-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathexperiment.py
203 lines (168 loc) · 6.5 KB
/
experiment.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
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
import numpy as np
import math
import matplotlib.pyplot as plt
from bandit import Bandit
from explore_then_exploit_agent import ExploreThenExploit
from MC_simulator import *
from epsilon_greedy_agent import EpsilonGreedy
from ubc1_agent import UBC1Agent
from report import plot
from successive_elimination_agent import SuccessiveEliminationAgent
from agent import Agent
from agent_factory import agentFactory
n_sim = 40
def createBanditInstancesAndSimulate(params,n_mc_sim):
n_sim = n_mc_sim
for i in range(len(params)):
results = []
time_horizon = params[i]['time_horizon']
number_of_arms = params[i]['number_of_arms']
number_of_exploration_per_arm = params[i]['number_of_exploration_per_arm']
exp_agent = ExploreThenExploit(time_horizon,number_of_arms,number_of_exploration_per_arm)
epsilon_greedy_constant_half_epsilonAgent = EpsilonGreedy(time_horizon,number_of_arms,[1/2]*time_horizon)
epsilon_greedy_constant_epsilonAgent = EpsilonGreedy(time_horizon,number_of_arms,[number_of_exploration_per_arm*number_of_arms/time_horizon]*time_horizon)
ubc_agent = UBC1Agent(time_horizon,number_of_arms)
se_agent = SuccessiveEliminationAgent(time_horizon,number_of_arms)
random_agent = Agent()
bandit = Bandit(time_horizon,number_of_arms,random_agent)
results.append(mc_simulate(n_sim,bandit))
bandit = Bandit(time_horizon,number_of_arms,exp_agent)
results.append(mc_simulate(n_sim,bandit))
bandit = Bandit(time_horizon,number_of_arms,epsilon_greedy_constant_epsilonAgent)
results.append(mc_simulate(n_sim,bandit,"constant-epsilon=rate-of-explore-exploit"))
bandit = Bandit(time_horizon,number_of_arms,epsilon_greedy_constant_half_epsilonAgent)
results.append(mc_simulate(n_sim,bandit,"constant-epsilon=0.5"))
bandit = Bandit(time_horizon,number_of_arms,ubc_agent)
results.append(mc_simulate(n_sim,bandit))
bandit = Bandit(time_horizon,number_of_arms,se_agent)
results.append(mc_simulate(n_sim,bandit))
plot(results,time_horizon,params[i])
def experiment1():
params = [{
"time_horizon" : 1000,
"number_of_arms" : 5
},
{
"time_horizon" : 10000,
"number_of_arms" : 5
},
{
"time_horizon" : 1000,
"number_of_arms" : 10
},
{
"time_horizon" : 10000,
"number_of_arms" : 10
},
{
"time_horizon" : 1000,
"number_of_arms" : 20
},
{
"time_horizon" : 10000,
"number_of_arms" : 20
},
]
for i in range(len(params)):
results = []
time_horizon = params[i]["time_horizon"]
number_of_arms = params[i]["number_of_arms"]
agent1 = agentFactory("explore-then-exploit",time_horizon,number_of_arms,5)
agent2 = agentFactory("explore-then-exploit",time_horizon,number_of_arms,time_horizon/10)
agent3 = agentFactory("explore-then-exploit",time_horizon,number_of_arms,time_horizon/100)
bandit = Bandit(time_horizon,number_of_arms,agent1)
results.append(mc_simulate(n_sim,bandit,"N=5"))
bandit = Bandit(time_horizon,number_of_arms,agent2)
results.append(mc_simulate(n_sim,bandit,"N=T/10"))
bandit = Bandit(time_horizon,number_of_arms,agent3)
results.append(mc_simulate(n_sim,bandit,"N=T/100"))
plot(results,time_horizon,params[i])
def experiment1():
params = [{
"time_horizon" : 1000,
"number_of_arms" : 5
},
{
"time_horizon" : 10000,
"number_of_arms" : 5
},
{
"time_horizon" : 1000,
"number_of_arms" : 10
},
{
"time_horizon" : 10000,
"number_of_arms" : 10
},
{
"time_horizon" : 1000,
"number_of_arms" : 20
},
{
"time_horizon" : 10000,
"number_of_arms" : 20
},
]
for i in range(len(params)):
results = []
time_horizon = params[i]["time_horizon"]
number_of_arms = params[i]["number_of_arms"]
agent1 = agentFactory("explore-then-exploit",time_horizon,number_of_arms,5)
agent2 = agentFactory("explore-then-exploit",time_horizon,number_of_arms,time_horizon/10)
agent3 = agentFactory("explore-then-exploit",time_horizon,number_of_arms,time_horizon/100)
bandit = Bandit(time_horizon,number_of_arms,agent1)
results.append(mc_simulate(n_sim,bandit,"N=5"))
bandit = Bandit(time_horizon,number_of_arms,agent2)
results.append(mc_simulate(n_sim,bandit,"N=T/10"))
bandit = Bandit(time_horizon,number_of_arms,agent3)
results.append(mc_simulate(n_sim,bandit,"N=T/100"))
plot(results,time_horizon,params[i])
def experiment2():
params = [{
"time_horizon" : 500,
"number_of_arms" : 5
},
{
"time_horizon" : 5000,
"number_of_arms" : 5
},
{
"time_horizon" : 500,
"number_of_arms" : 10
},
{
"time_horizon" : 5000,
"number_of_arms" : 10
},
{
"time_horizon" : 500,
"number_of_arms" : 20
},
{
"time_horizon" : 5000,
"number_of_arms" : 20
},
]
for i in range(len(params)):
results = []
time_horizon = params[i]["time_horizon"]
number_of_arms = params[i]["number_of_arms"]
agent1 = agentFactory("random",time_horizon,number_of_arms)
epsilons = []
for j in range(time_horizon):
epsilons.append(math.pow((j+1)*number_of_arms*math.log(j+1),1/3))
agent2 = agentFactory("epsilon-greedy",time_horizon,number_of_arms,epsilons)
agent3 = agentFactory("explore-then-exploit",time_horizon,number_of_arms,time_horizon/100)
agent4 = agentFactory("ucb1",time_horizon,number_of_arms)
agent5 = agentFactory("successive-elimination",time_horizon,number_of_arms)
bandit = Bandit(time_horizon,number_of_arms,agent1)
results.append(mc_simulate(n_sim,bandit))
bandit = Bandit(time_horizon,number_of_arms,agent2)
results.append(mc_simulate(n_sim,bandit))
bandit = Bandit(time_horizon,number_of_arms,agent3)
results.append(mc_simulate(n_sim,bandit,"N=T/100"))
bandit = Bandit(time_horizon,number_of_arms,agent4)
results.append(mc_simulate(n_sim,bandit))
bandit = Bandit(time_horizon,number_of_arms,agent5)
results.append(mc_simulate(n_sim,bandit))
plot(results,time_horizon,params[i])