Disentangling QE and Pixel Size Variation

[cpadavis]

On a per pixel level, the two look the same. How can we tell the two apart across the field?

I was thinking you would expect the centroid of a non-uniform source to behave differently to QE and pixel size variation. In other words, you could disentangle this using several star flats.

To do that, I need to

• [ ] Look at how the stats (primarily the pixel coordinate centroid) of a star varies when traveling through QE variations vs pixel size variations. Let's hope there actually is a difference!
• [ ] Which means I need to put in a QE mover class that takes fluxes and multiplies some sort of QE against them.
• [ ] And I need to come up with some sort of reasonable model for what a pixel variation field would look like vs QE. I could probably get these from fits to LSST flats?

[mbaumer]

Within LSST, Stubbs and I have done some work on this topic (disentangling PSV from QE) by looking at sinusoidally-illuminated flats. Check out this LSST confluence page if your account lets you view it. Let me know if not and I'll get you the info by other means...

[cpadavis]

I have access to LSSTDESC but not LSSTCAM :(

[cpadavis]

(related to this is that incorporating this code should make it easier to test whether the sinusoidally-illuminated flats are actually worth the effort)