Please confirm or correct my evaluation of this code

[masijade]

    [ignore,maxind] = max(abs(U),[],1);
    negloc = (U(maxind + (0:N_f:(N_f-1)*N_f)) < 0);
    U(:,negloc) = -U(:,negloc);

(N_f is the number of columns in the matrix (actually the number of rows, but it is a square matrix, so same-same))

I am assuming that these two line multiply all the elements of the columns where the maximum absolute value comes from a negative number by -1, without modifying or eliminating any other columns. Is this correct?

    [U,D] = eig((correlation_matrix+correlation_matrix')/2);
    D = diag(D);
    norma=diag(1./sqrt(U.^2*D));

Okay, my evaluation of the last line above
Square the elements of Matrix U, multiply that D, and then "squareroot" all the elements, then apply the diag method.
My question is, is D a vector or a "square" matrix at that point. I am assuming a Vector as norma is used as a square matrix later in the program, and D should be, after "eig" a "square" matrix, and so a vector after the call to diag. I find it very bad form to use "D" as the variable for both a vector and a matrix, if that's the case though.

I cannot run this code, as I do not have a matlab environment. I am porting this code to something else, and I just want to make sure that my "assumptions" are correct, or get them corrected, if they are not.

I am also posting this question here and here.

 

[masijade]

Okay, I guess you can't edit posts here (or maybe you can and I'm just blind, neither would surprise here) In any case, here are the two "links"

http://www.matlabtutorials.com/math...ect-my-evaluation-of-this-code/msg434/#msg434

http://www.kluid.com/mlib/viewtopic.php?f=13&t=5071&p=8972#p8972

 

[masijade]

Nevermind, got help from somewhere else. Thanks anyway.