-
Notifications
You must be signed in to change notification settings - Fork 0
/
test_muot.py
55 lines (46 loc) · 1.71 KB
/
test_muot.py
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
import numpy as np
import matplotlib.pyplot as plt
from MUOT import *
import pdb
n = 3
dims = [2]*3
T = [10]*3
C = np.random.uniform(low=1, high=50, size=dims)
C = (C + C.T) / 2
R = []
for d in dims:
r = np.random.uniform(low=0.1, high=1, size=(d, 1))
R.append(r)
info = sinkhorn_muot(C, R, eta=0.1, T=T, n_iter=500, early_stop=False)
# inf_norms = {}
# for i in range(n):
# inf_norms[i] = []
#
# for i in range(info['stop_iter']):
# for j in range(n):
# inf_norms[j].append(supnorm(info['U_all'][i][j] - info['U_all'][-1][j]))
fig, ax = plt.subplots(1,3)
min_primal = np.min(info['f_primal_val_list'])
min_dual = np.min(info['f_dual_val_list'])
min_unregularized = np.min(info['unreg_f_val_list'])
conv_primal = info['f_primal_val_list'][-1]
conv_dual = info['f_dual_val_list'][-1]
conv_unregularized = info['unreg_f_val_list'][-1]
ax[0].plot(np.arange(info['stop_iter']+1), info['f_primal_val_list'], label=f'primal, {min_primal:.3f}, {conv_primal:.3f}')
ax[2].plot(np.arange(info['stop_iter']+1), info['f_dual_val_list'], label=f'dual, {min_dual:.3f}, {conv_dual:.3f}')
ax[1].plot(np.arange(info['stop_iter']+1), info['unreg_f_val_list'], label=f'unregularized, {min_unregularized:.3f}, {conv_unregularized:.3f}')
ax[0].legend()
ax[1].legend()
ax[2].legend()
# for j in range(n):
# ax.plot(np.arange(info['stop_iter'])+1, inf_norms[j], label='marginal i={}'.format(j))
# ax.set_title('$||u^t_i - u^*_i||$')
# U = np.concatenate(info['U_all'], axis=-1).transpose(0, 2, 1) # [3, 10000, 2]
# for i in range(3):
# for j in range(info['stop_iter']):
# U[i, j] = U[i, j, :] - U[i, -1, :]
# print(U[0, -1, :])
# for i in range(3):
# ax[i].scatter(U[i][:,0], U[i][:,1], s=5)
# ax[i].legend()
plt.show()