[Rd] strange lme errors (PR#8477)

[Prof Brian Ripley]

lme is not part of R, but of the contributed package nlme.  So if you 
thought this is an lme error, you reported it in the wrong place.  See the 
BUGS section in the FAQ.

On Thu, 12 Jan 2006 wb at arb-phys.uni-dortmund.de wrote:

> Full_Name: Wilhelm Bernhard Kloke
> Version: R-2.2.1
> OS: FreeBSD-5.3-i386
> Submission from: (NULL) (195.253.22.63)
>
>
> Since 2.2.0 I am getting strange lme errors like
>
>> lme(ampl ~ gapf * bl,hframe2,~1|VP)
> Fehler in lme.formula(ampl ~ gapf * bl, hframe2, ~1 | VP) :
>        See PORT documentation.  Code (27)
>
> I have no clue how to understand this. The specification worked well until
> 2.1.0.
> I tried a second dataset with the same error message.

Try traceback().  It will probably tell you the message comes from nlminb, 
in which case try ?nlminb.  The problem here is the change in lme from 
optim to nlminb which seems to have broken many examples.

Searching for that message would have lead you to suggestions that it 
comes from a bug in your compiler.  RSiteSearch() lead me to

http://finzi.psych.upenn.edu/R/Rhelp02a/archive/60118.html

And googling showed that your OS has gcc 3.4.2 with its broken Fortran 
compiler.  So it seems this is not a bug in lme nor in R but in your OS.

> At least the said "PORT documentation" should be included in the R 
> distribution, just in case.

The reference is!  More specifically the URL given points you to
http://netlib.bell-labs.com/cm/cs/cstr/153.pdf
section 3.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595