Rob Williams <[email protected]> is looking for code similar to that
used by -mprobit- and -asmprobit- to perform quadrature and integration by
simulation.
As Kit Baum <[email protected]> noted, official policy is not to release C or Mata
source code used to implement Stata commands. Besides the obvious reasons,
that code is very specialized, documenting it would be a major effort, and we
want the freedom to change it.
On the other hand, our policy is to provide source code on the Mata side of
things, and to make sure that that code is general purpose and well
documented.
Concerning Rob's question, we have been planning on releasing
general-purpose routines in Mata to evaluate multivariate normals by
simlulation. Richard Gates <[email protected]> is working on that
and he is sending to Rob premlinary versions of that code.
Concerning quadrature, We recommend Dunnett (1989), Algorithm AS 251,
"Multivariate normal probability integrals with product correlation
structures," JRSS, vol. 38, pp 564-579. The FORTRAN program is available on
on Statlib at, for instance, http://sunsite.univie.ac.at/statlib/apstat/251.
I (Bill Gould) will be talking on Wednesday in Boston about translating
FORTRAN programs into Mata. It is easy to do.
-- Bill -- Richard
[email protected] [email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
