[R-sig-ME] Model specification help

[Douglas Bates]

I am experimenting with a version of lmer2 that should not use more
memory after the initial setup. This version requires a recent
development version of R-2.5.0.  I'll check the sources for the
package into the SVN archive.

https://svn.r-project.org/R-packages/trunk/lme4

The SVN archive for the development version of R is

https://svn.r-project.org/R/trunk

Please be warned that I may need to make an adjustment in the way that
lme4 gets its header files.  If so I will announce on this list that
you need to update both lme4 and R-devel.


On 3/12/07, Andrew Perrin <clists at perrin.socsci.unc.edu> wrote:
> Sadly, the parsimonious model seems to have overwhelmed memory, both using
> lmer and lmer2:
>
> > grades.lmer<-lmer(grade.pt ~ (1|stud.id) + (1|instr.id) + (1|cour.dep),
> +                   newgrades.stripped.df,
> +                   control =
> +                   list(gradient = FALSE, niterEM = 0, msVerbose = 1)
> +                   )
>    0  3.66128e+06: 0.0865649 0.0125233 0.000161387
>    1  3.57932e+06: 0.110177 0.0457235 0.999331
>    2  3.44933e+06: 0.603073 0.129709 0.999302
>    3  3.44882e+06: 0.604352 0.206648 0.997351
>    4  3.44777e+06: 0.681062 0.200279 0.997066
>    5  3.44748e+06: 0.736691 0.200143 0.989167
>    6  3.44746e+06: 0.744199 0.198462 0.988935
>    7  3.44744e+06: 0.759364 0.195889 0.988278
>    8  3.44744e+06: 0.762290 0.196680 0.987739
>    9  3.44744e+06: 0.764978 0.196422 0.986260
>   10  3.44744e+06: 0.764321 0.197399 0.983415
>   11  3.44744e+06: 0.763853 0.196192 0.980621
>   12  3.44744e+06: 0.758416 0.196809 0.946310
>   13  3.44744e+06: 0.758501 0.196269 0.911570
>   14  3.44739e+06: 0.756907 0.193748 0.355670
>   15  3.44738e+06: 0.761644 0.197153 0.320580
>   16  3.44738e+06: 0.762348 0.195416 0.202269
>   17  3.44738e+06: 0.765301 0.195340 0.219071
>   18  3.44738e+06: 0.762698 0.197282 0.235819
>   19  3.44738e+06: 0.764292 0.196909 0.226505
>   20  3.44738e+06: 0.764221 0.196833 0.226059
> Error: cannot allocate vector of size 505003 Kb
>
>
>
>
> > grades.lmer2<-lmer2(grade.pt ~ (1|stud.id) + (1|instr.id) +
> (1|cour.dep),
> + newgrades.stripped.df,
> + control =
> + list(gradient = FALSE, niterEM = 0, msVerbose = 1))
>    0  3.66128e+06: 0.294219 0.111907 0.0127038
>    1  3.44990e+06: 0.759028 0.360622 0.862465
>    2  3.44785e+06: 0.933722 0.418394 0.797191
>    3  3.44743e+06: 0.872076 0.462189 0.792549
>    4  3.44742e+06: 0.876257 0.448197 0.791479
>    5  3.44741e+06: 0.873351 0.445076 0.790576
>    6  3.44741e+06: 0.874786 0.443924 0.789611
>    7  3.44741e+06: 0.873415 0.443482 0.788113
>    8  3.44741e+06: 0.874736 0.443449 0.786510
>    9  3.44741e+06: 0.873514 0.443352 0.782539
>   10  3.44740e+06: 0.874955 0.434554 0.666426
>   11  3.44738e+06: 0.874233 0.446203 0.550557
>   12  3.44738e+06: 0.876339 0.439541 0.434311
>   13  3.44738e+06: 0.872474 0.444317 0.481580
>   14  3.44738e+06: 0.875574 0.444679 0.485606
>   15  3.44738e+06: 0.874206 0.444010 0.483564
>   16  3.44738e+06: 0.874194 0.443919 0.478097
>   17  3.44738e+06: 0.874232 0.443418 0.475652
>   18  3.44738e+06: 0.874182 0.443707 0.474796
>   19  3.44738e+06: 0.874195 0.443675 0.475530
>   20  3.44738e+06: 0.874195 0.443669 0.475407
> Error: cannot allocate vector of size 538066 Kb
>
>
>
>
> I'm going to try to start the same job on the solaris machine, but expect
> to have some problems because I don't have root access there.  Will let
> you know how it goes.
>
> Thanks,
> Andy
>
> ----------------------------------------------------------------------
> Andrew J Perrin - andrew_perrin (at) unc.edu - http://perrin.socsci.unc.edu
> Assistant Professor of Sociology; Book Review Editor, _Social Forces_
> University of North Carolina - CB#3210, Chapel Hill, NC 27599-3210 USA
> New Book: http://www.press.uchicago.edu/cgi-bin/hfs.cgi/00/178592.ctl
>
>
>
>