def test1():
env = Environment(RandomPlayer('O'))
agent = RLPlayer('X')
episode = Episode(agent, env)
board = p.empty_board()
agent, final_board = episode.execute(board)
return agent, final_board
def play_game(p1, p2, file=None):
board = p.empty_board()
players = [p1, p2]
current_player_index = 0
winner = None
move_count = 0
while (True):
print("Current move is for player: ", players[current_player_index].player)
if (file is not None):
file.write("PRIOR TO MOVE " + str(move_count) + " ------------\n")
file.write(to_display_string(board))
if p.is_cat_game(board):
if (file is not None):
file.write("RESULT IS CAT GAME")
break
m = players[current_player_index].pick_next_move(board)
board = p.add_move(m[0], m[1], board)
p.display_board(board)
move_count += 1
if p.is_winner(board, players[current_player_index].player):
winner = players[current_player_index]
if (file is not None):
file.write("FINAL BOARD AFTER MOVE " + str(move_count) + " WINNER IS: " + winner.player + "\n")
file.write(to_display_string(board))
break
# alternate players
if (current_player_index == 0):
print("Switching to player 1...")
current_player_index = 1
else:
print("Switching to player 0...")
current_player_index = 0
if (winner is None):
print("CAT GAME")
else:
print("WINNER IS PLAYER: ", winner.player)
return winner
def generate_tic_tac_toe_episode(policy):
current_board = p.empty_board()
opponent = GreedyRandomPlayer('O')
game_complete = False
transitions = []
while not game_complete:
previous_state = current_board
print("PRIOR TO MOVE:")
print(game.to_display_string(current_board))
file.write("PRIOR TO MOVE:\n")
file.write(game.to_display_string(current_board))
selectedAction = sample_tic_tac_toe_policy(current_board, policy)
# model the environment --> returns a next state and a reward
next_state, reward, game_complete = model_environment(opponent, current_board, selectedAction)
# append the episode
transition = {}
transition['from_state'] = current_board
transition['to_state'] = next_state
transition['action'] = selectedAction
transition['reward'] = reward
transitions.append(transition)
file.write("ADDING TRANSITION: " + str(transition) + "\n")
current_board = next_state
# now figure out the reward
print("BOARD AT END OF EPISODE")
print(game.to_display_string(current_board))
file.write("BOARD AT END OF EPISODE\n")
file.write(game.to_display_string(current_board))
return transitions
def test_sample_policy():
board = p.empty_board()
policy = {}
move = sample_tic_tac_toe_policy(board,policy)
print("SELECTED MOVE: ", move)