使用python实现Strassen矩阵乘法算法

先将矩阵打包成一个类，重载加法和减法运算，就可以很简单地实现了，目前这个程序中还有一些冗余的地方，比如将行数和列数分别讨论，其实可以归为一个矩阵size。类的定义感觉需要改进，不过懒得做下去了

# -*- coding: utf-8 -*-def matrixproduct(a, b):    def matrixproductmask(mat_a, mat_b):        if mat_a.row == 1:            c11 = [[mat_a.mat_list[mat_a.row_list[0]][mat_a.col_list[0]] *                    mat_b.mat_list[mat_b.row_list[0]][mat_b.col_list[0]]]]            return Martrix(c11)        else:            mat_a11 = mat_a.divide('11')            mat_a12 = mat_a.divide('12')            mat_a21 = mat_a.divide('21')            mat_a22 = mat_a.divide('22')            mat_b11 = mat_b.divide('11')            mat_b12 = mat_b.divide('12')            mat_b21 = mat_b.divide('21')            mat_b22 = mat_b.divide('22')            s1 = mat_b12 - mat_b22            s2 = mat_a11 + mat_a12            s3 = mat_a21 + mat_a22            s4 = mat_b21 - mat_b11            s5 = mat_a11 + mat_a22            s6 = mat_b11 + mat_b22            s7 = mat_a12 - mat_a22            s8 = mat_b21 + mat_b22            s9 = mat_a11 - mat_a21            s10 = mat_b11 + mat_b12            p1 = matrixproductmask(mat_a11, s1)            p2 = matrixproductmask(s2, mat_b22)            p3 = matrixproductmask(s3, mat_b11)            p4 = matrixproductmask(mat_a22, s4)            p5 = matrixproductmask(s5, s6)            p6 = matrixproductmask(s7, s8)            p7 = matrixproductmask(s9, s10)            c11 = (p5 + p4 - p2 + p6)            c12 = (p1 + p2)            c21 = (p3 + p4)            c22 = (p5 + p1 - p3 - p7)            return matrixmerge(c11, c12, c21, c22)    mat_a = Martrix(a)    mat_b = Martrix(b)    product = matrixproductmask(mat_a, mat_b)    return product.mat_listdef matrixmerge(c11, c12, c21, c22):    mat_list = []    for i in c11.row_list:        mat_list.append(c11.mat_list[i]+c12.mat_list[i])    for i in c21.row_list:        mat_list.append(c21.mat_list[i]+c22.mat_list[i])    return Martrix(mat_list)class Martrix(object):    def __init__(self, *args):        if len(args) == 1 and isinstance(args[0], list):            self.mat_list = args[0]            self.row = len(args[0])            self.col = len(args[0][0])            self.row_list = range(self.row)            self.col_list = range(self.col)    def __add__(self, mat2):        mat_list = [[self.mat_list[self.row_list[i]][self.col_list[j]]+mat2.mat_list[mat2.row_list[i]][mat2.col_list[j]]                     for j in range(self.col)] for i in range(self.row)]        return Martrix(mat_list)    def __sub__(self, mat2):        mat_list = [[self.mat_list[self.row_list[i]][self.col_list[j]]-mat2.mat_list[mat2.row_list[i]][mat2.col_list[j]]                     for j in range(self.col)] for i in range(self.row)]        return Martrix(mat_list)    def divide(self, block):        result = Martrix()        result.mat_list = self.mat_list        result.row = self.row/2        result.col = self.col/2        dic = {'11': [self.row_list[:result.row], self.col_list[:result.col]],               '12': [self.row_list[:result.row], self.col_list[result.col:]],               '21': [self.row_list[result.row:], self.col_list[:result.col]],               '22': [self.row_list[result.row:], self.col_list[result.col:]]}        result.row_list = dic[block][0]        result.col_list = dic[block][1]        return resulta = [[1,4,8,7],[5,7,9,13],[3,6,8,11],[-1,-3,5,3]]b = [[4,8,-12,5],[2,1,9,4],[12,45,-21,5],[5,-1,4,7]]c = matrixproduct(a, b)print c