[R-sig-ME] Purpose of .nloptr object?

[Phillip Alday]

This seems to be the docs getting slightly out of date in some sections
(as happens with any big piece of software). The actual nloptwrap
function in lme4 (now the default optimizer, IIRC) doesn't include that:

https://github.com/lme4/lme4/blob/c3d0c643d444785fad5e554380c028706f14e274/R/utilities.R#L1151-L1169

A wild guess on my part (until Ben Bolker has a few): that code was
there to debug writing a wrapper / show how you can further examine the
return value from nloptr to get more information and got left in, even
once the further debugging/examination was removed from the example.


Phillip

On 08/06/2021 14:19, Michael Chirico via R-sig-mixed-models wrote:
> I see this code snippet suggested in the performance vignette:
>
> https://cran.r-project.org/web/packages/lme4/vignettes/lmerperf.html
>
> nlopt <- function(par, fn, lower, upper, control) {
>     *.nloptr <<-* res <- nloptr(par, fn, lb = lower, ub = upper,
>         opts = list(algorithm = "NLOPT_LN_BOBYQA", print_level = 1,
>         maxeval = 1000, xtol_abs = 1e-6, ftol_abs = 1e-6))
>     list(par = res$solution,
>          fval = res$objective,
>          conv = if (res$status > 0) 0 else res$status,
>          message = res$message
>     )
> }
>
> That's writing an .nloptr object to .GlobalEnv (usually), but there's no
> mention of why and I don't see any usage of this object anywhere else in
> the code base:
>
> grep -Fr ".nloptr" lme4/R
>
> Am I missing something?
> Mike C
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-mixed-models using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models