CODEWITHSUNDEEP
matlabmatrixvectorizationbsxfun
user1285419
user1285419

Reputation: 2225

alternative way for element-by-element multiplication for matrices in matlab

I am working on a big matrix multiplication. I have a big matrix A (at least 5000x5000) and a column vector V (5000x1). In my code, each V is going to multiply each column of A element by element. I did it with a loop

K = zeros(5000, 5000);
for n=1:5000
  K(:, n) = V.*A(:, n);
end

but it is so slow. So I create a big matrix with each column assigned as V such that 

MV=repmat(V,1,5000);
K = MV.*A;

it is fast but it waste too much of memory. When size of the matrix increase, it uses too much of memory. Is that any idea to use less memory but fast?

Upvotes: 0

Views: 436

Answers (1)

Shai
Shai

Reputation: 114786

classic bsxfun

K = bsxfun( @times, A, V );

Alternatively, you might want to look at James Tursa's MTIMESX (found in FEX).

Upvotes: 4

Related Questions