[Rd] R crashes using the em function of package mclust (PR#7719)

Brian D Ripley ripley at stats.ox.ac.uk
Mon Mar 7 10:15:04 CET 2005


Please read the FAQ and report this in the correct place:

   Bug reports on contributed packages should be sent first to the package
maintainer, and only submitted to the R-bugs repository by package
maintainers, mentioning the package in the subject line.

You are not the named maintainer.


On Mon, 7 Mar 2005 becker at kfs.oeaw.ac.at wrote:

> Hi,
>
> I got the same problem like
> http://tolstoy.newcastle.edu.au/R/devel/04/11/1204.html
>
> R crashes when I use the em function from the mclust package on
> univariate data and on a special case on bivariate data (when the matrix
> is not provided as written in the manual).
> It seems as if the problem is the format of the data to be analyzed.
>
> Operating System: Windows XP (SP2)
> R version: R-2.0.1
>
> The following example causes a crash of R:
>
> # univariate example
> require(mclust)
> X <- c(rnorm(100, mean=1), rnorm(100, mean=5))
> # I also tried: X <- as.matrix(c(rnorm(100, mean=1), rnorm(100, mean=5)))
> Xmap <- cbind(c(rep(1, 100), rep(0, 100)), c(rep(0, 100), rep(1, 100)))
> Xm <- mstep(modelName="V", data=X, z=Xmap)
> # CRASH:
> em(modelName=Xm$modelName, data=X, mu=Xm$mu, sigmasq=Xm$sigmasq, pro=Xm$pro)
>
> For bivariate data the following example works:
>
> # bivariate example
> require(mclust)
> Y <- as.matrix(cbind(c(rnorm(100, mean=1), rnorm(100, mean=5)),
> c(rnorm(100, mean=1), rnorm(100, mean=5))))
> Ymap <- cbind(c(rep(1, 100), rep(0, 100)), c(rep(0, 100), rep(1, 100)))
> Ym <- mstep(modelName="EEE", data=Y, z=Ymap)
> # NO CRASH:
> em(modelName=Ym$modelName, data=Y, mu=Ym$mu, sigmasq=Ym$Sigma, pro=Ym$pro)
>
> But when the variables are not columns (as written in the mclust manual)
> but rows, R crashes again:
>
> # CRASH:
> em(modelName=Ym$modelName, data=t(Y), mu=Ym$mu, sigmasq=Ym$Sigma,
> pro=Ym$pro)
>
> Perhaps this is a hint to the origin of the problem.
> I could not use the em function for univariate data successfully yet.
> Perhaps anyone can provide a working example for univariate data?
>
> Thanks in advance.
>
> ______________________________________________
> R-devel at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>

-- 
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-devel mailing list