Strassen矩阵乘法

一、思路

假设n是2的幂。将矩阵A，B和C中每一矩阵都分块成4个大小相等的子矩阵，每个子矩阵都是的方阵。由此可将方程C=AB重写为

定义

则

时间复杂度

二、C++代码：

//C++
#include 

using namespace std;

//矩阵类
class matrix {
private:
    int **mp;//矩阵数组
    int n;//矩阵的阶
public:
    //创建零矩阵
    explicit matrix(int n) {
        this->n = n;
        mp = new int *[n];
        for (int i = 0; i < n; ++i) {
            mp[i] = new int[n];
            for (int j = 0; j < n; ++j) {
                mp[i][j] = 0;
            }
        }
    }

    //使用数组创建矩阵
    matrix(int n, int **mp) {
        this->n = n;
        this->mp = new int *[n];
        for (int i = 0; i < n; ++i) {
            this->mp[i] = new int[n];
            for (int j = 0; j < n; ++j) {
                this->mp[i][j] = mp[i][j];
            }
        }
    }

    //以矩阵A的1/4部分创建矩阵
    matrix(matrix A, int p1, int p2) {
        n = A.n / 2;
        mp = new int *[n];
        for (int i = 0; i < n; ++i) {
            mp[i] = new int[n];
            for (int j = 0; j < n; ++j) {
                mp[i][j] = A.mp[i + (p1 - 1) * (n)][j + (p2 - 1) * (n)];
            }
        }
    }

    matrix operator+(const matrix &b) {
        matrix c(this->n);
        for (int i = 0; i < n; ++i) {
            for (int j = 0; j < n; ++j) {
                c.mp[i][j] = this->mp[i][j] + b.mp[i][j];
            }
        }
        return c;
    }

    matrix operator-(const matrix &b) {
        matrix c(this->n);
        for (int i = 0; i < n; ++i) {
            for (int j = 0; j < n; ++j) {
                c.mp[i][j] = this->mp[i][j] - b.mp[i][j];
            }
        }
        return c;
    }

    void show() {
        for (int i = 0; i < n; ++i) {
            for (int j = 0; j < n; ++j) {
                cout << mp[i][j] << " ";
            }
            cout << endl;
        }
    }

    //四个子矩阵合并成一个矩阵
    void merge(matrix a11, matrix a12, matrix a21, matrix a22) {
        for (int i = 0; i < n / 2; i++) {
            for (int j = 0; j < n / 2; j++) {
                mp[i][j] = a11.mp[i][j];
            }
            for (int j = n / 2; j < n; j++) {
                mp[i][j] = a21.mp[i][j - n / 2];
            }
        }
        for (int i = n / 2; i < n; i++) {
            for (int j = 0; j < n / 2; j++) {
                mp[i][j] = a12.mp[i - n / 2][j];
            }
            for (int j = n / 2; j < n; j++) {
                mp[i][j] = a22.mp[i - n / 2][j - n / 2];
            }
        }
    }

    //乘法
    static matrix multiply(matrix a, matrix b) {
        matrix c(a.n);
        if (a.n == 1) {
            c.mp[0][0] = a.mp[0][0] * b.mp[0][0];
        } else {
            matrix a11(a, 1, 1), a12(a, 1, 2), a21(a, 2, 1), a22(a, 2, 2);
            matrix b11(b, 1, 1), b12(b, 1, 2), b21(b, 2, 1), b22(b, 2, 2);
            matrix m1 = multiply(a11, b12 - b22);
            matrix m2 = multiply(a11 + a12, b22);
            matrix m3 = multiply(a21 + a22, b11);
            matrix m4 = multiply(a22, b21 - b11);
            matrix m5 = multiply(a11 + a22, b11 + b22);
            matrix m6 = multiply(a12 - a22, b21 + b22);
            matrix m7 = multiply(a11 - a21, b11 + b12);
            c.merge(m5 + m4 - m2 + m6, m1 + m2, m3 + m4, m5 + m1 - m3 - m7);
        }
        return c;
    }
};

int main() {
    int n = 4;
    int array[4][4] = {{1,  2,  3,  4},
                       {5,  6,  7,  8},
                       {9,  10, 11, 12},
                       {13, 14, 15, 16}};

    //将array转换成动态数组
    int **a = new int *[n];
    for (int i = 0; i < n; ++i) {
        a[i] = new int[n];
        for (int j = 0; j < n; ++j) {
            a[i][j] = array[i][j];
        }
    }

    //创建矩阵
    matrix m(n, a);
    //自己乘自己
    matrix::multiply(m, m).show();
}

三、运行结果

90 202 314 426
100 228 356 484
110 254 398 542
120 280 440 600