alphabetaengine.cpp

中国象棋源程序,该程序采用多种搜索算法实现了人机对弈的功能

// AlphaBetaEngine.cpp: implementation of the CAlphaBetaEngine class.
//
//////////////////////////////////////////////////////////////////////

#include "stdafx.h"
#include "AlphaBetaEngine.h"

//////////////////////////////////////////////////////////////////////
// Construction/Destruction
//////////////////////////////////////////////////////////////////////

CAlphaBetaEngine::CAlphaBetaEngine()
{

}

CAlphaBetaEngine::~CAlphaBetaEngine()
{

}

CAlphaBetaEngine::SearchAGoodMove(BYTE position[][9])
{
	memcpy(CurPosition,position,90);    //将当前局面复制到CurPosition

	m_nMaxDepth=m_nSearchDepth;			//设定搜索深度
	AlphaBeta(m_nMaxDepth,-20000,20000);//进行alphabeta搜索
	m_umUndoMove.cmChessMove=m_cmBestMove;
	m_umUndoMove.nChessID=MakeMove(&m_cmBestMove);	

	memcpy(position,CurPosition,90);	//将走过的棋盘传出
}

int CAlphaBetaEngine::AlphaBeta(int nDepth,int alpha,int beta)
{
	int score;
	int Count,i;
	BYTE type;

	i=IsGameOver(CurPosition,nDepth);//检查是否游戏结束
	if(i!=0)
		return i;//结束，返回估值

	//叶子节点取估值
	if(nDepth<=0)
		return m_pEval->Eveluate(CurPosition,(m_nMaxDepth-nDepth)%2,m_nUserChessColor);

	//此函数找出当前局面所有可能的走法，然后放进m_pMG ->m_MoveList当中
	Count=m_pMG->CreatePossibleMove(CurPosition,nDepth,(m_nMaxDepth-nDepth)%2,m_nUserChessColor);
	if(nDepth==m_nMaxDepth)
	{
		//在根节点设定进度条
		m_pThinkProgress->SetRange(0,Count);
		m_pThinkProgress->SetStep(1);
	}

	//对所有可能的走法
	for(i=0;i<Count;i++)
	{		
		if(nDepth==m_nMaxDepth)
			m_pThinkProgress->StepIt();//走进度条

		type=MakeMove(&m_pMG->m_MoveList[nDepth][i]);  //将当前局面应用此走法，变为子节点的局面
		score=-AlphaBeta(nDepth-1,-beta,-alpha);       //递归搜索子节点
		UnMakeMove(&m_pMG->m_MoveList[nDepth][i],type);//将此节点的局面恢复为当前节点

		if(score>alpha)
		{
			alpha=score;//保留极大值

			//靠近根节点时保留最佳走法
			if(nDepth==m_nMaxDepth)
				m_cmBestMove=m_pMG->m_MoveList[nDepth][i];
		}
		
		if(alpha>=beta)
			break;//剪枝，放弃搜索剩下的节点
	}

	return alpha;//返回极大值
}