7楼: Originally posted by zzzx2008 at 2012-04-23 23:19:12:
兄弟的想法 不错  值得考虑
看看下面 这个 如何？
clear;
clc;
N=10;%二维 矩阵行列值
zero_num=5;%小于零的个数
a=randint(N,N);%0 1 随机
b=reshape(a,1,N*N);% 转化一维 处理方便
c=b*2-1;%+ - 1
...

function A = MyMatrix(M, N, K)
    A = ones(1, M*N);
    t = randperm(M*N);
    A(t(1:K)) = -1;
    A = reshape(A, M, N);

MyMatrix(100, 100, randi(100))