首页 > 编程 > C++ > 正文

C++回溯法实例分析

2020-01-26 15:17:53
字体:
来源:转载
供稿:网友

本文实例讲述了C++的回溯法,分享给大家供大家参考之用。具体方法分析如下:

一般来说,回溯法是一种枚举状态空间中所有可能状态的系统方法,它是一个一般性的算法框架。

解向量a=(a1, a2, ..., an),其中每个元素ai取自一个有限序列集Si,这样的解向量可以表示一个排列,其中ai是排列中的第i个元素,也可以表示子集S,其中ai为真当且仅当全集中的第i个元素在S中;甚至可以表示游戏的行动序列或者图中的路径。

在回溯法的每一步,我们从一个给定的部分解a={a1, a2, ..., ak}开始,尝试在最后添加元素来扩展这个部分解,扩展之后,我们必须测试它是否为一个完整解,如果是的话,就输出这个解;如果不完整,我们必须检查这个部分解是否仍有可能扩展成完整解,如果有可能,递归下去;如果没可能,从a中删除新加入的最后一个元素,然后尝试该位置上的其他可能性。

用一个全局变量来控制回溯是否完成,这个变量设为finished,那么回溯框架如下,可谓是回溯大法之精髓与神器

bool finished = false;void backTack(int input[], int inputSize, int index, int states[], int stateSize){ int candidates[MAXCANDIDATE]; int ncandidates; if (isSolution(input, inputSize, index) == true) { processSolution(input, inputSize, index); } else { constructCandidate(input, inputSize, index, candidates, &ncandidates); for (int i = 0; i < ncandidates; i++) {  input[index] = candidates[i];  backTack(input, inputSize, index + 1);  if (finished)  return; } }}

不拘泥于框架的形式,我们可以编写出如下代码:

#include <iostream>using namespace std;char str[] = "abc";const int size = 3;int constructCandidate(bool *flag, int size = 2){ flag[0] = true; flag[1] = false; return 2;}void printCombine(const char *str, bool *flag, int pos, int size){ if (str == NULL || flag == NULL || size <= 0) return;  if (pos == size) { cout << "{ "; for (int i = 0; i < size; i++) {  if (flag[i] == true)  cout << str[i] << " "; } cout << "}" << endl; } else { bool candidates[2]; int number = constructCandidate(candidates); for (int j = 0; j < number; j++) {  flag[pos] = candidates[j];  printCombine(str, flag, pos + 1, size); } }}void main(){ bool *flag = new bool[size]; if (flag == NULL) return; printCombine(str, flag, 0, size); delete []flag;}

采用回溯法框架来计算字典序排列:

#include <iostream>using namespace std;char str[] = "abc";const int size = 3;void constructCandidate(char *input, int inputSize, int index, char *states, char *candidates, int *ncandidates){ if (input == NULL || inputSize <= 0 || index < 0 || candidates == NULL || ncandidates == NULL) return;  bool buff[256]; for (int i = 0; i < 256; i++) buff[i] = false; int count = 0; for (int i = 0; i < index; i++) { buff[states[i]] = true; } for (int i = 0; i < inputSize; i++) { if (buff[input[i]] == false)  candidates[count++] = input[i]; } *ncandidates = count; return;}bool isSolution(int index, int inputSize){ if (index == inputSize) return true; else return false;}void processSolution(char *input, int inputSize){ if (input == NULL || inputSize <= 0) return; for (int i = 0; i < inputSize; i++) cout << input[i]; cout << endl;}void backTack(char *input, int inputSize, int index, char *states, int stateSize){ if (input == NULL || inputSize <= 0 || index < 0 || states == NULL || stateSize <= 0) return;  char candidates[100]; int ncandidates; if (isSolution(index, inputSize) == true) { processSolution(states, inputSize); return; } else { constructCandidate(input, inputSize, index, states, candidates, &ncandidates); for (int i = 0; i < ncandidates; i++) {  states[index] = candidates[i];  backTack(input, inputSize, index + 1, states, stateSize); } }}void main(){ char *candidates = new char[size]; if (candidates == NULL) return; backTack(str, size, 0, candidates, size); delete []candidates;}

对比上述两种情形,可以发现唯一的区别在于全排列对当前解向量没有要求,而字典序对当前解向量是有要求的,需要知道当前解的状态!
八皇后回溯法求解:

#include <iostream>using namespace std;int position[8];void constructCandidate(int *input, int inputSize, int index, int *states, int *candidates, int *ncandidates){ if (input == NULL || inputSize <= 0 || index < 0 || candidates == NULL || ncandidates == NULL) return;  *ncandidates = 0; bool flag; for (int i = 0; i < inputSize; i++) { flag = true; for (int j = 0; j < index; j++) {  if (abs(index - j) == abs(i - states[j]))  flag = false;  if (i == states[j])  flag = false; } if (flag == true) {  candidates[*ncandidates] = i;  *ncandidates = *ncandidates + 1; } }/* cout << "ncandidates = " << *ncandidates << endl; system("pause");*/ return;}bool isSolution(int index, int inputSize){ if (index == inputSize) return true; else return false;}void processSolution(int &count){ count++;}void backTack(int *input, int inputSize, int index, int *states, int stateSize, int &count){ if (input == NULL || inputSize <= 0 || index < 0 || states == NULL || stateSize <= 0) return;  int candidates[8]; int ncandidates; if (isSolution(index, inputSize) == true) { processSolution(count); } else { constructCandidate(input, inputSize, index, states, candidates, &ncandidates); for (int i = 0; i < ncandidates; i++) {  states[index] = candidates[i];  backTack(input, inputSize, index + 1, states, stateSize, count); } }}void main(){ //初始化棋局 for (int i = 0; i < 8; i++) position[i] = i; int states[8]; int count = 0; backTack(position, 8, 0, states, 8, count); cout << "count = " << count << endl;}

希望本文所述对大家C++程序算法设计的学习有所帮助。

发表评论 共有条评论
用户名: 密码:
验证码: 匿名发表