vectorization help

[William Krekeler]

I'm trying to vectorize an operation on a three dimensional matrix and would appreciate suggestions for improvements. I included an example of the original code and my poor initial attempt at vectorization. The input matrices are random to provide an easily replicable test case. The alternative vectorization case results in the wrong size matrix as a cell array, and before calls to cell2mat and reshape to fix the size my alternative is slower than my original code set.

 

% original code set

RC = rand( 1, 448 );

Cube = rand( 160,512,448 );

tic

for lamda = 1: size( Cube, 3 )

   Cube(:,:,lamda) = ( RC(lamda)./ squeeze( Cube(:,:,lamda) ) );

end

toc

 

% incomplete vectorization attempt

tic

new = arrayfun( @(i)(bsxfun(@rdivide,RC(i),Cube(:,:,i))), 1:size(Cube,3), 'UniformOutput',false); 

toc

 

new is not equal to Cube in this case.  Please suggest better vectorization code or other improvements.

 

William Krekeler