This repository has been archived by the owner on Oct 6, 2021. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 0
/
game_agent.py
368 lines (296 loc) · 12.8 KB
/
game_agent.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
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
"""This file contains all the classes you must complete for this project.
You can use the test cases in agent_test.py to help during development, and
augment the test suite with your own test cases to further test your code.
You must test your agent's strength against a set of agents with known
relative strength using tournament.py and include the results in your report.
"""
import random
class Timeout(Exception):
"""Subclass base exception for code clarity."""
pass
def heuristic1(game, player):
"""Heuristic #1
Multiply opponent's moves by 2 if legal moves still available
Returns
-------
float
A score as a float value
"""
if game.is_loser(player):
return float("-inf")
if game.is_winner(player):
return float("inf")
own_moves = len(game.get_legal_moves(player))
opp_moves = len(game.get_legal_moves(game.get_opponent(player)))
score = float(own_moves - (2 * opp_moves))
return score
def heuristic2(game, player):
"""Heuristic #2
Multiply opponent's moves by .5 if legal moves still available
Returns
-------
float
A score as a float value
"""
if game.is_loser(player):
return float("-inf")
if game.is_winner(player):
return float("inf")
own_moves = len(game.get_legal_moves(player))
opp_moves = len(game.get_legal_moves(game.get_opponent(player)))
score = float(own_moves - (.5 * opp_moves))
return score
def heuristic3(game, player):
"""Heuristic #3
Difference of player's and opponent's move(s), then divide total by all remaining legal moves
Parameters
----------
game : `isolation.Board`
An instance of `isolation.Board` encoding the current state of the
game (e.g., player locations and blocked cells).
player : object
A player instance in the current game (i.e., an object corresponding to
one of the player objects `game.__player_1__` or `game.__player_2__`.)
Returns
-------
float
A score as a float value
"""
if game.is_loser(player):
return float("-inf")
if game.is_winner(player):
return float("inf")
own_moves = len(game.get_legal_moves(player))
opp_moves = len(game.get_legal_moves(game.get_opponent(player)))
score = float((own_moves - opp_moves) / (own_moves+opp_moves))
return score
def custom_score(game, player):
"""Calculate the heuristic value of a game state from the point of view
of the given player.
Note: this function should be called from within a Player instance as
`self.score()` -- you should not need to call this function directly.
Parameters
----------
game : `isolation.Board`
An instance of `isolation.Board` encoding the current state of the
game (e.g., player locations and blocked cells).
player : object
A player instance in the current game (i.e., an object corresponding to
one of the player objects `game.__player_1__` or `game.__player_2__`.)
Returns
-------
float
The heuristic value of the current game state to the specified player.
"""
# TODO: finish this function!
#Use heuristic 1
# return heuristic1(game, player)
#Use heuristic 2
# return heuristic2(game, player)
#Use heuristic 3
return heuristic3(game, player)
class CustomPlayer:
"""Game-playing agent that chooses a move using your evaluation function
and a depth-limited minimax algorithm with alpha-beta pruning. You must
finish and test this player to make sure it properly uses minimax and
alpha-beta to return a good move before the search time limit expires.
Parameters
----------
search_depth : int (optional)
A strictly positive integer (i.e., 1, 2, 3,...) for the number of
layers in the game tree to explore for fixed-depth search. (i.e., a
depth of one (1) would only explore the immediate sucessors of the
current state.) This parameter should be ignored when iterative = True.
score_fn : callable (optional)
A function to use for heuristic evaluation of game states.
iterative : boolean (optional)
Flag indicating whether to perform fixed-depth search (False) or
iterative deepening search (True). When True, search_depth should
be ignored and no limit to search depth.
method : {'minimax', 'alphabeta'} (optional)
The name of the search method to use in get_move().
timeout : float (optional)
Time remaining (in milliseconds) when search is aborted. Should be a
positive value large enough to allow the function to return before the
timer expires.
"""
def __init__(self, search_depth=3, score_fn=custom_score,
iterative=True, method='minimax', timeout=10.):
self.search_depth = search_depth
self.iterative = iterative
self.score = score_fn
self.method = method
self.time_left = None
self.TIMER_THRESHOLD = timeout
def get_move(self, game, legal_moves, time_left):
"""Search for the best move from the available legal moves and return a
result before the time limit expires.
This function must perform iterative deepening if self.iterative=True,
and it must use the search method (minimax or alphabeta) corresponding
to the self.method value.
**********************************************************************
NOTE: If time_left < 0 when this function returns, the agent will
forfeit the game due to timeout. You must return _before_ the
timer reaches 0.
**********************************************************************
Parameters
----------
game : `isolation.Board`
An instance of `isolation.Board` encoding the current state of the
game (e.g., player locations and blocked cells).
legal_moves : list<(int, int)>
DEPRECATED -- This argument will be removed in the next release
time_left : callable
A function that returns the number of milliseconds left in the
current turn. Returning with any less than 0 ms remaining forfeits
the game.
Returns
-------
(int, int)
Board coordinates corresponding to a legal move; may return
(-1, -1) if there are no available legal moves.
"""
self.time_left = time_left
# TODO: finish this function!
# Implemented by Tom M
# set which search method to call in try block for best move found
search_method = self.minimax if self.method is 'minimax' else self.alphabeta
# Perform any required initializations, including selecting an initial
# move from the game board (i.e., an opening book), or returning
# immediately if there are no legal moves
try:
# The search method call (alpha beta or minimax) should happen in
# here in order to avoid timeout. The try/except block will
# automatically catch the exception raised by the search method
# when the timer gets close to expiring
# TODO: Tom M - check maximizing_player arg for each search function
if self.iterative:
depth = 0
while True:
depth += 1
_, cur_best_move = search_method(game, depth) # _ is discarded
else:
_, cur_best_move = search_method(game, self.search_depth) # _ is discarded
except Timeout:
# Handle any actions required at timeout, if necessary
return cur_best_move
# Return the best move from the last completed search iteration
return cur_best_move
def minimax(self, game, depth, maximizing_player=True):
"""Implement the minimax search algorithm as described in the lectures.
Parameters
----------
game : isolation.Board
An instance of the Isolation game `Board` class representing the
current game state
depth : int
Depth is an integer representing the maximum number of plies to
search in the game tree before aborting
maximizing_player : bool
Flag indicating whether the current search depth corresponds to a
maximizing layer (True) or a minimizing layer (False)
Returns
-------
float
The score for the current search branch
tuple(int, int)
The best move for the current branch; (-1, -1) for no legal moves
Notes
-----
(1) You MUST use the `self.score()` method for board evaluation
to pass the project unit tests; you cannot call any other
evaluation function directly.
"""
if self.time_left() < self.TIMER_THRESHOLD:
raise Timeout()
# TODO: finish this function!
# Tom M Implementation
if maximizing_player: # Not using this arg, logic in max and min functions
return self.maxvalue(game, depth)
else:
return self.minvalue(game, depth)
def minvalue(self, game, depth):
possible_moves = {}
if len(game.get_legal_moves()) == 0:
return (float('inf'), (-1, -1))
for move in game.get_legal_moves():
game_copy = game.forecast_move(move)
if depth == 1:
possible_moves[move] = self.score(game_copy, game.inactive_player)
else:
possible_moves[move] = self.maxvalue(game_copy, depth - 1)[0]
best_move = min(possible_moves, key=possible_moves.get)
return (possible_moves[best_move], best_move)
def maxvalue(self, game, depth):
possible_moves = {}
if len(game.get_legal_moves()) == 0:
return (float('-inf'), (-1, -1))
for move in game.get_legal_moves():
game_copy = game.forecast_move(move)
if depth == 1:
possible_moves[move] = self.score(game_copy, game.active_player)
else:
possible_moves[move] = self.minvalue(game_copy, depth - 1)[0]
best_move = max(possible_moves, key=possible_moves.get)
return (possible_moves[best_move], best_move)
def alphabeta(self, game, depth, alpha=float("-inf"), beta=float("inf"), maximizing_player=True):
"""Implement minimax search with alpha-beta pruning as described in the
lectures.
Parameters
----------
game : isolation.Board
An instance of the Isolation game `Board` class representing the
current game state
depth : int
Depth is an integer representing the maximum number of plies to
search in the game tree before aborting
alpha : float
Alpha limits the lower bound of search on minimizing layers
beta : float
Beta limits the upper bound of search on maximizing layers
maximizing_player : bool
Flag indicating whether the current search depth corresponds to a
maximizing layer (True) or a minimizing layer (False)
Returns
-------
float
The score for the current search branch
tuple(int, int)
The best move for the current branch; (-1, -1) for no legal moves
Notes
-----
(1) You MUST use the `self.score()` method for board evaluation
to pass the project unit tests; you cannot call any other
evaluation function directly.
"""
if self.time_left() < self.TIMER_THRESHOLD:
raise Timeout()
# TODO: finish this function!
# Tom M. Implementation
legal_moves = game.get_legal_moves()
if depth == 0 or not legal_moves:
return self.score(game, self), (-1, -1)
if maximizing_player:
max_score = float('-inf')
max_m = None
for move in legal_moves:
score = self.alphabeta(game.forecast_move(move), depth - 1, alpha, beta, False)[0]
if score > max_score:
max_score = score
max_m = move
if score >= beta: # prune node
return score, move
alpha = max(alpha, score)
return max_score, max_m
else: # min node
min_score = float('inf')
min_m = None
for move in legal_moves:
score = self.alphabeta(game.forecast_move(move), depth - 1, alpha, beta, True)[0]
if score < min_score:
min_score = score
min_m = move
if score <= alpha: # prune node
return score, move
beta = min(beta, score)
return min_score, min_m