search

barnabytprowe / great3-public

Public repository for the Third Gravitational Lensing Accuracy Testing Challenge
BSD 3-Clause "New" or "Revised" License
9 stars 11 forks source link

multiepoch optical psf variation between epochs #2

Closed dmargala closed 10 years ago

dmargala commented 10 years ago

Looking at the multiepoch-space-(constant,variable) psf images, the optical psf model seems to change quite significantly (the number of struts for example) from epoch to epoch. This does not seem physical and the logic demonstrated in the example coadd doesn't seem to account for this. Is this correct?

rmandelb commented 10 years ago

Hi Daniel -

I think there are actually two aspects to your question, one related to why does the PSF model do this and the second related to whether the coaddition script can handle it. I will answer these in reverse order:

The coaddition script (which GitHub is persistently trying to change to 'coalition' - thanks, GitHub! - so I keep having to correct its helpful non-corrections) does not formally assume the PSF is the same per epoch, and indeed, in the ground images the PSF FWHM differs from epoch to epoch, which is representative of real data. The coaddition script does the simplest possible mathematically correct process, which is a mean coadd, and if you use the same weights when coadding the PSFs (which it does) then the coadded PSF is, mathematically speaking, the proper PSF for the coadded galaxy images. This may not be the optimal way of handling images with different PSFs, but our goal was not to distribute something optimal, just something mathematically correct as a basic demonstration, as in the caveats given in the docstring for that example script.

With that said: The intention was that the optical PSFs would be fixed to the same telescope model for all epochs in a subfield (i.e., same telescope diameter, number of struts) with only the additional / stochastic aberrations varying from epoch to epoch. And of course, for ground sims, the atmospheric PSFs will vary from epoch to epoch. You are correct that the telescope model is also varying from epoch to epoch, contrary to my intent. This is not a completely impossible situation - one might want to combine images from two different systems - but that's more of an edge case rather than the norm.

We are discussing now what to do about it, will post again once there's a decision.

Thanks for bringing this to our attention!

rmandelb commented 10 years ago

Update: we're regenerating these.

rmandelb commented 10 years ago

@dmargala : these are ready. Please look out for our e-mail announcement later today.