I have uploaded release 0.6-1 of the lme4 package to the incoming area
on CRAN. I expect it will be transferred to the archive within a
couple of days, after which Uwe's Windows build daemon should be able
to get it to build a Windows version. The lme4 package itself is a
pure R package (i.e. it does not contain any code to be compiled) but
it depends on the Matrix_0.8-7 package which has a considerable amount
of C code in it.
This package contains the version of lme that I spoke of at useR!2004.
Using sparse matrix methods - in particular, Tim Davis' LDL package -
we are able to fit models with crossed random effects quickly and
effectively. This package also contains an implementation of GLMM for
Generalized Linear Mixed Models using either Penalized
Quasi-Likelihood (PQL) or ML estimation using the Laplacian
approximation to the marginal likelihood. At present method 'Laplace'
is considerably slower than method = 'PQL'. We would
recommend using PQL for model building and determining final parameter
estimates using method = 'Laplace'.
Feedback is welcome.
--
Douglas Bates bates at stat.wisc.edu
Statistics Department 608/262-2598
University of Wisconsin - Madison http://www.stat.wisc.edu/~bates/
_______________________________________________
R-packages mailing list
R-packages at stat.math.ethz.ch
https://www.stat.math.ethz.ch/mailman/listinfo/r-packages
