File size: 5,547 Bytes
079c32c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
import time
import torch
from hpc_rll.origin.td import dist_nstep_td_error, dist_nstep_td_data
from hpc_rll.rl_utils.td import DistNStepTD
from testbase import mean_relative_error, times

assert torch.cuda.is_available()
use_cuda = True

T = 128
B = 128
N = 128
gamma = 0.95
v_min = -10.0
v_max = 10.0
n_atom = 51


def dntd_val():
    ori_dist = torch.randn(B, N, n_atom).abs()
    ori_next_n_dist = torch.randn(B, N, n_atom).abs()
    ori_action = torch.randint(0, N, size=(B, ))
    ori_next_n_action = torch.randint(0, N, size=(B, ))
    ori_reward = torch.randn(T, B)
    ori_done = torch.randn(B)
    ori_weight = torch.randn(B)

    hpc_dist = ori_dist.clone().detach()
    hpc_next_n_dist = ori_next_n_dist.clone().detach()
    hpc_action = ori_action.clone().detach()
    hpc_next_n_action = ori_next_n_action.clone().detach()
    hpc_reward = ori_reward.clone().detach()
    hpc_done = ori_done.clone().detach()
    hpc_weight = ori_weight.clone().detach()
    hpc_dntd = DistNStepTD(T, B, N, n_atom)

    if use_cuda:
        ori_dist = ori_dist.cuda()
        ori_next_n_dist = ori_next_n_dist.cuda()
        ori_action = ori_action.cuda()
        ori_next_n_action = ori_next_n_action.cuda()
        ori_reward = ori_reward.cuda()
        ori_done = ori_done.cuda()
        ori_weight = ori_weight.cuda()

        hpc_dist = hpc_dist.cuda()
        hpc_next_n_dist = hpc_next_n_dist.cuda()
        hpc_action = hpc_action.cuda()
        hpc_next_n_action = hpc_next_n_action.cuda()
        hpc_reward = hpc_reward.cuda()
        hpc_done = hpc_done.cuda()
        hpc_weight = hpc_weight.cuda()
        hpc_dntd = hpc_dntd.cuda()

    ori_dist.requires_grad_(True)
    ori_loss, ori_td_err = dist_nstep_td_error(
        dist_nstep_td_data(ori_dist, ori_next_n_dist, ori_action, ori_next_n_action, ori_reward, ori_done, ori_weight),
        gamma, v_min, v_max, n_atom, T
    )
    ori_loss = ori_loss.mean()
    ori_loss.backward()

    hpc_dist.requires_grad_(True)
    hpc_loss, hpc_td_err = hpc_dntd(
        hpc_dist, hpc_next_n_dist, hpc_action, hpc_next_n_action, hpc_reward, hpc_done, hpc_weight, gamma, v_min, v_max
    )
    hpc_loss = hpc_loss.mean()
    hpc_loss.backward()

    mre = mean_relative_error(
        torch.flatten(ori_loss).cpu().detach().numpy(),
        torch.flatten(hpc_loss).cpu().detach().numpy()
    )
    print("dntd fp mean_relative_error: " + str(mre))
    mre = mean_relative_error(
        torch.flatten(ori_td_err).cpu().detach().numpy(),
        torch.flatten(hpc_td_err).cpu().detach().numpy()
    )
    print("dntd fp td_err mean_relative_error: " + str(mre))
    mre = mean_relative_error(
        torch.flatten(ori_dist.grad).cpu().detach().numpy(),
        torch.flatten(hpc_dist.grad).cpu().detach().numpy()
    )
    print("dntd bp mean_relative_error: " + str(mre))


def dntd_perf():
    ori_dist = torch.randn(B, N, n_atom).abs()
    ori_next_n_dist = torch.randn(B, N, n_atom).abs()
    ori_action = torch.randint(0, N, size=(B, ))
    ori_next_n_action = torch.randint(0, N, size=(B, ))
    ori_reward = torch.randn(T, B)
    ori_done = torch.randn(B)
    ori_weight = torch.randn(B)

    hpc_dist = ori_dist.clone().detach()
    hpc_next_n_dist = ori_next_n_dist.clone().detach()
    hpc_action = ori_action.clone().detach()
    hpc_next_n_action = ori_next_n_action.clone().detach()
    hpc_reward = ori_reward.clone().detach()
    hpc_done = ori_done.clone().detach()
    hpc_weight = ori_weight.clone().detach()
    hpc_dntd = DistNStepTD(T, B, N, n_atom)

    if use_cuda:
        ori_dist = ori_dist.cuda()
        ori_next_n_dist = ori_next_n_dist.cuda()
        ori_action = ori_action.cuda()
        ori_next_n_action = ori_next_n_action.cuda()
        ori_reward = ori_reward.cuda()
        ori_done = ori_done.cuda()
        ori_weight = ori_weight.cuda()

        hpc_dist = hpc_dist.cuda()
        hpc_next_n_dist = hpc_next_n_dist.cuda()
        hpc_action = hpc_action.cuda()
        hpc_next_n_action = hpc_next_n_action.cuda()
        hpc_reward = hpc_reward.cuda()
        hpc_done = hpc_done.cuda()
        hpc_weight = hpc_weight.cuda()
        hpc_dntd = hpc_dntd.cuda()

    ori_dist.requires_grad_(True)
    for i in range(times):
        t = time.time()
        ori_loss, ori_td_err = dist_nstep_td_error(
            dist_nstep_td_data(
                ori_dist, ori_next_n_dist, ori_action, ori_next_n_action, ori_reward, ori_done, ori_weight
            ), gamma, v_min, v_max, n_atom, T
        )
        ori_loss = ori_loss.mean()
        ori_loss.backward()
        if use_cuda:
            torch.cuda.synchronize()
        print('epoch: {}, origin dntd cost time: {}'.format(i, time.time() - t))

    hpc_dist.requires_grad_(True)
    for i in range(times):
        t = time.time()
        hpc_loss, hpc_td_err = hpc_dntd(
            hpc_dist, hpc_next_n_dist, hpc_action, hpc_next_n_action, hpc_reward, hpc_done, hpc_weight, gamma, v_min,
            v_max
        )
        hpc_loss = hpc_loss.mean()
        hpc_loss.backward()
        if use_cuda:
            torch.cuda.synchronize()
        print('epoch: {}, hpc dntd cost time: {}'.format(i, time.time() - t))


if __name__ == '__main__':
    print(
        "target problem: T = {}, B = {}, N = {}, gamma = {}, v_min = {}, v_max = {}, n_atom = {}".format(
            T, B, N, gamma, v_min, v_max, n_atom
        )
    )
    print("================run dntd validation test================")
    dntd_val()
    print("================run dntd performance test================")
    dntd_perf()