PLAYER)
self.ai_oppoenet = MCTS(None, P2, 128, OPPONENT)
self.current_user = None
def select_action(self, state):
if self.current_user == PLAYER:
self.ai_player.current_user = PLAYER
action = self.ai_player.simulation(state)
else:
self.ai_oppoenet.current_user = OPPONENT
action = self.ai_oppoenet.simulation(state)
return action
if __name__ == '__main__':
env = tictactoe_env.TicTacToeEnv()
manager = AiVsAi()
result = {-1: 0, 0: 0, 1: 0}
for game in range(GAMES):
print('########## Game: {} ##########\n'.format(game + 1))
player_color = (MARK_O + game) % 2
state = env.reset(player_color=player_color)
done = False
step_play = 0
while not done:
current_user = ((PLAYER if player_color == MARK_O else OPPONENT) +
step_play) % 2
print('- BOARD -')
print(env.board[PLAYER] + env.board[OPPONENT] * 2)
state_memory.appendleft(self.root)
pi_memory.appendleft(pi.flatten())
return tuple(action)
if __name__ == '__main__':
start = time.time()
train_dataset_store = []
state_memory = deque(maxlen=102400)
pi_memory = deque(maxlen=102400)
z_memory = deque(maxlen=102400)
env_game = tictactoe_env.TicTacToeEnv()
env_simul = tictactoe_env.TicTacToeEnv()
result_game = {-1: 0, 0: 0, 1: 0}
win_mark_o = 0
step_game = 0
step_total_simul = 0
print("=" * 30, " Game Start ", "=" * 30, '\n')
for game in range(GAMES):
player_color = (MARK_O + game) % 2
state_game = env_game.reset(player_color=player_color)
mcts = MCTS()
done_game = False
step_play = 0