WAC Notes December 10 2020

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 December 3 2020 << >> WAC Notes January 7 2021

Organization

Date and time: December 10, 2020, 11:00am
Bluejeans ID: 454783784
Meeting link: https://bluejeans.com/454783784
Runlist spreadsheet

December 10th notes, updates from Dec 3

• Pedestal update (a temporary placeholder) summarized in HAPLOG 4417
• Prompt data set summarized in HAPLOG 4418
• Cuts and other Respin updates summarized in HAPLOG 4419
  • Added bcm_an_diff device to all CREX runs, resolving issue that was especially bad in runs 8350-8370: https://logbooks.jlab.org/entry/3845576
    • Added bcm_an_diff cut to all CREX runs already having reasonable current cuts

combinedbcm, bcm_an_diff,       -10,     +10,    g,      0,   0.15

• • Removed burp cuts from BPM wires for all CREX part 2 runs
  • Added BPM 12X burp cut to all CREX part 2 runs
  • no conclusion yet on burp cut extent - needs test runs and plot evidence
  • No cuts yet on target X, Y limits, wait until after respin 1 with a scan of the damage (along with BPM 12X) - see Devi's plot for summary of runwise jumps in monitors: https://prex.jlab.org/analysis/crex/grand_summary//slopes/det_sam_slope_yield_crexslugs.pdf
  • Added global cuts to ATs for all of CREX
    • Ignoring AT outs for 8444->
    • Ignoring ATs before slug 107, all runs before 5636 - http://ace.phys.virginia.edu/HAPPEX/4058
    • Fixed AT pedestals for blinded period to match - https://logbooks.jlab.org/entry/3858061
• Weibin added a bcm_an_diff plot to ifarm install of prompt plots
  • He'll update the cut to only do can events when (bcm_an_diff.Device_ErrorCode&0x40001c00)!=0 && (ErrorFlag&0x8000000)==0
  • If possible edit the asyms to diffs in their plots
• Cameron updated the Error Counter macro on the ifarm install
  • Will change the general cut information to show Beam Trip and All Else separately -> (ErrorFlag&0x8000000)==0 cuts out beam trip data
  • And change percentages to N events
• The target BPMs are not easily diagnosable from minirun plots (thanks Weibin), but will be analyzed similarly to the BPM12X data after respin1
  • Note that at run 6500 we have a major change in our optics and beam delivery, which produces similar beam instability as we saw at the beginning of the experiment in slug 100 or so
  • Some documentation of pedestal investigations during that time here: https://logbooks.jlab.org/entry/3787428
• Some commits of changes on adaq computer may conflict with committed or uncommitted changes on the ifarm (due to push pull privileges and edits testing)
• Weibin created a run list of runs included in prompt runlist text files but disagreeing in the RCDB: https://docs.google.com/spreadsheets/d/1l5qA0u9KhLIFA-MbcEpI3ezUD9saq6g1kzB_eTJD_po/edit#gid=214727289
• Robert made a run list for respin 1 with generous filters: respin 1 spreadsheet

December 3, Updates on action items from 19 November

• Plotting scripts - the ifarm respin plots get a GitHub PR 5
  • ifarm version lives in /u/group/halla/parity/software/japan_offline/prompt/prex-prompt on the feature-farm-prompting-tests branch of www.github.com/JeffersonLab/prex-prompt
  • operations version lives in ~/PREX/prompt/ and the operations branch
  • 38 commits different, likely large number of conflicts in local file paths and script implementation, some useful plot changes were made in CREX preompt which should be easy to copy over manually in isolation if needed
  • Paul King has push/pull privileges
  • ifarm directory group permissions are an issue
• BCM event level differences - re: https://logbooks.jlab.org/entry/3845576
  • Victoria says the 0.15 (0.3uA) level an_us-an_ds difference cut is sufficient for all concerning runs she can see, though there may be less-aggressive jumps elsewhere that are missed
  • Weibin added a prompt plot (tested for run 5459, see page 7 of BCM plots)
  • This plot needs updated cuts - don't draw beam trip data, don't draw BPM cut data -> Look for bits == bcm_diff object burp active && NOT beam trip & NOT BPM jumps
• Cameron will set up an error code counter to add to plots
• Cameron and Robert will work to get map files all up to speed
  • Pedestal map files (including information coming from beam off, ATs, and blinded tube study periods)
  • Global error definitions from bcm_diff and AT global cuts, etc.
  • Make sure all cuts are appropriate (what about the burp cuts extent changing midway through the run - see what that change was and verify with Paul King)
• Devi has a HAPLOG looking at large fluctuation BPM 12X jumps in all of CREX
• Sakib will update his run list more over the next week, anticipate adding some BPM 4aX target fluctuation cuts, but leaving 12X cuts for post-respin1
• Check-ins over SLACK preparing for respin 1 on ~ Monday

Updates on action items from 12 November

• Paul: Event-level BCM difference combination
  • New variable bcm_an_diff defined as (bcm_an_us-bcm_an_ds)/2, initial commit just has it starting at run 7502 to test it. This should be pushed back to all CREX runs.
    • Weibin will add a plot to the ifarm prompt plots scripts (and show Cameron where these live)
  • Event cut on this channel (initially defined in prexCH_beamline_eventcuts.8301-.map to test it): must be in -10 to +10 range (so up to 20 uA difference possible; could be tighter if desired); no stability cut; burp cut level of 0.15 (so a change in the disagreement by 0.3 uA triggers the burp). This was initially tested using run 8354.
  • Victoria has done further testing.
  • Robert will propagate this variable to all map files and do spot checks
  • Need to add this variable to our monitoring plots
  • Do we have a way to keep track of how many event are cut due to this reason?
• Back-propagating pedestals to start of run2 period
• Cuts monitoring
  • Can be tracked with the error counter macro
  • Needs to be uncommented
  • Add a text file printing for this info
  • Track which devices are triggers for non-beam trip cuts
  • Include BPM 12X and BCM difference (above definition)

November 5th notes

• Weibin: stability cuts map files
• Robert and Paul: burp cuts
• Special cuts for runs 6852, 7860, 8321, 8507, and more? Coming from injector conditions, etc. These should be verified. Also run 7402 needs an individual file due to IHWP change. Make sure we have a list of any runs that need "chainfiles" set since submission script needs to be special too.
  • 6852 - /adaqfs/home/apar/PREX/japan_WAC_only/Parity/prminput/eventNumberCut.6852.conf # Cuts to keep only evnts with Ca-48 target in position - cut out event=20000:
  • 7402 - /adaqfs/home/apar/PREX/japan_WAC_only/Parity/prminput/prex_bad_events.7402.map #HWP changed after starting the run - cut out 0:2000
  • 7842 - split two 12X domains, needs revisiting
  • 7860 (blinder.force-target-blindable=1)
  • 8321 (chainfiles=1)
  • 8507 - /adaqfs/home/apar/PREX/japan_WAC_only/Parity/prminput/eventNumberCut.8507.conf # Cuts to keep only events with Ca-40 target in position - cut out event=20000:, blinder.force-target-out=1
  • and more? Coming from injector conditions, etc.
• Cameron and Amali: new pedestals
• Other updates

Suggested Changes

Each WAC will suggest changes for their run period, which will be discussed at this meeting:

Cameron

Beginning of the second run period. It is in identical conditions to how we left in March.

See full notes from last week here.

• Main takeaway from that discussion is the questionable cuts runs will be addressed after respin 1 when checking how the burp cut changes do for removing 12X jumps.
• Otherwise there were some notes about potential BMOD slope segmentations that should be checked out (usually denoted in italics or bold).
• Runs 7626 and 7657 and 7798 should get marked as good, or at least get included in the respin to check again with ideal pedestals and cuts.

Sakib

• Slug 195- 7865 and 7866 is flagged as need cut. Slightly higher asym_us_avg rms. WAC comment indicates dp/p fluctuation. No other obvious issues. Ok to reclassify or needs a closer look?
• Slug 196- 7867 and 7869 has the same issue of dp/p fluctuations.
• Some runs from August 21 onwards might need bpm12X cuts. Still sorting through them.
• Run 7944 look like it could use a BPM 4aX yield cut

Weibin

• Runs 8032, 8034, 8370, 8637 were erroneously marked as production (are junk) and some marked Good (weren't)
• Run 6647 was too long and split

General

Tracking events and calibrations:

• Global run list - https://docs.google.com/spreadsheets/d/1l5qA0u9KhLIFA-MbcEpI3ezUD9saq6g1kzB_eTJD_po/edit#gid=192113375
• Special runs and events list - https://docs.google.com/spreadsheets/d/1mt39fCuIW3Pr8TO7fzhtFjmMFt4Op30jMm5wMQke6Xo/edit#gid=0
• Calibration runs analysis list - https://docs.google.com/spreadsheets/d/1mt39fCuIW3Pr8TO7fzhtFjmMFt4Op30jMm5wMQke6Xo/edit#gid=598758666
• CREX PQB sheet - https://docs.google.com/spreadsheets/d/1IXYl7AnaxCbTfQwMgTbbUlGdmMqnLQiTY1WADiEt-yE/edit#gid=179750759
• Full Experiment Aggregator chained file: ~/PREX/prompt/aggRootfiles/slugRootfiles/grandRootfile/CREX-full-miniruns.root
• PREX version: ~/PREX/prompt/aggRootfiles/PREXII_slugRootfiles/grandRootfile/PREX-full-miniruns.root (maybe 20% of the data got messed up somehow :/ )
• Kent's collab meeting list of tasks

Concerns:

• Need global cuts on ATs (probably also including at the end when they get blinded)
• Run 8107 is likely a "good" production run, but was labelled as a pedestal run
• Any run which had special cuts applied to segment BPM 12X regions will need to be checked on after respin 1 to retune those cuts

Weibin's stability/burp cuts info:

• Weibin harvested the cut levels from map files and sees that there is a only one BCM with stability and no burp cuts: HAPLOG
• And the BPMs burp cut changes from 2000 to 0 only at some point (around run 7600), but it needs to be clarified.
• Victoria says this has to do with BMOD having some data get cut that shouldn't have.
• Paul says he and Robert were working on getting the stability/burp cut map files to work correctly, which will require some spot checking to see if anything changes severely
• Recall why changes in cuts took place during running.
• We should, before respin 1, spot check a consistent set of burp cuts, scan over current good-events statistics, count how many 12X events we expect to get cut out, and then see if that is matched by a respin 1.
  • Checking the runs in slug 186 to see if it works there would be a good testing grounds
  • Also 7701 and 7718

Runs for checking later

Get pedestals set properly for initial part of the run - decide after respin whether to keep 7626

• 7626 - decide after respin
• 7627 - good
• 7701, 7718 need burp cut checks for sure
• 8157 looks to have a hardware issue in BPM 11
• Burp cuts and 12X cuts will be further scrutinized after the respin 1 data is made available (though we should make a systematic scan of current ErrorFlag statistics and 12X suspicious spikes).

Respin needs to consider specially:

• 6852 - /adaqfs/home/apar/PREX/japan_WAC_only/Parity/prminput/eventNumberCut.6852.conf # Cuts to keep only evnts with Ca-48 target in position - cut out event=20000:
• 7402 - /adaqfs/home/apar/PREX/japan_WAC_only/Parity/prminput/prex_bad_events.7402.map #HWP changed after starting the run - cut out 0:2000
• 7842 - split two 12X domains, needs revisiting
• 7860 (blinder.force-target-blindable=1)
• 8321 (chainfiles=1)
• 8507 - /adaqfs/home/apar/PREX/japan_WAC_only/Parity/prminput/eventNumberCut.8507.conf # Cuts to keep only events with Ca-40 target in position - cut out event=20000:, blinder.force-target-out=1

How to use Aggregator

Now updated to include "rcdb_[run_condition]" branches (converted to doubles, string entries have a lookup table)

• n_runs == Slug number branch. Was originally intended to denote how many runs had been TChained to derive a given file's outputs, but we just do miniruns and such these days instead.
• run_number == Run number branch
• minirun_n == Minirun ID branch (starts at 0 for first minirun of each run)
• split_n == Nothing for now. Was originally intended to denote split runs in JAPAN analysis, but other scripts never fully supported it and it rarely comes up
• All of the data in these aggregator files is exactly ErrorFlag==0 passing data, meaning it cuts out BMOD ON data and also fails to properly consider AT and SAM error flags (in prompt outputs, see respins for proper global error flag on those).
• The individual data fields for a given device are as follows:
  • _mean == mean value for a given minirun
  • _mean_error == the error on the mean
  • _rms == the RMS from the histogram for a given minirun
  • _rms_error == the error on RMS calculation (probably not really useful)
  • _nentries == the number of Multiplet entries used in the minirun for the histogram to obtain the above numbers
  • Some of the EPICS and slopes and such have _slope and maybe _slope_error or _mean and are what you would expect

Attendence

Robert Radloff, Cameron Clarke, Weibin Zhang, Devi Adhikari, Ryan Richards, Paul King, Sakib Rahman