-
-
Notifications
You must be signed in to change notification settings - Fork 56
/
testing_(HumanvsAI)_ReinforcementLearning.py
159 lines (133 loc) · 6.09 KB
/
testing_(HumanvsAI)_ReinforcementLearning.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
import numpy as np
from math import inf as infinity
import itertools
import random
game_state = [[' ',' ',' '],
[' ',' ',' '],
[' ',' ',' ']]
players = ['X','O']
def play_move(state, player, block_num):
if state[int((block_num-1)/3)][(block_num-1)%3] is ' ':
state[int((block_num-1)/3)][(block_num-1)%3] = player
else:
block_num = int(input("Block is not empty, ya blockhead! Choose again: "))
play_move(state, player, block_num)
def copy_game_state(state):
new_state = [[' ',' ',' '],[' ',' ',' '],[' ',' ',' ']]
for i in range(3):
for j in range(3):
new_state[i][j] = state[i][j]
return new_state
def check_current_state(game_state):
# Check horizontals
if (game_state[0][0] == game_state[0][1] and game_state[0][1] == game_state[0][2] and game_state[0][0] is not ' '):
return game_state[0][0], "Done"
if (game_state[1][0] == game_state[1][1] and game_state[1][1] == game_state[1][2] and game_state[1][0] is not ' '):
return game_state[1][0], "Done"
if (game_state[2][0] == game_state[2][1] and game_state[2][1] == game_state[2][2] and game_state[2][0] is not ' '):
return game_state[2][0], "Done"
# Check verticals
if (game_state[0][0] == game_state[1][0] and game_state[1][0] == game_state[2][0] and game_state[0][0] is not ' '):
return game_state[0][0], "Done"
if (game_state[0][1] == game_state[1][1] and game_state[1][1] == game_state[2][1] and game_state[0][1] is not ' '):
return game_state[0][1], "Done"
if (game_state[0][2] == game_state[1][2] and game_state[1][2] == game_state[2][2] and game_state[0][2] is not ' '):
return game_state[0][2], "Done"
# Check diagonals
if (game_state[0][0] == game_state[1][1] and game_state[1][1] == game_state[2][2] and game_state[0][0] is not ' '):
return game_state[1][1], "Done"
if (game_state[2][0] == game_state[1][1] and game_state[1][1] == game_state[0][2] and game_state[2][0] is not ' '):
return game_state[1][1], "Done"
# Check if draw
draw_flag = 0
for i in range(3):
for j in range(3):
if game_state[i][j] is ' ':
draw_flag = 1
if draw_flag is 0:
return None, "Draw"
return None, "Not Done"
def print_board(game_state):
print('----------------')
print('| ' + str(game_state[0][0]) + ' || ' + str(game_state[0][1]) + ' || ' + str(game_state[0][2]) + ' |')
print('----------------')
print('| ' + str(game_state[1][0]) + ' || ' + str(game_state[1][1]) + ' || ' + str(game_state[1][2]) + ' |')
print('----------------')
print('| ' + str(game_state[2][0]) + ' || ' + str(game_state[2][1]) + ' || ' + str(game_state[2][2]) + ' |')
print('----------------')
# Initialize state values
player = ['X','O',' ']
states_dict = {}
all_possible_states = [[list(i[0:3]),list(i[3:6]),list(i[6:10])] for i in itertools.product(player, repeat = 9)]
n_states = len(all_possible_states) # 2 players, 9 spaces
n_actions = 9 # 9 spaces
state_values_for_AI = np.full((n_states),0.0)
print("n_states = %i \nn_actions = %i"%(n_states, n_actions))
for i in range(n_states):
states_dict[i] = all_possible_states[i]
winner, _ = check_current_state(states_dict[i])
if winner == 'O': # AI won
state_values_for_AI[i] = 1
elif winner == 'X': # AI lost
state_values_for_AI[i] = -1
def update_state_value(curr_state_idx, next_state_idx, learning_rate):
new_value = state_values_for_AI[curr_state_idx] + learning_rate*(state_values_for_AI[next_state_idx] - state_values_for_AI[curr_state_idx])
state_values_for_AI[curr_state_idx] = new_value
def getBestMove(state, player):
'''
Reinforcement Learning Algorithm
'''
moves = []
curr_state_values = []
empty_cells = []
for i in range(3):
for j in range(3):
if state[i][j] is ' ':
empty_cells.append(i*3 + (j+1))
for empty_cell in empty_cells:
moves.append(empty_cell)
new_state = copy_game_state(state)
play_move(new_state, player, empty_cell)
next_state_idx = list(states_dict.keys())[list(states_dict.values()).index(new_state)]
curr_state_values.append(state_values_for_AI[next_state_idx])
print('Possible moves = ' + str(moves))
print('Move values = ' + str(curr_state_values))
best_move_idx = np.argmax(curr_state_values)
best_move = moves[best_move_idx]
return best_move
# PLaying
#LOAD TRAINED STATE VALUES
state_values_for_AI = np.loadtxt('trained_state_values_O.txt', dtype=np.float64)
play_again = 'Y'
while play_again == 'Y' or play_again == 'y':
game_state = [[' ',' ',' '],
[' ',' ',' '],
[' ',' ',' ']]
current_state = "Not Done"
print("\nNew Game!")
print_board(game_state)
player_choice = input("Choose which player goes first - X (You - the petty human) or O(The mighty AI): ")
winner = None
if player_choice == 'X' or player_choice == 'x':
current_player_idx = 0
else:
current_player_idx = 1
while current_state == "Not Done":
curr_state_idx = list(states_dict.keys())[list(states_dict.values()).index(game_state)]
if current_player_idx == 0: # Human's turn
block_choice = int(input("Oye Human, your turn! Choose where to place (1 to 9): "))
play_move(game_state ,players[current_player_idx], block_choice)
else: # AI's turn
block_choice = getBestMove(game_state, players[current_player_idx])
play_move(game_state ,players[current_player_idx], block_choice)
print("AI plays move: " + str(block_choice))
print_board(game_state)
winner, current_state = check_current_state(game_state)
if winner is not None:
print(str(winner) + " won!")
else:
current_player_idx = (current_player_idx + 1)%2
if current_state is "Draw":
print("Draw!")
play_again = input('Wanna try again?(Y/N) : ')
print('GG!')