[R-SIG-Mac] bug in R.app GUI 1.40 for Mac when running R (Under development (unstable)

[Prof Brian Ripley]

Another issue: I believe that tail-area computations in mvtnorm use 
the random number generator: that calculation certainly sets 
.Random.seed.  So to be reproducible, you need to specify the seed.

We've over the years encountered occasional random segfaults in 
mvtnorm on several platforms (certainly x86_64 Linux and both Windows 
builds) that only occur for certain random number seeds.

But I ran that example 1000 times under valgrind (under my own R-devel 
build, which is very similar to the nightly builds), without any 
problem.

On Sun, 10 Jul 2011, Simon Urbanek wrote:

> Bill,
>
> I highly doubt that this has anything to do with the GUI as it's not 
> even involved in the trace. Also I can't reproduce it with the 
> latest nightly build even on 10.5.8:
>
>> library(mvtnorm)
>> sessionInfo()
> R Under development (unstable) (2011-07-09 r56345)
> Platform: i386-apple-darwin9.8.0/i386 (32-bit)
>
> locale:
> [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
>
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base
>
> other attached packages:
> [1] mvtnorm_0.9-9991
>> qmvnorm(.95,sigma=diag(2),tail="both")
> $quantile
> [1] 2.236497
>
> $f.quantile
> [1] 2.5831e-06
> attr(,"error")
> [1] 1e-15
> attr(,"msg")
> [1] "Normal Completion"
>
> $iter
> [1] 11
>
> $estim.prec
> [1] 6.103516e-05
>
>
> However, you setup seems fishy - there is no 1.41 GUI build for R-devel, so are you sure you installed the correct files? If you are not using the nightly builds, then you will need to provide a lot more detail - compilers used, exact configure, BLAS etc. - and then then you may have to look at the backtrace yourself since the CRAN setup doesn't seem at have that issue.
>
> Cheers,
> Simon
>
>
> On Jul 10, 2011, at 11:41 AM, William Revelle wrote:
>
>> Simon and the other developers of the wonderful R.Gui for the  Mac,
>>
>> The following code (anything using mvtnorm) reliably crashes R (under development)  when running from the GUI, but not when running  from the X11 window or under R-Studio.
>>
>>
>> I reported this several months ago but was told to wait for a newer release.   I believe I have the latest versions of R (unstable), the R GUI, and mvtnorm.
>>
>>> library(mvtnorm)
>>> sessionInfo()
>>
>>
>> R Under development (unstable) (2011-07-08 r56336)
>> Platform: i386-apple-darwin9.8.0/i386 (32-bit)
>>
>> locale:
>> [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
>>
>> attached base packages:
>> [1] stats     graphics  grDevices utils     datasets  methods   base
>>
>> other attached packages:
>> [1] mvtnorm_0.9-9991 psych_1.0.99
>>> qmvnorm(.95,sigma=diag(2),tail="both")
>>
>> *** caught bus error ***
>> address 0x0, cause 'non-existent physical address'
>>
>> Traceback:
>> 1: .Fortran("mvtdst", N = as.integer(n), NU = as.integer(df), LOWER = as.double(lower),     UPPER = as.double(upper), INFIN = as.integer(infin), CORREL = as.double(corrF),     DELTA = as.double(delta), MAXPTS = as.integer(x$maxpts),     ABSEPS = as.double(x$abseps), RELEPS = as.double(x$releps),     error = as.double(error), value = as.double(value), inform = as.integer(inform),     PACKAGE = "mvtnorm")
>> 2: probval.GenzBretz(algorithm, n, df, lower, upper, infin, corr, corrF, delta)
>> 3: probval(algorithm, n, df, lower, upper, infin, corr, corrF, delta)
>> 4: mvt(lower = lower, upper = upper, df = 0, corr = corr, delta = mean,     algorithm = algorithm, ...)
>> 5: pmvnorm(lower = low, upper = upp, mean = args$mean, corr = args$corr,     sigma = args$sigma, algorithm = algorithm)
>> 6: f(upper, ...)
>> 7: uniroot(pfct, interval = interval)
>> 8: qmvnorm(0.95, sigma = diag(2), tail = "both")
>>
>>
>> Good luck and thanks for all the great work.
>>
>> Bill
>>
>> --
>> William Revelle		http://personality-project.org/revelle.html
>> Professor			http://personality-project.org
>> Department of Psychology             http://www.wcas.northwestern.edu/psych/
>> Northwestern University	http://www.northwestern.edu/
>> Use R for psychology                       http://personality-project.org/r
>> It is 6 minutes to midnight	http://www.thebulletin.org
>>
>> _______________________________________________
>> R-SIG-Mac mailing list
>> R-SIG-Mac at r-project.org
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>>
>>
>
> _______________________________________________
> R-SIG-Mac mailing list
> R-SIG-Mac at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>

-- 
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