Fix Version/s: None
Sprint:AP F21-5 (October)
Once the chi2 and npixels fields are being output by PsfFluxAlgorithm, we can get them into the apdb output. This will require modifying the apdb schema and likely some changes to ap_association, among other things.
DM-32197 Use psFlux chi2 values to investigate image subtraction quality
- is blocked by
DM-29346 Compute PSFlux chi squared metric
- is triggering
DM-32265 ap_verify_hits2015 detectors 2 and 61 broken in gen3
- Won't Fix
- relates to
DM-37911 Write dipoleNdata, dipoleLnl, dipoleCov fields to APDB
- To Do
I've uploaded a histogram of the chi2/ndata values for my ap_verify_hits2015 run. These seem reasonable, though there are also 15 chi2 values < 0 (this worries me a bit!) and 26954 NaNs (no infinities).
Chris Morrison [X]: Please review my changes to the ap_association yaml file on the following PR. Do I need to change anything else to call this complete? It looks like that was all I needed to get this value into the association.db file when running ap_verify.
Krzysztof Findeisen: Please review these three PRs. I made some changes to the ap_verify docs to reflect the hits2015 processing trouble. I also added a note about using `-j`, since I wouldn't have thought to do so until I was told about it on slack.
I also tweaked the API of one of the functions I added to analysis_ap: I'd forgotten to remove the `return`, which we don't want because it modifies the table in place.
It looks good. If you see that values are being stored in this column in the Apdb after running some data through it, that's basically enough to confirm it's working.
Adding psChi2 and psNdata to the DiaSource.yaml file resulted in an ap_verify_ci_hits2015 run having those fields filled out in the assocation.db.
Here's a jenkins run with ci_hsc to see if anything else breaks. https://ci.lsst.codes/blue/organizations/jenkins/stack-os-matrix/detail/stack-os-matrix/35140/pipeline