递归解法:s2[0...j]是否可以由s1[0...i]转换 isScramble(s2[0...j], s1[0...i]),可以分解成 i 个子问题(i 其实等于j,因为两个字符串长度不一样,肯定不能互相转换):
( isScramble(s2[0...k], s1[0...k]) && isScramble(s2[k+1...j], s1[k+1...i]) ) || ( isScramble(s2[0...k], s1[i-k...i]) && isScramble(s2[k+1...j], s1[0...i-k-1]) ),(k = 0,1,2 ... i-1,k相当于字符串的分割点)
只要一个子问题返回ture,那么就表示两个字符串可以转换。代码如下:
class Solution {
public:
bool isScramble(string s1, string s2) {
return isScrambleRecur(s1,s2);
}
bool isScrambleRecur(string &s1, string &s2)
{
string s1cop = s1, s2cop = s2;
sort(s1cop.begin(), s1cop.end());
sort(s2cop.begin(), s2cop.end());
if(s1cop != s2cop)return false;//两个字符串所含字母不同
if(s1 == s2)return true;
int len = s1.size();
for(int i = 1; i < len; i++)//分割位置
{
string s1left = s1.substr(0, i);
string s1right = s1.substr(i);
string s2left = s2.substr(0, i);
string s2right = s2.substr(i);
if(isScrambleRecur(s1left, s2left) && isScrambleRecur(s1right, s2right))
return true;
s2left = s2.substr(0, len-i);
s2right = s2.substr(len-i);
if(isScrambleRecur(s1left, s2right) && isScrambleRecur(s1right, s2left))
return true;
}
return false;
}
};
动态规划解法:
递归解法有很多重复子问题,比如s2 = rgeat, s1 = great 当我们选择分割点为0时,要解决子问题 isScramble(reat, geat),再对该子问题选择分割点0时,要解决子问题 isScramble(eat,eat);而当我们第一步选择1作为分割点时,也要解决子问题 isScramble(eat,eat)。相同的子问题isScramble(eat,eat)就要解决2次。
动态规划用数组来保存子问题,设dp[k][i][j]表示s2从j开始长度为k的子串是否可以由s1从i开始长度为k的子串转换而成,那么动态规划方程如下
- 初始条件:dp[1][i][j] = (s1[i] == s2[j] ? true : false)
- dp[k][i][j] = ( dp[divlen][i][j] && dp[k-divlen][i+divlen][j+divlen] ) || ( dp[divlen][i][j+k-divlen] && dp[k-divlen][i+divlen][j] ) (divlen = 1,2,3...k-1, 它表示子串分割点到子串起始端的距离) ,只要一个子问题返回真,就可以停止计算
代码如下:
class Solution {
public:
bool isScramble(string s1, string s2) {
if(s1.size() != s2.size())
return false;
const int len = s1.size();
bool dp[len+1][len][len];
for(int i=0;i<len;i++) {
for(int j=0;j<len;j++) {
dp[1][i][j] = (s1[i] == s2[j]) ? true:false;
}
}
for(int k=2;k<=len;k++) {
for(int i=0;i<=len-k;i++) {
for(int j=0;j<=len-k;j++) {
dp[k][i][j] = false;
for(int div = 1;div<k && !dp[k][i][j];div++) {
dp[k][i][j] = (dp[div][i][j] && dp[k-div][i+div][j+div]) || (dp[div][i][j+k-div] && dp[k-div][i+div][j]);
}
}
}
}
return dp[len][0][0];
}
};