I received 3 responses to my request for a pmvnorm like function
that might approximate mvn orthant probabilities for general
covariance and general dimension:
1) Eric Zivot --
As part of the next release of FinMetrics, Insightful will be implementing
the GHK simulator with various options and improvements by Alan Genz to
compute multivariate normal probabilities. You might want to contact
Insightful directly to about using this code now.
2) Spencer Graves --
From "www.r-project.org" -> search -> R site search, I found
documentation of a different "pmvnorm" function that seems not to have
this restriction. However, I have not used it, so I can't say.
It accesses Fortran code, to it won't be completely
straightforward to port to S-Plus. If you are able to make it work in
S-Plus, please be so kind as to observe the GNU license requirements by
posting your modified code back to someone like someone like Thorston
Hothorn, listed on the help page as having made the R port.
3) Jason Hsu -- pointed to Frank Bretz' website
http://www.bioinf.uni-hannover.de/%7ebretz/publ.html
which has
** SAS/IML programs for exact calculations of orthant probabilities for
arbitrary dimensions.
and
** Methods for the computation of multivariate t-probabilities.
Thank you all for your help.
Joe Verducci
|