
Date: Wed, 13 Jul 2005 15:58:42 -0500 (CDT)
From: Jun Yan <jyan@stat.uiowa.edu>
To: Torsten Hothorn <Torsten.Hothorn@rzmail.uni-erlangen.de>
Subject: pmvnorm 

Dear Torsten,

I have been using your package mvtnorm and found it very useful. However, 
I'd like to request for a functionality of pmvnorm (and pmvt): Could you 
make it available to compute the cdf given a matrix, each row of which is 
an observation? For example, let u be such a matrix of dim n by p, then

pmvnorm(lower = matrix(Inf, n, p), upper = u, sigma=sigma)

returns a vector of length n.

This way, the cdf function behaves more like the univariate cdf functions. 
I could have used apply function to do such batch jobs, but I think it is 
cleaner if pmvnorm can handle this in the compiled code for the sake of 
speed and elegancy.

Thanks!

Jun
-------
Jun Yan
Assistant Professor
Department of Statistics and               
 Actuarial Science                             Voice: 319-335-0824
University of Iowa                               Fax: 319-335-3017
219 Schaeffer Hall                          Email: j-yan@uiowa.edu
Iowa City, IA 52242           Web: http://www.stat.uiowa.edu/~jyan
