BlueJeans: https://bluejeans.com/426716450
attending: Eric Bellm Unknown User (cmorrison) Unknown User (emorganson) Meredith Rawls Ian Sullivan Krzysztof Findeisen
Topics for discussion:
- HiTS reprocessing (Meredith Rawls)
- Made "goodSeeing" coadds with a fwhm cutoff rather than an n-images cutoff ( - DM-16191Getting issue details... STATUS ). The mysterious images in the middle of nowhere have disappeared. Also generating a naive coadd of everything for comparison.
- Getting ready to start ap_pipe kicked off on these coadds.
- Eric Bellm asked for thoughts on #dm-science-pipelines about how to handle - DM-16191Getting issue details... STATUS ; got a range of orthogonal ideas, including RHL suggesting
- DES processing (Unknown User (emorganson))
- Now working for real on gri bands; alert db on his repo is now getting filled as we speak. Just needed to get config.warpType correct.
- Will work with Meredith Rawls on looking the outputs.
- Going to start working with Jim Parsons at NCSA to give him more realistic inputs
- Also poking at Docker/Kafka/Kubernetes.
- Right now it's 110 exposures (no z-band), but have two more years we could add later if we wanted more scale.
- calibration database he made should be ~universal and work for all DECam datasets
- going to try to make some fringe calibration inputs; not clear the correction will work
- CI (Krzysztof Findeisen )
- we are running in CI nightly, so far with no failures
- running on a minimal subset of HiTS 2015
- We are computing #s of DIASources, # of direct sources, timing of major subtasks. Not currently uploading anything to SQuASH. Eric will ping Josh/Simon/etc. to find out how we can do this. Not computing any of the DRP metrics; is that a problem?
- DCR (Ian Sullivan )
- now recovering input source spectra, modeling reports better convergence.
- including a restriction to detected sources, which allows use of larger number of subfilters
- some DCR was existing within subfilters; larger number of subfilters helps, other algorithmic fixes
- now applying the LSST throughput in a different place–applied to measurements at the end.
- one more scaling bug to check on; otherwise ready for running on HiTS. Need to think a little bit about input configuration parameters based on what's in HiTS data.
- longitude bug to be fixed in obs_decam by next week thanks to Tim Jenness.
- now recovering input source spectra, modeling reports better convergence.
ap_
* development- ap_association: all the old hooks are removed; talking directly to PPDB and writing the SQLite. Found and fixed a bug with summary stats. But can't merge yet because has adjusted to only handle dpdd-ified objects; so writing a ticket to do the dpdd-ification. Will eventually have a ticket on ap_pipe for handing in the database object.
- expect some database changes
- AOB/table round