Python井字棋Minimax算法无法选择最优落子问题排查求助

我仔细看了你的井字棋代码，发现几个关键问题导致Bot无法选出最优解，尤其是你说的无法阻止你走1-5-9获胜的情况。咱们一步步来解决：

核心问题1：平局判断完全错误 #

你的checkdraw函数逻辑搞反了！当前它是当有玩家获胜时返回True，但平局的定义是「棋盘已满且没有任何人获胜」。这个错误会导致Minimax算法在平局时无法正确返回0分，进而让评分逻辑混乱，Bot无法做出正确的决策。

修复后的checkdraw应该是这样：

def checkdraw(board):
    # 先检查有没有人赢，有赢的就不是平局
    if checkForWin2():
        return False
    # 再检查棋盘是否还有空位
    for key in board.keys():
        if board[key] == ' ':
            return False
    # 没人赢且棋盘满了，就是平局
    return True

核心问题2：Minimax里多余的打印干扰（非逻辑错误但影响调试） #

在minimax函数的最大化玩家分支里，你加了一个printboard()，这会在递归过程中疯狂打印棋盘，不仅看不清正常输出，还会拖慢算法。把这行删掉就行。

次要问题：冗余参数导致混淆 #

botgo函数的possibles参数完全没被用到（函数里直接遍历了board.keys()），可以删掉这个参数，让代码更清晰：

def botgo(mark):  # 移除possibles参数
    bestscore = -800
    bestmove = 0
    for key in board.keys():
        if (board[key] == ' '):
            board[key] = mark
            score = minimax(board,0,False)
            board[key] = ' '
            if(score > bestscore):
                bestscore = score
                bestmove = key
    insert(bestmove,mark='O')
    return

同时记得修改start函数里的调用：botgo(mark='O')

修复后的完整代码 #

board = {1: ' ', 2: ' ', 3: ' ', 4: ' ', 5: ' ', 6: ' ', 7: ' ', 8: ' ', 9: ' '}
win = False
turn = 1
depth = 1
nodeindex = 0
possibles= []
moves = []
depth = 0
targetdepth = 3
movesdone = []

def checkForWin(mark):
    if board[1] == board[2] == board[3] == mark:
        return True
    elif board[4] == board[5] == board[6] == mark:
        return True
    elif board[7] == board[8] == board[9] == mark:
        return True
    elif board[1] == board[4] == board[7] == mark:
        return True
    elif board[2] == board[5] == board[8] == mark:
        return True
    elif board[3] == board[6] == board[9] == mark:
        return True
    elif board[1] == board[5] == board[9] == mark:
        return True
    elif board[7] == board[5] == board[3] == mark:
        return True
    else:
        return False

def checkForWin2():
    if board[1] == board[2] == board[3] != ' ':
        return True
    elif board[4] == board[5] == board[6] != ' ':
        return True
    elif board[7] == board[8] == board[9] != ' ':
        return True
    elif board[1] == board[4] == board[7] != ' ':
        return True
    elif board[2] == board[5] == board[8] != ' ':
        return True
    elif board[3] == board[6] == board[9] != ' ':
        return True
    elif board[1] == board[5] == board[9] != ' ':
        return True
    elif board[7] == board[5] == board[3] != ' ':
        return True
    else:
        return False

def possiblemoves(board):
    possibles.clear()
    for i in board:
        if board[i] == " ":
            possibles.append(i)
    return possibles

def botgo(mark):
    bestscore = -800
    bestmove = 0
    for key in board.keys():
        if board[key] == ' ':
            board[key] = mark
            score = minimax(board, 0, False)
            board[key] = ' '
            if score > bestscore:
                bestscore = score
                bestmove = key
    insert(bestmove, mark='O')
    return

def printboard():
    print(board[1] + '|' + board[2] + '|' + board[3])
    print('-----')
    print(board[4] + '|' + board[5] + '|' + board[6])
    print('-----')
    print(board[7] + '|' + board[8] + '|' + board[9])

def start():
    turn = 1
    count = 0
    while count != 9:
        humango()
        printboard()
        possiblemoves(board)
        botgo(mark='O')
        printboard()
        count += 1

def spacefree(space):
    return board[space] == ' '

def insert(space, mark):
    if spacefree(space):
        board[space] = mark
        if checkForWin(mark):
            printboard()
            print("human win" if mark == 'X' else "BOT WIN")
            exit()
    else:
        print("cannot insert there!!!")
        space = int(input("Enter position"))
        insert(space, mark)

def checkdraw(board):
    if checkForWin2():
        return False
    for key in board.keys():
        if board[key] == ' ':
            return False
    return True

def humango():
    global turn
    space = int(input("Enter position"))
    insert(space, mark='X')
    turn += 1
    printboard()

def minimax(board, depth, ismax):
    if checkForWin(mark='O'):
        return 1
    elif checkForWin(mark='X'):
        return -1
    elif checkdraw(board):
        return 0
    if ismax:
        bestscore = -800
        for key in board.keys():
            if board[key] == ' ':
                board[key] = 'O'
                score = minimax(board, depth + 1, False)
                board[key] = ' '
                if score > bestscore:
                    bestscore = score
        return bestscore
    else:
        bestscore = 800
        for key in board.keys():
            if board[key] == ' ':
                board[key] = 'X'
                score = minimax(board, depth + 1, True)
                board[key] = ' '
                if score < bestscore:
                    bestscore = score
        return bestscore

start()

额外优化建议 #

• 我简化了一些条件判断（比如board[1] == board[2] and board[1] == board[3]改成board[1] == board[2] == board[3]），让代码更简洁。
• spacefree函数可以直接返回表达式，不用if-else，更简洁。
• 全局变量太多了，比如possibles、depth这些其实很多没用到，可以考虑重构，把需要的变量作为参数传递，避免全局变量带来的潜在问题。

现在你再测试一下，Bot应该能正确阻止你走1-5-9的获胜路线了！

内容的提问来源于stack exchange，提问作者RotatingConsistency