Need some clarification

Hey all, I've got some code that's trying to solve a system of 6 equations using fsolve. It runs a loop, because the six equations represent variables in a data set, so multiple iterations are desired. Here is the code itself:

%data.csv should be defined as the file with measured values.
data = csvread('data_c.csv');
assert (mod(size(data, 1), 6) == 0, ...
      'Input data must have an integer multiple of 6 rows');
assert (size(data, 2) == 2, ...
      'Input data must have exactly two columns.');
nsys = size(data, 1) / 6;
soln = zeros(nsys, 6);
options=optimset('MaxFunEvals',1e10,'MaxIter',25000);
for k = 1 : nsys,
  F = c_generate(data(6*(k-1) + (1:6), 1:end));
  guess = [1 1 1 1 1 1];
  soln(k, :) = fsolve(F, guess,options);   
end

fid=fopen('results.csv','w');
fprintf(fid,'%5.5f, %5.5f, %5.5f, %5.5f, %5.5f, %5.5f\n',soln);
fclose(fid);

And here is the function file:

function F = c_generate(data)
  assert (ndims(data) ==2, ...
          'System parameters ''p'' must be 2D matrix.');
    assert (all(size(data) ==[6,2]), ...
            'System parameters must be 6-by-2 matrix.');
    y = data(:,1);
    n = data(:,2);

      F = @(x) (x(1)+(x(2).*(y.^2))+(x(3)/(y.^2))+(x(4)/(y.^4))+(x(5)/(y.^6))+(x(6)/(y.^8))-n.^2);
end

When I run it, I get an error:

??? Error using ==> plus
Matrix dimensions must agree.

Error in ==>
c_generate>@(x)(x(1)+(x(2).*(y.^2))+(x(3)/(y.^2))+(x(4)/(y.^4))+(x(5)/(y.^6))+(x(6)/(y.^8))-n.^2)
at 9
    F = @(x)
    (x(1)+(x(2).*(y.^2))+(x(3)/(y.^2))+(x(4)/(y.^4))+(x(5)/(y.^6))+(x(6)/(y.^8))-n.^2);

Which leads to the usual red cascade. Since I'm using the skeleton of an older function file to create this new one, I don't know what it means by matrix sizes must agree. Since the x's are being solved for, don't they represent 1x1 matrices? How should I fix this, what does this mean?

Answer 1

[Matt Tearle]

Just from looking at the code, it looks as though you're dividing a scalar by a column vector:

y = data(:,1);
...
...x(3)/(y.^2)...

This will result in a matrix right-divide, which will return a row vector. But

x(2).*(y.^2)

will give a column vector. And adding these together, of course, gives our good friend "Matrix dimensions must agree".

I suspect you want to do ./ instead of /.

[Alex]

Ha, perfect! I can't believe I forgot the (.). Wow, always pays to have someone watching your back.

Thanks!

[Matt Tearle]

Heh. I once spent a couple of hours trying to debug an algorithm because I had * instead of .* Total n00b move, even after more years of MATLAB than I care to remember.