[AstroPy] How does ePSFBuilder work

[Larry Bradley]

Hi Peter,

The algorithm was originally designed for undersampled HST point sources.
You could try it with your sources, however, if your objects are resolved
the output won't represent a point source (true PSF).  If this happens to
be HST data, you could try using a library ePSF.

Cheers,
Larry

On Tue, Mar 21, 2023 at 2:31 PM Peter Dzwig <pdzwig at summaventures.com>
wrote:

> Larry,
>
> On 17/03/2023 16:02, Larry Bradley wrote:
> >> Questions: (i) is my understanding correct?;
> >
> > Yes.
>
> Thanks.
> >
>   Note that you can’t build an ePSF model from a
> > single input star. You need *many* stars (e.g., ~hundred or more) to
> > build a realistic oversampled ePSF model.  Having many empirical stars
> > (which will each have small subpixel shifts) ensures that the
> > oversampled output grid of the ePSF model is well sampled.
> >
>
> OK. Is it applicable in the case of a galaxy cluster?
>
> My problem is that I am looking to estimate the ePSF for a cluster in
> which many of the principal objects suffer (optical)cross-pollution.
> Unfortunately I don't have access to the original (e)PSF and am trying
> to approximate it.
>
> Because of orientations and types of objects in the image I have little
> that could be used in place of a single point object, so I am working
> with a few  "sufficiently isolated" spherical galaxies to generate an
> approximate ePSF.
>
> I'd appreciate any thoughts that you might have,
>
> Peter
>
> --
>
> Dr. Peter Dzwig
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/astropy/attachments/20230321/937ac8b4/attachment.html>