[R] Fitting mixed-effects models with lme with fixed error term variances

[Virgilio Gómez-Rubio]

Hello again,

And excuse me for not replying in real time. I am subscribed in digest
mode and the data that generated my question are in a computer not
connected to the Internet.

Douglas Bates replied to my original mail and he requested the
following:

"Can you be more specific about which parameters you want to fix and
which you want to vary in the optimisation?"


My model is

Y_i = X beta + u_i + e_i

where u_i ~ N(0, sigma2_u I) are the random effects and 
e_i ~ N(0, SIGMA) the error terms. What I would like to fix 
is matrix SIGMA, which, in addition, is diagonal. Hence, I only want to  
estimate beta and sigma2_u. The reason why I want to fix SIGMA is because,
otherwise, the model is unidentifiable (I have one observation per group)
and I can get an estimate of SIGMA by other means.

Is that possible? If not, could I use any trick to do so?

Many thanks.

Best regards,

Virgilio