[R] lmer() causes segfault

[Ben Bolker]

Bertolt Meyer <bmeyer <at> sozpsy.uzh.ch> writes:

> 
> Hello lmer() - users,
> 
> A call to the lmer() function causes my installation of R (2.11.1 on  
> Mac OS X 10.5.8) to crash and I am trying to figure out the problem.

  [snip snip]

> detach("package:nlme")
> library(lme4)
> 
> mod1 <- lmer(performance ~ time + (time | GroupID/StudentNumber), data  
> = dataset.long, na.action = na.omit)
> 
> However, this call results in a segfault:
> 
> *** caught segfault ***
> address 0x154c3000, cause 'memory not mapped'
> 
> and a lengthy traceback. I can reproduce this error. It also occurs  
> when I don't load nlme before lme4. Can someone tell me what I am  
> doing wrong? Any help is greatly appreciated.

   This may well be a bug in lmer.  There have been a number of
fussy computational issues with the lme4 package on the Mac platform.
If it is at all possible, please (1) post the results of sessionInfo()
[which will in particular specify which version of lme4 you are using];
(2) possibly try this with the latest development version of lme4, from
R-forge, if that's feasible (it might be necessary to build the package
from source), and most importantly:

 (3) create a reproducible (for others) example -- most easily by
posting your data on the web somewhere, but if that isn't possible
by simulating data similar to yours (if it doesn't happen with another
data set of similar structure, that's a clue -- it says it's some more
particular characteristic of your data that triggers the problem) and

 (4) post to to *either* the R-sig-mac or the R-sig-mixed-models list,
where the post is more likely to come to the attention of those who
can help diagnose/fix ...

  good luck
    Ben Bolker