def getPlayer(nn):
if nn.endswith(".json"):
return NNPlayer(json_tricks.loads(open(nn).read())['player'])
elif nn=='mm':
return MMPlayer()
elif nn[:4]=='mcts':
return MCTSPlayer(nPlay=int(nn[4:]), maxPlies=9999, bNegamax=True)
elif nn[:2]=='mc':
return MCPlayer(nPlay=int(nn[2:]))
elif nn=='rp':
return RandomPlayer()
elif nn=='hu':
return HumanPlayer()
elif nn[:2]=='oa':
return OmniscientAdversary(nPlay=int(nn[2:]))
else:
raise Exception("Unsupported player [%s]" % nn)
def init(screen):
if not pygame.get_init():
pygame.init()
# 将变量设为初始值
global WIN, background, pieceRecord, player, CUR_PIECE_COLOR, fontCount
WIN = False # 设置获胜标志为否
CUR_PIECE_COLOR = BLACK_PIECE # 设置当前玩家为黑棋
fontCount = pygame.font.Font(FONT_FILE, 20) # 读取字体
# 初始化两个玩家
# todo: 待AI玩家开发完毕，在这里将执白玩家设置成AI玩家
player1 = HumanPlayer(BLACK_PIECE)
player2 = AlgorithmPlayer(WHITE_PIECE)
player = (player1, player2) # 初始化玩家信息
# 初始化棋局数组
pieceRecord = np.full((LINES, LINES), -1)
# 绘制背景
global background
background = pygame.image.load(new_img_filename)
background = pygame.transform.scale(background, SCREEN)
screen.blit(background, (0, 0))
# 初始化棋子图层
global sufPiece, sufOthers, sufPieceOrder, pieceColor
pieceBlack = pygame.image.load("go_piece_black.png").convert_alpha()
pieceWhite = pygame.image.load("go_piece_white.png").convert_alpha()
pieceBlack = pygame.transform.scale(pieceBlack,
(PIECE_WIDTH, PIECE_HEIGHT))
pieceWhite = pygame.transform.scale(pieceWhite,
(PIECE_WIDTH, PIECE_HEIGHT))
pieceColor = (pieceBlack, pieceWhite) # 将棋子图层放在数组中，用于在展示时切换
sufPiece = screen.convert_alpha() # 创建棋子透明图层
sufPieceOrder = screen.convert_alpha() # 创建展示棋子顺序的透明图层
sufOthers = screen.convert_alpha() # 创展示其他信息的透明图层
# 显示棋盘
pygame.display.update()
from vector2 import Vector2
from humanPlayer import HumanPlayer
from board import Board
from game import Game
players = [HumanPlayer(i) for i in range(2)]
numShipsOfSize = {2: 1, 3: 2, 4: 1, 5: 1}
boardSize = Vector2(10, 10)
board = Board(boardSize, numShipsOfSize, players)
game = Game(board)
game.Play()
# Experience ReplayというDQNで用いる学習手法で使うバッファ
replay_buffer = chainerrl.replay_buffer.ReplayBuffer(capacity=10**6)
# Agentの生成(replay_bufferなどを共有する2つ)
agent = chainerrl.agents.DoubleDQN(q_func,
optimizer,
replay_buffer,
gamma,
explorer,
replay_start_size=500,
update_interval=1,
target_update_interval=100)
agent.load("result_100000")
humanPlayer = HumanPlayer()
board = Board()
for i in range(1):
board.reset()
dqn_first = np.random.choice([True, False])
while not board.done:
# DQN
if dqn_first or np.count_nonzero(board.board) > 0:
board.show()
action = agent.act(board.board.copy())
board.move(action, 1)
if board.done:
board.show()
if board.winner == 1:
from randomComputerPlayer import RandomComputerPlayer
from ticTacToe import TicTacToe
from humanPlayer import HumanPlayer
from gameController import GameController
if __name__ == "__main__":
x_player = RandomComputerPlayer("X")
o_player = HumanPlayer("O")
tic = TicTacToe()
GameController.play(tic, x_player, o_player, print_game=True)
from state import State
from player import Player
from humanPlayer import HumanPlayer
if __name__ == "__main__":
# training
training = str(input("Do you want to retrain the computer? Y/N?"))
if training == 'Y' or training == 'y':
p1 = Player("p1")
p2 = Player("p2")
st = State(p1, p2)
print("training...")
st.play(20000)
#st.play()
p1.savePolicy()
# play with human
p1 = Player("computer", exp_rate=0)
p1.loadPolicy("policy_p1")
p2 = HumanPlayer("Coco")
st = State(p1, p2)
st.play2()
def updateBoard(iboard, point, icon):
global board
x = int(point.split(',')[0])
y = int(point.split(',')[1])
if board[x][y] == '.':
board[x][y] = icon
else:
print("this cell is taken, nothing happened")
print("hello")
board_utils = BoardUtils()
score_keeper = Score()
h = HumanPlayer(score_keeper)
a = AIPlayer(score_keeper)
players = [h, a]
current_player = 0
board = [['.', '.', '.'], ['.', '.', '.'], ['.', '.', '.']]
while not board_utils.game_over(board):
print(f"now is the turn of {players[current_player].name}")
print("this is the board before the move")
point = players[current_player].next_move(board)
updateBoard(board, point, players[current_player].icon)
current_player = (current_player + 1) % len(players)
# print(current_player)
# point = players[current_player].next_move(board)
# print(point)
# board_utils.updateBoard(board, point, players[current_player])