[R] Rprof causing R to crash

[Prof Brian Ripley]

On 14/12/2012 13:22, Jon Olav Skoien wrote:
> Uwe,
>
> I am unfortunately not able to upgrade to R 2.15.2 right now, but I have

Why not?   Note that is part of the R-help contract: we only offer any 
support for the current version of R (see the posting guide).

The posting guide also asked for 'at a minimum' information: we do not 
know (amongst other things) if this is 32- or 64-bit R.  (Your example 
works for me in both.)

> seen a similar problem with several older R versions. If you want to
> test with a shorter script, you can try the lines below. These provoke a
> crash from a fresh R session on my machine (R 2.15.1 Windows 7):
>
> Rprof()
> z = 1
> for (i in 1:1e8) z = z+1/i
>
> This runs without problems without the call to Rprof(). I was far from
> using all my memory with this call, but you could also check with a
> higher number of loops in case you can run these lines and it is somehow
> memory related. Just for comparison, I tried 1e9 loops after a call to
> Rprof() on one of our Linux servers (R 2.14.0) without any problems.

Which is irrelevant: a different version of R and a completely different 
implementation of Rprof.

>
> Jon
>
> On 12-Dec-12 17:06, Uwe Ligges wrote:
>>
>>
>> On 12.12.2012 00:05, Marian Talbert wrote:
>>> I'm trying to use Rprof() to identify bottlenecks and speed up a
>>> particullary
>>> slow section of code which reads in a portion of a tif file and compares
>>> each of the values to values of predictors used for model fitting.  I've
>>> written up an example that anyone can run.  Generally temp would be a
>>> section of a tif read into a data.frame and used later for other
>>> processing.
>>> The first portion which just records the time works in about 6
>>> seconds the
>>> second part causes RGui to immediately close with no error or
>>> warning.  Any
>>> advice on how to get Rprof to work or how to speed up this code would be
>>> greatly appreciated.  I'm using Windows 7 (which might be my problem)
>>> and R
>>> version 2.15.0.
>>
>> The problem is rather the R version: I cannot reproduce errors with a
>> recent R.
>>
>> Uwe Ligges
>>
>>
>>>
>>> CalcMESS<-function(tiff.entry,pred.vect){
>>> f<-sum(pred.vect<tiff.entry)/length(pred.vect)*100
>>>                if(is.na(f)) return(NA)
>>>                if(f==0)
>>> return((tiff.entry-min(pred.vect))/(max(pred.vect)-min(pred.vect))*100)
>>>                if(0<f & f<=50) return(2*f)
>>>                if(50<=f & f<100) return(2*(100-f))
>>>                if(f==100)
>>> return((max(pred.vect)-tiff.entry)/(max(pred.vect)-min(pred.vect))*100)
>>>                else return(NA)
>>> }
>>>
>>> train.dat <-
>>> data.frame(a=runif(200),b=runif(200),c=runif(200),d=runif(200))
>>> temp <-
>>> data.frame(a=runif(130000),b=runif(130000),c=runif(130000),d=runif(130000))
>>>
>>> pred.rng<-temp
>>> vnames.final.mod <- names(train.dat)
>>> nvars.final <- length(vnames.final.mod)
>>>
>>> start.time<-Sys.time()
>>>               for(k in 1:nvars.final){
>>>
>>> pred.range<-train.dat[,match(vnames.final.mod[k],names(train.dat))]
>>>
>>> pred.rng[,k]<-mapply(CalcMESS,tiff.entry=temp[,match(vnames.final.mod[k],names(temp))],MoreArgs=list(pred.vect=pred.range))
>>>
>>>                           }
>>> Sys.time()-start.time
>>>
>>>
>>> Rprof("C:\\temp\\mapply.out")
>>>               for(k in 1:nvars.final){
>>>
>>> pred.range<-train.dat[,match(vnames.final.mod[k],names(train.dat))]
>>>
>>> pred.rng[,k]<-mapply(CalcMESS,tiff.entry=temp[,match(vnames.final.mod[k],names(temp))],MoreArgs=list(pred.vect=pred.range))
>>>
>>>                           }
>>> Rprof(NULL)
>>>
>>>
>>>
>>> --
>>> View this message in context:
>>> http://r.789695.n4.nabble.com/Rprof-causing-R-to-crash-tp4652846.html
>>> Sent from the R help mailing list archive at Nabble.com.
>>>
>>> ______________________________________________
>>> 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.
>>>
>>
>> ______________________________________________
>> 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.
>
> ______________________________________________
> 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