[R-sig-ME] Contribution to lme4Eigen

Joehanes, Roby (NIH/NHLBI) [F] roby.joehanes at nih.gov
Thu Mar 1 20:11:30 CET 2012


Hi lme4 developers,

I hope I can contribute to something positive to the community. I noticed that even in the latest version of lme4Eigen the lmer function only use Nelder-Mead optimizer. I would like to get Bobyqa optimizer back as an option (because I really like Bobyqa optimizer). So, I patched the code a little bit. I used the (hopefully) latest revision from SVN version 1631. I also added control for xst and xt factor multiplier (a FIXME item list). I hope this change is also acceptable. If you feel I am ignorant of your code style, I apologize.

Here is the code (only the lmer function). I clearly marked my changes. If you need a diff file with the current HEAD, I'll be happy to provide you with one. The function seems to be running happily without error or warning on my machine. I hope the changes can make it to the main trunk.

Sincerely,
Roby



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