AP Pipeline QA Meeting, 2021-01-25

4pm Pacific Time

BlueJeans: https://bluejeans.com/426716450

attending:

• Eric Bellm
• Leanne Guy
• Andrew Connolly
• Kenneth Herner
• Jeffrey Carlin
• Ian Sullivan
• Meredith Rawls
• Zeljko Ivezic
• Bryce Kalmbach
• Krzysztof Suberlak
• Unknown User (gkovacs)
• Colin Slater

regrets:

Topics for discussion:

• Precursor processing (Meredith Rawls ):
  • DM-27370: HiTS reprocessing in gen 3 from raws! (no linearity or crosstalk possible in Gen3 right now)
  • some differences in templates vs Gen2–have to use same visits for all chips in Gen3 right now which is a limitations
  • there are some gaps that are at patch boundaries–need to drill down to understand why
  • some bad columns creeping through
  • psFlux = 0 are dipoles, thinks Chris (or noise sources)
  • totflux on the calexp has negative values–this is odd
  • more sources than past processing, even with the missing chunks–many more edge sources came through in Gen3–changing in flag behavior?  some edge/suspect flags that were caught in Gen2 but not in Gen3
  • wants to wait to chase some of these issues until we have better templates & ISR
    • let's spawn tickets for these issues and hold them and see if they can be won'tfixed once the Gen3 processing is improved
    • we could replicate Gen2 with the same template selections, linearity and crosstalk turned off–but we don't think this is a great use of time
• Crowded fields processing (Ian Sullivan ):
  • showing draft of DMTN-171
  • 2015 seeing was much better than 2013, which is not ideal
  • PSFEx did well in g-band, but in i-band some visits have defects in the PSF.  This is not fatal for running diffim, though
  • showing source counts vs. magnitude–a few discrepant visits that seem to be failing photometric calibration
  • coaddition and image differencing pipelines ran effectively
  • Colin: are the Saha catalogs available?  Eric: Paper claims the catalogs would be made available via datalab
    • Ian Sullivancheck if the Saha Bulge catalogs are available, either through datalab or from Abi directly 22 Feb 2021 
• SITCOM team updates (Andrew Connolly  et al.):
  • cleaning up wavefront code to remove sims code, just use DM code
  • running into PhoSim trouble assuming that all devices are ITL–need to simulate mixed focal plane
  • Gaia EDR3 completeness in densest regions is only 20-30%, even at 15th-17th mag (trying to get down to 19th mag)–looking at SkyMapper/PS1 as supplements
  • next month starting to think about deblending donuts–welcome ideas
• PreOps V&V (Colin Slater ):
  • still getting ramped up
• Review CI (https://chronograf-demo.lsst.codes/):
  • Unknown User (gkovacs) to check for changes that might result in a jump in the ratio of DIASources/Science sources between Jan 10 & 13 01 Feb 2021 
• Next meetings (all at 4pm Pacific):
  • choose a regular week?
    • we'll use first Mondays of the month going forward–so March 1 is our next meeting
• AOB
  • Chris reminds us that fakes are now being produced in Gen3 HSC
    • 80% completeness at 27th, 90% at lower mags for ~30 sources inserted per frame
    • will start showing up in Squash once we enable Gen3 uploads

Action Items