Why does the inner dimensions of the last line of my code not agree?

% subtract off the mean for each dimension
mn = mean(data,2);
mn2 = mean(data2,2);
data = double(data);
data2 = double(data2);
data = data - mn;
data2 = data2 - mn2;
% calculate the covariance matrix
covariance = cov(data,data2);
%covariance2 = (1 / (N-1)) * (data2'*data2);
% find the eigenvectors and eigenvalues
[PC, V] = eig(covariance);
% [PC2, V2] = eig(covariance2);
% extract diagonal of matrix as vector(so extracting the diagonal of 
%the diagonal matrix)
%V = diag(V);
%V2 = diag(V2);
% sort the variances in decreasing order this will allow the PCs to be
% ordered the -1 makes it in descending, otherwise standard is ascending order
%[junk, rindices] = sort(-1*V);
%The value of rindices is the number of columns in V
%V = V(rindices); 
%Making a PC matrix
%PC = PC(:,rindices);
% project the original data set
signals = PC * data;

I don't think anything is missing from my calculations and knowledge with PCA, but I get an error because the last line of my code does not allow me to get the signals once I project the zero meaned image on the image matrix that I want the signal of. Can someone explain me why? Cheers - Neo Cornel

[Image Analyst]

Don't double space code - just learn how to format it properly: http://www.mathworks.com/matlabcentral/answers/13205-tutorial-how-to-format-your-question-with-markup

So, what are the sizes of PC and data? If PC is d1-by-d2, then data must be d2-by-d3 (though it's possible d3 might equal d1 because there's no restriction on that.)

[Neo]

Sorry Anaylst. You sent the link before, forgot to use it. I hope you can read it well this time.

Let me include the workspace (attached). I SEE what the dimensions are and thus no why there is an error but my problem is, I don't see anything missing from this code, that will allow this error to occur.

[Walter Roberson]

Nothing is attached.

[Neo]

I'm looking at the attachment. But here it is again

[Neo]

It's a bit big, let me know if you can see it now.

[Neo]

the data matrix is 262441x1 double PC is [1,0;0,1] V is [0,0;0,0]

Answer 1

[Image Analyst]

You have:

signals = PC * data;

And PC is 2x2 and data is 262144x1. So you have a matrix multiplication of 2x2*262144x1. The inner dimensions are 2 and 262144 <== those are not the same numbers so you can't do that. Did you see my comment above about the d2 must be the same? Review your algorithm to make sure your equations are correct.

[Neo]

I mean I know that the error was the multiplication, I am asking where did the error occur in the code to make this happen?

[Neo]

The equations are correct and I don't believe I am missing anything so can someone tell me what is it that gives me the error? I should have stated that before accepting the answer.