| Re: Problems with CHISQR_PDF when moving from IDL 5.2 to 5.4 [message #28035] |
Fri, 16 November 2001 12:34  |
Craig Markwardt
Messages: 1869
Registered: November 1996
|
Senior Member |
|
|
David Williams <d.williams@qub.ac.uk> writes:
> Our data often consist of thousands of points, and for any more than a
> few hundred degrees of freedom (and we have thousands of degrees) IDL
> 5.4 runs into difficulties. The problem we're encountering is that
> CHISQR_PDF can't return an answer, and the only major difference I can
> immediately see -- between the version of CHISQR_PDF bundled with 5.2
> and that bundled with 5.4 -- is that the newer version calls IGAMMA
> rather than IGAMMA_PDF. In fact, it's IGAMMA that fails to converge to a
> value and echoes a message saying this. It's pretty frustrating because
> 5.2 handled our routines just fine.
Hi Dave--
If you want to go ahead and take Wayne's suggestion, that's a good
one. I've also just put up a set of routines for hypothesis testing
and confidence interval estimation on my web page. This includes
chi-square hypothesis testing.
One major difference is that I took my code from Stephen Moshier's
CEPHES special function library, so there's no dependence on the
implementation of special functions by IDL. So far I haven't had
convergence problems with this code.
Good luck,
Craig
http://cow.physics.wisc.edu/~craigm/idl/ (under fitting)
--
------------------------------------------------------------ --------------
Craig B. Markwardt, Ph.D. EMAIL: craigmnet@cow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response
------------------------------------------------------------ --------------
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
