[R] Timer on a function

[Prof Brian Ripley]

On Thu, 15 Mar 2012, Bert Gunter wrote:

> Bill et. al:
>
> 1. This is new to me. Thanks.
>
> 2. As I read the man page, this is not guaranteed to work if the model
> fitting function does not contain sufficient interrupts. Is that
> correct?

Yes.  If someone wrote a model fitting package in Fortran with no 
callbacks then it is not interruptible, including not by timers.  In 
that case all you can do is to kill the R session.

The author of setTimeLimit()

>
> -- Bert
>
> On Thu, Mar 15, 2012 at 4:14 PM, William Dunlap <wdunlap at tibco.com> wrote:
>> There is a setTimeLimit function in base.  It could be encapsulated into
>> the following to limit the time spent on an expression:
>>
>> timeOut <- function (expr, ...)  {
>>    on.exit(setTimeLimit())
>>    setTimeLimit(...)
>>    expr
>> }

Using setTimeLimit(transient = TRUE) is slightly simpler here.

>> E.g., with the following slow way to compute Euler's phi
>>   f <- function(n) sum(sapply(seq_len(n), function(i)1/i)) - log(n)
>> I get
>>  > timeOut(f(1e5), elapsed=1)
>>  [1] 0.5772207
>>  > timeOut(f(1e6), elapsed=1)
>>  Error in FUN(1:1000000[[711624L]], ...) : reached elapsed time limit
>> Use try() or tryCatch() to check for the error.  E.g.,
>>  > sapply(1:7, function(n)tryCatch(timeOut(f(10^n), elapsed=1), error=function(e)-1))
>>  [1]  0.6263832  0.5822073  0.5777156  0.5772657  0.5772207 -1.0000000 -1.0000000
>> You could look at 'e' in the error handler to see if it is a time out problem.
>>
>> Bill Dunlap
>> Spotfire, TIBCO Software
>> wdunlap tibco.com
>>
>>> -----Original Message-----
>>> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org] On Behalf
>>> Of Bert Gunter
>>> Sent: Thursday, March 15, 2012 3:05 PM
>>> To: Ramiro Barrantes
>>> Cc: r-help at r-project.org
>>> Subject: Re: [R] Timer on a function
>>>
>>> On Thu, Mar 15, 2012 at 2:24 PM, Ramiro Barrantes
>>> <ramiro at precisionbioassay.com> wrote:
>>>>    Hello,
>>>>
>>>> I have a program that consists of a loop fitting a function over many
>>> models.  Sometimes the fitting on a particular model takes minutes to converge.  Is there
>>> a way that I can limit the amount of time that R spends on a given model:
>>>
>>> AFAIK, no -- this is an OS level issue.
>>>
>>> Of course, most iterative fitting procedures have controls for the
>>> number of iterations, convergence criteria, etc. , but there is no
>>> awareness of timing except when the OS is interrogated, e.g. by
>>> ?proc.time or ?system.time . Such calls would have to be built into
>>> the fitting function or OS level services would have to be invoked to
>>> run the R process with timing limitations built in. See e.g. ?Rscript
>>> for one possible approach.
>>>
>>> Corrections or clever tricks to get around these perceived limitations
>>> welcomed, of course.
>>>
>>> -- Bert
>>>
>>> Cheers,
>>> Bert
>>>>
>>>> say if my line is:
>>>>
>>>> fittingFunction( func, model.1)
>>>>
>>>> can I have some function:
>>>>
>>>> stopIfUnderTime(  fittingFunction( func, model.1) , 5 )
>>>>
>>>> where stopIfUnderTime will return the result if it finishes under 5 seconds, or NA
>>> otherwise.
>>>>
>>>> Is there anything like this?  (just looked through the web but did not find anything)
>>>>
>>>> Thanks,
>>>> Ramiro
>>>>
>>>>
>>>>
>>>>        [[alternative HTML version deleted]]
>>>>
>>>> ______________________________________________
>>>> 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.
>>>
>>>
>>>
>>> --
>>>
>>> Bert Gunter
>>> Genentech Nonclinical Biostatistics
>>>
>>> Internal Contact Info:
>>> Phone: 467-7374
>>> Website:
>>> http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-
>>> biostatistics/pdb-ncb-home.htm
>>>
>>> ______________________________________________
>>> 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.
>
>
>
> -- 
>
> Bert Gunter
> Genentech Nonclinical Biostatistics
>
> Internal Contact Info:
> Phone: 467-7374
> Website:
> http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm
>
> ______________________________________________
> 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