WAC Notes February 23 2021

From PREX Wiki
Revision as of 14:14, 23 February 2021 by Cameronc (Talk | contribs) (→‎February 23rd 2021)

Jump to: navigation, search
RCDB HALOG HAPLOG DocDB Online Prompt BlueJeans Link Runlist spreadsheet, respin 1 spreadsheet

PREX Main << Weekly Analysis Coordinator << WAC Notes CREX << WAC Post-Run_Meeting Notes

WAC Notes February 16 2021 << >> WAC Notes March 2 2021

Organization

Date and time: February 23, 2021, 11am
Bluejeans ID: 564945377
Meeting link: https://bluejeans.com/564945377
Runlist spreadsheet

February 23rd 2021

Preparing for respin2

  • Paul fixed the Burp Cut -> Trip Cut for BCMs, to resolve the ~ run 7865 large burp wipe outs
    • Example output rootfiles: /chafs1/work1/apar/pking/rootFiles/respin1_newburp_7865.000.root
  • We are satisfied with the shortened updated burp cut length, stability ring length, beam trip cut length
  • A bunch of cut map updates are done, a few remain
    • Github branch crex-respin2-alpha
    • /adaqfs/home/apar/cameronc/WAC/japan-crex-respin2/
  • A bunch of RCDB updates are listed, a double check would be a good idea
    • /u/group/halla/parity/software/japan_offline/prompt/prex-prompt/PVDB/prex/scripts/respin2_changes.csv
  • Weibin determined that the Burp cuts outside of beam trip times are not a significant source of cut out data
    • Burp cut N runs N Events
    • Cut = Burp cut is on, Beam trip off, BMW off, and User bad events off
    • 620823552 75 2745 - Hex = 0x25010400
    • 604046336 73 691 - Hex = 0x24010400
    • 604046464 22 130
    • 620823680 12 513
    • 604046400 8 31
    • 889259008 4 84
    • 620823872 2 2
    • 604048513 1 6
    • 604048385 1 6
  • Beam off pedestals and further pedestal studies are underway

February 16th 2021

AT respin data is not reproducible immediate from Operations branch.

  • /chafs2/work1/apar/aggRootfiles/PREX_regression_AT
  • CREX Respin1 was done on the ifarm over the summer. Where is that data?
  • Why are there 200 additional JAPAN evt_tree entries in all respin1 root files? Sounds like a ring buffer change or engine change (to not toss that 200 event ring). Changed ring size from 400 to 200?
  • Regenerate prompt and respin1 AT data on adaq computers as a double check of that data. The map files look good after August 14th, so no December respin1 changes should affect AT. The crex-respin1 branch should match ifarm respin1, and it should not have any meaningful pedestal or cut changes after August 14th. The prompt roll-back commits should match the existing prompt rootfiles.

February 9th 2021

Special investigations to finish cleaning up the runlist and cuts:

  • Some effort needs to be applied to reading the triage runlist
    • Making sure that our understanding of cuts, etc. of the last few days is applied to the runs looked at a month ago
    • Adding cuts and changes to the repository
    • Do a respin 1.2 with small changes in cuts and RCDB classifications, then later respin 2
      • If the change is unambiguous and clearly going to be sufficient - Don't need to immediately respin
      • If the change is difficult to accomplish, do a quick respin to check the result
      • Put changes into ifarm install, new crex-respin2 branch: /u/group/halla/parity/software/japan_offline/operations/Parity/prminput
      • Do a respin 1.2 after all changes from triage runlist have been implemented
      • Personal checkouts of JAPAN can be set up for spot-checking of changes
    • For RCDB changes use the script to change all at once, prepare a big csv first
      • /u/group/halla/parity/software/japan_offline/prompt/prex-prompt/PVDB/prex/scripts/update_conditions.py
      • /u/group/halla/parity/software/japan_offline/prompt/prex-prompt/PVDB/prex/scripts/respin2_changes.csv
  • 12X Burp cuts too tight around 7800 run range, and other ranges too?
    • Check the rootfile to find out what cut is actually tripping (BCM? BPM?) (Weibin)
  • 4eX and 12X Limit cuts slice data off after pedestal shifts in respin1 (Victoria already looked at 1000 runs)
  • Double checking the noise-removed BPMs (from Devika's BPM resolution analysis) don't see any helicity pickup or have excess noise that pops out at us - especially 12X and 1X (Cameron)
  • Investigate runs with significant changes in event numbers from prompt to respin1 - http://ace.phys.virginia.edu/HAPPEX/4454 (Robert, Cameron)
  • Advanced comparisons between prompt and respin data sets at the minirun level (Do later)
  • Strange BPM Y noisy behavior - trip recoveries (possibly letting data in before BPM gains have stabilized) (Robert, Victoria)
  • Pedestal analyses (especially beam off pedestals)
    • Beam off pedestals (Weibin, Cameron)

February 5th 2021

Finishing up looking at BPM issues and Aq issues

  • 7790-7794 run range has a BPM Y hairy noise that is in all BPMs
    • Check if the beam trip recovery is at fault
    • If needed add a Y hard limit cut to some of these runs (maybe 4eY)
    • See also: Run 7259 also has X noise in all the BPMs for no clear reason
  • 7719 has some low current data during beam trip times that make it past > 80 uA cut
    • We should try to cut out low current data, particularly if it is during very trippy periods
    • They don't matter a whole lot
  • Large Aq runs: http://ace.phys.virginia.edu/HAPPEX/4457
    • The bad ones are already under suspicion, remaining runs will be checked for bad beam positions or other fatalities (large Aq alone is not fatal)

Convenient Agg scans:

  • Aq scan (2 sigma cut)
    • agg->Scan("rcdb_run_type:rcdb_run_flag:run_number:minirun_n:n_runs:yield_bcm_target_mean:yield_bpm12X_mean:1:asym_bcm_target_mean/ppm:asym_bcm_target_mean_error/ppm:asym_bcm_target_rms/ppm:(asym_bcm_target_mean/asym_bcm_target_mean_error)","abs(asym_bcm_target_mean/asym_bcm_target_mean_error)>2")
  • BCM yields per minirun
    • agg->Scan("run_number:minirun_n:yield_bcm_target_mean:yield_bcm_target_rms:rcdb_run_type:rcdb_run_flag","rcdb_run_flag==1 && yield_bcm_target_mean<120")
    • Change the low current cut to get an idea

February 2nd 2021

BPM 1X and 12X look stable and have reasonable noise, coming from Devika's analysis of BPMs vs. eachother.

Looking at 12X and Target BPM jumps/shift runs: http://ace.phys.virginia.edu/HAPPEX/4428

  • BPM Target matter for avoiding copper
  • BPM E matter for avoiding inelastics and preserving regression stability

Need to finish looking at BPM jumps on Friday

February 1st 2021

Finish looking over runs:

  • Finished looking at NeedCut and Suspicious runs - suggestions live in spreadsheet
  • Pedestals for BPMs are all over the place, and cause yield shifts
    • These yield shifts bump into limit-cuts for target BPMs at least once, and likely more - see HAPLOG 4454
    • BPM 12X sees a 0.3mm shift during CREX part 2 vs. it's prompt value, this likely causes cut issues too
    • Reverify pedestal values used in respin1
    • Recheck pedestal analyses for quality


January 29th 2021

We looked at the run lists generated by WACs over the last few weeks and blessed suggestions or suggested new things to do to fix runs with obvious issues or solutions:

https://docs.google.com/spreadsheets/d/1l5qA0u9KhLIFA-MbcEpI3ezUD9saq6g1kzB_eTJD_po/edit?pli=1#gid=480769793

We'll meet again on Monday (starting at 7994 in the NeedCut list) and also look at the BPM jump run list (http://ace.phys.virginia.edu/HAPPEX/4437)

January 26th 2021

Last day of going over issues and populating the WAC lists. A number of comments and notes are listed here, including possible merging of slugs, etc.

Attendence

Cameron, Victoria, Paul King, AJ Zec, Juliette, Sakib, Weibin, Ryan

Retrieved from "https://prex.jlab.org/wiki/index.php?title=WAC_Notes_February_23_2021&oldid=12507"