Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Already on GitHub? Sign in to your account

REF: handling of sigma in GLS #310

Open
josef-pkt opened this Issue Jun 12, 2012 · 0 comments

Comments

Projects
None yet
1 participant
Owner

josef-pkt commented Jun 12, 2012

see #295 handling of scalar sigma

I think the right solution should avoid cholsigmainv for sigma.ndim 0 or 1

future: FGLS might need a different way to handle sigma, using (nobs,nobs) matrices is not very efficient
WLS and GLSAR avoids going through the sigma handling in GLS.init
but that should be a different issue

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment