath331 where @xmath332 is stated in the hypothesis of this theorem and @xmath333 is the flow capacity of link @xmath146 . without loss of generality , we let @xmath334 . throughout * step 3 * , we assume that @xmath335 and @xmath336 ( the other cases will be treated in * step 4 * ) . as demonstrated at the end of [ secgtv ] , in order to estimate the increase in the tangent vector norms it suffices to consider the following type of sequence in which @xmath177-waves interacts with the junction : @xmath303 then @xmath316 ; in other words , the recipient of the previous @xmath177-wave interaction is the source of the next @xmath177-wave interaction . for each @xmath337 , we let @xmath338 be the recipient of the last interacting @xmath177-wave during @xmath327 . this gives rise to the sequence @xmath339 ; see figure [ figikfig ] for an illustration . s represent the times when @xmath176-waves interact with the junction , changing the turning ratios @xmath153 and @xmath154 . the crosses represent the times at which @xmath177-waves interact with the junction . the circles indicate the last @xmath177-wave interaction within the interval during which @xmath176 is constant . @xmath340 s denote the recipients of the last interacting @xmath177-waves . ] we make the following crucial observation . consider any three elements in the sequence of the form @xmath341 , @xmath342 and @xmath343 where @xmath344 . by definition , the product of the multiplication factors within @xmath345 is @xmath346 where we use the superscripts to indicate dependence on a specific time interval . the significance of is that the multiplication factor can be decomposed into a term @xmath347 with a very specific structure ( to be further elaborated below ) , and a term that would have been the multiplication factor as if the middle link @xmath348 were removed . clearly , the argument above applies equally to the case where @xmath349 , and we use @xmath350 to represent the term factored out if the middle link @xmath349 is removed . by repeating this procedure , one may eliminate all links @xmath147 and @xmath148 from the sequence @xmath351 except when they are the first or possibly the last in this sequence . as a result , the multiplicative terms @xmath352 and @xmath353 are factored out . therefore , the entire multiplication factor for the tangent norm is bounded from above by @xmath354 where @xmath355 and @xmath356 are subsequences of @xmath357 such that @xmath358 and @xmath359 . the first two terms in are results of eliminating @xmath147 and @xmath148 from the sequence @xmath351 ; the third term takes care of the first element and possibly the last element in the sequence , and is derived from the observation that @xmath360 for all @xmath344 and @xmath361 . it remains to show that the first and second terms of are bounded , and hence the entire is bounded . we write , without referring explicitly to the multiplication indices , that @xmath362 this leads to the following . @xmath363}{1\over |\alpha| } \cdot \sum_k \big|\alpha_{1,2}^{k+1}-\alpha_{1,2}^{k+2}\big| \right ) } \cdot \exp{\left ( \sup_{\alpha\in[\gamma,\,1-\gamma ] } { 1\over |1-\alpha| } \cdot \sum_k \big|\alpha_{1,2}^{k+2}-\alpha_{1,2}^{k+1}\big| \right ) } \\ ~\leq~ & \exp{\left({1\over \gamma } \cdot tv(\mu ) \right)}\cdot \exp{\left ( { 1\over \gamma}\cdot tv(\mu ) \right ) } ~<~\infty\end{aligned}\ ] ] where @xmath364 is the bounded total variation of the path disaggregation variable @xmath176 . + * [ step 4 ] . * in this part of the proof we deal with the situation where @xmath365 ; the other case where @xmath366 is entirely similar . we begin with the following observation . @xmath367 since @xmath368 . in other words , the minimum is never attained at @xmath369 , and any @xmath177-wave interacting from @xmath147 will not generate new waves on @xmath146 or @xmath148 since @xmath370 does not change before and after the interaction . so the only interactions that may change the tangent vector norm are @xmath371 and @xmath372 , and the corresponding multiplication factors are respectively @xmath373 and @xmath374 . consider any sequence of interaction times @xmath375 of the @xmath176-waves . since the interactions @xmath376 or @xmath377 do not generate any increase in the tangent norm , we only need to consider the following sequence of interactions : @xmath378 . the resulting multiplication factor is bounded by constant of the form @xmath379 where the second multiplicative term deals with the first and the last element in the sequence since @xmath380 for all @xmath106 . we may then proceed in the same way as * step 3 * and conclude that this multiplication factor is bounded , provided that @xmath381 . since @xmath382 is defined in terms of the supplies , it suffices for us to focus only on those densities such that @xmath383 for @xmath141 , @xmath76\times[a_i,\,b_i]$ ] . it is also useful to keep in mind that densities beyond the critical density always propagate backwards in space . the proof is divided into several steps . + * [ step 1 ] . * @xmath384 . we have @xmath385 . since the network is initially empty , all the supplies @xmath386 , @xmath387 $ ] , are maximal and equal to the respective flow capacities at @xmath388 . afterwards , a higher - than - critical density or a lower - than - maximum supply can only emerge from the downstream end of a link and propagate backwards along this link . moreover , these backward waves can never reach the upstream end of the link within @xmath389 since @xmath390 is the minimum link traversal time for backward waves . * * case ( 1 ) . * we use the same notation shown in figure [ figtwojunc ] . according to the reason provided above , @xmath391 , @xmath392 for @xmath385 . let the time of the wave interaction be @xmath393 . recall from that @xmath394 if the minimum is attained at @xmath395 , then the entrance of @xmath146 will remain in the uncongested phase , i.e. @xmath396 after the interaction . hence , no lower - than - maximum supply is generated . on the other hand , if say @xmath397 is the smallest among the three , then @xmath398 in summary , the supply values corresponding to the backward waves generated at @xmath146 , if any , are uniformly above @xmath399 . * * case ( 2 ) . * we turn to the merge junction shown in figure [ figtwojunc ] , and note @xmath400 for @xmath385 . as usual , we let @xmath393 be the time of interaction . there are two further cases for the merge junction model as shown in figure [ figmerge ] . we first consider the situation illustrated in figure [ figmerge](a ) . clearly , we have that @xmath401 that is , the supply values of the backward waves generated at the downstream ends of @xmath159 and @xmath160 are uniformly above @xmath402 . + for the situation depicted in figure [ figmerge](b ) , we first note that the coordinates of @xmath403 is @xmath404 , and the coordinates of the solution @xmath405 either satisfy @xmath406 as shown exactly in figure [ figmerge](b ) , or @xmath407 taking as an example ( the case can be treated similarly ) , we have @xmath408 and no increase in density is produced at the downstream end of link @xmath159 since its exit flow is equal to the demand . + to sum up , the supply values corresponding to the backward waves generated at @xmath159 or @xmath160 , if any , are uniformly above @xmath402 . * * case ( 3 ) . * for each destination @xmath223 , let @xmath225 be its supply and denote by @xmath409 the virtual link connected to @xmath224 . if @xmath410 then no backward waves can be generated on this link and the supply value on the link is always equal to @xmath411 . if @xmath412 , then only one higher - than - critical density can exist on this virtual link , that is , @xmath177 such that @xmath413 and @xmath414 . clearly , its supply value is equal to @xmath225 . + to sum up , the supply value corresponding to the backward waves generated at any virtual link , if any , is bounded below by @xmath415 . finally , we notice that all the backward waves exhaustively described above originate from the downstream end of a link , and they can not reach the upstream end of the same link within period @xmath389 . thus , these backward waves can not bring further reduction in the supply values by means of interacting with the junctions . we conclude that during period @xmath416 , the minimum supply in the network , @xmath417 , is bounded below ; that is , @xmath418 * [ step 2 ] . * we move on to @xmath337 . in addition to case ( 1)-(3 ) , which do not bring any supply values below @xmath419 , two more cases may arise in which higher - than - critical densities may be generated as a result of backward waves interacting with junctions : * * case ( 4 ) . * without loss of generality , we assume that the backward wave that interacts with the diverge junction is coming from @xmath147 , and has the density value @xmath420 $ ] . let @xmath393 be the time of the interaction . in view of , if the minimum is attained at @xmath395 , then the interaction does not bring any increase in density at the downstream end of @xmath146 , hence no decrease in the supply there . + if the minimum is attained at @xmath421 , we deduce in a similar way as that @xmath422 the last inequality is due to the fact that a backward wave such as @xmath423 must be created at the downstream end of @xmath147 at a time earlier than @xmath424 , thus its supply value @xmath425 must be bounded below by @xmath426 . if the minimum is attained at @xmath427 , we have @xmath428 the last inequality is because any lower - than - maximum supply on @xmath148 must be created in the previous time interval . + to sum up , the supply values corresponding to the backward waves generated at @xmath146 , if any , are bounded below by @xmath426 . * * case ( 5 ) . * for the merge junction , we begin with the case illustrated in figure [ figmerge](a ) . assuming the backward wave that interacts with the merge junction from @xmath161 has the density value @xmath429 $ ] . similar to - , we have @xmath430 where the last inequalities are due to the same reason provided in * case ( 4)*. the case illustrated in figure [ figmerge](b ) can be treated in the same way as - . + to sum up , the supply values corresponding to the backward waves generated at @xmath159 or @xmath160 , if any , must be bounded below by @xmath431 . so far , we have shown that for @xmath432 where @xmath337 , the presence of higher - than - critical densities , as exhaustively illustrated through * case ( 1)- case ( 5 ) * , brings supply values throughout the entire network that are bounded below by @xmath433 the proof is divided into a few steps . + * [ step 1 ] . * notice that assumption 3 implies that the departure rates @xmath259 are non - zero on a finite set of intervals . indeed , let @xmath30 be the number of intervals where @xmath261 does not vanish then @xmath441 , which implies that @xmath30 is finite . + * [ step 2 ] . * using the assumption that each fundamental diagram @xmath59 has nonvanishing derivative , we have , at each origin , that the bounded total variation of the density must imply bounded total variation of the flow , and vise versa . therefore , the assumption on the bounded variation of @xmath261 implies that the density and path disaggregation variable are of bounded variation on the virtual link incident to that origin . + * [ step 3 ] . * assumption 1 guarantees that , on each link , all waves of the type @xmath442 or @xmath443 are contact discontinuities for @xmath177 sufficiently small and , in particular , travel with a constant speed . moreover , all @xmath176-waves travel with the same speed for low densities . therefore , taking into account * step 1 * , whenever @xmath176 is non - zero , it is uniformly bounded away from zero on the entire network . in particular there exists @xmath444 so that @xmath445 $ ] for every @xmath446 and @xmath14 . therefore , at diverging junctions , the coefficients @xmath447 and @xmath373 satisfy the same properties . + * [ step 4 ] . * let us now turn to the total variation of the path disaggregation variables . we know , from * step 2 * , that @xmath176 has bounded variation on virtual links incident to the origins . we also know from * step 3 * that @xmath176 is bounded away from zero whenever it is non - zero ; and the same holds for the turning ratios at diverge junctions . consider a @xmath176-wave @xmath448 , then its variation @xmath449 can change only upon interaction with diverge junctions . more precisely , denote by @xmath450 and @xmath451 the wave respectively before and after the interaction , we have @xmath452 where @xmath453 or @xmath454 and is bounded away from zero . since the @xmath176-waves travel only forward on the links with uniformly bounded speed , we must have that the interactions with diverge junctions can occur only finite number of times . thus we conclude that @xmath176 has bounded total variation . let @xmath455 be the shift of the @xmath96-th jump of @xmath261 , which occurs at time @xmath456 , then the expression of @xmath255 is possibly affected on the interval @xmath457 $ ] ( assuming @xmath458 ) . more precisely , if @xmath459 then no wave is generated on the virtual link while a shift in the queue is generated with @xmath460 , where @xmath461 is the jump in @xmath261 occurring at time @xmath456 . on the other hand , if @xmath462 then a wave @xmath463 is produced at time @xmath456 on the virtual link with shift @xmath464 where @xmath465 is the speed of the wave @xmath463 . we can then compute : @xmath466 moreover , @xmath467 , thus the norm of the tangent vector generated is the same as before . to prove that the norm of the tangent vector @xmath252 is bounded for all times , let us first consider a backward - propagating wave @xmath463 interacting with the queue @xmath468 , and with a shift @xmath469 . then we can write : @xmath470 where @xmath471 and @xmath472 are the time - derivatives of @xmath255 before and after the interaction , respectively . therefore , denoting @xmath465 the speed of the wave @xmath463 , we get : @xmath473 thus the norm of the tangent vector is bounded . aboudolas , k. , papageorgiou , m. , komsmatopoulous , e. , 2009 . store - and - forward based methods for the signal control problem in large - scale congested urban road networks . transportation research part c , 17 ( 2 ) , 163 - 174 . han , k. , friesz , t.l . , szeto , w.y . , liu , h. , 2015a . elastic demand dynamic network user equilibrium : formulation , existence and computation . transportation research part b , doi : 10.1016/j.trb.2015.07.008 . han , k. , friesz , t.l . , yao , t. , 2013a . a partial differential equation formulation of vickrey s bottleneck model , part i : methodology and theoretical analysis . transportation research part b 49 , 55 - 74 . han , k. , piccoli , b. , szeto , w.y . continuous - time link - based kinematic wave model : formulation , solution existence , and well - posedness . transportmetrica b : transport dynamics , doi : 10.1080/21680566.2015.1064793 . yperman , i. , logghe , s. , immers , l. , 2005 . the link transmission model : an efficient implementation of the kinematic wave theory in traffic networks , advanced or and ai methods in transportation , proc . 10th ewgt meeting and 16th mini - euro conference , poznan , poland , 122 - 127 , publishing house of poznan university of technology .

this paper establishes the continuity of the path delay operators for _ dynamic network loading _ ( dnl ) problems based on the lighthill - whitham - richards model , which explicitly capture vehicle spillback . the dnl describes and predicts the spatial - temporal evolution of traffic flow and congestion on a network that is consistent with established route and departure time choices of travelers . the lwr - based dnl model is first formulated as a system of _ partial differential algebraic equations _ ( pdaes ) . we then investigate the continuous dependence of merge and diverge junction models with respect to their initial / boundary conditions , which leads to the continuity of the path delay operator through the _ wave - front tracking _ methodology and the _ generalized tangent vector _ technique . as part of our analysis leading up to the main continuity result , we also provide an estimation of the minimum network supply without resort to any numerical computation . in particular , it is shown that gridlock can never occur in a finite time horizon in the dnl model . ( 1,0)469 path delay operator , continuity , dynamic network loading , lwr model , spillback , gridlock

collimated outflows / jets are produced from active galactic nuclei , compact stars in x - ray binaries , and young stellar objects ( ysos ) ( e.g. , krolik 1999 ; mirabel & rodriguez 1999 ; fender 2003 ; reipurth & bally 2001 ) . over the last two decades , evidence for jet precession in these systems has steadily increased . the clearest example is the galactic source ss 433 , whose jet direction varies with an approximate 164 day period ( margon 1984 ; eikenberry et al . the black hole binary gro j1655 - 40 shows jet precession with a period of 3 days ( tingay et al . 1995 ; hjellming & rupen 1995 ) . the jets from the super soft source cal83 may be precessing with a period of @xmath0 days ( cowley et al . 1998 ) . in agns , indirect evidence for jet precession is seen in the morphology of the radio hot spots , which show `` fossil '' components offset from the `` present '' component positions ( e.g. , cygnus a ) . finally , the changes in the flow directions of several yso jets have been interpreted in terms of jet precession ( e.g. terquem et al . 1999 ; bates et al . 2000 ) . a natural cause for jet precession is the changes of orientation of the underlying accretion disk . in addition , the super - orbital variabilities observed in a number of x - ray binaries ( e.g. , the 35-day periodicity in her x-1 ; see priedhorsky & holt 1987 ; ogilvie & dubus 2001 ) have long been interpreted as due to precession of a tilted accretion disk . in both x - ray binaries and ysos , the tidal force from the companion star could play a role in driving the precession ( e.g. , katz 1973 ; wijers & pringle 1999 ; terquem et al . 1999 ; bate et al . 2000 ; ogilvie & dubus 2001 ) , although it may not be the only or the dominant effect ( e.g. , the 3 day precession period of the gro j1655 - 40 jet is too short to be explained by the tidal effect for a 2.6 day binary ) . most importantly , for the precession to operate , disk tilt needs to be excited and maintained . in accreting binary systems , the disk plane is expected to be aligned with the orbital plane since the disk angular momentum originates from the binary motion . for ysos in binaries , the disk plane may be initially misaligned with the orbital plane . however , if we consider the disk as a collection of circular rings , different rings will have different precession rates ; it has been recognized that the combined effects of differential precession and internal disk stress / dissipation tend to damp the the disk tilt , so that the disk settles into the binary plane ( lubow & ogilvie 2000 ) . several driving mechanisms for disk tilt / warp have been proposed . schandl & meyer ( 1994 ) showed ( in the context of her x-1 ) that irradiation - driven wind from the outer parts of the disk can carry away momentum flux and make the disk unstable to warping . pringle ( 1996 ) showed that even without wind loss , radiation pressure itself can induce warping instability in the outer region of the disk . futher studies indicated this radiation - driven warping instability may indeed operate in x - ray binaries ( e.g. , maloney , begelman & nowak 1998 ; wijers & pringle 1999 ) , although it does not provide a generic explanation for the long - term variabilities in all x - ray binaries ( ogilvie & dubus 2001 ) . quillen ( 2001 ) showed that a wind passing over the disk surface may induce warping via kelvin - helmholtz instability . finally , in the case of disk accretion onto magnetic stars ( e.g. , neutron stars , white dwarfs and t tauri stars ) , the stellar magnetic field can induce disk warping and precession ( lai 1999 ; see also aly 1980 ; lipunov & shakura 1980 ; terquem & papaloizou 2000 ) ; this may explain several observed features of quasi - periodic oscillations in low - mass x - ray binaries ( shirakawa & lai 2002a ) , milli - hertz variabilities in accreting x - ray pulsars ( shirakawa & lai 2002b ) , and variabilities of t tauri stars ( terquem & papaloizou 2000 ; see also agapitou et al . 1997 ) . in this paper we suggest a new disk warping mechanism that is directly tied to the production of magnetically driven outflows / jets . using an idealized setup ( 2 ) , we show that a disk threaded by a large - scale magnetic field may experience a warping instability and precess around the central object ( 3 ) . these magnetically driven disk warping and precession arise from the interaction between the large - scale magnetic field and the induced electric current in the disk . while more studies are needed , we suggest that these effects may provide a natural explanation for the procession of jets / outflows and other variabilities observed in various systems ( 4 ) . the current paradigm for the origin of astrophysical jets / outflows involves a large - scale magnetic field threading the accretion disk around a central object ( star or black hole ) ; this ordered magnetic field plays a crucial role in extracting / channeling mass , energy and angular momentum from the disk . the energy outflow can be either hydromagnetic ( with significant mass flux ) ( blandford & payne 1982 ) or electromagnetic ( dominated by poynting flux ) ( blandford 1976 ; lovelace 1976 ) . the origin of the disk - threading magnetic field is not completely clear : the field could be advected inwards by accretion , or generated locally by dynamo processes . in the case of protostellar outflows , the stellar magnetic field may play an important role ( shu et al . 1994 , 2000 ) . many theoretical / numerical studies have been devoted to understanding magnetically driven outflows / jets from accretion disks ( e.g. , see recent reviews by lovelace et al . 1999 ; knigl & pudritz 2000 ; meier et al . 2001 ) . figure 1 shows an idealized sketch of the magnetic field disk configuration in which hydromagnetic outflows are produced . a geometrically thin disk is threaded by the poloidal magnetic field @xmath1 ( where @xmath2 are unit vectors in cylindrical coordinates ) . this poloidal field is of course curved ( with @xmath3 at the disk midplane ) , and we use the superscript `` + '' to denote the field at the upper disk surface . for a sufficiently inclined field ( radial pitch angle @xmath4 ) , centrifugally driven outflows are possible ( blandford & payne 1982 ) . the twisting of @xmath5 by the disk and the outflow gives rise to a toroidal field @xmath6 , which has different signs above and below the disk plane . we introduce the azimuthal pitch @xmath7 , so that @xmath8 at the upper disk surface . that @xmath7 is positive reflects the fact that the disk always rotates faster than the outflow , where @xmath9 is the the poloidal velocity of the outflow and @xmath10 its rotation rate , @xmath11 is the disk rotation at the midplane . alternatively , in the absence of outflow , @xmath6 near the disk is governed schematically by the equation @xmath12 , where @xmath13 is the disspation timescale . in steady state , we find @xmath14 , with @xmath15 ( where @xmath16 is the vertical scale height in which @xmath17 in the magnetosphere varies . ] . the discontinuities in @xmath6 and @xmath18 across the disk imply a net disk surface current ( integrated over the disk thickness ) @xmath19 . thus , for an unpertutbed ( unwarped ) disk , the only torque on the disk ( per unit area ) is the usual magnetic braking torque t_z=12rb_zb_^+=-12rb_z^2 . [ eqtz]angular momentum is extracted out of the disk by the open field lines , and can be carried away by torsional alfven waves or a centrifugally driven wind ( for sufficiently inclined poloidal fields ) . in addition , the radial magnetic force , f_r = b_zb_r^+/(2 ) , [ eqfr]provides partial support of the disk against gravity . the configuration shown in fig . 1 lies at the heart of many theoretical models of magnetic disk outflows . the real situation is certainly much more complicated . nevertheless , in 3 we shall adopt this simple picture in our consideration of magnetically driven disk warping and precession our goal is to discuss these physical effects in a way that is as transparent as possible , so as to expose their robustness and potential pitfall . we now consider warping perturbations to the disk . the disk can be considered as a collection of circular rings which interact with each other via internal stresses . we specify the disk warping by the unit normal vector @xmath20 for the ring at radius @xmath21 . consider the situation ( see fig . 2 ) in which the ring around radius @xmath21 is tilted by an angle @xmath22 while most of the disk remains flat ( @xmath23 , where @xmath24 is the normal vector of the unperturbed disk ; see fig . 1 ) . since the magnetic field @xmath25 is produced by all the currents in the disk - outflow system , we posit that @xmath25 at radius @xmath21 is not modified by such a local perturbation . for the ring at @xmath21 , the field @xmath26 acts as an `` external '' magnetic field . now set up a coordinate system with the @xmath27-axis along the @xmath28 ( see the lower panel of fig . 2 ) and the associated cylindrical coordinate system @xmath29 . as in the case of the unperturbed disk , twisting of @xmath30 by the disk rotation produces a toroidal field ( in the new coordinate system ) @xmath31 ( where the subscript `` t '' serves as a reminder that this field is due to twisting of @xmath32 ) , whose value ( at radius @xmath21 ) at the upper disk plane is @xmath33 , and at the lower disk plane @xmath34 . note that @xmath32 is necessarily continuous across the disk plane in this perturbation . the radial surface current associated with @xmath35 is k_r =- c2 b_,t^+=c2b_z . interaction of this @xmath36 with @xmath32 gives an azimuthal force on the disk , and results in the magnetic braking torque ( per unit area ) @xmath37 , analogous to eq . ( [ eqtz ] ) . unlike the unperturbed disk , however , the `` extrenal '' field @xmath38 now has a @xmath39-component , @xmath40 , and the interaction of @xmath36 with @xmath41 gives rise to a vertical force ( per unit area ) f_z=12b_z^2 . this vertical force can also be seen as arising from the imbalance of magnetic stresses above and below the disk plane : @xmath42 in the upper disk plane while @xmath43 in the lower disk plane , and thus @xmath44/(8\pi)$ ] . obviously , averaging the force over the azimuthal angle , @xmath45 , gives zero . but since @xmath46 is unevenly distributed along the ring , it gives rise to a net torque on the ring . averaging over @xmath39 , the torque per unit area is _ warp&= & -14r b_z^2*y * + & = & -14r b_z^2 . clearly , this torque tends to pull the vector @xmath28 away from `` external '' magnetic field direction @xmath47 . this is the _ magnetically driven warping instability _ : for small tilt , the angle @xmath22 tends to grow ( in the absence of other forces ) at the rate _ warp = b_z^24r_d , [ gwarp]where @xmath48 is the surface density , and @xmath11 the angular velocity of the disk . in a hypothetical situation , if @xmath26 had a fixed direction for all @xmath22 , and if there were no coupling between different rings , the vector @xmath28 would evolve toward being perpendicular to @xmath26 , i.e. , the plane of the disk ( ring ) would prefer to lie along the `` external '' magnetic field ( cf . lai 1999 ) . in the perturbed disk , the azimuthal surface current @xmath49 ( for small @xmath22 ) also interacts with the `` external '' field @xmath38 , which can be decomposed into @xmath50 . the interaction between @xmath51 and @xmath30 gives a radial force , as in eq . ( [ eqfr ] ) . for the perturbed disk ( @xmath52 ) , a new vertical force arises from the interaction between @xmath51 and @xmath53 : f_z=-1ck_b_z . this force is also equal to @xmath54/(8\pi)$ ] , where @xmath55 . again , @xmath56 , but the net torque is non - zero . averaging over @xmath39 , we find the torque per unit area _ prec=-12ck_b_z r*x * = -12ck_b_z r**. clearly , the effect of this torque is to make the ring ( at radius @xmath21 ) precess around the @xmath57-axis ( the unperturbed disk normal vector ) . the angular frequnecy of the _ magnetically driven precession _ is _ prec =- k_b_z2cr_d * * = -b_z^24r_d * * . [ oprec]the negative sign implies that the pecession is retrograde , i.e. , opposite to the disk rotation . to get an idea on the warping and precession timescales associated with eqs . ( [ gwarp ] ) and ( [ oprec ] ) , we note the disk angular momentum equation ( for an unwarped disks ) can be written as v_rddr(_d r^2)=-12rb_z^2+t_visc , where @xmath58 is the radial velocity and @xmath59 is the viscous torque . if we set @xmath60 and use @xmath61 , we find _ warp=|_prec| = -v_r4fr . for typical parameters ( @xmath62 ) , the warping timescale and precession period are of order the disk infall time @xmath63 . it is important to note that the magnetically driven disk warping and precession discussed in 3 are secular effects which operate on timescales much longer than the disk dynamical time . in deriving eqs . ( [ gwarp ] ) and ( [ oprec ] ) , we have implicitly assumed that the perturbed disk ( see fig . 2 ) reaches a new steady state so that the magnetic field behaves in a similar way as in the unperturbed disk ( e.g. , @xmath64 ) . these effects are not captured in a dynamical perturbation analysis of the disk . clearly , our derivation of the magnetically driven warping and precession is far from being rigorous . nevertheless , it suggests that the warped , precessing disk - outflow may be an alternative ( and preferred ) state for the accretion disk threaded by large - scale magnetic fields . the effects of magnetically driven warping and precession discussed in this paper are analogous to the similar effects that exist in the accretion disks around magnetic stars ( lai 1999 ; shirakawa & lai 2002a , b ; pfeiffer & lai 2003 ; see also aly 1980 ; lipunov & shakura 1980 ) . in those systems , the central star provides the `` external '' magnetic field for the disk . the warping and precessional torques arise from the interaction between the surface current on the disk and the horizontal magnetic field ( parallel to the disk ) produced by the stellar magnetic dipole . here , in the absence of the central magnetic star , the `` external '' magnetic field ( for the warped region of the disk ) is generated by the other currents in the disk - outflow system , and we have argued that similar torques exist : the warping torque relies on the surface current generated by the twisting of the vertical field threading the disk , while the precessional torque relies on the azimuthal screening current due to the diamagnetic response of the disk . the simplistic nature of our arguments ( perhaps even to the point of being wrong ) in 3 and the complexity of the real astrophysical systems ( see references in 2 ) preclude us from addressing many important issues such as the extent and location of the warped disk region . nevertheless , we are not aware of any previous discussion of these effects , and we think it is useful to present them so that they may be studied further by the astrophysics community . the magnetically driven warping and precession effects discussed in this paper potentially have advantages over the other disk warping mechanisms that are needed to explain jet precession ( see 1 ) . for example , the radiation driven warping instability ( pringle 1996 ) , while probably important for x - ray binaries ( ogilvie & dubus 2001 ) , is irrelevant for ysos ( since the instability operates at large disk radii ) . the magnetic effects discussed in this paper are directly connected to the large - scale magnetic field associated with the jet / outflow , and the model naturally predicts retrograde precession of the warped disk ( and thus the jet ) , in agreement with observations in x - ray binaries ( e.g. , her x-1 ) . the idea for this paper was motivated by the questions from chris mckee and frank shu following a seminar that the author gave at uc berkeley in 2001 . i thank richard lovelace for useful discussion . this work is supported in part by nsf grant ast 9986740 and nasa grant nag 5 - 12034 , as well as by a research fellowship from the alfred p. sloan foundation . lovelace , r.v.e . , et al . 1999 , in active galactic nuclei and related phenomena ( proceedings of iau symposium 194 ) , eds . y. terzian , et al . ( asp : san francisco ) lubow , s.h . , & ogilvie , g.i . 2000 , apj , 538 , 326 shu , f. h. , et al . 2000 , in protostars and planets iv , eds . v , manning et al . ( univ . of arizona press ) terquem , c. , et al . 1999 , apj , 512 , l131 terquem , c. , & papaloizou , j.c.b . 2000 , a&a , 1031 tingay , s.j . , et al . 1995 , nature , 374 , 141

current theoretical models for the outflows / jets from agn , x - ray binaries and young stellar objects involve large - scale magnetic fields threading an underlying accretion disk . we suggest that such a disk is subjected to warping instability and retrograde precession driven by magnetic torques associated with the outflow . the growth timescale for the disk warp and the precession period are of order the radial infall time of the disk . these effects may be relevant to jet precession and other variabilities observed in many systems .