20190708-Analyzer-Mtg

Back to Main Page >> Analyzer_Meeting

previous meeting << >> following meeting

Logistic information

Meeting room at JLab: CH Conference room (on the accelerator side).

 BlueJeans calling instructions:
 Toll-Free Number (U.S.&  Canada):  888-240-2560
 International toll number:         408-740-7256
 Bluejeans CODE:                  385 600 867
 Bluejeans link: https://bluejeans.com/385600867

Agenda

Analyzer code updates

1. Recent branch activity & pull requests
   • Feature aggregator io changes
   • Adding a "grid" option to panguin
   • Merge from operations to develop, then merge to master: we should then verify against operations, and prepare to update operations
     • Tuning of burp cuts
     • Activating miniruns?
       • Miniruns aren't quite ready (as of this morning)
2. Feedback
   • Status of outputing to the new EPICS variables?
     • Amali had been having trouble merging her recetn developments with the branch which had a merge from develop. Rakitha will follow up later today
3. New feature plans
   • Continuing development of minirun functionality
   • Data-handlers which can process results from other data-handlers
     • Specific case: correlation analysis of combined results
     • This is why data-handlers have "priority". But so far the field is not used at all.
       • The data handlers are prioritized by the order they are listed in the config file, so we could use that ordering for now, without getting prioritization
     • Data handlers would need to participate in the "request/publication" type mechanic
       • Maybe pull these type of functions into a mixin class, to use in both the subsystems and the datahandlers
       • Quinn will make a list of the functions involved, and we'll discuss later in the week
   • Recheck the separator for the event cuts files: should be commas: Robert
   • BMW first pass extractor
     • Hooks are in place for an event-level data handler.
     • Checks the BMW decoder, and then fill a copy of the event tree (in its own file) if the BMW is active
     • Perhaps better, write a general-purpose data extractor (selector): it just requests either specific variables or looks at the ErrorFlag and writes to a tree based on the values. Can be used for beammod, or other things we haven't thought of yet
     • Wouter will set up a prototype, then hand it off to Cameron
   • Change of internal and rootfile base of asymmetry/differences?
     • This would end up impacting many scripts and locations in the code
     • We'll keep this as a idea for now, not start development yet
   • Issue #207: Cameron will follow up

Hardware/DAQ tasklist

1. Synchronization-check system. w/help from Ed Jastrzembski
   • New digital system is available basically now; when to install?
   • Cameron will follow up with Bob and Ed about what changes in the DAQ, and then how to interpret the results
2. Alarms like end-of-run voice
   • Data quality alarm concept: a realtime engine will report problems via text output on a periodic basis (~few seconds), the alarm display will poll this and warn the shift crew.
   • Cameron's alarm display is generally working, but has some small issues
   • "End-of-run" voice would be tied to coda transitions
     • Cameron will set up an alarm for about an hour after a run starts, to remind the crew to start a new one
3. Document everything. Circuit diagrams, cable map, "how to".
   • CC: timing diagrams need to be updated for the free clock modes.

Prompt analysis toolchain

• The engine is run twice, and then several scripts are run to produce plots
• Plots got to a web directory
• What tools are we missing for the WAC?
  • Autoprompt and the "prompt from list" are working fine so far
  • WAC should be using the RCDB commenting tools to store the state of the runs: Adding comments to the RCDB

Other issues

• Wouter has found a difference between the subblock correlations and the hw_sum correlations at up to 10-sigma level. Subblock slopes are all lower than the hw_sum slope; this can also be seen by plotting correlations between variables in the mul tree. See comment in HALOG entry 3699374. Disagreements between the blocks and the hw_sum can all be the same sign, instead of being scattered both positive and negative.
  • In the mul tree, he found yields of the BPM hw_sum compared to the sum of the sublocks shows non-zero results.... Actually, this was a plotting error, as one needs to divide the sum of the subblocks by four to compare to the HW sum (all of the quantities are normalized to the number of samples in that measurement).
  • Wouter will try using the phasemonitor to see if the subblocks are picking up different 60Hz contributions

Present

Cameron, Wouter, Rakitha, Weibin, Quinn Phone: Devika, Robert R., Sakib, Paul, Sanghwa