[R-sig-ME] Fwd: [R-pkgs] New package: ordinal

Paul Johnson pauljohn32 at gmail.com
Wed Mar 17 19:15:35 CET 2010


On Tue, Mar 16, 2010 at 8:43 PM, David Duffy <David.Duffy at qimr.edu.au> wrote:
> On Tue, 16 Mar 2010, Gregor GORJANC wrote:
>
>> Hi,
>> This should be of interest to many on this list.
>
> It does look yummy ;)  I might add, for people who are not aware of it, that
> you can fit ordinal mixed models with crossed effects in R using the
> OpenMx package (http://openmx.psyc.virginia.edu/).  It doesn't seem
> to be on CRAN (it is Apache licensed, so maybe that is the reason).
>

I've been monitoring OpenMx too.  That project has very high-reaching
aspirations.  If it all comes together, it will offer the only truly
comprehensive R package to replace commercial SEM software like Mplus.

I don't think OpenMx can be distributed on CRAN, though.  OpenMx
relies on a proprietary library npsol that they distribute as a
compiled library (no source).  As far as I can tell, this means it can
never be distributed over CRAN.  If you need to see what I mean, grab
the tar.gz for OpenMx and look

./inst/npsol
./inst/npsol/windows/x86/rtools2.10/libnpsol.dll
./inst/npsol/windows/x86/rtools2.9/libnpsol.dll
./inst/npsol/osx/libnpsol.a
./inst/npsol/linux/x86/gcc3.9/libnpsol.a
./inst/npsol/linux/x86/gcc4.4/libnpsol.a
./inst/npsol/linux/x86/gcc4.3/libnpsol.a
./inst/npsol/linux/x86/gcc4.1/libnpsol.a
./inst/npsol/linux/x86/gcc3.4/libnpsol.a
./inst/npsol/linux/x86_64/gcc4.2/libnpsol.a
./inst/npsol/linux/x86_64/gcc3.9/libnpsol.a
./inst/npsol/linux/x86_64/gcc4.4/libnpsol.a
./inst/npsol/linux/x86_64/gcc4.3/libnpsol.a
./inst/npsol/linux/x86_64/gcc4.1/libnpsol.a
./inst/npsol/linux/x86_64/gcc3.4/libnpsol.a

If they have not built a version of libnpsol.a for your system, then
OpenMx cannot be built on your system.  Unless you buy npsol,  you
can't compile OpenMx.   I don't know what benefits npsol has over
other R system solvers.

I'd be interested to hear back from other people who've tested out
OpenMx.  I suppose r-sig-mixed is as good as any other venue for that
discussion.

:)

-- 
Paul E. Johnson
Professor, Political Science
1541 Lilac Lane, Room 504
University of Kansas




More information about the R-sig-mixed-models mailing list