我一直在学习递归思考，第一个打动我的自然解决方案如下。一个更简单的问题是找到一个短一个字母的字符串的排列。我将假设，并相信我的每一根纤维，我的函数可以正确地找到一个字符串的排列，比我目前正在尝试的字符串短一个字母。

Given a string say 'abc', break it into a subproblem of finding permutations of a string one character less which is 'bc'. Once we have permutations of 'bc' we need to know how to combine it with 'a' to get the permutations for 'abc'. This is the core of recursion. Use the solution of a subproblem to solve the current problem. By observation, we can see that inserting 'a' in all the positions of each of the permutations of 'bc' which are 'bc' and 'cb' will give us all the permutations of 'abc'. We have to insert 'a' between adjacent letters and at the front and end of each permutation. For example

我们有bc

“a”+“bc”=“abc”

“b”+“a”+“c”=“bac”

“b”+“a”=“b”

对于'cb'我们有

a + b = acb

“c”+“a”+“b”=“cab”

“cb”+“a”=“cb”

下面的代码片段将说明这一点。下面是该代码片段的工作链接。

def main():
    result = []
    for permutation in ['bc', 'cb']:
        for i in range(len(permutation) + 1):
            result.append(permutation[:i] + 'a' + permutation[i:])
    return result


if __name__ == '__main__':
    print(main())

完整的递归解将是。下面是完整代码的工作链接。

def permutations(s):
    if len(s) == 1 or len(s) == 0:
        return s
    _permutations = []
    for permutation in permutations(s[1:]):
        for i in range(len(permutation) + 1):
            _permutations.append(permutation[:i] + s[0] + permutation[i:])
    return _permutations


def main(s):
    print(permutations(s))


if __name__ == '__main__':
    main('abc')

Java中一个非常基本的解决方案是使用递归+设置(以避免重复)，如果你想存储和返回解决方案字符串:

public static Set<String> generatePerm(String input)
{
    Set<String> set = new HashSet<String>();
    if (input == "")
        return set;

    Character a = input.charAt(0);

    if (input.length() > 1)
    {
        input = input.substring(1);

        Set<String> permSet = generatePerm(input);

        for (String x : permSet)
        {
            for (int i = 0; i <= x.length(); i++)
            {
                set.add(x.substring(0, i) + a + x.substring(i));
            }
        }
    }
    else
    {
        set.add(a + "");
    }
    return set;
}

我一直在学习递归思考，第一个打动我的自然解决方案如下。一个更简单的问题是找到一个短一个字母的字符串的排列。我将假设，并相信我的每一根纤维，我的函数可以正确地找到一个字符串的排列，比我目前正在尝试的字符串短一个字母。

Given a string say 'abc', break it into a subproblem of finding permutations of a string one character less which is 'bc'. Once we have permutations of 'bc' we need to know how to combine it with 'a' to get the permutations for 'abc'. This is the core of recursion. Use the solution of a subproblem to solve the current problem. By observation, we can see that inserting 'a' in all the positions of each of the permutations of 'bc' which are 'bc' and 'cb' will give us all the permutations of 'abc'. We have to insert 'a' between adjacent letters and at the front and end of each permutation. For example

我们有bc

“a”+“bc”=“abc”

“b”+“a”+“c”=“bac”

“b”+“a”=“b”

对于'cb'我们有

a + b = acb

“c”+“a”+“b”=“cab”

“cb”+“a”=“cb”

下面的代码片段将说明这一点。下面是该代码片段的工作链接。

def main():
    result = []
    for permutation in ['bc', 'cb']:
        for i in range(len(permutation) + 1):
            result.append(permutation[:i] + 'a' + permutation[i:])
    return result


if __name__ == '__main__':
    print(main())

完整的递归解将是。下面是完整代码的工作链接。

def permutations(s):
    if len(s) == 1 or len(s) == 0:
        return s
    _permutations = []
    for permutation in permutations(s[1:]):
        for i in range(len(permutation) + 1):
            _permutations.append(permutation[:i] + s[0] + permutation[i:])
    return _permutations


def main(s):
    print(permutations(s))


if __name__ == '__main__':
    main('abc')

public class Permutation 
{ 
public static void main(String[] args) 
{ 
    String str = "ABC"; 
    int n = str.length(); 
    Permutation permutation = new Permutation(); 
    permutation.permute(str, 0, n-1); 
} 

/** 
* permutation function 
* @param str string to calculate permutation for 
* @param l starting index 
* @param r end index 
*/
private void permute(String str, int l, int r) 
{ 
    if (l == r) 
        System.out.println(str); 
    else
    { 
        for (int i = l; i <= r; i++) 
        { 
            str = swap(str,l,i); 
            permute(str, l+1, r); 
            str = swap(str,l,i); 
        } 
    } 
} 

/** 
* Swap Characters at position 
* @param a string value 
* @param i position 1 
* @param j position 2 
* @return swapped string 
*/
public String swap(String a, int i, int j) 
{ 
    char temp; 
    char[] charArray = a.toCharArray(); 
    temp = charArray[i] ; 
    charArray[i] = charArray[j]; 
    charArray[j] = temp; 
    return String.valueOf(charArray); 
} 

} 

使用递归。

依次尝试每个字母作为第一个字母，然后使用递归调用找到剩余字母的所有排列。 基本情况是，当输入是空字符串时，唯一的排列就是空字符串。

倒计时Quickperm算法的通用实现，表示#1(可伸缩，非递归)。

/**
 * Generate permutations based on the
 * Countdown <a href="http://quickperm.org/">Quickperm algorithm</>.
 */
public static <T> List<List<T>> generatePermutations(List<T> list) {
    List<T> in = new ArrayList<>(list);
    List<List<T>> out = new ArrayList<>(factorial(list.size()));

    int n = list.size();
    int[] p = new int[n +1];
    for (int i = 0; i < p.length; i ++) {
        p[i] = i;
    }
    int i = 0;
    while (i < n) {
        p[i]--;
        int j = 0;
        if (i % 2 != 0) { // odd?
            j = p[i];
        }
        // swap
        T iTmp = in.get(i);
        in.set(i, in.get(j));
        in.set(j, iTmp);

        i = 1;
        while (p[i] == 0){
            p[i] = i;
            i++;
        }
        out.add(new ArrayList<>(in));
    }
    return out;
}

private static int factorial(int num) {
    int count = num;
    while (num != 1) {
        count *= --num;
    }
    return count;
}

它需要list，因为泛型不能很好地使用数组。