Is this because we ensure that is true using the image rotator?
because the source may be variable but it will always have the same shape?
-
the direction of the dcr is aligned with the pixel grid, as mentioned in my assumptions above. However even if it is not, a trivial rotation of the frame in software will produce the needed alignment with extremely minimal correlations.
-
This was answered in my parenthetical before hand. If the source is not moving or variable, then the only change in flux will come from a redistribution of flux due to dcr. If the flux levels do change due to the nature of the object, this will be evident in the optimization, and difference image. However as this is what we are attempting to detect this is a net positive. Additionally, any trouble in optimization due to āmissingā or āexcessā flux, could help characterize what has changed, i.e. changing SED of super nova or something.
I think I have our DCR problem solved! My notebook is a bit messy right now, but I will send it around soon! It should be pretty easy to include in the stack, pretty much as is too, with only minor changes from my cartoon example for specifics of our setup. Stay tuned.
1 Like
Iāve just realised this isnāt on the DM publications/talks list. Were there other DM talks at this joint ZTF-LSST meeting?
Sorry Tim, it was fairly informal. There must have been an overview talk by Mario at least.
Informal is not a problem, especially when itās a talk given to people outside the project (which I think meets my remit on tracking DM publications). Iāll upload it to Zenodo (and I can add anything @mjuric has as well).