[R-sig-ME] lme4

Ben Bolker bbolker at gmail.com
Fri Oct 24 01:10:44 CEST 2014

On 14-10-23 02:07 PM, Alejandra Tapia wrote:
> Dear Prof. Ben,
> I'm writing you because I have the following question.
> For R package lme4 version 0.999999-2 the glmer function uses by default
> |bobyqa| for first phase and Nelder-Mead for second phase?
> Sincerely,
> -- 
> Alejandra

  I believe that old versions of lme4 (pre 1.0) used the nlminb
optimizer for all optimizations (there was no choice of optimizer).

  I would suggest being extremely careful with versions of lme4 prior to
1.0 used on R 3.1.0 or greater -- people have reported problems, which
we haven't had the opportunity to track down.

  Ben Bolker

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