Sprint 009 agenda - March 29, 2023

• Strategy check in (to be completed by product owner prior to meeting)
  • Look in on recent issue activity
  • look over the epic list and the road map to make sure we are still happy with the big picture
  • Add last minute issues to epics
• Sprint 008 review
  • check out the sprint management dashboard and go over issues that were closed
• Sprint 008 retrospective
  • Screenshot burndown
  • go over the burndowns
  • check out the sprint management dashboard and go over issues that were not closed
• Sprint 009 planning:
  • Consult product owner notes below
  • go over the sprint planning dashboard and move tasks into the sprint 008 board.
    • At Risk Issues
    • Needs attention Issues
    • On track
    • unprioritized The Scrum Master should work with the team to assign a health for these in the coming sprint
  • go around the room and assign tasks
    • Assign ROTA members
    • Does everyone have a task?
    • Does everyone’s hours add up to what is reasonable?
  • choose a scrum master

Personnel

• Product Owner: Leo Tsukada
• Scrum master: Becca
• Stakeholders?: TBD

Minutes

Attendance

Strategy check in and product owner notes (done pre meeting by Chad)

I am increasingly worried about the following

• Online MDC - things to be fixed/changed for MDC12
  • singles penalty
  • dt dphi tweaks (jacobian)
  • signal model tweaks. Now the mismatch range goes up to 30%.
  • pastro. It is not classifying as well as it should be and probably the template weights are part of it
  • iDQ gating.
• Online real data
  • A discovery is possible and we want to be ready
  • See if we can deploy the range check while waiting on undisturbed bit. It is a good check anyway
  • start thigns over and get a good background collection. Be ready to upload to gracedb and to setup alerts if possible
  • updates on PSD convergence patch?
• Code versions
  • Do we need to compile our own ligolw from the release branch in a container via CI?
  • Do we need to branch gstlal at this point for O4 and get that built into our container via CI?
  • Do we need to hack lalsuite for the coinc rates thing and can we get that into our container via CI?
  • What else? All code versions and tweaks need to be captured in a gitlab CI to build a production container. Lets make that a requirement for MDC 12.
  • I would still like configurations to be in these containers too…
• Offline analysis
  • We MUST confirm injection distributions with R&P. Can we handle the 6s rate in one file? Should we ask for 2 to be safe?
  • We should work with MBTA and PyCBC to finalize the injection distribution details.
  • We must get the online rerank working ASAP
  • We must get the injection dag with OSG enhancements working asap
    • This should have one SVD bank per job
    • This should read injection frames
    • We need to check that the clustering built into gstlal inspiral is “good enough” to produce small files for the rerank (it seems to be true online)
    • More broadly we need to get the full OSG scalable offline analysis working.
    • Perhaps we can practice on MDC data??? (there are already injection frames that we can pretend are the R&P frames) lets try to see if we can get a dag that filters all 40 days in 2 days.
• Still worried about the cluster transition!
• Calibration monitoring
• Online.ligo.org

ISSUES OF NOTE

• Make Renee and Esmee ready for discovery
• Pastro classification
• Reneesmee single time
• Offline rework meeting
• signal model mismatch
• Injection splitting
• dt dphi SNR threshold for covariance
• Whitener range check
• R&P injection plan
• Meet about O4 containers
• Launch MDC-12