How to do a quick division between a column array and a matrix?

Hello, All,

I want to divid a column array b (N x 1) by each column in a big sparse matrix A (N x M). My code is as follows:

dividCol = zeros(N,M);

for i = 1 : M

nonCol = find(abs(A(:,i))>0);

dividCol(nonCol,i) = b(nonCol)./A(nonCol,i);

end

But it is very slow. I do know if there exist a faster approach to write this code.

Thanks a lot in advance.

Benson

Answer 1

[Omer Yasin Birey]

Hi Benson, try this

% b = randi(25,25,1);
% A = randi(25,25,5);
[row,col] = find(abs(A(:,:))>0);
dividCol = bsxfun(@rdivide,b(row,1),A(row,col));

[Benson Gou]

Hi, Omer,

Thanks for your help. I tested your code and found the following:

A is 7 x 14, b is 7 x 1. But I got dividCol is 61 x 61. The size of dividCol is not correct.

Thanks a lot again,

Benson

[Omer Yasin Birey]

Yes I checked the code and it does every operation twice. I had to use loop and updated the code. Try it instead.

% b = randi(25,7,1);
% A = randi(25,7,14);
for i = 1:M
nonCol = find(abs(A(:,i))>0);
dividCol = bsxfun(@rdivide,b(nonCol),A(nonCol,i));
end

[Benson Gou]

Omer,

Thanks for your prompt reply.

Is it possible to do this without using a loop? because loop normally takes more time.

Benson