python跳棋,agent对打

import random, re, datetime,time class Agent(object): def __init__(self, game): self.game = game def getAction(self, state): raise Exception("Not implemented yet") class RandomAgent(Agent): def getAction(self, state): legal_actions = self.game.actions(state) self.action = random.choice(legal_actions) class SimpleGreedyAgent(Agent): # a one-step-lookahead greedy agent that returns action with max vertical advance def getAction(self, state): legal_actions = self.game.actions(state) self.action = random.choice(legal_actions) player = self.game.player(state) if player == 1: max_vertical_advance_one_step = max([action[0][0] - action[1][0] for action in legal_actions]) max_actions = [action for action in legal_actions if action[0][0] - action[1][0] == max_vertical_advance_one_step] else: max_vertical_advance_one_step = max([action[1][0] - action[0][0] for action in legal_actions]) max_actions = [action for action in legal_actions if action[1][0] - action[0][0] == max_vertical_advance_one_step] self.action = random.choice(max_actions) class TeamNameMinimaxAgent(Agent): def getAction(self, state): legal_actions = self.game.actions(state) self.action = random.choice(legal_actions) player = self.game.player(state) ### START CODE HERE ### ### END CODE HERE ### class C_zihao_Agent(Agent): def getAction(self, state): start = time.time() legal_actions = self.game.actions(state) self.action = random.choice(legal_actions) player = self.game.player(state) ### START CODE HERE ### global staticDepth, step, repeat # store the initial value for depth for minimax depth = 2 # starting from 2 staticDepth = depth # store original depth max_depth = 20 # breadth for each layer, max is 42 board = state[1] player_status = board.getPlayerPiecePositions(player) ##occupy last_flag = (player_status in status_queue) ##true or false? all false?? player_lists = [(2, 1), (2, 2), (3, 2)] status_queue.append(player_status) step += 1 ##step++ max_error=player_status[9][0]-player_status[0][0]+1 mean_error=0 for i in range(0,10): mean_error=mean_error+player_status[i][0] mean_error=mean_error/10 #print(mean_error) #print(board.board_status) special_flag=False if(mean_error==3.1): x_index=player_status[9][1] if(x_index>1): if(board.board_status[4,x_index-1]==0): self.action=((player_status[9][0],player_status[9][1]),(4,x_index-1)) special_flag=True if (x_index<4): if(board.board_status[4, x_index + 1] == 0): self.action = ((player_status[9][0], player_status[9][1]), (4, x_index + 1)) special_flag=True if (board.board_status[(3, 2)] == player+2 and board.board_status[(1,1)]==player): if (board.board_status[(2, 1)] == player+2 and board.board_status[(2, 2)] == 0): self.action = ((3, 2), (2, 2)) special_flag=True elif (board.board_status[(2, 2)] == player+2 and board.board_status[(2, 1)] == 0): self.action = ((3, 2), (2, 1)) special_flag=True if (board.board_status[(3, 1)] == player+2 and board.board_status[(2, 1)] == player+2 and board.board_status[(3, 2)] == player+2): self.action = ((3, 2), (2, 2)) special_flag = True if (board.board_status[(3, 3)] == player+2 and board.board_status[(2, 2)] == player+2 and board.board_status[(3, 2)] == player+2): self.action = ((3, 2), (2, 1)) special_flag = True ''' if (board.board_status[(2, 1)] == player+2 and board.board_status[(3, 2)] == player+2 and board.board_status[(4, 3)] == player+2): self.action = ((3, 2), (2, 2)) special_flag = True if (board.board_status[(2, 2)] == player+2 and board.board_status[(3, 2)] == player+2 and board.board_status[(4, 2)] == player+2): self.action = ((3, 2), (2, 1)) special_flag = True ''' if(board.board_status[(2,1)]==player+2 and board.board_status[(2,2)]==player+2 and board.board_status[(3,2)]==0): if(board.board_status[(4,4)]==player+2 and board.board_status[(3,3)]==player): self.action=((4,4),(5,4)) special_flag=True elif(board.board_status[(4, 1)] == player+2 and board.board_status[(3, 1)] == player): self.action = ((4, 1), (4, 2)) special_flag = True if(board.board_status[(3,2)]==player): if(board.board_status[(4,2)]==player): self.action=((3,2),(5,2)) special_flag=True elif(board.board_status[(4,3)]==player): self.action=((3,2),(5,4)) special_flag=True second=time.time() print('second time: %s Seconds' % (second - start)) if step >= 15 and last_flag and special_flag==False: # in case of "stuck" situation, use "greedy" status_queue.remove(status_queue[0]) repeat += 1 #if repeat == 2: # in case of repeated "stuck" situation if repeat>5: self.action = random.choice(legal_actions) else: if player == 1: # playing from bottom to top, hence later vertical value is smaller max_vertical_advance_one_step = max([action[0][0] - action[1][0] for action in legal_actions]) max_actions = [action for action in legal_actions if action[0][0] - action[1][0] == max_vertical_advance_one_step] else: # playing from top to bottom, hence later vertical value is larger max_vertical_advance_one_step = max([action[1][0] - action[0][0] for action in legal_actions]) max_actions = [action for action in legal_actions if action[1][0] - action[0][0] == max_vertical_advance_one_step] random_flag=False random_cnt=0 while(random_flag==False): random_cnt=random_cnt+1 if(random_cnt<3): self.action = random.choice(max_actions) else: self.action=random.choice(legal_actions) random_flag=True if (board.board_status[self.action[0]] == player+2): if (self.action[0] in player_lists): random_flag=False if (self.action[1] == (1, 1)): random_flag = False # continue if (board.board_status[self.action[0]] == player): if (self.action[1] in player_lists): random_flag = False elif(special_flag==False): ##care if step >= 3: status_queue.remove(status_queue[0]) repeat = 0 #print(step) tran_var = -float('inf') order = PriorityQueue() # search with preference for action in legal_actions: order.put((-(3 - 2 * player) * (action[0][0] - action[1][0]), action)) order_next = PriorityQueue() while True: count = 0 ##max while (not order.empty()) and (max_depth > count):