def computer(board, o_s, turn_no):
if turn_no == 1:
if o_s[-1] in [(0, 0), (2, 2), (2, 0), (0, 2)]:
board[1][1] = 'X'
else:
i, j = minimax(board, 0, True, 8)[1:]
board[i][j] = 'X'
elif turn_no == 3:
s1 = ''.join(board[i][i] for i in range(3))
s2 = ''.join(board[i][2 - i] for i in range(3))
if any(i == 'OXO' for i in [s1, s2]):
board[2][1] = 'X'
else:
ch = checkConsecutive(board)
if ch != -1:
i, j = ch
else:
i, j = minimax(board, 0, True, 9 - turn_no - 1)[1:]
board[i][j] = 'X'
return board
else:
try:
ch = checkConsecutive(board)
if ch != -1:
i, j = ch
else:
i, j = minimax(board, 0, True, 9 - turn_no - 1)[1:]
board[i][j] = 'X'
return board
except:
print('Its a Draw')
return board
def run(self, agent):
'''
This function specifies the framework for the play -
changing turns and continuing the game with provided search.
'''
self.noOfLegalActionsPerPlay=[]
self.noOfFaceUpCardsPerPlay=[]
sequence_of_actions = [] # initializing the sequence of actions
opponent = self.get_enemy(agent) # getting the opponent
print("*********************************\nInitial Table: " + str(self.table)) # shows the dominoes on the initial table
print(str(agent.role) + " is playing first")
print(str(agent.role) + "'s hand: " + str(agent.hand)) # shows the initial dominoes in the current player's hand
print("Dominos " + str(agent.role) + " collected: " + str(agent.collected)) # shows the cards collected by the current player
print(str(opponent.role) + "'s hand: " + str(opponent.hand)) # the dominoes on the opponent's hand
print("Dominos " + str(opponent.role) + " collected: " + str(opponent.collected)) # shows the cards collected by the opponent
print("*********************************\n\n")
while not self.isGameOver():
best_action = minimax.minimax(self, agent) # identifying the best action according to the search algorithm
print("main_action1: " + str(best_action))
if len(best_action) > 1: # picked_last = the last player who performed an action
self.picked_last = agent.role
self.makeMove(best_action[0], best_action[1:], agent) # the agent performs the best option
self.noOfFaceUpCardsPerPlay.append(len(self.table))
self.print_board(agent) # prints the board after the move
sequence_of_actions.append(best_action)
# if the game is not over => switch to the other player
if not self.isGameOver():
best_action = minimax.minimax(self, opponent) # identifying the best action according to the search algorithm
print("main_action2: " + str(best_action))
if len(best_action) > 1:
self.picked_last = agent.role
self.makeMove(best_action[0], best_action[1:], opponent)
self.noOfFaceUpCardsPerPlay.append(len(self.table))
self.print_board(opponent)
sequence_of_actions.append(best_action)
else:
break
#print("Number of times legal actions generation was executed=%d" % globals.countLegalActionsGenerationExecution)
print("Score of the Max at the end:" + str(self.computeScore(agent)))
print("sequence of actions played:" + str(sequence_of_actions))
return sequence_of_actions
def test_time():
state = State()
start_time = time.time()
minimax_full(state)
execution_time = round(time.time() - start_time, 2)
print(f'Full minimax: {execution_time}s')
for i in range(1, 10):
start_time = time.time()
minimax(state, i)
execution_time = round(time.time() - start_time, 2)
print(f'Minimax depth = {i}: {execution_time}s')
def play(self, gameState, timeLimit):
nextStates = gameState.findNextStates()
max = minimax(nextStates[0], self.depth, True)
i = 0
rangMax = 0
for state in nextStates:
if minimax(state, self.depth, True) > max:
max = minimax(state, self.depth, True)
rangMax = i
i += 1
return (nextStates[rangMax])
#use minimax here to return the next state with higher score
raise NotImplementedError() # remove this error once it is done
def defendStrategy(self, Board, colour, oppo_colour, type, no_turns):
FIRST_SHRINK = 128
SECOND_SHRINK = 192
#weights for the evaluation function and depth
evalWeights = [3, 2, 0, 1, 1, 3]
depth = 3
#set distance line evaluating function to zero since its after placing phase
if (type == "MP"):
evalWeights[0] = 0
#Increase weights for eval from center when board get close to shrinking
if (no_turns <= FIRST_SHRINK and no_turns >= 96):
evalWeights[3] = evalWeights[3] + (0.6) * (no_turns - 96)
if (no_turns <= SECOND_SHRINK and no_turns >= 152):
evalWeights[3] = evalWeights[3] + (0.6) * (no_turns - 152)
#increase depth when we shrink for the last time
if (no_turns > SECOND_SHRINK):
depth = 4
#switch this evaluation if its in this number of ranges: done to keep
#time constraint
if (no_turns > 100 or no_turns < 20):
evalWeights[2] = 2
return (minimax.minimax(Board, colour, oppo_colour, type, evalWeights,
depth))
def play_ai():
print("AI IS PLAYING")
global board, turn
# score = 0
best_score = -inf
pos = None
pos_to_play = None
for i in range(8):
for j in range(8):
b = board[i][j]
if (b != ""):
if (b[0] != "w"):
score = minimax(board, i, j, 2, True)
# print(score)
if (score[1] > best_score):
pos = [i, j]
pos_to_play = [score[0][0], score[0][1]]
if (pos):
print(pos, pos_to_play)
mv = board[pos[0]][pos[1]]
print(mv)
if (rules(mv, pos[0], pos[1], pos_to_play[0], pos_to_play[1], board)):
board[pos[0]][pos[1]] = ''
board[pos_to_play[0]][pos_to_play[1]] = mv
initialize()
turn = "b" if turn == "w" else "w"
def main():
board = [
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
]
display(board)
while not mm.board_filled(board) and mm.over(board) == 0:
player_move = int(input("Move: ")) - 1
mm.move(player_move, 1, board) # fix
display(board)
if (mm.board_filled(board)) or mm.over(board) != 0:
break
print("Thinking...")
next_move = mm.minimax(-1, 1, board)
sleep(2)
mm.move(next_move % 7, -1, board) # fix
display(board)
print("Your Turn:")
def ai_decision_making():
ai_temp_hand = []
for i in ai_hand: #get playable tiles
if i.topval == tiles_on_field[0].openval:
ai_temp_hand.append(i)
if i.bottomval == tiles_on_field[0].openval:
ai_temp_hand.append(i)
if i.topval == tiles_on_field[-1].openval:
ai_temp_hand.append(i)
if i.bottomval == tiles_on_field[-1].openval:
ai_temp_hand.append(i)
if len(ai_temp_hand) == 0: #if none, get random tile
if len(set_of_tiles) > 0:
get_random_tile(0)
ai_decision_making()
else:
print("AI cannot get tile")
return
utility = -inf #utility value
utile = None #uitility tile
temp_field = tiles_on_field[:]
for t in ai_temp_hand: #minimax to make decision
maxi = minimax.minimax(temp_field,ai_hand,'a',field,set_of_tiles,t)
if maxi > utility:
utility = maxi
utile = t
player_place_tile(utile,ai_hand,0) #place tile
return
def findBestMove():
s.found_pattern = False
bestScore = -math.inf
bestMove = s.Vec2(0, 0)
for y in range(s.board.size):
for x in range(s.board.size):
if (s.board.state[y][x].val == '0'):
s.board.state[y][x].val = '1'
score = minimax(0, -math.inf, math.inf, False)
s.board.state[y][x].val = '0'
# if (x == 9 and y == 5):
# print("score when 9,5: ", score)
# print("best =", bestScore)
if (score > bestScore):
# print("new score:", score)
# print("x = " + str(x) + ", y = " + str(y))
bestScore = score
bestMove.x = x
bestMove.y = y
if (bestScore == 0 and s.found_pattern == False):
bestMove = s.board.randomCoord()
# print('random')
return bestMove
def minimax_move(self, depth=1, eval_func='pos_score'):
# return the move with max minimax score
# minimax(board, depth, player, alpha, beta) -> int:
move_eval_dict = {}
possible_moves = self.find_all_valid_moves()
if possible_moves:
for move in possible_moves:
game_copy = deepcopy(self)
game_copy.take_move(move[0], move[1])
move_eval_dict[move] = mm.minimax(game_copy.board,
depth=depth,
player=opposite(
self.current_player),
eval_func=eval_func)
move_eval_dict = mm.shuffle_dict(
move_eval_dict
) # shuffle the dict, or always choose the same move
if self.current_player == BLACK:
return max(move_eval_dict, key=move_eval_dict.get
) # return the move with max minimax score
else: # White is the minimizing player, the less the better.
return min(move_eval_dict, key=move_eval_dict.get)
# Initially when I adapted from Sebestian's Youtube code, I forgot the above two lines
# and Black wins 95% even white uses minimax and black uses 'random'
else:
return None
def test_bestposiblemove_number(supply_board_bestpossiblemove2):
import metainfo
metaObj = metainfo.metainfo(4)
import minimax
obj = minimax.minimax(metaObj)
board = supply_board_bestpossiblemove2
pos = obj.findBestMove(board, '2', '1')
assert pos == (2, 0), "Computer's Best Possible Move Test Case Failed"
def test_win_detection(self):
z = [0, 0, 1, 1, 2, 2] # black wins if plays in 3
for i in z:
self.board = self.board.move(i)
aiPlay = minimax.minimax(self.board,
maximizing=True,
original_player=connectfour.C4Piece.B,
max_depth=4)
self.assertEqual(aiPlay, 3)
def auto_alg():
init_game()
best_move = -2
game_over = False
depth = 4
start_timer = 0
end_timer = 0
timer_array = []
times_2048 = 0
while (not (game_over)):
start_timer = timer()
temp_board = []
temp_board = Game.board
empty_cells = []
false_counter = 0
for x in range(0, 4):
for y in range(0, 4):
if (temp_board[x][y] == 0):
empty_cells.append(0)
test_depth = len(empty_cells)
depth_map = [9, 9, 8, 7, 7, 7, 7, 6, 6, 6, 6, 6, 5, 4, 4, 4]
depth = depth_map[test_depth]
best_move = minimax.minimax(temp_board, depth, "PLAYER", -np.inf,
np.inf)[0]
Game.board = game_logic.main_loop(temp_board, best_move)[1]
print("New board with move: " + str(best_move))
print(Game.board)
for i in range(0, 4):
if (game_logic.main_loop(temp_board, i)[0] != True):
false_counter += 1
if (false_counter == 4):
print("NOOOOO GAME OVAAAHHR")
game_over = True
end_timer = timer()
print("TIME ELAPSED THIS MOVE:")
print(end_timer - start_timer)
print("-")
timer_array.append(end_timer - start_timer)
print("TOTAL TIME ELAPSED:")
print(sum(timer_array))
print("AVERAGE TIME EACH MOVE:")
print(np.mean(timer_array))
for x in range(0, 4):
for y in range(0, 4):
if (Game.board[x][y] >= 2048):
times_2048 += 1
return np.mean(timer_array), sum(timer_array), times_2048
def init_game_minimax(grid=1):
grid = gr.create_grid(grid)
start_location = gr.get_start_location()
game = Game(grid, start_location)
score, moves = minimax(game)
if score < game.max_steps():
print("Score:", score)
else:
print("Score: infinity")
return score, moves
def on_get(self, req, resp):
boardString = req.params['board']
player = int(req.params['player'])
board = list(map(int, boardString.split(', ')))
move = minimax(board, player)
resp.status = falcon.HTTP_200 # This is the default status
resp.body = json.dumps(move)
def player_machine_mini():
move = minimax(board)
print move
board[move[0]][move[1]] = 'M'
printBoard(board)
green_circle(150 + move[1] * 100, 400 - move[0] * 100)
if isGameOver(board)[0] == True:
print "Machine wins"
t.onscreenclick(None)
player = 'H'
def test(root, solution, maximize):
''' Test si minimax appliqué à root renvoie bien la solution '''
result = minimax(root, 10, maximize=maximize)
if not result == solution:
print(
str.format("Error lors du test (maximize = {:b}) sur l'arbre : ",
maximize))
print(root)
print('minimax returned : ', result)
print('but the solution was : ', solution)
raise Exception()
def get_ready_state(data):
board = data['board']
move = minimax(board, checkOpens(board), 20, True, data['player_turn_id'],
-1000, 1000, ['efficiency_heuristic'])[1]
sio.emit(
'play', {
'tournament_id': 5000,
'player_turn_id': data['player_turn_id'],
'game_id': data['game_id'],
'movement': move
})
def get_agent_piece_and_move(self):
'''
here we will call minimax from minimax.py. This will return the agent piece that will move along with the coordinate
it will move to
'''
start = time.time()
_, _, piece, pos = minimax(self, depth, float('-inf'), float('inf'),
True)
print('time passed:', time.time() - start)
return piece, pos
def massacre(self, Board, colour, oppo_colour, type, no_turns):
SECOND_SHRINK = 192
#weights for the evaluation function and depth
evalWeights = [0, 2, 0, 1, 0, 1]
depth = 3
#increase depth when we shrink for the last time
if (no_turns > SECOND_SHRINK):
depth = 4
return (minimax.minimax(Board, colour, oppo_colour, type, evalWeights,
depth))
def test_win_prevention(self):
self.board = self.board.move(3)
self.board = self.board.move(3)
self.board = self.board.move(4)
# In this position, black needs to occupy 2 or 5, otherwise black will occupy any of them
# and win next move. This is actually harder than preventing a direct win.
recommendation = minimax(self.board,
maximizing=True,
original_player=C4Piece.B,
max_depth=4)
self.assertIn(recommendation, [2, 5])
def minimaxStep(self, graph):
if len(self.strategy) < 2:
self.strategy = minimax(graph, self.currentPosition, self.otherPos,
self.limit, self.type)
if self.strategy == None:
print("No Strategy!")
return
# self.calcTime = exp * T
print("New Strategy: ", self.strategy)
if len(self.strategy) < 2:
return ""
self.strategy = self.strategy[1:]
# print('Next step: ',self.strategy[0])
return self.strategy[0]
def test_win_prevention(self):
"""
testing to see if the ai play is the right one to prevent the other player from wining
:return:
"""
z = [0, 0, 1, 1,
2] # the ai play must be 3 ot stop the other player from wining
for i in z:
self.board = self.board.move(i)
aiPlay = minimax.minimax(self.board,
maximizing=True,
original_player=connectfour.C4Piece.B,
max_depth=4)
self.assertEqual(aiPlay, 3)
def opponent_move(self):
"""
Computer movement - minimax algorithm. Changes the state of the object.
"""
print('\nKolej przeciwnika.')
best_payment = minimax(self.state, self.search_depth, self.successor_fun, self.heuristics)
next_states = self.state['next']
out_states = list()
for next_state in next_states:
if next_state['payment'] == best_payment:
out_states.append(next_state)
new_state = rd.choice(out_states)
self.state = new_state
def test_win_detection(self):
self.board = self.board.move(0)
self.board = self.board.move(1)
self.board = self.board.move(0)
self.board = self.board.move(1)
self.board = self.board.move(0)
self.board = self.board.move(1)
# Red will win if played in 0
recommendation = minimax(self.board,
maximizing=True,
original_player=C4Piece.B,
max_depth=4)
self.assertEqual(recommendation, 0)
def computer_player_turn(board, computer_player):
print("It is the computer's turn. Computer is making a move.")
print()
board.pretty_print()
(move, value, nodes_evaluated,
time_taken) = minimax(board, computer_player, 7)
board.place_marker(computer_player, move)
print(
str(nodes_evaluated) + " game states evaluated in " + str(time_taken) +
" seconds.")
print("Computer selected column " + str(move + 1) + " for its marker.")
print()
def ai_play(board: Board, search_depth: int = SEARCH_DEPTH):
best_move = None
best_move_score = None
for i in board.available_squares:
board_clone = copy.deepcopy(board)
board_clone.place_mark(i)
score = minimax(board_clone, search_depth)
if best_move is None:
best_move = i
best_move_score = score
else:
is_better = best_move_score < score if board.playing == Mark.NOUGHT else best_move_score > score
if is_better:
best_move = i
best_move_score = score
board.place_mark(best_move)
def __main__():
max_player = True
s = State(max_player)
while (True):
print("\n\ntwoj ruch!")
s = s.make_move()
s.show_board()
if game_over(s):
break
print("\n\nruch przeciwnika!")
U = s.get_successors()
index = minimax(s, 4, not max_player, 4)
s = State(not U[index].max_player, copy.deepcopy(U[index].board))
s.show_board()
if game_over(s):
break
def getScores(board, depth):
scores = []
for value in prange(BOARD_SIZE**2):
x = math.floor(value / 5)
y = value % 5
if board[x][y] == 0:
copiedArray = np.copy(board)
copiedArray[x][y] = 1
score = minimax(copiedArray, depth, -999999, 999999, False)
scores.append([x, y, score])
print(x, y, score)
return scores
def ready(data):
gameID = data['game_id']
playerTurnID = data['player_turn_id']
board = data['board']
movementCoordinates = [0, 0]
#print("Ready")
#print(board)
movementCoordinates = minimax(board, True)
sio.emit(
'play', {
'tournament_id': tournamentID,
'player_turn_id': playerTurnID,
'game_id': gameID,
'movement': movementCoordinates
})
def computerGame():
#--Make human move(s) and computer reply/replies.
copyOldBoardToScreenInMiniturizedForm(c,r)
move=minimax.minimax(COMPUTER, M, PW, PB, MAXDEPTH)
print(move)
makeMove(move[1], move[0], move[2], COMPUTER)
canvas.create_rectangle(655, 330, 870,370, width = 0, fill = 'grey30')
if legalMove(HUMAN) and not legalMove(COMPUTER): return
#--Make computer reply/replies (1 = BLACK = human, -1 = computer = WHITE)
if legalMove(COMPUTER): makeComputerReply() # <--This is the computer's strategy (IMPORTANT!).
while legalMove(COMPUTER) and not legalMove(HUMAN):
makeComputerReply()
#displayAllLegalMovesForHumanPlayer('BLACK')
if not legalMove(HUMAN) and not legalMove(COMPUTER): quit()
#-- Note: legal move for human must now necessarily exist.
return
def computer_play(self):
minimax.PLAYER = self.game.turn.color
print '------------------------------'
print 'executing minimax with:'
print self.game.board
print self.game.turn.color
print '------------------------------'
position = minimax.minimax(self.game.board,
self.difficulty,
self.game.turn.color,
self.heuristic)[1]
print 'minimax finished with choice: %s' % str(position)
self.game.play(position)
message = "%s's turn." % (self.game.turn.color)
self.update_status(message)
self.update_board()
self.update_score()
self.update_pass_turn()
self.check_next_turn()
def gameover(self, game, last_move):
pass
if __name__ == "__main__":
import othello
import minimax
# Experiment 1:
# Player 1 and Player 2 are evenly matched with 3-ply deep search
# player 2 wins with a final score of 28
# player 1 0.2 s per ply player 2 0.4 s per ply
play(othello.game(), player_epsilon(lambda x: minimax.minimax(x, 3)),
player_epsilon(lambda x: minimax.minimax(x, 3)), False)
# Experiment 2:
# now we show the significance of an evaluation function
# we weaken player1 to 2 ply deep but use the edge eval fun
# player 1 now beats player 2 with a score of 58!
# player 1 0.1 s per ply player 2 0.4 s per ply
# play(othello.game(), player(lambda x: minimax.minimax(x, 2, othello.edge_eval)),player(lambda x: minimax.minimax(x, 3)), False)
# Experiment 1 (with alpha-beta):
# player 1 0.1 s per ply, player 2 0.1 s per ply
# play(othello.game(), player(lambda x: minimax.alphabeta(x, 3)),player(lambda x: minimax.alphabeta(x, 3)), False)
# Experiment 2 (with alpha-beta):
# player 1 0.0 s per ply player 2 0.1 s per ply
circle = not circle
win[1][2] = first
print("H")
elif (3 * width) / 3 > mouse[0] > (2 * width) / 3 and (3 * height) / 3 > mouse[1] > (
2 * height) / 3 and 0 == win[2][2]:
pygame.draw.circle(tic, blue, i, rad, 7)
circle = not circle
win[2][2] = first
print("I")
if minimax.isGameOver(win) or minimax.isTie(win):
print("GAME OVER")
playing = False
if not circle and playing:
print("Xs Turn")
win = list(minimax.minimax(list(win), movefirst))
if win[0][0] == second:
pygame.draw.line(tic, red, (a[0] - xch, a[1] - xch), (a[0] + xch, a[1] + xch), 7)
pygame.draw.line(tic, red, (a[0] - xch, a[1] + xch), (a[0] + xch, a[1] - xch), 7)
print("A")
if win[1][0] == second:
pygame.draw.line(tic, red, (b[0] - xch, b[1] - xch), (b[0] + xch, b[1] + xch), 7)
pygame.draw.line(tic, red, (b[0] - xch, b[1] + xch), (b[0] + xch, b[1] - xch), 7)
print("B")
if win[2][0] == second:
pygame.draw.line(tic, red, (c[0] - xch, c[1] - xch), (c[0] + xch, c[1] + xch), 7)
pygame.draw.line(tic, red, (c[0] - xch, c[1] + xch), (c[0] + xch, c[1] - xch), 7)
print("C")
## Rich's playing around with functions....
#game2.play(othello.game(), game2.player(lambda x: minimax.minimax(x, 3)), game2.player(lambda x: random_move_policy(x)), True)
#t = UCT_tree.Tree(5, random_policy, 1)
#game2.play(othello.game(), game2.player(lambda x: minimax.minimax(x, 3)), game2.player(t.policy), True)
#game2.play(othello.game(), game2.player(t.policy), game2.player(lambda x: minimax.minimax(x, 4)), True)
policies = {"random": random_policy}#, "greedy": greedy_policy}
budgets = [1]
c_vals = [1, 5]#, 20, 50]
# opponents = {"random": game2.player(random_policy),
# "greedy": game2.player(greedy_policy),
# "minimax-2": game2.player(lambda x: minimax.minimax(x, 2)),
# "minimax-3": game2.player(lambda x: minimax.minimax(x, 3)),
# "minimax-4": game2.player(lambda x: minimax.minimax(x, 4))}
opponents = {"minimax-4": game2.player(lambda x: minimax.minimax(x, 4))}
# number of games going first/second with each configuration
n = 10
# run this algorithm with different hard-coded sizes
# for each policy
# for each budget (i.e. 1,2,5 seconds)
# for each opponent
# run n trials with us first, n trials with them first
for pol_key in policies:
for b in budgets:
for c in c_vals:
for opp_key in opponents:
t = UCT_tree.Tree(b, policies[pol_key], c)
uct_player = game2.player(t.policy)
def computerMove(self):
index = minimax.minimax(game, game.turn)
game.makeMove(index)
runs = 10
steps = np.zeros((episodes))
for i in range(runs):
print "run",i
steps += Qlearning(episodes, [(0,0)], epsGreedyPolicy,alpha_pred=0.2,alpha_prey=0.2)[0]
steps /= runs
steps = ndimage.filters.gaussian_filter(steps,4)
pl.plot(range(len(steps)), steps)
steps = np.zeros((episodes))
for i in range(runs):
print "run",i
steps += minimax(episodes,[(0,0)],0.1, 0.99999, 0.9)[0]
steps /= runs
steps = ndimage.filters.gaussian_filter(steps,4)
pl.plot(range(len(steps)), steps)
pl.legend(["Q-learning","Minimax"])
pl.title('Number of iterations per episode using Minimax and Q-learning.')
pl.ylabel('Iterations')
pl.xlabel('Episodes')
pl.savefig("plots/equalminimax.png")
pl.ylim(0,200)
pl.savefig("plots/equalminimaxzoom.png")
pl.close()
""" if game_elem==-1: return 2 elif game_elem==1: return 0 else: return 1 if __name__ == "__main__": """ Creates a main player """ nTuplesSystematicObject = nTuplesSystematic() # nTuplesSystematic - Black # Minimax - White # game2.play(othello.game(),game2.player(lambda x: nTuplesSystematicObject.play_next_move(x)), game2.player(lambda x: minimax.minimax(x, 0)), False) # Minimax - Black # nTuplesSystematic - White # game2.play(othello.game(), game2.player(lambda x: minimax.minimax(x, 0)),game2.player(lambda x: nTuplesSystematicObject.play_next_move(x)),False) # nTuplesSystematic - Black # Minimax Edge Eval - White # game2.play(othello.game(),game2.player(lambda x: nTuplesSystematicObject.play_next_move(x)), game2.player(lambda x: minimax.minimax(x, 0,othello.edge_eval)), True) # Minimax Edge Eval - Black # nTuplesSystematic - White game2.play(othello.game(), game2.player(lambda x: minimax.minimax(x, 0,othello.edge_eval)),game2.player(lambda x: nTuplesSystematicObject.play_next_move(x)), True)
def calculate_move(self, board, whites_turn, depth):
value, move = minimax.minimax(board, None, depth,
whites_turn, evaluator.evaluate_advanced)
return move
def makeComputerReply():
UpdateThePointMatrices()
move=minimax.minimax(COMPUTER, M, PW, PB, MAXDEPTH)
print(move.shape)
makeMove(move[0], move[1], LocateTurnedPieces(move[0],move[1],COMPUTER, M), COMPUTER)
displayAllLegalMovesForHumanPlayer('BLUE')
def play(self, game, opp_move):
return self.play_fn(game)
def gameover(self, game, last_move):
pass
if __name__ == "__main__":
import othello
import minimax
# Experiment 1:
# Player 1 and Player 2 are evenly matched with 3-ply deep search
# player 2 wins with a final score of 28
# player 1 0.2 s per ply player 2 0.4 s per ply
play(othello.game(), player(lambda x: minimax.minimax(x, 3)),
player(lambda x: minimax.minimax(x, 3)), True)
# Experiment 2:
# now we show the significance of an evaluation function
# we weaken player1 to 2 ply deep but use the edge eval fun
# player 1 now beats player 2 with a score of 58!
# player 1 0.1 s per ply player 2 0.4 s per ply
play(othello.game(), player(lambda x: minimax.minimax(x, 2, othello.edge_eval)),
player(lambda x: minimax.minimax(x, 3)), False)
# Experiment 1 (with alpha-beta):
# player 1 0.1 s per ply, player 2 0.1 s per ply
play(othello.game(), player(lambda x: minimax.alphabeta(x, 3)),
player(lambda x: minimax.alphabeta(x, 3)), False)
def main():
global m,n, surface
#start screen
font = pygame.font.Font('Font.ttf', 72)
text = font.render("Red", 1, (255,0,0))
textpos = text.get_rect(centerx = 25*n, centery = 12*m)
surface.blit(text, textpos)
text = font.render("Green", 1, (0,255,0))
textpos = text.get_rect(centerx = 25*n, centery = 36*m)
surface.blit(text, textpos)
font = pygame.font.Font('Font.ttf', 12)
text = font.render("Choose a Color", 1, (100,100,100))
textpos = text.get_rect(centerx = 25*n, centery = 25*m)
surface.blit(text, textpos)
pygame.display.update()
this_loop = True
while this_loop:
for event in pygame.event.get():
if event.type == pygame.MOUSEBUTTONDOWN:
y = pygame.mouse.get_pos()[1]
if y < 25*m:
player_first = True
else:
player_first = False
this_loop = False
#depth screen
surface.fill((0,0,0))
font = pygame.font.Font('Font.ttf', 12)
text = font.render("How deep should I look?", 1, (100,100,100))
textpos = text.get_rect(centerx = 25*n, centery = 12*m)
surface.blit(text, textpos)
font = pygame.font.Font('Font.ttf', 48)
depth = 3
text = font.render(str(depth),1,(255,255,0))
textpos = text.get_rect(centerx = 25*n, centery = 25*m)
surface.blit(text, textpos)
pygame.display.update()
this_loop = True
while this_loop:
for event in pygame.event.get():
if (event.type == pygame.KEYDOWN and event.key == pygame.K_RETURN) or event.type == pygame.MOUSEBUTTONDOWN:
this_loop = False
elif event.type == pygame.KEYDOWN:
if event.key < 256 and chr(event.key) in '1234567890':
rect = pygame.Rect(12*m,25*n,100,100)
pygame.draw.rect(surface,(0,0,0),rect,0)
pygame.display.update()
depth = int(chr(event.key))
text = font.render(str(depth),1,(255,255,0))
textpos = text.get_rect(centerx = 25*n, centery = 25*m)
surface.blit(text, textpos)
pygame.display.update()
#rows screen
surface.fill((0,0,0))
font = pygame.font.Font('Font.ttf', 12)
text = font.render("How many rows?", 1, (100,100,100))
textpos = text.get_rect(centerx = 25*n, centery = 12*m)
surface.blit(text, textpos)
font = pygame.font.Font('Font.ttf', 48)
rows = 9
text = font.render(str(rows),1,(255,255,0))
textpos = text.get_rect(centerx = 25*n, centery = 25*m)
surface.blit(text, textpos)
pygame.display.update()
this_loop = True
while this_loop:
for event in pygame.event.get():
if (event.type == pygame.KEYDOWN and event.key == pygame.K_RETURN) or event.type == pygame.MOUSEBUTTONDOWN:
this_loop = False
elif event.type == pygame.KEYDOWN:
if event.key < 256 and chr(event.key) in '1234567890':
rect = pygame.Rect(12*m,25*n,100,100)
pygame.draw.rect(surface,(0,0,0),rect,0)
pygame.display.update()
rows = int(chr(event.key))
text = font.render(str(rows),1,(255,255,0))
textpos = text.get_rect(centerx = 25*n, centery = 25*m)
surface.blit(text, textpos)
pygame.display.update()
#columns screen
surface.fill((0,0,0))
font = pygame.font.Font('Font.ttf', 12)
text = font.render("How many columns?", 1, (100,100,100))
textpos = text.get_rect(centerx = 25*n, centery = 12*m)
surface.blit(text, textpos)
font = pygame.font.Font('Font.ttf', 48)
columns = 6
text = font.render(str(columns),1,(255,255,0))
textpos = text.get_rect(centerx = 25*n, centery = 25*m)
surface.blit(text, textpos)
pygame.display.update()
this_loop = True
while this_loop:
for event in pygame.event.get():
if (event.type == pygame.KEYDOWN and event.key == pygame.K_RETURN) or event.type == pygame.MOUSEBUTTONDOWN:
this_loop = False
elif event.type == pygame.KEYDOWN:
if event.key < 256 and chr(event.key) in '1234567890':
rect = pygame.Rect(12*m,25*n,100,100)
pygame.draw.rect(surface,(0,0,0),rect,0)
pygame.display.update()
columns = int(chr(event.key))
text = font.render(str(columns),1,(255,255,0))
textpos = text.get_rect(centerx = 25*n, centery = 25*m)
surface.blit(text, textpos)
pygame.display.update()
#some initialization code
m, n = rows, columns
surface = pygame.display.set_mode((50*n, 50*m))
pygame.display.set_caption('Chain Reaction')
board = structure.Board(m=m,n=n)
total_moves = 0
#game screen
drawBoard(board)
if not player_first:
new_move = minimax.minimax(board)[0]
lock.acquire()
thread.start_new_thread(slowMove, (board, new_move))
board = structure.move(board, new_move)
total_moves += 1
this_loop = True
while this_loop:
for event in pygame.event.get():
if event.type == pygame.MOUSEBUTTONDOWN:
x,y = pygame.mouse.get_pos()
x,y = x/50,y/50
if not (board.new_move == structure.sgn(board[(y,x)]) or 0 == structure.sgn(board[(y,x)])):
print "Illegal Move!"
continue
show_move((y,x))
lock.acquire()
thread.start_new_thread(slowMove, (board,(y,x)))
board = structure.move(board,(y,x))
total_moves += 1
if total_moves >= 2:
if structure.score(board,board.new_move*(-1)) == 10000:
winner = board.new_move*(-1)
this_loop = False
break
new_move = minimax.minimax(board,depth)[0]
show_move(new_move)
lock.acquire()
thread.start_new_thread(slowMove, (board, new_move))
board = structure.move(board, new_move)
total_moves += 1
if total_moves >= 2:
if structure.score(board,board.new_move*(-1)) == 10000:
winner = board.new_move*(-1)
this_loop = False
break
#winning screen
while lock.locked():
continue
m, n = 9, 6
surface = pygame.display.set_mode((50*n, 50*m))
font = pygame.font.Font('Font.ttf', 72)
pygame.display.set_caption('Chain Reaction')
if winner == 1:
text = font.render("Red", 1, (255,0,0))
else:
text = font.render("Green", 1, (0,255,0))
textpos = text.get_rect(centerx = 25*n, centery = 12*m)
surface.blit(text, textpos)
font = pygame.font.Font('Font.ttf', 48)
text = font.render("Wins!", 1, (100,100,100))
textpos = text.get_rect(centerx = 25*n, centery = 25*m)
surface.blit(text, textpos)
pygame.display.update()
def play_game(game, heuristic1, heuristic2):
while not end_game(game.board):
turn = game.turn.color
board = game.board
if has_valid_position(board, turn):
minimax.PLAYER = turn
if turn == "B":
position = minimax.minimax(board, 1, turn, heuristic1)[1]
else:
position = minimax.minimax(board, 1, turn, heuristic2)[1]
game.play(position)
else:
game.change_turn()
return game.winning_side(formatted=False)
def main():
print "Terceira questao"
tabuleiro = [' ']*9
raiz = arvore.No(tabuleiro)
print " "
arvore.geraArvore(tabuleiro,raiz)
print "Abaixo segue a arvore, com nos raiz, filhos e netos"
print " "
arvore.imprimirArvore(raiz)
print " "
print "Avaliacao das folhas"
arvore.avaliarArvore(raiz)
print " "
minimax_value = minimax(raiz, 2, True)
print 'Minimax melhor valor: ' + str(minimax_value)
for i,filho in enumerate(raiz.filhos):
if filho.avaliacao == minimax_value:
print 'Melhor jogada vai para:'
filho.imprimir()
break
