-
Notifications
You must be signed in to change notification settings - Fork 0
/
game_agent.py
executable file
·688 lines (544 loc) · 25.7 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
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
"""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
import math
class Timeout(Exception):
"""Subclass base exception for code clarity."""
pass
def distance(p0, p1):
return math.sqrt((p0[0] - p1[0])**2 + (p0[1] - p1[1])**2)
def custom_score(game, player):
"""The "Improved" evaluation function discussed in lecture that outputs a
score equal to the difference in the number of moves available to the
two players.
Parameters
----------
game : `isolation.Board`
An instance of `isolation.Board` encoding the current state of the
game (e.g., player locations and blocked cells).
player : hashable
One of the objects registered by the game object as a valid player.
(i.e., `player` should be either game.__player_1__ or
game.__player_2__).
Returns
----------
float
The heuristic value of the current game state
"""
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)))
if(hasattr(player, 'q1')):
available_spaces = game.get_blank_spaces()
available_q1 = len([x for x in available_spaces if x in player.q1])
available_q2 = len([x for x in available_spaces if x in player.q2])
available_q3 = len([x for x in available_spaces if x in player.q3])
available_q4 = len([x for x in available_spaces if x in player.q4])
better_place = max([(available_q1, player.b1), (available_q2, player.b2), (available_q3, player.b3), (available_q4, player.b4)])
return float(own_moves - opp_moves) + \
distance(better_place[1], game.get_player_location(game.get_opponent(player))) - \
distance(better_place[1], game.get_player_location(player))
else:
return float(own_moves - opp_moves)
def custom_score_center(game, player):
"""The "Improved" evaluation function discussed in lecture that outputs a
score equal to the difference in the number of moves available to the
two players.
Parameters
----------
game : `isolation.Board`
An instance of `isolation.Board` encoding the current state of the
game (e.g., player locations and blocked cells).
player : hashable
One of the objects registered by the game object as a valid player.
(i.e., `player` should be either game.__player_1__ or
game.__player_2__).
Returns
----------
float
The heuristic value of the current game state
"""
if game.is_loser(player):
return float("-inf")
if game.is_winner(player):
return float("inf")
d = distance(game.get_player_location(player), (float(game.height)/2, float(game.width)/2))
d2 = distance(game.get_player_location(game.get_opponent(player)), (float(game.height)/2, float(game.width)/2))
return float(len(game.get_legal_moves(player)) - \
len(game.get_legal_moves(game.get_opponent(player)))) - d + d2
def custom_score_distance(game, player):
"""The "Improved" evaluation function discussed in lecture that outputs a
score equal to the difference in the number of moves available to the
two players.
Parameters
----------
game : `isolation.Board`
An instance of `isolation.Board` encoding the current state of the
game (e.g., player locations and blocked cells).
player : hashable
One of the objects registered by the game object as a valid player.
(i.e., `player` should be either game.__player_1__ or
game.__player_2__).
Returns
----------
float
The heuristic value of the current game state
"""
if game.is_loser(player):
return float("-inf")
if game.is_winner(player):
return float("inf")
d = distance(game.get_player_location(player), game.get_player_location(game.get_opponent(player)))
return d + float(len(game.get_legal_moves(player)) - len(game.get_legal_moves(game.get_opponent(player))))
def custom_score_h1_full_distance(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!
if game.is_loser(player):
return float("-inf")
if game.is_winner(player):
return float("inf")
available_spaces = game.get_blank_spaces()
available_q1 = len([x for x in available_spaces if x in player.q1])
available_q2 = len([x for x in available_spaces if x in player.q2])
available_q3 = len([x for x in available_spaces if x in player.q3])
available_q4 = len([x for x in available_spaces if x in player.q4])
better_place = max([(available_q1, player.b1), (available_q2, player.b2), (available_q3, player.b3), (available_q4, player.b4)])
own_moves = len(game.get_legal_moves(player))
opp_moves = len(game.get_legal_moves(game.get_opponent(player)))
"""return float(own_moves - opp_moves) + \
distance(better_place[1], game.get_player_location(game.get_opponent(player))) - \
distance(better_place[1], game.get_player_location(player))"""
return distance(better_place[1], game.get_player_location(game.get_opponent(player))) - \
distance(better_place[1], game.get_player_location(player))
def custom_score_h1_full_mixed(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!
if game.is_loser(player):
return float("-inf")
if game.is_winner(player):
return float("inf")
available_spaces = game.get_blank_spaces()
available_q1 = len([x for x in available_spaces if x in player.q1])
available_q2 = len([x for x in available_spaces if x in player.q2])
available_q3 = len([x for x in available_spaces if x in player.q3])
available_q4 = len([x for x in available_spaces if x in player.q4])
better_place = max([(available_q1, player.b1), (available_q2, player.b2), (available_q3, player.b3), (available_q4, player.b4)])
own_moves = len(game.get_legal_moves(player))
opp_moves = len(game.get_legal_moves(game.get_opponent(player)))
return float(own_moves - opp_moves) + \
distance(better_place[1], game.get_player_location(game.get_opponent(player))) - \
distance(better_place[1], game.get_player_location(player))
def custom_score_h1_begin_mixed(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!
if game.is_loser(player):
return float("-inf")
if game.is_winner(player):
return float("inf")
available_spaces = game.get_blank_spaces()
own_moves = len(game.get_legal_moves(player))
opp_moves = len(game.get_legal_moves(game.get_opponent(player)))
if(len(available_spaces) >= (game.width * game.height) / 2):
available_q1 = len([x for x in available_spaces if x in player.q1])
available_q2 = len([x for x in available_spaces if x in player.q2])
available_q3 = len([x for x in available_spaces if x in player.q3])
available_q4 = len([x for x in available_spaces if x in player.q4])
better_place = max([(available_q1, player.b1), (available_q2, player.b2), (available_q3, player.b3), (available_q4, player.b4)])
return float(own_moves - opp_moves) + \
distance(better_place[1], game.get_player_location(game.get_opponent(player))) - \
distance(better_place[1], game.get_player_location(player))
else:
return float(own_moves - opp_moves)
def custom_score_h1_begin_distance(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!
if game.is_loser(player):
return float("-inf")
if game.is_winner(player):
return float("inf")
available_spaces = game.get_blank_spaces()
own_moves = len(game.get_legal_moves(player))
opp_moves = len(game.get_legal_moves(game.get_opponent(player)))
if(len(available_spaces) >= (game.width * game.height) / 2):
available_q1 = len([x for x in available_spaces if x in player.q1])
available_q2 = len([x for x in available_spaces if x in player.q2])
available_q3 = len([x for x in available_spaces if x in player.q3])
available_q4 = len([x for x in available_spaces if x in player.q4])
better_place = max([(available_q1, player.b1), (available_q2, player.b2), (available_q3, player.b3), (available_q4, player.b4)])
return distance(better_place[1], game.get_player_location(game.get_opponent(player))) - \
distance(better_place[1], game.get_player_location(player))
else:
return float(own_moves - opp_moves)
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.)
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).
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
self.q1 = None
self.q2 = None
self.q3 = None
self.q4 = None
self.b1 = None
self.b2 = None
self.b3 = None
self.b4 = None
def create_quarter(self, width, height):
split_width = int(width/2)
if( (width % 2) == 1): split_width = int(width/2) +1
split_height = int(height/2)
if( (height % 2) == 1): split_height = int(height/2) +1
q1 = [(i,j) for i in range(0, split_width) for j in range(0, split_height)]
q3 = [(i,j) for i in range(int(width/2), width) for j in range(0, split_height)]
q2 = [(i,j) for i in range(0, split_width) for j in range(int(height/2), height)]
q4 = [(i,j) for i in range(int(width/2), width) for j in range(int(height/2), height)]
b1 = tuple([sum(x) for x in zip(*q1)])
b2 = tuple([sum(x) for x in zip(*q2)])
b3 = tuple([sum(x) for x in zip(*q3)])
b4 = tuple([sum(x) for x in zip(*q4)])
b1 = tuple([x/len(q1) for x in b1])
b2 = tuple([x/len(q2) for x in b2])
b3 = tuple([x/len(q3) for x in b3])
b4 = tuple([x/len(q4) for x in b4])
return q1, q2, q3, q4, b1, b2, b3, b4
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)>
A list containing legal moves. Moves are encoded as tuples of pairs
of ints defining the next (row, col) for the agent to occupy.
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!
# 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
# create positions for 4 quarter
if(self.q1 == None):
self.q1, self.q2, self.q3, self.q4, self.b1, self.b2, self.b3, self.b4 = \
self.create_quarter(game.width, game.height)
if(len(legal_moves) == (game.width * game.height)):
# opening move of player 1: take the center of the board
return ( int(game.width/2), int(game.height/2))
if(len(legal_moves) == (game.width * game.height - 1)):
# opening move of player 2: not required to be implemented for the project submission
return (0,0)
if(len(legal_moves) == 0):
# end game
return (-1,-1)
else:
best_move = legal_moves[0]
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
if(self.iterative):
for idx in range(1, game.width * game.height):
if(self.method == 'minimax'):
score, best_move = self.minimax(game, idx)
if(self.method == 'alphabeta'):
score, best_move = self.alphabeta(game, idx)
else:
depth = game.width * game.height # max possible depth
if(self.search_depth >= 1): depth = self.search_depth
if(self.method == 'minimax'):
score, best_move = self.minimax(game, depth)
if(self.method == 'alphabeta'):
score, best_move = self.alphabeta(game, depth)
except Timeout:
# Handle any actions required at timeout, if necessary
if( (len(legal_moves) > 0) and (best_move == (-1, -1))):
print('Timeout error condition')
return best_move
# Return the best move from the last completed search iteration
if( (len(legal_moves) > 0) and (best_move == (-1, -1))):
print('Normal error condition ', self.iterative)
return best_move
def minimax_max(self, game, depth, maximizing_player=True):
if self.time_left() < self.TIMER_THRESHOLD:
raise Timeout()
# exit condition1 : leaf node
if(len(game.get_legal_moves()) == 0):
exit_score = self.score(game, self)
if(exit_score == float('inf')):
return exit_score, game.get_player_location(self)
else:
return exit_score, (-1, -1)
#exit condition 2 : max fixed depth reached
if(depth == 0):
return self.score(game, self), game.get_player_location(self)
score = float("-inf")
move = game.get_player_location(self)
legal_move = game.get_legal_moves()
for m in legal_move:
new_score, notmove = self.minimax_min(game.forecast_move(m), depth - 1, not maximizing_player)
if( (new_score > score) or (new_score == float('inf'))):
score = new_score
move = m
return score, move
def minimax_min(self, game, depth, maximizing_player=True):
if self.time_left() < self.TIMER_THRESHOLD:
raise Timeout()
# exit condition 1 : leaf node
if(len(game.get_legal_moves()) == 0):
exit_score = self.score(game, self)
if(exit_score == float('inf')):
return exit_score, game.get_player_location(self)
else:
return exit_score, (-1, -1)
#exit condition 2 : max fixed depth reached
if(depth == 0):
return self.score(game, self), game.get_player_location(self)
score = float("inf")
move = game.get_player_location(self)
legal_move = game.get_legal_moves()
for m in legal_move:
new_score, notmove = self.minimax_max(game.forecast_move(m), depth - 1, not maximizing_player)
if( (new_score < score) or (new_score == float('-inf'))):
score = new_score
move = m
return score, 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()
if(len(game.get_legal_moves()) == 0):
exit_score = self.score(game, self)
if(exit_score == float('inf')):
return exit_score, game.get_player_location(self)
else:
return exit_score, (-1, -1)
# TODO: finish this function!
return self.minimax_max(game, depth, maximizing_player)
def alphabeta_max(self, game, depth, alpha, beta, maximizing_player=True):
if self.time_left() < self.TIMER_THRESHOLD:
raise Timeout()
# exit condition 1 : leaf node
if(len(game.get_legal_moves()) == 0):
exit_score = self.score(game, self)
if(exit_score == float('inf')):
return exit_score, game.get_player_location(self)
else:
return exit_score, (-1, -1)
#exit condition 2 : max fixed depth reached
if(depth == 0):
return self.score(game, self), game.get_player_location(self)
score = float("-inf")
move = game.get_player_location(self)
legal_move = game.get_legal_moves()
for m in legal_move:
new_score, notmove = self.alphabeta_min(game.forecast_move(m), depth - 1, alpha, beta, not maximizing_player)
if( (new_score > score) or (new_score == float('inf'))):
score = new_score
move = m
if(score >= beta):
return score, move
alpha = max(alpha, score)
return score, move
def alphabeta_min(self, game, depth, alpha, beta, maximizing_player=True):
if self.time_left() < self.TIMER_THRESHOLD:
raise Timeout()
# exit condition 1 : leaf node
if(len(game.get_legal_moves()) == 0):
exit_score = self.score(game, self)
if(exit_score == float('inf')):
return exit_score, game.get_player_location(self)
else:
return exit_score, (-1, -1)
#exit condition 2 : max fixed depth reached
if(depth == 0):
return self.score(game, self), game.get_player_location(self)
score = float("inf")
move = game.get_player_location(self)
legal_move = game.get_legal_moves()
for m in legal_move:
new_score, notmove = self.alphabeta_max(game.forecast_move(m), depth - 1, alpha, beta, not maximizing_player)
if( (new_score < score) or (new_score == float('-inf'))):
score = new_score
move = m
if(score <= alpha):
return score, move
beta = min(beta, score)
return score, 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()
if(len(game.get_legal_moves()) == 0):
exit_score = self.score(game, self)
if(exit_score == float('inf')):
return exit_score, game.get_player_location(self)
else:
return exit_score, (-1, -1)
return self.alphabeta_max(game, depth, alpha, beta, maximizing_player)