Π polarizability Software Meeting Feb 10, 2022

From GlueXWiki
Revision as of 09:41, 11 February 2022 by Davidl (Talk | contribs)

Jump to: navigation, search

Meeting Time And Location

12:30 EST (JLab time) Thursdays Weekly Meeting

Connecting

Join ZoomGov Meeting
https://jlab-org.zoomgov.com/j/1613177947?pwd=MUtiRnZjWnF4SzFjL0NwRW83NC9uZz09
Meeting ID: 161 317 7947
Passcode: 287835

David Lawrence is inviting you to a scheduled ZoomGov meeting.

Topic: CPP Software Meeting
Time: This is a recurring meeting Meet anytime

Join ZoomGov Meeting
https://jlab-org.zoomgov.com/j/1613177947?pwd=MUtiRnZjWnF4SzFjL0NwRW83NC9uZz09

Meeting ID: 161 317 7947
Passcode: 287835
One tap mobile
+16692545252,,1608628791# US (San Jose)
+16468287666,,1608628791# US (New York)

Dial by your location
       +1 669 254 5252 US (San Jose)
       +1 646 828 7666 US (New York)
       +1 669 216 1590 US (San Jose)
       +1 551 285 1373 US
       833 568 8864 US Toll-free
Meeting ID: 161 317 7947
Find your local number: https://jlab-org.zoomgov.com/u/acZK7hDWFf
 
Join by SIP
1608628791@sip.zoomgov.com

Join by H.323
161.199.138.10 (US West)
161.199.136.10 (US East)
Meeting ID: 160 862 8791
Passcode: 112160


Agenda

  1. Previous Meeting
  2. Pending Pull Requests and Reviews (CPP branch)
  3. FMWPC Library
    • DFMWPCHit - Calibrated hits
    • DFMWPCCluster
    • DFMWPCMatchedTrack
  4. PID:
    • e+e- PID
    • π+π- PID
    • πμ classification
  5. Event viewer
  6. FMWPC_online monitoring plugin
  7. CPP Simulations
  8. AOT

Useful Links

Minutes

Attendees: David L., Mark I., Simon T., Andrew S., Albert F.?, Nikhil K., Elton S., Rory M., Ilya L.

  • Pull Requests
    • Alex merged in David's pull request. Initially build failed from it but after deleting the src/.Linux_XXX directory and rebuilding, it succeeded. This is likely due to the change in the SBMS build system that was included in PR. This changed caused ROOT dictionary files with new names to be created, but any old dictionaries left over from previous builds would still be compiled.
    • Elton reported failed builds from the current CPP branch. It was thought (and later confirmed) that this was due to the same issue Alex found and so Elton could try the same fix.
    • Elton reported looking into his PR from last Nov. that was never merged. This was a plugin that should not interfere with anything else and so should still be merged in. Elton will confirm once he is able to compile the CPP branch as it is again.
  • FMWPC library
    • Simon is working on the change to remove dE and properly fill q in the DFMWPCHit objects. Expects it to be done by early next week. Will submit PR once it is ready.
    • Alex is still looking into the conversion of the DFMWPCCluster objects to provide values in the global lab coordinate system. There was some discussion on how to handle x(y) values for horizontal(vertical) wires where there is no actual value. Suggestions where whether to use 0 or some other value (e.g. -1000 or 9999). Alex will decide.
    • David has not yet modified DFMWPCMatchedTrack to use the DFMWPCCluster. He plans to do that over the next week.
    • Some discussion was had over how to access DFCALCluster information that was matched to the tracks.
      • Simon noted that this is already available in the DChargedTrack class.
      • David noted that he used DTrackTimeBased since he needed separate matching info for each mass hypothesis.
      • The FCAL_E_3x3 and FCAL_E_5x5 members of DFMWPCMatchedTrack do not include depth corrections (and possibly others) that are in the DFCALCluster objects. This may be important for the actual EM showers in the FCAL. Specifically as part of input features in the e+e- classification model. Andrew will look into whether he could retrain his e+e- classification model using these rough values and how much difference it makes.
    • Iliya suggested adding another data member to DFMWPCMatchedTrack that would indicate the likelihood of the shower matched to a track as being an EM shower.
  • AI PID
    • Andrew reported that he contacted Matt Shepherd and learned that the weights used in the FCAL TMVA model are embedded in a class generated by TMVA to implement the model. The generated source is compiled into the FCAL library so the weights are compiled right into the executable (e.g. hd_root)
    • Andrew will follow this method with his classifier.
    • Some discussion on how to get Nikhil's model into recon code since he uses Python/Keras. David suggested we may be able to transfer the model/weights into a TMVA form after it is trained. We will wait and see what Andrew learns by implementing his e+e- model first and then use that to guide how we implement the pi/mu model.
  • FMWPC_online plugin
    • Alex reported that he has added histograms for calibrated FMWPC hits to the plugin based on Elton's suggestion.
    • A macro to display them was also added. Example plots drawn using the new macro were shown. (The x-axis label was noted as being incorrect)
    • The changes have already been merged into master.
  • hdview2
    • David showed hdview2 with some of the updates suggested at the CPP meeting Wed.
    • Suggestion to add momentum of track at first plane as a draw option.
    • Better visualization of FCAL response is needed. This includes drawing DFCALCluster info.

Action Items

  1. Look into failing builds of CPP branch. See if Alex's fix resolves for Elton as well. (Elton)
  2. Merge or close Elton's PR from Nov. 2021 (Elton)
  3. Remove dE and fill in q for DFMWPCHit (Simon)
  4. Add conversion to lab coordinates in DFMWPCCluster (Alex)
  5. Replace FMWPC clusterizer in DFMWPCMatchedTrack with DFMWPCCluster objects (David)
  6. Implement class for accessing results of e+e- model (Andrew)
  7. Fix x-axis label for calibrated hit histos in DFMWPC_online (Alex)
  8. Add data member to DFMWPCMatchedTrack to indicate likelihood track matches to EM shower (David)
  9. Improves FCAL visualization on hdview2 (David)
  10. Add option to draw momentum at first FMWPC plane to hdview2 (David)