[AstroPy] Python spectral fitting package - ala Sherpa

Paul Barrett barrett at stsci.edu
Wed Feb 4 08:39:54 CST 2004

<x-flowed>Tom Loredo wrote:
> Hi Paul-
> I'm sorry to have delayed response to your question about this back in
> November.  I hoped to have a test version of the PIE available around
> the New Year, but I'm still stuck on some of the "big picture" aspects
> of it.  Lots of little pieces of the package exist, but I haven't
> wanted to release them until I settled on the interface.  Part of
> the problem is that I've spent so much time trying to find funding
> for the future that I've had less time to spend on my few actually
> funded projects than I hoped.  But the real problem is that every
> design I've come up has so far been okay for one set of problems but
> clumsy for others.  I've nearly reached the end of my creativity
> on it, so if I haven't converged on it within a couple weeks, I will
> just dump what I have on you and Alanna and Travis and beg for
> help.  My problem may be that I want it to be too general; I may
> have to settle for a handful of specific solutions instead of one
> general framework.

Maybe you should let us take a look at it now.  Sometimes just 
presenting a subject can lead to a solution.

> Regarding AISR, I wasn't planning on submitting anything this year.
> If the Inference package goes well, I hope to submit something asking
> for a reduced amount to maintain and expand it next year.  But if
> there is something specific you were thinking of proposing that you
> think I could contribute to, do let me know.

I still think there is a great need for an enhanced spectral fitting 
package ala XSPEC and Sherpa.  The problems that I see with these two 
packages are:

1. they are not object oriented,

2. their spectral model is flawed (i.e. multiplicative and additive
    models are a needless constraint),

3. they only handle 1 dimensional response matrices (see e.g. FUSE,
    which could use a 2D response matrix), and

4. they are primarily implemented using compiled languages, so
    development is slow.  (Sherpa's implementation is simply
    bass-ackwards, i.e. imbeding an interpreted language in a compiled
    language.  In addition, Numeric/numarray syntax is more powerful than
    SLANG and hence, numerically more efficient.)

If PIE can solve these problems, then I'm willing to wait, otherwise it 
would be nice to get started on the successor to XSPEC and Sherpa.

> The NSF has a new program targeting statistical and mathematical
> research in astronomy.  If the kind of projects you were thinking
> of have a strong statistical focus, I can give you the NSF URL
> for this.  Here at Cornell we're putting together a time series
> proposal for it.

Yes, I'm aware of this program.


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

AstroPy mailing list  -  astropy at stsci.edu

More information about the AstroPy mailing list