Replacing 1s in a sparse matrix by a random int.

Hi everyone I have a adjacency matrix from which I want to create a weight matrix. For each nonzero entry I'll put a random number in the weight matrix. The reason I want to this way is to reduce memory requirements. for instance if n=100'000. I create a sparse binary adjacency matrix with no problem. If I create my Weight matrix by W=randi(maxweight,n,n) it will consume a lot memory. Instead, i want it to be sparse as the adjacency matrix.

Answer 1

[Teja Muppirala]

If "A" is your sparse matrix, then:

A(find(A)) = randi(maxweight,nnz(A),1);

will replace every element with a random integer.

[Mehmet Candemir]

Thanks that was really fast