骰子游戏中的Expectiminimax

Question

我已经为骰子游戏实施了expectiminimax算法，但它肯定无法正常工作，你可以验证我是否正确实现了它？

至于现在，我可以选择随机骰子选择，我的AI的'智能'也会类似，但我不知道我的算法可能出错。

游戏的流程如下（对于那些不了解骰子游戏的人）：

1. 玩家掷骰子
2. 玩家选择他们想再投掷的骰子
3. 玩家掷骰子

我的算法如下，我将其分为4部分

maxDice = [x,x,x,x,x]
minDice = [x,x,x,x,x]

Max:
for combination in each possible combination of maxDice that max can throw again (including none):
 max = max(MaxChance(combination),max)

MaxChance(combination):
expected = 0
for actualCombination in each possible result of max dice throw:
 maxHand = maxDice - combination + actualCombination //these are set operations of course
 expected += ProbablityOfThrowing(actualCombination)*Min(maxHand)
return expected

Min(maxHand):
for combination in each possible combination of minDice that min can throw again (including none):
 min = min(MinChance(combination,maxHand),min)
return min

MinChance(minCombination,maxHand):
expected = 0
for actualCombination in each possible result of min dice throw:
 minHand = minDice-minCombination+actualCombination
 expected += ProbablityOfThrowing(actualCombination)*Evaluate(maxHand,minHand)
return expected

//for example
Evaluate(maxHand,minHand):
return ValueOfDiceHand(maxHand)-ValueOfDiceHand(minHand)