GlueX Offline Meeting, December 11, 2013
GlueX Offline Software Meeting
Wednesday, December 11, 2013
1:30 pm EST
JLab: CEBAF Center F326/327
- 1 Agenda
- 2 Communication Information
- 3 Minutes
- 3.1 Announcements
- 3.2 Software Review Re-Cap
- 3.3 Review of Minutes from the November 13 Meeting
- 3.4 Track Reconstruction Holes: Single Track Reconstruction
- 3.5 MC Track Genealogy
- 3.6 Long-term code maintainability
- 3.7 REST format and Track Swimming
- 3.8 Reconstruction Job Crash Status
- 3.9 Early Adjournment
- 3.10 Action Items
- Review of minutes from the last meeting: all
- Software Review Re-Cap
- Track Reconstruction Holes: Single Track Reconstruction
- Simon: "So far all I have been able to determine is that it seems to depend on the choice of mother volume for the straws PCON for the current geometry, TUBS for older versions). For tracks going in certain phi regions in the back angles, the straws seem to be invisible in GEANT. In fact these dead regions appear to be periodic in phi."
- MC Track Genealogy
- Long-term code maintainability (Paul): separate sim-recon into mcsim (genr8, bggen, hdgeant, mcsmear), recon (up to REST), and analysis?
- REST format and Track Swimming (Paul)
- JANA job crash status
- Event Store Overview Documentation
- Mantis Bug Tracker Review
- Review of recent repository activity: all
- ESNet: 8542553
You can view the computer desktop in the meeting room at JLab via the web.
- Go to http://esnet.readytalk.com
- In the "join a meeting" box enter the Hall D code: 1833622
- Fill in the participant registration form.
To connect by telephone:
- US and Canada: (866)740-1260 (toll free)
- International: (303)248-0285 (toll call) or look up toll-free number at http://www.readytalk.com/intl
- enter access code followed by the # sign: 1833622#
Talks can be deposited in the directory
/group/halld/www/halldweb/html/talks/2013-4Q on the JLab CUE. This directory is accessible from the web at https://halldweb.jlab.org/talks/2013-4Q/ .
- CMU: Paul Mattione
- FSU: Aristeidis Tsaris
- IU: Kei Moriya, Matt Shepherd
- JLab: Mark Ito (chair), David Lawrence, Dmitry Romanov, Simon Taylor, Elliott Wolin, Beni Zihlmann
- Northwestern: Sean Dobbs
This meeting was recorded (the recording will be deleted after a month).
Yesterday there was a meeting at JLab to plan deployment of database and web servers for Hall D, both in the counting house and in the computer center. This was the second meeting we had. Paul Letta presented a design which we batted back and forth. We came to a general consensus; Paul has circulated a document presenting the consensus design.
Software Review Re-Cap
Mark had given his summary of the review at two previous meetings and participants agreed to forgo a repeat performance.
Elliott noted that the committee had expressed mild encouragement about the use of C++-11.
Review of Minutes from the November 13 Meeting
We went over the minutes.
Dmitry is alpha-testing David's new SCONS build scheme.
Track Reconstruction Holes: Single Track Reconstruction
So far all I have been able to determine is that it seems to depend on the choice of mother volume for the straws PCON for the current geometry, TUBS for older versions). For tracks going in certain phi regions in the back angles, the straws seem to be invisible in GEANT. In fact these dead regions appear to be periodic in phi.
Richard Jones has requested a "before" plot in addition to the "after" plot he already has to help diagnose.
Paul told us that the last time this worked was circa the summer Analysis Workshop.
MC Track Genealogy
Mark reported that he has a working revision to how particle parentage is report in the simulation data model. The method, suggested by Richard, is to keep track of the correlation between our global particle ID number (usually no more than 20 per event) and the Geant track number (which can go to several thousand) using an STL map, behind the scenes.
He will check it into the trunk and would like others to look at it and report any problems or irregularities.
Long-term code maintainability
Paul introduced a proposal to separate sim-recon into mcsim (genr8, bggen, hdgeant, mcsmear), recon (up to REST), and analysis.
The overall issue is that, as the years pass, older data may require old versions of the software to reconstruct and/or analyze, but may still benefit from new versions of algorithms. How to maintain multiple, consistent versions is not a straightforward problem. Some of the problems can be solved by breaking our software packages into smaller, quasi-independent pieces. This introduces confusion due to multiple packages with multiple versions. Also only certain combinations of these versions will work together since the packages will never be truly independent.
Matt laid out how CLEO dealt with this issue in an email to the list. It frames the problem in more detail and provides a specific strategy.
Mark is working on a system to specify a combination of various packages in an XML configuration file that can be used both to drive a complete build of our software and generate set-up scripts to use this build. He showed an example file. This would help by providing a functional document of valid package combinations.
Matt suggested that to move forward, we need a specific proposal on a system that we can discuss and modify. Mark volunteered to propose something based on his XML configuration scheme.
REST format and Track Swimming
Paul noted that when analyzing REST format, a significant amount of time is spent re-swimming charged tracks to do matching with hits in the TOF and calorimeters. This matching is already done during reconstruction, we just do not save those results in the REST files. This gives us the option of using different matching algorithms post-reconstruction.
We decided to go ahead and save the results of matching in the REST file. As an option, the track can be re-swum and matching done again, overwriting the saved results. In either case all clusters from the calorimeters are saved in REST. Paul will implement the changes.
Reconstruction Job Crash Status
Simon checked in a change to fix a problem with events that never finish processing]].
Kei reported that he was still having problems with events of this type two weeks ago. Simon and Kei will check and see if this is due to the use of an old version or a different type of problem.
We had to adjourn after an hour; another group had the meeting room. Remaining agenda items will be addressed at the next meeting.
- Send Richard a "before" plot of the single-track test results. -> Simon
- Check Mark's decay chain reporting scheme. -> Kei
- Propose a long-term code maintenance scheme. -> Mark
- Implement storage of track matching information in REST. -> Paul
- Check for source of Kei's never-ending events. -> Kei, Simon