From barrett at comet.stsci.edu  Fri Apr 25 08:57:43 2003
From: barrett at comet.stsci.edu (Paul Barrett)
Date: Fri, 25 Apr 2003 08:57:43 -0400
Subject: [AstroPy] Python spectral fitting package - ala Sherpa
Message-ID: <3EA930C7.5060604@comet.stsci.edu>

<x-flowed>Has anyone on this list used the spectral fitting application, Sherpa, 
that is part of Chandra's CIAO data analysis package?

I've been using it to analyze/fit FUSE spectra.  I like the design of 
this application, because it separates the minimization algorithm from 
the statistical fit. I routinely use the Bayesian statistic instead of 
the chi-square, since it produces faster and better fits to my spectra. 
On the negative side, Sherpa uses the S-language which is smaller that 
Python and Numeric, but is also less advanced. The user can also add his 
own models to Sherpa, but is limited to only one model written in C. All 
others must be written in S-lang which I find cumbersome for complicated 
models.

The purpose of this message is to see if anyone is interested in 
developing a Python spectral fitting module based on Sherpa's design of 
separating the minimization and statistical algorithms. The use of 
Python classes would make for a simple, yet robust spectral fitting module.

Any intereste out there?

 -- Paul

-------------------------------------------------------------------------
Paul Barrett, PhD      Space Telescope Science Institute
Phone: 410-338-4475    ESS/Science Software Group
FAX:   410-338-4767    Baltimore, MD 21218


_____________________________________________________
AstroPy mailing list  -  astropy at stsci.edu
http://lheawww.gsfc.nasa.gov/~bridgman/AstroPy/
</x-flowed>

From jaime at iaa.es  Fri Apr 25 10:32:15 2003
From: jaime at iaa.es (Jaime Perea)
Date: Fri, 25 Apr 2003 16:32:15 +0200
Subject: [AstroPy] Python spectral fitting package - ala Sherpa
In-Reply-To: <3EA930C7.5060604@comet.stsci.edu>
References: <3EA930C7.5060604@comet.stsci.edu>
Message-ID: <200304251632.15744.jaime@iaa.es>

An embedded and charset-unspecified text was scrubbed...
Name: not available
URL: <http://mail.python.org/pipermail/astropy/attachments/20030425/9a680faa/attachment.ksh>

From loredo at astro.cornell.edu  Fri Apr 25 12:59:33 2003
From: loredo at astro.cornell.edu (Tom Loredo)
Date: Fri, 25 Apr 2003 12:59:33 -0400 (EDT)
Subject: [AstroPy] Python spectral fitting package - ala Sherpa
Message-ID: <200304251659.h3PGxXi10702@laplace.astro.cornell.edu>

Hi Paul-

I'm developing something along these lines as part of the statistics
package I'm working on with Alanna Connors (associated with the
astrostat group behind Sherpa) and Travis Oliphant.  Also, I wrote
the algorithm for the Bayesian statistic that Peter Freeman put
in Sherpa, so that will be part of the package.  We call it a
Parametric Inference Engine (PIE).  Besides minimization, it
includes tools to help with other tasks, such as quadrature over
uninteresting params, maximization over subsets, calculation of
Hessian matrix, and we hope some basic capability for Markov
Chain Monte Carlo (MCMC), though there's a certain limit to
how much that can be automated.

We should have a prototype in the next couple months; I'll run it
by you.

This may be a package of its own, but at the moment we're leaning
toward making it a subpackage of SciPy.

-Tom
_____________________________________________________
AstroPy mailing list  -  astropy at stsci.edu
http://lheawww.gsfc.nasa.gov/~bridgman/AstroPy/