In a for loop matrix dimension changes?

I'm having a problem. Here is my code:

UFR = 0.042;
alpha = 0.1;
N = 20;

%Wilson function
for p = 1:N
    for q = 1:N
        W = SWfunction(p,q,alpha,UFR);
    end
end

w = inv(W);

for t = 1:s
    for h = 1:N
    M(:,h) = exp(-(h)*R(t,h));
    MU(:,h) = exp(-UFR*h);
    Z(:,h) = zeros;

      m = transpose(M);
      mu = transpose(MU);
      z = transpose(Z);

      z(:,t) = w.*(m-mu);

      end
  end

When I run my code matlab tells me matrix dimensions do not agree when calculating z(:,t). But here's what confuses me, when I run the code without that line the matrix dimensions are fine for multiplication (20x20 with 20x1) but when I have that line it changes the variables m and mu to 1x1 matrices... I need z to be a matrix that performs that calculation for all t. Can someone help me?

Answer 1

[Roger Stafford]

If 'SWfunction' produces a scalar value, then 'w' will be a scalar, that is 1 x 1, since it is being overwritten 400 times in your first nested loops. That would produce an error at the line

   z(:,t) = w.*(m-mu);

You need to use p and q to index 'w' values, as in w(p,q), so as to produce a 20 x 20 matrix.

[Catarina]

Thank you! That worked to some extent - now it no longer has a problem with the matrix dimension mismatch and creates a matrix z with the size I need. But it only populates the last column of the matrix, the rest of the matrix is zeros. Any idea why this might be happening?

UFR = 0.042;
alpha = 0.1;
N = 20;

%Wilson function
for p = 1:N
    for q = 1:N
        W = SWfunction(p,q,alpha,UFR);
    end
end

w = inv(W);

for t = 1:s
    for h = 1:N
        M(t,h) = exp(-(h)*R(t,h));
        MU(t,h) = exp(-UFR*h);
        Z(t,h) = zeros;

          m = transpose(M);
          mu = transpose(MU);
          z = transpose(Z);

          mmu(h,t) = m(h,t)-mu(h,t);

          z(:,t) = w(p,q)*mmu(:,t);

      end
  end

Thanks again for taking the time to read and reply! :)

[Roger Stafford]

In the code you show in the above comment the quantity 'w' is still a 1 x 1 scalar. You need W(p,q) instead of W in the line

   W = SWfunction(p,q,alpha,UFR);

A second problem is that in the nested for loops that create M and MU you are taking the transpose of arrays that are only partially computed. That makes no sense. You need to be creating the 'm', 'mmu' and 'z' arrays in a later action after M and MU are fully computed.

Also the line

   z(:,t) = w(p,q)*mmu(:,t);

makes no sense. The p and q indices have long since become obsolete since they refer to indices in the first set of for loops.

[Catarina]

Ok I've figured it out now! Thank you for your help! :)