对于给定的棋盘游戏,我使用alpha-beta修剪编写了以下minimax search algorithm
:
class AlphaBeta(SearchAlgos):
iter_time = timelib()
if self.goal(state):
return self.utility(state), state.move
if depth == 0:
return self.heuristicFunc(state), state.move
children = self.succ(state)
agent_to_move = self.turn(state)
if agent_to_move == maximizing_player:
cur_max = -np.inf
direction = None
for child in children:
if timelib() - iter_time > sent_time:
raise TimeoutError()
val = self.search(child, depth - 1, maximizing_player, sent_time - (timelib() - iter_time,alpha,beta ))
if val[0] > cur_max :
cur_max = val[0]
direction = child.move
alpha = max(cur_max,alpha)
if cur_max >= beta:
return (np.inf, direction)
to_ret = (cur_max, direction)
else:
cur_min = np.inf
direction = None
for child in children:
if timelib() - iter_time > sent_time:
raise TimeoutError()
val = self.search(child, depth - 1, maximizing_player, sent_time - (timelib() - iter_time,alpha ,beta ))
if val[0] < cur_min:
cur_min = val[0]
direction = child.move
beta = min(cur_min, beta)
if cur_min <= alpha:
return (-np.inf, direction)
to_ret = (cur_min, direction)
return to_ret
该算法同时返回极小极大值和最佳方向。我现在试图弄清楚以下几点:
成功运行后,返回的方向被删除(意外),我现在只有返回的值。 这就给我留下了一个问题:如果我已经有了返回值,在再次运行以找到方向之前,我可以对算法进行哪些更改以改进它
我考虑在alpha-beta参数中发送返回值,这样只会到达相关的最终移动,但是这似乎不起作用
目前没有回答
相关问题 更多 >
编程相关推荐