Slicing and assigning values to a 3D array in matlab

Question

I try to assign values for the first 2 dimensions of a 2D array, while keeping the third dimension fixed. But I am getting an error.

Assignment has fewer non-singleton rhs dimensions than non-singleton subscripts

See the code snippet below (Note that I use allcomb() from here ).

Any idea why and how to fix?

A = zeros(2, 94, 10);
combs = allcomb(1:2,1:94);
A(combs(:,1), combs(:,2), 1) = 1:(2*94);

Note that my intention was to write a vectorized form of:

A = zeros(2, 94, 10);
combs = allcomb(1:2,1:94)
vals = 1:(2*94);
for k=1:length(vals)
    A(combs(k,1), combs(k,2), 1) = vals(k);
end

Answer 1

The rhs , right hand side, of the equation needs to have the same dimensionality as the lhs . in this case, you're trying to assign an N x 1 vector to an 2 x N x 1 array.

Try,

A = zeros(2, 94, 10);
combs = allcomb(1:2,1:94);
val = (1:2*94)';
A(combs(:,1), combs(:, 2), 1) = [val, val];

Good luck!

Answer 2

You can try something like this:

A = zeros(2, 94, 10);
B = A;
combs = allcomb(1:2, 1:94);
vals = 1:(2*94);
for k=1:length(vals)
      A(combs(k, 1), combs(k, 2), 1) = vals(k);
end

% here it starts
val = repmat(1:94, 94, 2);
B(combs(:, 1), combs(:, 2), 1) = [val; val + 94];

isequal(A, B)   % returns 1

This should give you the desired result.

Answer 3

I found the short and elegant implementation I was looking for, using sparse() initialization:

A = zeros(2, 94, 10);
combs = allcomb(1:2,1:94);
vals = 1:(2*94);

%% Here it is: 
A(:,:,1) = sparse(combs(:,1), combs(:,2), vals);