[R-sig-ME] Problem with lmer crashing on x64 server

Douglas Bates bates at stat.wisc.edu
Thu Dec 16 13:55:52 CET 2010


Sorry for the delayed response (end of semester and all that).  You
must bear in mind that you are estimating many variance-covariance
parameters (at least 21 and perhaps more if cohort90 is a factor).  It
is quite likely that the estimates are converging to a singular
variance-covariance matrix and the estimates are not meaningful in
either case.

On Wed, Nov 17, 2010 at 5:21 AM, Camille Szmaragd
<Camille.Szmaragd at bristol.ac.uk> wrote:
> Dear all,
>
> I am fitting a fairly simple 2 levels hierarchical  normal model with random
> slopes:
>
>
>
> lmer(score ~ cohort90 + female + sclassf + (1 + cohort90 + sclassf |
> schoolid), data = mydata, REML = FALSE), with sclassf being a 4 category
> factor.
>
>
>
> I can fit this model without a problem on a Windows Desktop (and R 2.11.0)
> and get sensible parameter estimates, matching the estimates obtained with
> other software packages.

>
>
> However when I tried to fit the same model on a 64 bit server, running R
> x64 2.12.0 (and the latest update of lme4), I get the following error
> message at the end of the model output:
>
>
>
> #####
>
> Warning message:
>
> In mer_finalize(ans) : iteration limit reached without convergence (9)
>
> #####
>
>
>
> I appreciate if anybody has a suggestion/idea as to why this is happening.
>
>
>
>  Many thanks in advance
>
>
>
> Camille
>
>
>
>
>
>
>
>
>
> ****---Save Paper - Do you need to print this message?---****
>
>
>
>  Dr Camille Szmaragd
>
>  Research Associate
>
>  School of Veterinary Science
>
>  University of Bristol
>
>  Langford BS40 5DU
>
>
>
>  Camille.Szmaragd at bristol.ac.uk
>
>
>
>
>        [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>




More information about the R-sig-mixed-models mailing list