Is there a way to optimise following loop using 1:end notation?

for ii = 1:NumberElements
    for jj = 1:NumberElements
        E(:,ii,jj) = fourier(A(:,ii,jj));
        E1(:,ii,jj) = fft(A(:,ii,jj));
    end
end

Answer 1

[Bjorn Gustavsson]

You should get an identical fft in one step if you do this:

E1mat = fft(A,[],1);

Here the third argument is the dimension of A to Fourier-transform along. As for the first expression the only fourier I find in my matlab-installation is in the symbolic toolbox, and that cannot be the one you're calling above. One thing you definitely should do is pre-allocate E - when you loop from 1 and assign values to a new row or column matlab needs to reallocate memory-space every step in the loop. If you cannot figure out how big the array will be but can safely run the loops in opposite direction that is a lazy-man's alternative:

for ii = NumberElements:-1:1
  for jj = NumberElements:-1:1
    E(:,ii,jj) = fourier(A(:,ii,jj));
  end
end

That way you make space for thefull array first time around the loops.

HTH