[R] quinModel S != R

[Douglas Bates]

"Dieter Menne" <dieter.menne at menne-biomed.de> writes:


> Running quinModel (Pinheiro/Bates page 380) on R (current release, windows)
> gives:
> 
> Nonlinear mixed-effects model fit by maximum likelihood
>   Model: conc ~ quinModel(Subject, time, conc, dose, interval, lV, lKa,
> lCl)
>   Data: Quinidine
>   Log-likelihood: -497
>   Fixed: lV + lKa + lCl ~ 1
>     lV    lKa    lCl
>  5.382 -0.273  2.470
> 
> Random effects:
>  Formula: list(lV ~ 1, lCl ~ 1)
>  Level: Subject
>  Structure: Diagonal
>               lV   lCl Residual
> StdDev: 4.43e-12 0.321     0.76
> 
> lV is confusing, since the book examples says it should be 0.31173.

There are different optimization algorithms being used in R and S.
The value from S is more sensible.  Trying to prevent the optimization
algorithm in R from taking unreasonably large steps inside nlme is one
of the things we will work on.

> Other question: is the source of nlme_one_comp_open (or similar) available?
> I would like to learn how to do these models in c, but could not find the
> code. (To avoid RTFM: I know about r-ext.pdf, but a special modelxXX example
> would make life easier).

You will need the source for the nlme package, which is available from
the CRAN archives.  That routine is defined in nlme/src/nlme.c
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._