AIEC Weekly Meeting Sep. 29, 2022

From epsciwiki
Jump to navigation Jump to search

Agenda:

  1. Previous Meeting
  2. Activities:
    • Grafana Update:
      • Diana working on fixing data access issues.
    • Torri working on EPICS, RCDB, CCDB documentation for others, Naomi working on getting specific documentation on tables in the CCDB
    • Diana/Thomas working on plot look books to make model evaluation and behavior easier to understand, with plans to turn control back on maybe next Thursday pending Naomi's sign off
    • Neurips ML for Physical Sciences paper will be submitted tonight
    • Naomi suggested submitting an abstract (due 12/14/22 at 7:59 PM ET) for HEP2023 in January
    • Gathering some preliminary data from Primex corresponding to extreme low pressure, low temperature runs that are out of our training domain
    • CPP
      • CPP Calibrations (Naomi)
    • Quantification of bad gain calibrations
      • Looking at initial results comparing traditional GCF compared to roboCDC GCF in progress
      • HVB Currents during initialization are not consistent with what we see during actual running, but are consistent with data from the HVBs from EPICS
      • Naomi identified why our model never learned anything useful from HVB current, related to the CDC alcohol swap between 2018 and 2020 run periods. This affects the gains. See also the data in these plots GCF_gasdensity that Diana's model used for training. GCFs vs gas density for 2018 and 2020 should not overlap.
      • CPP plans
      • CPP preparations
    • Capstone Students
      • Prepare for Oct 13th presentation to capstone student candidates.
      • Oct. - Dec. will purely be EDA and basic project understanding
        • 4 - 6+ hours of meetings to be expected in this timeframe
      • January will really get started.
      • Diana (and Torri?): Proof of concept retrieval of all calibrations, EPICs data, RCBD(?) data for a single run in preparation for when capstone students get going.
      • Is there any need to alert MIS that we expect new User logins to be needed by end of October?
      • What will they need access to?
        • JupyterHub resources?
        • ssh login to farm?
        • AIEC GitHub repo?
        • anything else?
    • FDC calibration
      • Reach out to Lubomir alternative FDC for cosmics- David.
      • 2020 CDC-FDC model results: CDC-FDC model slides
      • Potentially give capstone students FDC calibration model development
      • Discuss with Simon how useful FDC calibration constants are for physics analyses
    • Next steps
      • CLAS12 Polarized Targets
        • Update from James (8/25): concern about poor fits that give polarization (i.e. Diana's eventual target values). Can begin initial data exploration whenever we have time.
        • data from fits to obtain polarization AND relevant EPICS data are stored in json files located in /group/poltar/HallB/RGC and should be readable
      • **AIEC Auto Control Wish List:**
      • GlueX TOF
      • BCAL baseline (slides) - David follows up:
        • Do we need another sensor?
        • What are the limitations on existing sensor and logging of temp value?
  3. Workshops
  4. Talks/Presentations
    • Torri and Diana presenting at QNP (QNP)
    • Diana presenting at ACAT2022 (15 min talk + 5 min questions)
  5. Publications: