AP Pipeline Meeting, 2022-08-22

2pm Pacific Time

Zoom: https://washington.zoom.us/j/98846655020

attending:

• Eric Bellm 
• Ian Sullivan 
• Brianna Smart 
• Erin Howard 
• John Parejko 
• Krzysztof Findeisen 
• Kenneth Herner 
• Yusra AlSayyad 
• Meredith Rawls 
• Nima Sedaghat 

regrets

Topics for discussion:

• Project updates (Eric Bellm , Ian Sullivan ):
  
  • DM meeting-free week coming up September 19 - 23
  • 2023 DM SIT-COM Joint Subsystem meeting (JTM)  Week of March 13, 2023
    • Yursa: sending a ~registration later today
  • analysis_tools sprint in early September
    • Yusra can give a tour in-person Tuesday 3pm Pacific.
  • sprint planning
• USDF transition status
  • only HSC repo right now, with some missing sky datasets
  • takes 5 days for SLAC IT to respond, and that to your SLAC email...
• Pipelines Infrastructure (Krzysztof Findeisen , John Parejko , Brianna Smart ):
  • Krzysztof Findeisen : focused on
• Precursor processing (Meredith Rawls, Kenneth Herner , Erin Howard ):
  • Meredith Rawls : built some temporary templates with Brianna Smart and Erin Howard on HSC–useful for teaching but not much more than that
  • Kenneth Herner :
    • saha bulge was last run in January 2022, so should show up "for free"–do we want the previous one?  Ian: would be nice to bring over fall 2021 if we can/it's not too much work.  remembers some substantial changes between the two
    • tract 4431: likewise
• Image differencing algorithms, DCR  ( Ian Sullivan ):
  • nothing new since PCW
  • needs to find a dataset/time to really look at convolve science.  unit tests look good, but DECam results are not as great as he'd expect (away from the patch boundaries)–maybe this is just because the templates are definitionally worse?  Will look at matching kernel
  • Ken–there's still an excess of negative sources
  • Eric: status of patch boundary variance plane fix?  Ian: still has the ticket open.
  • Eric showing an outline slide of perceived risks for the upcoming PST talk
    • JP: could we revisit this ~monthly?  Eric: not a bad idea, maybe at QA meeting?
• transinet (Nima Sedaghat, Eric Bellm, John Parejko ):
  • RB classifier: starting to enter planning stage for domain adaptation
  • end-to-end classifier: working on negative transients, which is the bottleneck for now
  • available to help on DC CI
  • pytorch RFC approved, awaiting implementation by K-T.  maybe we can jump-start the process
  • Eric will need a few numbers for the PST talk
• alert distribution (Eric Bellm , Brianna Smart ):
  • nothing yet
• Review CI (https://chronograf-demo.lsst.codes/)
  • very slight increase in raw DIASource counts when turning compatibility mode off.  But fakes metrics also improved, so maybe this is more accurate?  main change is in how scaleVariance is applied.
• Review outstanding action items
• QA meeting in October:
  • PST diffim talk: risks outline table
  • analysis_tools sprint results
• AOB

Action Items