Matlab: Product Support 1109 - Code Vectroization Guide

Original post here.

This section might be useful:

Suppose you want to multiply each column of a very large matrix by the same vector. There is a way to do this using sparse matrices that saves space and can also be faster than matrix construction...

F = rand(1024,1024);
X = rand(1024,1);

Y = F * diag(sparse(X));
Y = diag(sparse(X)) * F;

On the other hand, most of times bsxfun would be a more efficient solution.