GlueX Offline Meeting, September 17, 2014

From GlueXWiki
Jump to: navigation, search

GlueX Offline Software Meeting
Wednesday, September 17, 2014
1:30 pm EDT
JLab: CEBAF Center F326/327


  1. Announcements
  2. Collaboration Meeting
  3. Review of minutes from September 3 (all)
  4. agenda for collaboration meeting
  5. commissioning simulations
  6. offline analysis of online data and coordination thereof
  7. 100 ns shift issue; effective time cut
  8. Action Item Review

Communication Information

Remote Connection


Talks can be deposited in the directory /group/halld/www/halldweb/html/talks/2014-3Q on the JLab CUE. This directory is accessible from the web at .


Present (compiled from the secretary's memory; omissions likely, please add yourself if you were omitted):

  • CMU: Paul Mattione, Curtis Meyer
  • JLab: Mark Ito (chair), David Lawrence, Kei Moriya, Simon Taylor
  • IU: Matt Shepherd
  • NU: Sean Dobbs
  • UConn: Richard Jones


  • CCDB 1.03 released David put in a fix to avoid conflicts in database access for

multi-threaded jobs. See the email announcing the release for links to more details.

  • Nightly Build Versions The nightly build is now using CCDB 1.03 and JANA 0.7.2.

Collaboration Meeting

We reviewed the agenda Mark posted on the collaboration meeting page. It looks like this:

  • 16:05 Session IVb (120) --- Offline - (Organizer: Mark Ito) - Chair:
    • 16:05 (25) --- Overview - Mark
    • 16:30 (25) --- Tracking - Simon
    • 16:55 (15) --- BCAL Calibrations - Will McGinley
    • 17:10 (25) --- Commissioning Simulations - Sean
    • 17:35 (25) --- Offline Analysis During Run - Paul
  • Curtis suggested the addition of the BCAL calibration talk from Will McG.
  • Matt noted that Ryan will give a talk on his multi-pion analysis in the physics session.

[Added in press: at Zisis's suggestion we will add a talk on BDTs and BCAL reconstruction.]

Review of Minutes from September 3

We looked at the minutes of the last meeting.

Realistic Tagger Timing Ambiguity

  • The trunk seems to be working after the merge of Richard's tagger timing development branch.
    • Paul reported that the Analysis Library is now working with the ambiguity in the chosen tagger hit. It picks the correct hit 95% of the time [at a beam rate of ?].
  • Richard reported that mcsmear has been updated to smear tagger hit timing.
  • Paul explained the method for picking the RF bucket. There are two levels:
    1. Event-level pick: based on start counter times.
    2. Combo-level pick: based on propagating (a) TOF and BCAL times for charged tracks and (b) photon shower times back to the vertex. Note that within a single event, this pick may vary from combo to combo.
    • Level (1) is useful for diagnostics. Within the context of an analysis using the Analysis Library, level (2) is the only one that counts.

Comparing Results between HDDM and EVIO Data Formats

Sean had further comments.

  • The doubling of wire-based track multiplicity from EVIO compared to HDDM has been traced to a rounding error in the Flash-125 emulation. A fix had not been implemented yet.
    • Richard had noted this effect in his debugging of the tagger-timing branch. Of the many such discrepancies he had to address, this one was left for the FADC experts.
      • Sean noted that of the many discrepancies addressed by Richard, he had independently addressed many of them. He is pleased that they have been resolved on the current trunk.

Commissioning Simulations

Detector Simulations

Simon has checked in the commissioning target geometry. He has also updated other beamline elements as well as correcting the start counter geometry. None of the elements have been changed in a way that modify hits. Richard has also been working on the pair spectrometer.

We are very close to being a position to start large-scale simulations.

Tagger Hall Simulations

Richard made an appeal for manpower to help run tagger Hall simulation. We decided that when setting up the detector simulation run we would try to set up the tagger hall simulation at the same time. Other folks could then be called in to baby sit the jobs.



With all of the activity centered around the commissioning run, Mark was skeptical if he and others have the time to fully launch Data Challenge 3 at this time. Its start will likely be delayed a month or so.

Offline Analysis of Online Data and Coordination Thereof

Paul has volunteered to organize this effort.

The template for the effort is the [list of commissioning items] Curtis put together some time ago.

Kei has volunteered to help with the effort as well. In particular he is interested in picking a few specific channels and doing the standard analysis on them.

Matt asked for guidance on how to get a file of Pythia events in EVIO format. David has a wiki page with instructions and Richard has written scripts that do exactly this.

Paul will look at the overall issue and gather volunteers.

Monitoring Histograms

David and Sean will work on documenting the process for implementing online monitoring histograms. These are done with plugins and use ROOTSpy to collect statistics from multiple nodes. Elliott Wolin started this effort, but it there have been changes in recent weeks. Sean also has done some development on the ROOTSpy package, as has David.

100 ns Shift Issue: an effective time cut

The 100 ns introduced was to give positive TDC times. A side effect is that times earlier than 0, after shifting, are dropped from the data stream. The "right" thing to do may be detector dependent. Richard pointed this out and suggested that this offset be separate for each detector system. Sean had already largely implemented this idea, putting the constants into the calibration database. The scheme depends on simulated EVIO data using "mc" variation, so that the shift introduced in conversion to EVIO from HDDM can be reversed. Of course the EVIO conversion program should use the same set of constants to apply the shift in the first place. Sean will complete the implementation along these lines.

Action Items

  • Reconcile understanding of pair spectrometer geometry changes. (Simon, Richard)
  • Update the documentation on how to implement monitoring histograms. (David, Sean)
  • Implement subsystem dependent time offsets for making TDC values positive all positive integers.