-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathsto_sim.py
executable file
·151 lines (117 loc) · 5.31 KB
/
sto_sim.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
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
import numpy as np
import pickle
import scipy.io as sio
from sto_adv_BA_algs import *
from sto_adv_BA_exp import *
playtime = 2000
K = 20
def set_mu():
mu = {}
u = 0.4 * np.ones([20, ])
u[0] = 0.5
mu[1] = u
u = 0.5 * np.ones([20, ])
for i in range(1, 10):
u[i] = 0.42
for i in range(10, 20):
u[i] = 0.38
mu[2] = u
u = 0.5 * np.ones([20, ])
u[1] = 0.48
for i in range(2, 20):
u[i] = 0.38
mu[3] = u
u = 0.5 * np.ones([20, ])
for i in range(1, 10):
u[i] = 0.5 - 1. / (5 * K)
for i in range(10, 20):
u[i] = 0.25
mu[4] = u
return mu
mu = set_mu() # mu is a dictionary contains different mu's for each simulation
Budgets = 3000
er_SH_sto = np.zeros([4, ])
er_AdUCBE_sto = np.zeros([4, ]) # c = 0.25
er_AdUCBE2_sto = np.zeros([4, ]) # c = 2
er_S3BA_sto = np.zeros([4, ]) # delta = 0.1
er_S3BA2_sto = np.zeros([4, ])
er_S3BA_sdelta_sto = np.zeros([4, ]) # delta = 0.05
er_S3BA_sdelta2_sto = np.zeros([4, ])
def set_params(C_w=0.5, C_3=1, C_init=10, C_gap=2):
parameters = {}
delta = 0.1
c = 0.25
C_w = 16
C_3 = 522
C_init = 100. / 9
C_gap = 60
parameters[1] = [delta, c, C_w, C_3, C_init, C_gap]
delta = 0.1
c = 2
parameters[2] = [delta, c, C_w, C_3, C_init, C_gap]
delta = 0.05
c = 0.25
parameters[3] = [delta, c, C_w, C_3, C_init, C_gap]
delta = 0.05
c = 2
parameters[4] = [delta, c, C_w, C_3, C_init, C_gap]
return parameters
parameters = set_params() # set parameters for S3BA alg.
loss_exp = {}
for i in range(4):
loss_exp[i] = np.zeros([playtime, K, Budgets])
for k in range(K):
for j in range(playtime):
np.random.seed(j + playtime * 3 * k)
loss_exp[i][j, k, :] = np.random.rand(int(Budgets)) > mu[i + 1][k]
sio.savemat('./sto_loss_exp' + str(i) + '.mat', {'sto_loss_exp' + str(i): loss_exp[i]})
print('Algorithm ' + str(1))
for i in range(4):
print('Group ' + str(i + 1))
er_SH_sto[i], _, _ = Exp(alg=Successive_Halving, playtimes=playtime, c=[], other_alg_parameters=[], N=Budgets, K=K,
loss_generate=False, losses=loss_exp[i], mu=[], var=[], best_arm=1, turn_bud_to_N=True, verbose=1000)
sio.savemat('./er_SH_sto.mat', {'er_SH_sto': er_SH_sto})
print('Algorithm ' + str(2))
for i in range(4):
print('Group ' + str(i + 1))
er_AdUCBE_sto[i], _, _ = Exp(alg=Ad_UCBE, playtimes=playtime, c=0.25, other_alg_parameters=[], N=Budgets, K=K,
loss_generate=False, losses=loss_exp[i], mu=[], var=[], best_arm=1, turn_bud_to_N=True, verbose=500)
sio.savemat('./er_AdUCBE_sto.mat', {'er_AdUCBE_sto': er_AdUCBE_sto})
print('Algorithm ' + str(3))
for i in range(4):
print('Group ' + str(i + 1))
er_AdUCBE2_sto[i], _, _ = Exp(alg=Ad_UCBE, playtimes=playtime, c=2, other_alg_parameters=[], N=Budgets, K=K,
loss_generate=False, losses=loss_exp[i], mu=[], var=[], best_arm=1, turn_bud_to_N=True, verbose=500)
sio.savemat('./er_AdUCBE2_sto.mat', {'er_AdUCBE2_sto': er_AdUCBE2_sto})
print('Algorithm ' + str(4))
for i in range(4):
print('Group ' + str(i + 1))
er_S3BA_sto[i], trigger_rate, trigger_time = Exp(alg=S3_BA, playtimes=playtime, c=[], other_alg_parameters=parameters[1], N=Budgets, K=K,
loss_generate=False, losses=loss_exp[i], mu=[], var=[], best_arm=1, turn_bud_to_N=True, verbose=500)
print trigger_rate, np.mean(np.array(trigger_time))
pickle.dump(trigger_time, open('trigger_time_sto_exp4' + str(i + 1) + '.pkl', 'wb'))
sio.savemat('./er_S3BA_sto.mat', {'er_S3BA_sto': er_S3BA_sto})
print('Algorithm ' + str(5))
for i in range(4):
print('Group ' + str(i + 1))
er_S3BA2_sto[i], trigger_rate, trigger_time = Exp(alg=S3_BA, playtimes=playtime, c=[], other_alg_parameters=parameters[2], N=Budgets, K=K,
loss_generate=False, losses=loss_exp[i], mu=[], var=[], best_arm=1, turn_bud_to_N=True, verbose=500)
print trigger_rate, np.mean(np.array(trigger_time))
pickle.dump(trigger_time, open('trigger_time_sto_exp5' + str(i + 1) + '.pkl', 'wb'))
sio.savemat('./er_S3BA2_sto.mat', {'er_S3BA2_sto': er_S3BA2_sto})'''
'''print('Algorithm ' + str(6))
for i in range(4):
print('Group ' + str(i + 1))
er_S3BA_sdelta_sto[i], trigger_rate, trigger_time = Exp(alg=S3_BA, playtimes=playtime, c=[], other_alg_parameters=parameters[3], N=Budgets, K=K,
loss_generate=False, losses=loss_exp[i], mu=[], var=[], best_arm=1, turn_bud_to_N=True, verbose=500)
print trigger_rate, np.mean(np.array(trigger_time))
pickle.dump(trigger_time, open('trigger_time_sto_exp6' + str(i + 1) + '.pkl', 'wb'))
sio.savemat('./er_S3BA_sdelta_sto.mat', {'er_S3BA_sdelta_sto': er_S3BA_sdelta_sto})
print('Algorithm ' + str(7))
for i in range(4):
print('Group ' + str(i + 1))
er_S3BA_sdelta2_sto[i], trigger_rate, trigger_time = Exp(alg=S3_BA, playtimes=playtime, c=[], other_alg_parameters=parameters[4], N=Budgets, K=K,
loss_generate=False, losses=loss_exp[i], mu=[], var=[], best_arm=1, turn_bud_to_N=True, verbose=500)
print trigger_rate, np.mean(np.array(trigger_time))
pickle.dump(trigger_time, open('trigger_time_sto_exp7' + str(i + 1) + '.pkl', 'wb'))
sio.savemat('./er_S3BA_sdelta2_sto.mat', {'er_S3BA_sdelta2_sto': er_S3BA_sdelta2_sto})