4pm Pacific Time
BlueJeans: https://bluejeans.com/426716450
attending:
- Eric Bellm
- Andrew Connolly
- Unknown User (cmorrison)
- Ian Sullivan
- Jeffrey Carlin
- Kenneth Herner
- Leanne Guy
- Meredith Rawls
- Bryce Kalmbach
- Unknown User (gkovacs)
- Krzysztof Suberlak
regrets:
Topics for discussion:
- Precursor processing (Meredith Rawls ):
- major effort last few weeks has been getting the pipelines running in Gen3
- built stack-produced calibs (biases & flats) in both Gen2 and Gen3 for HiTS
- gen2 comparison shown last month–no significant changes
- now building templates in gen3
- image selector we use in gen2 doesn't run in Gen2, so having to manually build the visit list for now
- we need to think about how we want to approach this in gen3
- (do note that there are some missing ISR steps–linearity and crosstalk not fully there for Gen3 DECam)
- will be restructuring repo using the RFC structure
- HSC COSMOS runs have been on the back burner–but community member Benjamin Racine has been trying this on his own and finding some "features" that he's shared with Meredith
- the duplicate diaSourceId problem, again
- some interesting plots with visit # vs. ccd number as a way to identify problems
- Crowded fields processing (Ian Sullivan ):
- all of the breakages happen in SFP
- using PSFEx gets ~99% of the data through
- processing time mostly 3-4 minutes, but a long tail–up to 4 hours
- change to astrometry catalog loading (with input from Lauren) sped up processing
- different years have very different seeing–2015 seeing better than 2013. For test, doing both 2013-2015 and 2015-2013
- minor issues with CR and background subtraction in a few places
- (general discussion of future plans for processing a variety of datasets)
- all of the breakages happen in SFP
- SITCOM team updates (Andrew Connolly et al.):
- working so AOS just uses DM code
- Bryce Kalmbach WCS issues resolved using DM code & the boresight
- working so phoSim simulated images are packaged correctly so obs_lsst can process them Krzysztof Suberlak
- finishing this will have removed all the sims code
- then will turn to looking at deblending of donuts in the AOS
- working so AOS just uses DM code
- Review CI (https://chronograf-demo.lsst.codes/):
- Next meetings (all at 4pm Pacific):
- Jan 25
- AOB
- No regular AP meetings next two mondays
- Leanne: what about using faro for the new fakes metrics?
- Eric: these are just MetricTasks, so pretty low-level–going to couple them into CI, primarily
- Colin: faro is collecting
Action Items
Description | Due date | Assignee | Task appears on |
---|---|---|---|
| 17 Jul 2023 | Eric Bellm | AP Pipeline Meeting, 2023-06-26 |
| 11 Sep 2023 | Eric Bellm | AP Pipeline Meeting, 2023-08-21 |