[R] Evaluation "conflict" in combination of replicate() and rexp()as variable inside a function

Prof Brian Ripley ripley at stats.ox.ac.uk
Sat May 24 18:02:50 CEST 2008


Don't use replicate() like this.  You have '...' in your *expression*, and 
it really should only be used in the body of a function.

But if you want to understand why

> debug(rexp)
> rdistr <- function(distr, ...) replicate(1, distr(n = 1, ...))
> rdistr(rexp)
debugging in: distr(n = 1, ...)
debug: .Internal(rexp(n, 1/rate))
Browse[1]> match.call()
distr(n = 1, rate = ..1)
Browse[1]> rate
[1] 0

The point being that in the context that ... is evaluated, it does not 
refer to your function arguments.

The behaviour is completely consistent: you called rnorm(n=1,0),
rexp(n=1, 0) and runif(n=1, 0).

replicate() is conveneint at the top level, but in programming you should 
manage replication yourself without the non-standard evaluation that use 
of eval.parent() (inside replicate()) implies.  However, if you must

rdistr <- function(distr, ...)
{
     d <- function() distr(n=1, ...)
     replicate(1, d())
}

will work.

On Fri, 23 May 2008, Gerrit Eichner wrote:

> Dear userRs,
>
> "playing around" with combinations of replicate() and random number 
> generating functions inside a self-defined "wrapper" function I encounterd a 
> puzzling behaviour.
>
> The following are intentionally simple (and rather nonsense-) examples to 
> isolate the relevant aspects. Please, note the seemingly "inconsistent" 
> behaviour for the second call of rdistr() (with distr = rexp) for which I 
> have not found any explanation (yet):
>
>> rdistr <- function( distr, ...) replicate( 1, distr( n = 1, ...))
>
>> rdistr( distr = rnorm)
> [1] -0.8889223
>
>> rdistr( distr = rexp)
> [1] NaN
> Warning message:
> In distr(n = 1, ...) : NAs produced
>
>> rdistr( distr = runif)
> [1] 0.8444856
>
>
> Trying to discover the reason for this observation, I looked into the code of 
> replicate() and realized that this seems to be an issue of evaluation schemes 
> (and maybe also of variable scoping), but I didn't delve deeper into it, yet. 
> Can anybody more competent enlight me or give me a hint where to search, 
> please? Thank you!
>
> Best regards  --  Gerrit
>
> PS:
>
>> sessionInfo()
> R version 2.6.2 (2008-02-08)
> x86_64-unknown-linux-gnu
>
> locale:
> LC_CTYPE=de_DE.UTF-8;LC_NUMERIC=C;LC_TIME=de_DE.UTF-8;LC_COLLATE=de_DE.UTF-8;LC_MONETARY=de_DE.UTF-8;LC_MESSAGES=de_DE.UTF-8;LC_PAPER=de_DE.UTF-8;LC_NAME=C;LC_ADDRESS=C;LC_TELEPHONE=C;LC_MEASUREMENT=de_DE.UTF-8;LC_IDENTIFICATION=C
>
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base
>
> loaded via a namespace (and not attached):
> [1] rcompgen_0.1-17
>
>
> ---------------------------------------------------------------------
> AR Dr. Gerrit Eichner                          Mathematical Institute
> gerrit.eichner at math.uni-giessen.de   Justus-Liebig-University Giessen
> Tel: +49-(0)641-99-32104          Arndtstr. 2, 35392 Giessen, Germany
> Fax: +49-(0)641-99-32029               http://www.math.uni-giessen.de
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

-- 
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



More information about the R-help mailing list