BlueJeans: https://bluejeans.com/426716450
attending: Eric Bellm Meredith Rawls Krzysztof Findeisen Unknown User (cmorrison) John Swinbank Unknown User (emorganson) Unknown User (gkovacs)
Topics for discussion:
- Gen 3 conversion (
):
- Gabor asking about what do if there are backwards-incompatable changes needed in ip_diffim
- HiTS processing (Meredith Rawls ):
- not much to report. We didn't make a ticket for mid-month processing yet–Meredith did so. Still using Ian's template resolution pending resolution of the optimization ticket.
- report from NCSA (Unknown User (emorganson)):
- can still run the pipeline
- had a day-long meeting on AP. have systems to simulate incoming images, humidity, etc.; now trying to put meaningful computation into those to understand effects of things like preloading, parallelization, etc.
ap_
association (Unknown User (cmorrison)):- finishing unit tests for multi-plugins
- plugins are too slow, even though the plugin system improved things. Using pandas built-ins will help some; others will need to drop to C. Timing both data access from pandas and feature computations
- DCR (Ian Sullivan ):
- (report via Slack): My report is that my DCR metric notebook is still in review (with Eric Bellm ), and I've reprocessed HiTS where I made DCR templates from 2015 data and used that to generate
ap_pipe
alerts from the 2014 data. The results look terrible for both comparewarp and DCR templates, so I've debugging what went wrong using a subset of the data. I haven't started DM-21917 yet, but plan to this week.
- (report via Slack): My report is that my DCR metric notebook is still in review (with Eric Bellm ), and I've reprocessed HiTS where I made DCR templates from 2015 data and used that to generate
- Image differencing (Unknown User (gkovacs) ):
- test run with scale variance (which actually scaled down the variance??)–in review with Yusra
- has been working on reviews; Eric and John urging to move to Gen 3 work.
- Review CI (https://chronograf-demo.lsst.codes/):
- Nothing
- AOB
- Gabor asking about discovery of private docstrings–advice is to ask on #dm-docs