Non-diagonal sampling covariance with lme4

Thanks a lot, Douglas and Ben. Right, pre-whitening will give me what I'm
looking for!

Asaf

You can "pre-whiten" the response and the model matrices by multiplying by
either the right or left inverse Cholesky factor of V.  (I always need to
write out the equations before i can determine if I should use the left or
the right factor.)

On Mon Nov 17 2014 at 9:11:45 AM Asaf Weinstein <
asafw.at.wharton at gmail.com> wrote:

Hi all,

I would like to obtain ML (or REML) estimates for theta, beta, sigsq in

Y|B=b ~ N( Zb + Xbeta, sigsq*V )
B ~ N( 0,Sigma(theta) )

where V is a known covariance matrix. lmer() does exactly that for V=I_n
(the n-by-n identity matrix); I wonder if there is a way to specify an
arbitrary covariance matrix.

Thanks so much,

Asaf

        [[alternative HTML version deleted]]

_______________________________________________
R-sig-mixed-models at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models

Non-diagonal sampling covariance with lme4

Thread (4 messages)