In sci.math MBALOVER <firstname.lastname@example.org> wrote: ... > thanks but no equation for computing length of the major axis.
It could be as simple as 2/lambda_1^2 where lambda_1 is the smallest eigenvalue. Depends on what you can easily do or have done to your data first in the way of centering and normalising. You can sometimes forget the normalising and use the sum of the eigenvalues (that may or may not be the same as the Tr of the original matrix, depending on application area) as a final scale factor. Usual lawyer talk if not all eigenvalues are real and positive.
-- > Kooks don't accept evaporation without a book reference? ROTFLOL! Huh? I can't find your cite .... -- BONZO@27-32-240-172 [daily nymshifter], 14 Oct 2011 11:56 +1100