求最长公共子序列长度 以及对应子序列
若给定序列 ABCDEFG , 则 ABC , CDE, ABCDEFG 等都是其自序列
package dynamic_programming;
import java.util.Scanner;
/**
* 求最长公共子序列长度 以及对应子序列
*
* 若给定序列 ABCDEFG , 则 ABC , CDE, ABCDEFG 等都是其自序列
*
* @author LiangGe
*
*/
public class LCSQuestion {
/**
* 计算最长公共子序列
* 由于在所考虑的子问题空间中,总共只有θ(m*n)个不同的子问题,因此,用动态规划算法自底向上地计算最优值能提高算法的效率
* 使用此方法每个子问题都会计算
*
* 时间复杂度为0(M*N)
*/
public static int getLcsLength(String str_x, String str_y, int lcsLengthArray[][], int lcsLengthMap[][]) {
int len_x = str_x.length();
int len_y = str_y.length();
// 边界,X字符串为空
for(int i=0; i<=len_x; i++) {
lcsLengthArray[i][0] = 0;
}
// 边界,Y字符串为空
for(int j=0; j<=len_y; j++) {
lcsLengthArray[0][j] = 0;
}
for(int i=1; i<=len_x; i++) {
for(int j=1; j<=len_y; j++) {
// 母问题对应三个子问题,求出三个子问题各自的最优解,再判断母问题最优解
if(str_x.charAt(i-1) == str_y.charAt(j-1)) {
lcsLengthArray[i][j] = lcsLengthArray[i-1][j-1] + 1;
lcsLengthMap[i][j] = 1;
} else {
if(lcsLengthArray[i-1][j] >= lcsLengthArray[i][j-1]) {
lcsLengthArray[i][j] = lcsLengthArray[i-1][j];
lcsLengthMap[i][j] = 2;
} else {
lcsLengthArray[i][j] = lcsLengthArray[i][j-1];
lcsLengthMap[i][j] = 3;
}
}
}
}
return lcsLengthArray[len_x][len_y];
}
/**
* 利用递归计算最长公共子序列,但是由于计算过的子问题会做备忘录,所以不会出现重复计算问题
* 使用此方法 不是所有子问题都需要计算一遍,只有具体问题需要用到的时候才会计算子问题
*
*/
public static int getLcsLength2(String str_x, String str_y, int len_x, int len_y, int lcsLengthArray[][], int lcsLengthMap[][]) {
int curLcsLength;
if(lcsLengthArray[len_x][len_y] != 0) {
curLcsLength = lcsLengthArray[len_x][len_y];
// 边界问题 len_x = 0 || len_y = 0
} else if(len_x == 0 || len_y == 0) {
lcsLengthArray[len_x][len_y] = 0;
curLcsLength = 0;
} else if(str_x.charAt(len_x-1) == str_y.charAt(len_y-1)) {
curLcsLength = getLcsLength2(str_x, str_y, len_x-1, len_y-1, lcsLengthArray, lcsLengthMap) + 1;
lcsLengthMap[len_x][len_y] = 1;
} else {
int len1 = getLcsLength2(str_x, str_y, len_x-1, len_y, lcsLengthArray, lcsLengthMap);
int len2 = getLcsLength2(str_x, str_y, len_x, len_y-1, lcsLengthArray, lcsLengthMap);
if(len1 >= len2) {
curLcsLength = len1;
lcsLengthMap[len_x][len_y] = 2;
} else {
curLcsLength = len2;
lcsLengthMap[len_x][len_y] = 3;
}
}
lcsLengthArray[len_x][len_y] = curLcsLength;
return curLcsLength;
}
/**
* 逆向求解具体最长子序列
*/
public static String getLcsString(int lcsLength, String str_x, String str_y, int lcsLengthMap[][]) {
int i = str_x.length();
int j = str_y.length();
char str[] = new char[lcsLength];
for(int k=lcsLength-1; k>=0;) {
if(lcsLengthMap[i][j] == 1) {
str[k--] = str_x.charAt(i-1);
i--; j--;
} else if(lcsLengthMap[i][j] == 2) {
i --;
} else {
j --;
}
}
return String.valueOf(str);
}
public static void main(String args[]) {
Scanner scanner = new Scanner(System.in);
int lcsLengthArray[][];
int lcsLengthMap[][];
String lcsString;
int lcsLength;
while(scanner.hasNext()) {
String curStrLine = scanner.nextLine();
String str[] = curStrLine.split(" ");
String str_x = str[0];
String str_y = str[1];
lcsLengthArray = new int[str_x.length()+1][str_y.length()+1];
lcsLengthMap = new int[str_x.length()+1][str_y.length()+1];
/* // 非递归求解
lcsLength = getLcsLength(str_x, str_y, lcsLengthArray, lcsLengthMap);
System.out.println("lcsLength : " + lcsLength);
lcsString = getLcsString(lcsLength, str_x, str_y, lcsLengthMap);
System.out.format("%s\n", lcsString);*/
// 递归求解
int len_x = str_x.length();
int len_y = str_y.length();
lcsLength = getLcsLength2(str_x, str_y, len_x, len_y, lcsLengthArray, lcsLengthMap);
System.out.println("lcsLength : " + lcsLength);
lcsString = getLcsString(lcsLength, str_x, str_y, lcsLengthMap);
System.out.format("%s\n", lcsString);
}
}
}