Evaluation function for minimax

Question

Hello everyone I am currently taking CS50AI course. 1st assignment is creating a tictactoe AI with minimax function. My problem is this: As far as I understand, there has to be a static evaluation for positions of the game. I was trying to write something like this in pseudocode :

If next move is a winning move:
    return 10 point
elif opponent is going to win stop him:
    return 8 point

kind of thing. But when I checked others minvalue - max value function there was no such a thing.

def minimax(board):
    """
    Returns the optimal action for the current player on the board.
    """
    currentactions = actions(board)
    if player(board) == X:
        vT = -math.inf
        move = set()
        for action in currentactions:
            v, count = maxvalue(result(board,action), 0)
            if v > vT:
                vT = v
                move = action
    else:
        vT = math.inf
        move = set()
        for action in currentactions:
            v, count = minvalue(result(board,action), 0)
            if v < vT:
                vT = v
                move = action
    print(count)
    return move

    def maxvalue(board, count):
        """
        Calculates the max value of a given board recursively together with minvalue
        """
    
        if terminal(board): return utility(board), count+1
    
        v = -math.inf
        posactions = actions(board)
    
        for action in posactions:
            vret, count = minvalue(result(board, action), count)
            v = max(v, vret)
        
        return v, count+1
    
    def minvalue(board, count):
        """
        Calculates the min value of a given board recursively together with maxvalue
        """
    
        if terminal(board): return utility(board), count+1
    
        v = math.inf
        posactions = actions(board)
    
        for action in posactions:
            vret, count = maxvalue(result(board, action), count)
            v = min(v, vret)
        
        return v, count+1

This is sikburn's tictactoe implementation's max - min functions. I could not understand what outcome will come from the max or min value functions. Can anyone clarify my logic please ? By the way, terminal() function checks if the game ended (has a winner or tie) and result() function takes a board and action as an input and returns the resultant board. Thanks for all the help.

Answer 1

In the utility function (not included in your code), you probably assigned 1 to X victory, -1 to O victory, and 0 otherwise. The minimax function calls minvalue and maxvalue recursively, for all possible moves, until it comes to the end of the game, be it a tie or a victory. Then it call utility to get the value. Both minvalue and maxvalue assure X and O will always select the best possible moves.

Don't forget to check if the board is terminal and return None if it is, before proceding, in the minimax function.

Swap the the calling of minvalue and maxvalue functions in minimax: for X, call minvalue (because X wants to know O next move) and for O, call maxvalue (for the same reason).

If you want to see the evalution at each iteration, you can print something like f"Minvalue: {v}, Iteration: {count+1}" and f"Maxvalue: {v}, Iteration: {count+1}" at the end of minvalue and maxvalue functions, just before returning those values. I think it would be easier to understand.

I hppe I clarified your doubts.