[Eeglablist] generalized EVD in CSP ?

[Vadlamani Samhitha]

Hi list , I am performing CSP on EEG data . I have computed covariance
matrices for class 1 and class 2 ( C1 and C2 ) . Now while obtaining
 filter coefficients , I  have been suggested to perform generalized
eigenvalue decomposition to get filter coefficients such that W^T C1 W = D
and W^T(C1+C2) W = I .(Identity matrix )
So , I have used [W,D]=eig(C1,C1+C2) ; where W is my filter matrix , D is
the  eigenvalue matrix for C1 .
after doing so , when I manually try to check if the above two conditions
are satisfied , they don't seem so . I have been struggling to find out
what went wrong . I would be really grateful if someone could  suggest what
I am missing here . Thank you in advance .
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20161018/6fef124f/attachment.html>