DRP Metrics Monitoring 2022-04-04

Mondays 12pm - 12:50pm ET

Yusra's Zoom: https://princeton.zoom.us/my/yusra

Attendees:

• Yusra AlSayyad 
• Colin Slater
• Nate Lust 
• Jim Bosch 
• Robert Lupton 
• Lauren MacArthur 
• Jeffrey Carlin 
• Lee Kelvin 
• Eli Rykoff 
• Leanne Guy 
• Fred Moolekamp 
• Dan Taranu 
• Christopher Waters 
• Andrés Alejandro Plazas Malagón 
• Arun Kannawadi 

Agenda:

• Meeting recorder? (last 5 meetings were: Fred, Lee, Keith, Clare, Jim)
• Announcements: 
• Review Action items from last week
  • Why did TE1/TE2 look bad in DC2 w10. Scarlet-lite change, "in that sense more correct", Fred was looking into building a "deblender-friendly star selector" (current selection includes things that were ruled out as primary sources).
  • Arun working on tract-level ellipticity residual plots. Goal to have it by weekly 13.
• Processing Status
• Review HSC RC2 rerun w_2022_12
  • What was noteworthy in this rerun? How did processing go?
    • DM-34125, error in a makeWarp tract (std::bad_alloc). Worked on Yusra's retry, will hopefully work on a rescue submission.
  • Gen2 reruns are officially done.

• • What changed in the metrics?
    • wperp dashboard broken due to "tellar_wPerp" misspelling. Ticket for Angelo.
    • Why there was a wPerp dispatch on Feb 16 and Feb 17, with different values? Have to change the date manually, could have been a typo(?).
    • TE1 y and i missing? Suspect a transient memory issue.
    • Where is bad_alloc coming from? Infrastructure issue, bad ram? Eli: I only see that if I accidentally try to allocate 10TB of ram...
    • Don't have metrics that would be sensitive to amp-to-amp matching that was turned on. Things did not get worse, but have to do more specific checks. Eli did a thing that showed no difference.
      • Does that mean that there's some remaining badness in BG subtraction? No, probably RC2 not big enough to test with Eli's test.
      • Also longer exposures make the effect less significant, and RC2 is all long exposures.
      • Would RMS of sky objects show this? Could, would be good to add this back into the gen3 dashboard.
      • Small change in nightly RC2 subset that might have corresponded to amp-amp offsets? Jeff added an annotation.
    • rc2_subset March 17th increase in wPerp? Then it went back down March 30, when piff was turned on.
      • Probably changed star selection, both through PSF mags and extendedness?
      • Will see this in DC2 in two weeks, RC2 in four weeks.
    • Plots are now part of the review
      • RHL would like vectors that show a star moving from one position on a plot to another position in a different run. Technically difficult.
      • Where did the Tier1/Tier2 plot list go? Yusra will post to the agenda.
    •  Annotate 

• • What do we expect from w_2022_16? (Tagged April  13)
    
    • What new metric or plot can we expect next time?
      • New features: PIFF
      • Metrics:  Sky Objects!  Maybe some deblender metrics (e.g. Number of parents skipped)
• Review DC2
  • w_2022_10 already reviewed 2 weeks ago
• AOB:
  • New flags getting added to deblender, could affect selections  used for metrics. Do we need to do something to distinguish these changes from new features like Piff?
    • No, if we need to do retrospective analysis, then we'll do retrospective.
• Next meeting 18 Apr 2022 

DM-34286

Action Items