-
Notifications
You must be signed in to change notification settings - Fork 0
/
play_matrix_game.py
132 lines (105 loc) · 4.24 KB
/
play_matrix_game.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
import json
from players import *
from game import *
from collections import defaultdict
import random
cooperative = open('LOLA.json', 'r').read()
defective = open('Q-Learning.json', 'r').read()
coop_player_network_load = json.loads(cooperative)
defect_player_network_load = json.loads(defective)
num_coop_network_states = 0
coop_player_network = defaultdict(lambda: random.random())
for key in coop_player_network_load:
num_coop_network_states += 1
formatted_key = key.split("-")
s = formatted_key[0]
a = formatted_key[1]
coop_player_network[s, a] = coop_player_network_load[key]
num_defect_network_states = 0
defect_player_network = defaultdict(lambda: random.random())
for key in defect_player_network_load:
num_defect_network_states += 1
formatted_key = key.split("-")
s = formatted_key[0]
a = formatted_key[1]
defect_player_network[s, a] = defect_player_network_load[key]
print "Cooperative Network has {} Q-values".format(num_coop_network_states)
print "Defective Network has {} Q-values".format(num_defect_network_states)
game_params = {
'invade_bonus': 10,
'invaded_penalty': -25,
'farming': True,
}
n_steps = 100000
board_size = 3
num_players = 2
defer_is_legal = True
setups = [
["Cooperative", "Cooperative"],
["Defective", "Defective"],
["Defective", "Cooperative"],
]
num_dilemmas = 10
dilemmas = []
while num_dilemmas > 0:
matchups = {}
for i, players in enumerate(setups):
rewards = {players[0]: 0.0, players[1]: 0.0}
player_ids = ["P0", "P1"]
print "Playing game {} vs {} now...".format(players[0], players[1])
game = StateOfNature(board_size, game_params, num_players)
player_types = []
if players[0] == "Cooperative":
coop_player = LOLAPlayer("P0", game.get_actions())
coop_player.set_Q(coop_player_network)
player_types.append(coop_player)
else:
defect_player = QLPlayer("P0", game.get_actions())
defect_player.set_Q(defect_player_network)
player_types.append(defect_player)
if players[1] == "Cooperative":
coop_player = LOLAPlayer("P1", game.get_actions())
coop_player.set_Q(coop_player_network)
player_types.append(coop_player)
else:
defect_player = QLPlayer("P1", game.get_actions())
defect_player.set_Q(defect_player_network)
player_types.append(defect_player)
last_two_moves = []
for step in range(1, n_steps + 1):
cur_state = game.get_cur_state()
cur_state = cur_state[:]
turn = game.get_cur_turn()
player_name = players[turn]
player = player_types[turn]
player_id = player_ids[turn]
if len(last_two_moves) == 2:
if last_two_moves.count("defer") == 2:
rewards[players[0]] += 25
rewards[players[1]] += 25
last_two_moves = []
a = player.act(cur_state, player_ids, board_size, defer_is_legal)
r, state_next = game.move(a)
rewards[player_name] += r
matchups[(players[0], players[1])] = \
(rewards[players[0]] / n_steps, rewards[players[1]] / n_steps)
if players[0] == players[1]:
matchups[(players[0], players[1])] = (matchups[(players[0], players[1])][0] / 2.0, matchups[(players[0], players[1])][1] / 2.0)
r = matchups[("Cooperative", "Cooperative")][0]
t = max(matchups[("Defective", "Cooperative")])
s = min(matchups[("Defective", "Cooperative")])
p = matchups[("Defective", "Defective")][0]
greed = t - r
fear = p - s
criteria = (r > p) and (2 * r > t + s) and ((t > r) or (p > s))
if criteria:
dilemmas.append((greed, fear))
num_dilemmas -= 1
print "---------------------------------------------"
print " C D "
print " C {} {} ".format((r, r), (s, t))
print " D {} {} ".format((t, s), (p, p))
print "---------------------------------------------"
print " Fear: {} Greed: {}".format(fear, greed)
print " Social Dilemma Criteria met: {}\n".format(criteria)
print dilemmas