MATLAB: How to solve this

parallel computing

a(:,:,1,j+1) = conv2(hj,hj,a(:,:,1,j),'same');
dx(:,:,1,j+1) = conv2(delta,gj,a(:,:,1,j),'same');
dy(:,:,1,j+1) = conv2(gj,delta,a(:,:,1,j),'same');

how do i convert this convolution function in parfor?

Best Answer

Excuse me sir,first of all can i know why and whats the meaning of:

Only the convolutions with respect to gj can be parallelized via parfor?
atmp=a(:,:,1,j); %excessive optimization perhaps, eliminate one subsref() call ?
hj(1:2^j:end)=h; ?
a(:,:,1,J+2)=a(:,:,1,1); %pre-allocation ?

Sorry for asking too much since i'm new learner.

Related Solutions

MATLAB: How to make the pool to parallel? “Error: The variable hj in a parfor cannot be classified.” We always get that error

The problem here is that you are not completely overwriting hj on each iteration of the parfor loop, and so the iterations are order dependent. parfor can only run if the iterations can be proved (by the MATLAB infrastructure that supports parfor) to be independent. You can convince parfor that your iterations are independent (at least with respect to hj) by completely overwriting it at the start of each loop iteration. I.e. replace

hj(1:lhj) = 0;

with

hj = zeros(1, lhj);

however this is not completely equivalent to the code you had previously, so it might not work as you expect.

MATLAB: Parfor error with 3d matrix

The parfor index variable can only appear in one index position inside the body of the loop.

Create a 2D array and extend it after the loop

G = ones(11,11,5);
G2 = ones(11, 5);
parfor i = 1 : 10
    v = zeros(1,5);
    for j = 1 : 5
        v(j) = rand(1);
    end
    G2(i, :) = v;
end
for i = 1 : 11;
    G(i, i, :) = G2(i, :);
end

Best Answer

Related Solutions

MATLAB: How to make the pool to parallel? “Error: The variable hj in a parfor cannot be classified.” We always get that error

MATLAB: Parfor error with 3d matrix

Related Question