def nextMove(self):
player = self.turn
# there are only 42 legal places for pieces on the board
# exactly one piece is added to the board each turn
if self.round > 42:
self.finished = True
# this would be a stalemate :(
return
# move is the column that player want's to play
move = player.move(self.board)
if move == None:
move = random.randint(0,6)
while self.board[0][move] == ' ':
move = random.randint(0,6)
for i in range(6):
if self.board[i][move] == ' ':
self.board[i][move] = player.color
m = Minimax(self.board, j)
print m.value(self.board, player.color)
self.switchTurn()
self.checkForFours()
self.printState()
return
# if we get here, then the column is full
print("Invalid move (column is full)")
return
def play(q_agent, human=0, helper=False, helper_depth=20):
"""
Play human game against the QAgent.
`human` can be set to 0 or 1 to specify whether
human moves first or second.
"""
connect4 = Connect4()
helper = Minimax(max_depth=helper_depth) if helper else None
while True:
for l in range(0, 42, 7):
row = ''.join([f"{connect4.board[l + i]}|" for i in range(7)])
print(row[:13])
print('-+-+-+-+-+-+-')
actions = connect4.available_actions(connect4.board)
if connect4.player == human:
print("Your Move.")
while True:
column = int(input().strip()) - 1
if not column in actions:
print(
'That place is already filled or invalid. Still your move.'
)
else:
break
# column, values = helper.get_move(connect4.board)
else:
print("QAgent's Move.")
if helper:
action, values = helper.get_move(connect4.board)
if values.count(1000) >= 1 or values.count(-1000) >= 1:
column = action
else:
column = q_agent.choose_action(connect4.board,
epsilon=False)
else:
column = q_agent.choose_action(connect4.board, epsilon=False)
print(f"QAgent put a chip in column {column + 1}.")
connect4.move(column)
if connect4.result is not None:
print("\nGAME OVER\n")
winner = "Human" if connect4.result == human else "QAgent"
print(f"Winner is {winner}")
break
if input("Play again?\n").lower() == "y":
play(q_agent)
def _run_minimax(self, max_depth) -> State:
minimax = Minimax(State(self._grid),
terminal_test_func=self._create_terminal_test(max_depth),
child_states_func=self._child_states,
eval_func=self._eval)
best_child, _ = minimax.decision()
return best_child
def move(self, state):
print("{0}'s turn. ".format(self.name))
m = Minimax(state)
best_move, value = m.alphaBetaBestMove(self.difficulty, state, self.color)
return best_move
def ai_vs_human(self, team):
self.menubar.add_command(label="AI Go", command=self.ai_move)
self.ai_team = team
self.only_allow_valid_moves = True
# AI time limit defined here
self.m = Minimax(self.time_limit, True)
self.master.update()
def __init__(self, game, depth, breadths, evaluate, tag=None):
assert depth == len(breadths)
self.game = game
self.depth = depth
self.evaluate = evaluate
self.tag = tag
self.minimax = Minimax(game, depth, breadths, evaluate)
def move(self, state):
print("Robot's beurt(" + self.color + ")")
# Voer minimax-algoritme uit
m = Minimax(state)
best_move, value = m.best_move(self.difficulty, state, self.color)
return best_move
class MinimaxPlayer:
def __init__(self, game, depth, breadths, evaluate, tag=None):
assert depth == len(breadths)
self.game = game
self.depth = depth
self.evaluate = evaluate
self.tag = tag
self.minimax = Minimax(game, depth, breadths, evaluate)
def choose_move(self):
self.minimax.run_search()
move = self.minimax.choose_move()
self.minimax.reset_tree()
return move
def make_move(self, move):
self.game.make_move(move)
def __repr__(self):
my_repr = ''
my_repr += 'Minimax Player with depth {} ' \
'and evaluation function {}'.format(self.depth, self.evaluate.__name__)
if self.tag:
my_repr += ' tag: {}.'.format(self.tag)
return my_repr
def move(self, state):
print("{0}'s turn. {0} is {1}, {2}".format(self.name, self.color,
self.ctr))
self.ctr += 1
m = Minimax(state)
best_move, value = m.bestMove(self.difficulty, state, self.color)
return best_move
def main(args):
if args.player1 == "human":
agent1 = Human(1, surface)
elif args.player1 == "minimax":
agent1 = Minimax(1, args.minimax_depth[0], args.variant)
elif args.player1 == "mcts":
agent1 = MCTS(1, args.mcts_depth[0], args.mcts_rollouts[0],\
args.variant, args.heuristic_rollouts[0], \
args.input_file[0] if args.input_file else None, args.output_file[0] if args.output_file else None, args.ucb_const[0])
if args.player2 == "human":
agent2 = Human(-1, surface)
elif args.player2 == "minimax":
agent2 = Minimax(-1, args.minimax_depth[1], args.variant)
elif args.player2 == "mcts":
agent2 = MCTS(1, args.mcts_depth[1], args.mcts_rollouts[1],\
args.variant, args.heuristic_rollouts[1], args.input_file[1] if len(args.input_file) == 2 else None,\
args.output_file[1] if len(args.output_file) == 2 else None, args.ucb_const[1])
for i in range(args.num_games):
play_game(agent1, agent2, surface, args.variant, args.wait_between)
if type(agent1) == MCTS:
agent1.reset(1)
if type(agent2) == MCTS:
agent2.reset(-1)
if args.alternate_sides:
agent1.switch_sides()
agent2.switch_sides()
temp = agent1
agent1 = agent2
agent2 = temp
if type(agent1) == MCTS:
agent1.store_root()
if type(agent2) == MCTS:
agent2.store_root()
class AIPlayer(Player):
""" AIPlayer object that extends Player
The AI algorithm is minimax, the difficulty parameter is the depth to which
the search tree is expanded.
"""
difficulty = None
def __init__(self, name, color, difficulty=5):
self.type = "AI"
self.name = name
self.color = color
self.difficulty = difficulty
self.m = Minimax()
def move(self, state):
print "{0}'s turn. {0} is {1}".format(self.name, self.color)
# sleeping for about 1 second makes it looks like he's thinking
#time.sleep(random.randrange(8, 17, 1)/10.0)
#return random.randint(0, 6)
self.m.set_board(state)
best_move, value = self.m.bestMove(self.difficulty, state)
print best_move
return best_move
def move(self, state):
super().move(self)
print("{0}'s turno. {0} es {1}".format(self.nombre, self.color))
m = Minimax(state)
mejor_mov, valor = m.mejorMov(self.dificultad, state, self.color)
return mejor_mov
async def _prepare_player(self, name):
async with self._session.post(f'{self._api_url}/game',
params={'team_name': name}) as resp:
res = (await resp.json())['data']
self._player_num = 1 if res['color'] == 'RED' else 2
self._player = {'color': res['color'], 'token': res['token']}
self.minimax = Minimax(self._player_num)
print("PLAYER_NUM", self._player_num)
def results_stats_memo():
memo = dict()
# evaluate the whole game tree
Minimax.minimax_memo(9, TTT(), memo)
counter = Counter(memo.values())
print(counter)
def move(self, state):
print("{0}'s turn. {0} is {1}".format(self.name, self.color))
# sleeping for about 1 second makes it looks like he's thinking
#time.sleep(random.randrange(8, 17, 1)/10.0)
#return random.randint(0, 6)
m = Minimax(state)
best_move, value = m.bestMove(self.difficulty, state, self.color)
return best_move
def move(self, state):
print("{0}'s turn. {0} is {1}".format(self.name, self.color))
m = Minimax(state)
best_move, _ = m.bestMove(self.difficulty, state, self.color,
self.heuristic)
#SetMoveMi(best_move)
print(self.name + ": " + str(best_move))
return best_move
def move(self, state):
print("{0}'s turn. {0} is {1}".format(self.name, self.color))
# sleeping for about 1 second makes it looks like he's thinking
#time.sleep(random.randrange(8, 17, 1)/10.0)
#return random.randint(0, 6)
m = Minimax(state)
best_move, value = m.bestMove(self.difficulty, state, self.color)
return best_move
def test_next_move_gives_the_best_move_with_diagonals(self):
minimax = Minimax(1)
game = Game(4, 4)
game.board = [
[0, -1, 1, -1],
[0, 1, -1, -1],
[0, -1, 1, -1],
[0, 0, 0, 1],
]
self.assertEqual(minimax.next_move(game), 0)
def ai_vs_ai(self):
m = Minimax(self.time_limit, True)
i = 0
while self.board.check_win() == Board.EMPTY:
if i % 2 == 0:
team = Board.GREEN
else:
team = Board.RED
self.board = m.search(self.board, team)[0]
i += 1
self.display_board(self.board)
self.master.update()
def test_next_move_gives_the_best_defence_move(self):
minimax = Minimax(1)
game = Game(4, 4)
game.board = [
[0, 0, 0, 0],
[0, 0, 0, 0],
[0, -1, -1, -1],
[0, 0, 1, 1],
]
self.assertEqual(minimax.next_move(game), 2)
def move(self, state):
print(f"{self.name}'s turn. {self.name} is {self.color}")
# time.sleep(random.randrange(8, 17, 1) / 10.0)
# return random.randint(0, 6)
m = Minimax(state)
best_move, value = m.best_move(state=state,
depth=self.difficulty,
alpha=-math.inf,
beta=math.inf,
maximizing_player=True)
return best_move
def make_move(self):
changed_board = self.change_board(settings.board)
m = Minimax(changed_board)
x = m.bestMove(self.depth, changed_board,
'x' if self.player == 1 else 'o')
x = x[0]
output = self.name + " played in column " + str(x) + "."
print(output)
return x
def alphaBeta(self, state, alpha, beta, depth):
if state in self.table:
return self.table[state].minimax_value
state.updateWinCheck()
if state.isTerminal():
u = state.stateUtility()
self.table[state] = Minimax(u, None)
return u
elif depth >= self.depth_cutoff:
e = state.stateEval()
self.table[state] = Minimax(e, None)
return e
elif state.turn == "MAX":
best_minimax_so_far = np.NINF
best_move_for_state = None
children_states = state.genNextBoards()
for i in range(len(state.moves)):
if state.moves[i]:
child_state = children_states[i]
minimax_of_child = self.alphaBeta(child_state, alpha, beta, depth + 1)
if minimax_of_child > best_minimax_so_far:
best_minimax_so_far = minimax_of_child
best_move_for_state = i
if best_minimax_so_far >= beta:
return best_minimax_so_far
alpha = max(alpha, best_minimax_so_far)
self.table[state] = Minimax(best_minimax_so_far, best_move_for_state)
return best_minimax_so_far
else:
best_minimax_so_far = np.Inf
best_move_for_state = None
children_states = state.genNextBoards()
for i in range(len(state.moves)):
if state.moves[i]:
child_state = children_states[i]
minimax_of_child = self.alphaBeta(child_state, alpha, beta, depth + 1)
if minimax_of_child < best_minimax_so_far:
best_minimax_so_far = minimax_of_child
best_move_for_state = i
if best_minimax_so_far <= alpha:
return best_minimax_so_far
beta = min(beta, best_minimax_so_far)
self.table[state] = Minimax(best_minimax_so_far, best_move_for_state)
return best_minimax_so_far
def move(self, state, timeLimit):
print("{0}'s turn. {0} is {1}".format(self.name, self.color))
if self.difficulty == 6:
m = AlphaBeta(state)
start = time.clock()
best_move, value, depth = m.bestMove(30, state, self.color,
timeLimit)
print("Alpha: ", value)
print("Elapsed:", time.clock() - start)
print("Depth Reached:", depth)
return best_move, depth
elif self.difficulty == 7:
m = Greedy(state)
time.sleep(randrange(8, 17, 1) / 10.0)
best_move = m.best_move(state, self.color)
print("guess greedy worked")
return best_move, 1
else:
m = Minimax(state)
start = time.clock()
best_move, value, depth = m.bestMove(30, state, self.color,
timeLimit)
print("Alpha: ", value)
print("Elapsed:", time.clock() - start)
print("Depth Reached:", depth)
return best_move, depth
def risanje(self, igralec, p, m):
if igralec==1:
barva="steel blue"
else: barva="chocolate"
r = deepcopy(self.stanje)
r = transponiraj(r)
r = obrni(r)
minimax1 = Minimax(r)
s = minimax1.vrednostStanja(r, igralec)
print(s)
v = 6
g=self.platno.create_rectangle(100*p+v, v, 100+100*p-1, 100-1, fill=barva, outline="white")
for i in range(0, m*100):
self.platno.move(g, 0, 1)
self.platno.update()
time.sleep(1/(50*i+5))
def ai(self,event): print "AI" self.new_game() ai = Minimax() while not self.board.has_lost(): ai.state = self.board move = ai.get_move() self.board.move(move) if self.board.valid_move: new_tile = Tile(None) new_tile.set_start_value() self.board.set_new_tile(new_tile) self.update_board() else: self.update_board() break self.game_ended()
def __compare_bots(self, iterations):
if self.state.playerToMove == self.player1:
assert self.state.playerToMove.algorithm == "ISMCTS"
return self.smart_ismcts.get_move(rootstate=self.state,
itermax=iterations,
verbose=False)
else:
assert self.state.playerToMove.algorithm == "Minimax"
return Minimax.get_move(self.state)
def minimax_player(
self,
state,
depth=2500000,
team=1,
heuristic_parameter=True
): # creates first successors to implement minimax algorithm
new_shape_x = np.asarray(state[1]).shape
player1 = Minimax(n=new_shape_x,
default_team=team,
advance_option=heuristic_parameter)
print('default_team', team, player1.default_team)
if team == -1:
state = player1.convert_board_state(state)
best_move = player1.decision_maker(state, depth)
chosen_succ, utility = best_move
if team == -1:
chosen_succ = player1.convert_board_state(chosen_succ)
return chosen_succ
def results_stats():
counter = defaultdict(int)
for index, board in enumerate(gen.all_positions_unique(TTT())):
hash = str(board)
evaluation = Minimax.minimax(9, board)
counter[evaluation] += 1
print(f"{index+1}: {hash} ==> {evaluation}")
print(counter)
def __init__(self, render=True, ros_node=None):
self.board = C4Board(render)
self.max = Minimax()
if not ros_node:
rospy.init_node("connect4", anonymous=True)
self.ur5_commander = rospy.Publisher("behaviors_cmd",
String,
queue_size=10)
self.token_grabber = rospy.Publisher("gripper", Int16, queue_size=10)
self.computer_vision = rospy.Publisher("c4_ready",
Int16,
queue_size=10)
rospy.Subscriber("arm_status", String, self.status_callback)
rospy.Subscriber("opponent_move", Int16, self.player_move)
self.status = True
self.turn = "Start"
self.player_action = None
def move(self, state, rounds):
if self.type.lower() == 'human':
print("{0}'s turn. {0} is {1}".format(self.name, self.color))
column = None
while column == None:
try:
choice = int(input("Enter a move (1-7): ")) - 1
except ValueError:
choice = None
if 0 <= choice <= 6:
column = choice
else:
print("Invalid choice, try again")
return column
else:
print("{0}'s turn. {0} is {1}".format(self.name, self.color))
#return random.randint(0, 6)
m = Minimax(state)
best_alpha, alpha = m.findColumn(self.difficulty, state, self.color, rounds)
return best_alpha
def randomai(self, r):
print("---------------")
g = deepcopy(self.stanje)
g = transponiraj(g)
g = obrni(g)
m = Minimax(g)
p = m.bestMove(self.tezavnost, g, 2)[0]
if self.preverjanje1(self.stanje)[0] != -1 or self.preverjanje1(self.stanje)[1] != -1:
if self.preverjanje1(self.stanje)[0] != -1:
p = self.preverjanje1(self.stanje)[0]
else: p = self.preverjanje1(self.stanje)[1]
v = [i for i in range(len(self.stanje[p][:])) if self.stanje[p][i]!=0]
v.append(6)
m = min(v)-1
self.steviloklikov[p] += 1
self.stanje[p][m] = 2
self.risanje(2, p, m)
self.platno.bind("<Button-1>", self.klik)
if self.zmaga(self.stanje):
self.aiscore += 1
self.risanjezmaga()
m1=tkinter.messagebox.askyesno("Jaesus","You lost! New game?")
if m1:
self.zacetek()
return
else:
return
if sum(self.steviloklikov)==42:
self.neodloceno+=1
m1=tkinter.messagebox.askyesno("Tie","It's a tie. New game?")
if m1:
self.zacetek()
return
else:
quit()
def __init__(self, mode, square_side, depth, ai_mode, omni):
self.square_side = square_side
player1 = Player(1, self.square_side)
player2 = Player(2, self.square_side)
self.player_turn = 1
self.players = [player1, player2]
self.mode = mode
self.min_pos = square_side / 2
self.max_pos = square_side * 8 - square_side / 2
self.turn = 1
if '3' == self.mode and omni:
ai2 = ai_mode[1]
else: ai2 = '3'
self.ai_player2 = Minimax(depth, ai_mode[0], ai2)
if '3' == self.mode:
if omni:
self.ai_player1 = Minimax(depth, ai2, ai_mode[0])
else: self.ai_player1 = Minimax(depth, ai2, '3')
class Computadora(object):
def __init__(self, color, prune=3):
self.limiteprofundidad = prune
evaluador = Evaluador()
self.minimaxObj = Minimax(evaluador.puntuacion)
self.color = color
def obtenertableroactual(self, tablero):
self.tablero_actual = tablero
def obtenermovida(self):
return self.minimaxObj.minimax(self.tablero_actual, None, self.limiteprofundidad, self.color, cambiarcolor(self.color))
class Computer( object ):
def __init__( self, color, prune=3 ):
self.depthLimit = prune
evaluator = Evaluator()
self.minimaxObj = Minimax( evaluator.score )
self.color = color
def get_current_board( self, board ):
self.currentBoard = board
def get_move( self ):
return self.minimaxObj.minimax( self.currentBoard, None, self.depthLimit, self.color, \
change_color(self.color) )
def move(self, state):
print("{0}'s turn. {0} is {1}".format(self.name, self.color))
# sleeping for about 1 second makes it looks like he's thinking
#time.sleep(random.randrange(8, 17, 1)/10.0)
#return random.randint(0, 6)
port = serial.Serial("COM12", 9600)
port.isOpen()
m = Minimax(state)
#m = Minimax(super_board)
best_move, value = m.bestMove(self.difficulty, state, self.color)
#print(state)
#input(best_move + 1)
print(best_move + 1)
inp = 7 - best_move
k = str.encode(str(inp))
#time.sleep(2)
time.sleep(1)
#k = bytes(inp)
port.write(k)
port.write(b'1')
#print(k)
i = 0
while True:
response = port.read()
if response == b'9':
print(response)
i = 1
if (i == 1):
break
port.close()
return best_move
def minimax(self, state):
if state in self.table:
return self.table[state].minimax_value
state.updateWinCheck()
if state.isTerminal():
u = state.stateUtility()
self.table[state] = Minimax(u, None)
return u
elif state.turn == "MAX":
best_minimax_so_far = np.NINF
best_move_for_state = None
children_states = state.genNextBoards()
for i in range(len(state.moves)):
if state.moves[i]:
child_state = children_states[i]
minimax_of_child = self.minimax(child_state)
if minimax_of_child > best_minimax_so_far:
best_minimax_so_far = minimax_of_child
best_move_for_state = i
self.table[state] = Minimax(best_minimax_so_far, best_move_for_state)
return best_minimax_so_far
else:
best_minimax_so_far = np.Inf
best_move_for_state = None
children_states = state.genNextBoards()
for i in range(len(state.moves)):
if state.moves[i]:
child_state = children_states[i]
minimax_of_child = self.minimax(child_state)
if minimax_of_child < best_minimax_so_far:
best_minimax_so_far = minimax_of_child
best_move_for_state = i
self.table[state] = Minimax(best_minimax_so_far, best_move_for_state)
return best_minimax_so_far
class Computer(object):
def __init__(self, color, depth=5, vec=None):
self.depthLimit = depth
evaluator = Evaluator(vec)
self.minimaxObj = Minimax(evaluator.evaluate_all_heuristics)
self.color = color
def get_current_board(self, board):
self.current_board = board
def get_move(self):
return self.minimaxObj.minimax(self.current_board, None,
self.depthLimit, self.color,
change_color(self.color))
class Computer(object):
def __init__(self, color, prune=3):
self.depthLimit = prune
evaluator = Evaluator()
self.minimaxObj = Minimax(evaluator.score)
self.color = color
def get_current_board(self, board):
self.current_board = board
def get_move(self):
return self.minimaxObj.minimax(self.current_board, None,
self.depthLimit, self.color,
change_color(self.color))
def __init__(self, color, prune=3):
self.depthLimit = prune
evaluator = Evaluator()
self.minimaxObj = Minimax(evaluator.score)
self.color = color
def move(self, state):
print("Computer turn")
m = Minimax(state)
best_move, value = m.bestMove(self.difficulty, state, self.color)
return best_move
def __init__(self, color, prune=3):
self.limiteprofundidad = prune
evaluador = Evaluador()
self.minimaxObj = Minimax(evaluador.puntuacion)
self.color = color
def move(self, state):
print("{0}'s turn. {0} is {1}, {2}".format(self.name, self.color, self.ctr))
self.ctr+=1
m = Minimax(state)
best_move, value = m.bestMove(self.difficulty, state, self.color)
return best_move