def train(self, opponent=RandomAgent(), n_episodes=1000):
board = TicTacToeBoard(self.size)
for episode_nb in range(n_episodes):
board.reset()
while not board.is_end_state():
self.train_one_step(board, opponent)
if episode_nb % 100 == 0:
self.eps *= self.eps_decay
print(self.eps)
def game_loop():
pygame.init()
screen = pygame.display.set_mode([720, 480])
screen.fill((0,0,0))
game_board = TicTacToeBoard(screen)
cpu_player = -1
toss = randint(1,2) # Simulating biased coin flip
if toss == 1: # In this case, CPU starts
cpu_state = TicTacToeState(game_board.state)
cpu_player = 1
state = cpu_state.monte_next_move(player=cpu_player)
game_board.update_to_state(state.state_list)
cpu_flag = 0
done = False
while not done:
for event in pygame.event.get():
if event.type == pygame.QUIT:
done = True
continue
elif event.type == pygame.MOUSEBUTTONDOWN and not cpu_flag:
mpos = pygame.mouse.get_pos()
if game_board.update_board(mpos):
cpu_flag = cpu_flag ^ 1
game_board.draw(screen)
pygame.display.flip()
end = TicTacToeState(game_board.state)
if end.is_end_game:
done = True
break
if cpu_flag:
cpu_state = TicTacToeState(game_board.state)
state = cpu_state.monte_next_move(player=cpu_player)
game_board.update_to_state(state.state_list)
cpu_flag = cpu_flag ^ 1
game_board.draw(screen)
end = TicTacToeState(game_board.state)
if end.is_end_game:
done = True
break
pygame.display.flip()
if end.winning_player == cpu_player:
print('CPU wins')
elif end.winning_player != 0:
print ('You win!')
else:
print ('Tie')
def test():
"""
Little test function for debug
"""
board = TicTacToeBoard()
tree = BinaryTree(board)
tree.create_tree()
print(tree._root)
print(tree._root.left)
print(tree._root.left.left)
print(tree._root.left.left.left)
print(tree._root.left.left.left.left)
print(tree._root.right)
def play_against_agent(agent):
board = TicTacToeBoard(size=agent.size)
while True:
board.reset()
while not board.is_end_state():
os.system('cls' if os.name == 'nt' else 'clear')
print(board)
x, y = [int(elt) for elt in input().split()]
board.do_move(x, y)
print(board)
print(agent.get_possible_move_values(board))
input("...")
board.do_move(*agent.best_move(board))
def start_game():
"""
Main game loop
Starts the game and leads to the end
"""
board = TicTacToeBoard()
answer = ""
rich.print("[bold blue]Welcome to Tic Tac Toe game![/bold blue]", ":game_die:")
while len(answer) < 1 or answer[0].lower() not in ["y", "n"]:
answer = input("Do you want to play first [Y/N]? ")
print(board)
if answer[0].lower() == "y":
# Enter the game loop which will exit once the
# board is filled or someone has won
# The player's turn is first
while True:
player_turn(board, "Player 1")
print(board)
if check_win(board, True):
break
rich.print("[bold magenta]Computer's turn:[/bold magenta]")
computer_turn(board, "Player 2")
print(board)
if check_win(board, True):
break
else:
# Enter the game loop which will exit once the
# board is filled or someone has won
# The player's turn is second
while True:
print("Computer's turn:")
computer_turn(board, "Player 1")
print(board)
if check_win(board, False):
break
player_turn(board, "Player 2")
print(board)
if check_win(board, False):
break
def test_agent(agent, opponent, n_episodes=10000):
from agents import RLAgent
board = TicTacToeBoard(agent.size)
n_wins = 0
for _ in range(n_episodes):
board.reset()
counter = 0
while not board.is_end_state():
to_play = opponent if counter % 2 == 0 else agent
if isinstance(to_play, RLAgent):
board.do_move(*(to_play.eps_greedy(board)[1]))
else:
board.do_move(*to_play.best_move(board))
counter += 1
n_wins += board.has_won(agent.sym)
print(f"{n_wins}/{n_episodes}")
return n_wins / n_episodes
def __init__(self):
self.board = TicTacToeBoard()
self.feature_vector_size = 28
def reset(self):
self.board = TicTacToeBoard()
def __init__(self):
self._board = TicTacToeBoard()
self._player = self.X
self._running = True
self._has_winner = False
self._moves = []