GlueX Offline Meeting, February 6, 2013
GlueX Offline Software Meeting
Wednesday, February 6, 2013
1:30 pm EST
JLab: CEBAF Center F326/327
- CCDB now the default
- New release of HDDS
- New release of sim-recon
- Review of minutes from the last meeting: all
- Collaboration Meeting
- Friday February 22, 2013
- 14:45 Session IVb (100) --- Offline Reconstruction - (Organizer: M. Ito)
- Data Challenge Status
- The Transition to GEANT4 (discussion).
- Closer review of b1pi/single-track jobs
- Action Item Review
- Review of recent repository activity: all
Talks can be deposited in the directory
/group/halld/www/halldweb/html/talks/2013-1Q on the JLab CUE. This directory is accessible from the web at https://halldweb.jlab.org/talks/2013-1Q/ .
- CMU: Paul Mattione, Curtis Meyer
- FSU: Nathan Sparks
- IU: Matt Shepherd
- JLab: Mark Ito (chair), David Lawrence, Yi Qiang, Sascha Somov, Simon Taylor, Elliott Wolin, Beni Zihlmann
- CCDB is now required to build sim-recon.
- There is a new release of HDDS, 1.5. This fixes some geometry issues and is needed to go along with the corresponding changes in the tracking code.
- The new release of sim-recon, sim-recon-2013-01-11, uses CCDB 0.06 and HDDS 1.5. Note that as of this writing, the recent bug fix from David for the REST format is not part of the release.
There was some discussion about the meaning of the date in the sim-recon releases. Note that the date corresponds to the date when the tag was first made from the trunk and not the date that the release is announced. In general, not all of the code in a release will correspond to the date in the tag name. Changes may have to be introduced to fix bugs.
The schedule for the offline session has been added to the agenda on the wiki.
Friday February 22, 2013
- 14:45 Session IVb (150) --- Offline Reconstruction - (Organizer: M. Ito) Chair:
- 14:45 (20) --- Overview and Data Challenge -- Mark Ito
- 15:05 (20) --- CCDB -- Dmitry Romanov
- 15:25 (20) --- Tracking -- Simon Taylor
- 15:45 (20) --- TBD -- David Lawrence
- 16:05 Coffee (30)
- 16:35 (20) --- Physics Analysis Tools -- Paul Mattione
Data Challenge Status
Not a lot has moved on analysis of data challenge data lately. Justin Stevens of MIT has been looking at them however. He expressed an interest in generating skims, to reduce processing time for selected channels. Paul is working on reconfiguring code in the REST plugin to facilitate skim creation.
The Transition to GEANT4 (discussion)
We reviewed the recent email discussion on the transition from GEANT3 to Geant4. In that discussion it was not unamimous that this is high priority. Points not raised in the emails:
- It should be possible to freeze G3 development for a time to allow Richard to work on a stable code base.
- Alternately, Richard could work on a branch, and changes to the trunk could be ported piecemeal, in a controlled fashion, to the branch if they occur.
- Richard has stated that there will be a lot of code re-use in the transition, so work done validating G3 code will not go completely to waste when validating G4 code.
- Discussion of this issue needs to involve Richard. Plans:
- Raise the issue informally with Richard at the collaboration meeting.
- Ask for a presentation on what is involved from Richard at a future offline meeting.
Closer review of b1pi/single-track jobs
At Beni's request, we discussed the problem of lack of monitoring of the results from the bi-weekly tests of the b1pi and single-track analyses. Recently there have been failures that have gone ignored. In particular, Beni has tracked down a issue with the proton yield in b1pi tests introduced a few weeks ago (January 16).
- Beni suggested that we show recent results at this meeting as a standing agenda item.
- Mark showed a prototype system for a light-weight email list system that interested collaborators could subscribe to on an opt-in basis. The idea is to have multiple result-nofication email lists.
We will try both approaches.
Review of recent repository activity
- Simon fixed a rare problems that was causing crashing in getting the fine-mesh magnetic field table. Junk position values from stopping particles were the cause of the problem.
- Paul has disable RF-time bunch selection.
- David has fixed an uninitiated pointer problem in the REST-format event generating code. He describes it as one of those I-wonder-how-it-ever-worked situations.
- David discovered a feature where bggen puts the vertex at (0,0,0), the default, and hdgeant assumes it really means (0,0,65 cm). This was discovered while introducing a new parameter that puts the vertex in an arbitrary place.