BlueJeans: https://bluejeans.com/426716450
attending: Eric Bellm Meredith Rawls Krzysztof Findeisen Ian Sullivan Unknown User (emorganson) Unknown User (gkovacs)John Swinbank
Topics for discussion
- HiTS reprocessing (Meredith Rawls):
- working on the excess detections due to background problem
- DES processing (Unknown User (emorganson)):
- ops rehearsal last week: data transfer, (simulated) images were processed through SFP. Main learning was about communication
- New faster computers that will eventually go to Chile; Eric will run ap_pipe timing estimates on them for comparison: faster disk reads, more RAM, etc.
ap_
association (Unknown User (cmorrison)):
- DCR (Ian Sullivan ):
- found & fixed bug when template is finer resolution than science . Probably this didn't affect HiTS processing since unlike the DCR sims the spatial sampling of HITS science and data are the same
- also wrote code to correctly calculate PSFs from DCR models
- gets somewhat cleaner results now (e.g. in total number of sources) since deconvolution artifacts have decreased
- image differencing (Unknown User (gkovacs) ):
- worked on backlog tickets
- wrote to Robert, who requested a short writeup with particular attention to preconvolution
- thinking some about interaction between polynomial modifications to the kernel
- Review CI (https://chronograf-demo.lsst.codes/):
- Minor decrease (→ improvement?) in # of unassociated DIAObjects, which we associate with DM-19660
- gradual (?!) improvement in runtime; Jenkins has migrated to faster machines at NCSA, but it's not an obvious step function
- AOB
- Ian requests another DCR run of DECam HiTS–Meredith will teach him to fish
- Meredith: needs an expert (maybe Yusra) to evaluate a possible config change. other option is to fix ip_diffim