Negamax Alpha Beta搜索
我正在寻求实施Negamax Alpha-Beta搜索算法的帮助。 我的C并不是最好的,但是我已经有了一个工作框架来玩这款游戏,而且我的Alpha-Beta搜索一直在返回似乎是最后一招的问题,而不是返回最佳举措找到。
以下是我的alpha测试搜索源文件。 如果任何人都可以帮助,我将不胜感激。 我不是在寻找任何人“做我的作业”,因为很多人害怕,因为我真的很感兴趣,而且我已经用其他语言实施过搜索技术。
再次,任何帮助将不胜感激。
最好的祝福
#include "alphabeta.h"
struct tree_node * create_root(struct tree_node * root, struct Position *pos)
{
root = (struct tree_node*) malloc(sizeof(struct tree_node));
root->pos = copy(pos);
root->first_child = NULL;
root->next_sibling = NULL;
root->best_move = NULL;
return root;
}
struct tree_node * create_node(struct tree_node *node)
{
struct tree_node *new_node = (struct tree_node*) malloc(sizeof(struct tree_node));
new_node->pos = copy(node->pos);
new_node->first_child = NULL;
new_node->next_sibling = NULL;
new_node->best_move = NULL;
return new_node;
}
struct tree_node * add_child(struct tree_node *parent)
{
if (parent->first_child == NULL)
{
parent->first_child = create_node(parent);
return parent;
}
// Loop through the list of existing
// children for the next gap
struct tree_node *ptr = NULL;
ptr = parent->first_child;
while (ptr->next_sibling != NULL)
{
ptr = ptr->next_sibling;
}
// create the new child and append to the list
ptr->next_sibling = create_node(parent);
return ptr;
}
struct tree_node * destroy_tree(struct tree_node * root)
{
if (root->first_child != NULL)
{
destroy_tree(root->first_child);
}
// loop through the siblings and delete
struct tree_node *ptr = root->next_sibling;
struct tree_node *ptr2 = root;
while (ptr != NULL)
{
free(ptr2);
ptr2 = ptr;
ptr = ptr->next_sibling;
}
return root;
}
int negamax(struct tree_node *node, int color, int *num_static,
int *num_cutoff, int height, int achievable, int hope)
{
printf("Called negamaxn");
// first costruct the list of moves
struct move_list *list = NULL;
list = moves(color, node->pos, list);
int num_moves = size_of_list(list);
int best_score = NEG_INF;
if (height == 0 || num_moves == 0)
{
*num_static = *num_static + 1;
return static_evaluator(node->pos, color, num_moves);
}
else
{
int temp = 0;
while (list != NULL)
{
// Make the new move
struct tree_node *new_node;
//new_node = add_child(node);
new_node = create_root(new_node, node->pos);
make_move(new_node->pos, list->move);
print_board(new_node->pos);
list = list->prev;
// Recurse over alpha beta
temp = -negamax(new_node, color, num_static,
num_cutoff, height - 1,
-hope, -achievable);
destroy(new_node->pos);
free(new_node);
//destroy_node(new_node);
//new_node = NULL;
//printf("Returned from the recursive calln");
// set the new best move
/*
if (temp > achievable)
{
best_score = temp;
node->best_move = list->move;
printf("Best move: %sn", node->best_move->str_rep);
}
*/
if (temp >= hope)
{
*num_cutoff = *num_cutoff + 1;
return temp;
}
// set to the max value
if (achievable <= temp)
{
node->best_move = list->move;
printf("Best move: %sn", node->best_move->str_rep);
achievable = temp;
}
}
}
return achievable;
}
/*
* This is my evaluation function. The parameters
* to the function are the attributes of the position;
* The guard count, the king count and the number of
* moves available. I chose to weight the parameters
* offering the highest weight to the number of moves
*
* If there are no moves available, I decided to remove
* points from the function. It will never go below zero,
* but this should permit it to give an extremely low
* value for positions which have no available moves
*/
int static_evaluator(struct Position *pos, int color, int num_moves)
{
int evaluation = 1;
// Calculate the value, while applying
// weights to the parameters
switch (color)
{
case BLACK_COL:
if (num_moves == 0)
{
evaluation += 1;
if (pos->num_black_kings > 0)
{
evaluation += pos->num_black_kings - 1;
}
}
else
{
evaluation += num_moves * 3;
evaluation += pos->num_black_kings * 2;
evaluation -= pos->num_white_kings * 2;
}
// Always count the guards
evaluation += pos->num_black_guards;
evaluation -= pos->num_white_guards;
break;
case WHITE_COL:
if (num_moves == 0)
{
evaluation += 1;
if (pos->num_white_kings > 0)
{
evaluation += pos->num_white_kings - 1;
}
}
else
{
evaluation += num_moves * 3;
evaluation += pos->num_white_kings * 2;
evaluation -= pos->num_black_kings * 2;
}
evaluation += pos->num_white_guards;
evaluation -= pos->num_black_guards;
break;
}
return evaluation;
}
链接地址: http://www.djcxy.com/p/9627.html
上一篇: Negamax Alpha Beta search
下一篇: How do I center text horizontally and vertically in a TextView?