Python停止基于时间的迭代加深搜索

Question

我目前正在开发迭代深化搜索，希望根据时间停止。我的代码如下所示：

def __init__(self, name='Agent', time_limit_s=5):
        self.name = name
        self.time_limit_s = time_limit_s
        self.stop_event = threading.Event()

def move(self, board):
        '''
        This method is called by the game engine to get the next move from the bot. It is time-limited,
        and uses the update_best_move method to get the best move.
        '''
        # Perform the iterative deepening search

        # Set up a shared variable to store the best result found
        best_move = None
        best_result_lock = threading.Lock()

        # Set up a separate thread to run the search loop
        search_thread = threading.Thread(target=self.run_search, args=(board, best_move, best_result_lock))

        # Start the search thread and run the search
        search_thread.start()

        # Wait for the search thread to finish or raise a TimeoutException
        search_thread.join(timeout=self.time_limit_s) # timeout does not end the execution, merely waits for this long

        if search_thread.is_alive():
            print('Thread still alive')
            self.stop_event.set() # Set the stop event to signal the search thread to stop running

        return best_move

    def run_search(self, board, best_move, best_result_lock):
        '''
        This method is called by the iterative_deepening_search method to run the search loop.
        '''
        cur_best_move = None
        depth = 1
        try:
            while not self.stop_event.is_set(): 
                cur_best_move, done = self.update_best_move(board, depth, cur_best_move)
                with best_result_lock:  # set the value of best_move in thread safe manner
                    best_move = cur_best_move
                if done or self.stop_event.is_set():
                    break

                depth += 1
        except TimeoutException:
            print('Time limit reached')
            with best_result_lock:  # set the value of best_move in thread safe manner
                best_move = cur_best_move

然而，以下情况不断发生：

时间限制：5秒

• 迭代 1：0 秒
• 迭代 2：0 秒
• 迭代 3：1 秒
• 迭代 4：3 秒

现在我们几乎用完了可用的 5 秒，但是因为 5 秒还没有过去，所以开始了另一个迭代

• 迭代 5：12 秒

我们现在总共让 16 秒过去了，而不是 5 秒。你可以想象情况可能更糟。我如何调整我的代码，以便它在设置的

time_limit_s

秒后明确停止？很高兴解决方案。

我尝试使用

time

和 threading.Event() 来指示代码停止，但它一直在超过时间限制，如上所述。