How to vectorize this code?

1 view (last 30 days)
UCL student
UCL student on 26 Oct 2014
Commented: UCL student on 27 Oct 2014
I try to use bsxfun, but maybe it does not suit this purpose.
C = bsxfun(@function,A(AIterator),cell2mat(B(BIterator)))
A is n*6, where the iterator runs on n, and B is a cell with 3*3 doubles (length m). Thanks for the help.
  5 Comments
Jan
Jan on 26 Oct 2014
I'm deeply confused. What is the meaning of:
The function calculates an array based on the 6; and 3*3 doubles.
?? Please post a minimal running example, which demonstrates your problem.
UCL student
UCL student on 27 Oct 2014
So, simplified:
A = zeros(100,6);
B = zeros(3,3);
celll = cell(28);
for l = 1:28
celll{l} = B;
end
outM = cell(size(A,1),size(B,1));
for AIterator = 1:size(A,1)
for BIterator = 1:size(B,1);
outM{AIterator,BIterator} = blackboxfunc(A(AIterator,:), cell2mat(celll(BIterator)));
end
end
Blackboxfunc is unknown for me, I try to test how to call that (computational time). I even would like to test it using the gpu.

Sign in to comment.

Answers (1)

Guillaume
Guillaume on 26 Oct 2014
bsxfun can only return matrices, not cell arrays, and only operates on matrices as well. If your output has to be a cell array, you can't use bsxfun.
Without knowing more about function, the best that you could do is vectorise the inner loop with:
outM(AIterator, :) = cellfun(@(b) function(A(AIterator, :), cell2mat(b), B, 'UniformOutput', false);

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!