GlueX Offline Meeting, May 27, 2015
GlueX Offline Software Meeting
Wednesday, May 27, 2015
1:30 pm EDT
JLab: CEBAF Center F326/327
- Geant4 update (David)
- HDGeant4 update (Richard)
- Review of minutes from April 29 (all)
- Offline Monitoring (Kei et al.)
- Spring 2015 Commissioning Simulations (Sean)
- New build manager (Nathan)
- Action Item Review
- The BlueJeans meeting number is 968 592 007 .
- Join the Meeting via BlueJeans
Talks can be deposited in the directory
/group/halld/www/halldweb/html/talks/2015 on the JLab CUE. This directory is accessible from the web at https://halldweb.jlab.org/talks/2015/ .
- CMU: Curtis Meyer
- FIU: Mahmoud Kamel
- FSU: Aristeidis Tsaris
- JLab: Mark Ito (chair), David Lawrence, Paul Mattione, Kei Moriya, Elton Smith, Nathan Sparks, Justin Stevens, Simon Taylor
- NU: Sean Dobbs
- Regina: Ahmed Foda, Zisis Papandreou
- UConn: Richard Jones
- gxproj5 account now available. Mark announced that a new shared production account has been created. Kei will be using it for offline monitoring of the Spring 2015 data.
- New Offline Software wiki page released. Mark reminded us that the Offline Software wiki page has been updated. The goal is better organization of the information. There are a still a few loose links at the bottom that need to be filed.
- Scratch git repositories on GitHub. Mark went through the email he sent announcing the repositories. He flashed through some of the features of the GitHub site. [Added in press] The repositories have been updated.
Geant4 and GlueX
David described his effort with a G4 simulation of GlueX as configured for the charged pion polarization experiment, CPPsim. See his slides for details. He discussed
- geometry implementation
- definition of the MWPC muon detector
- compatibility with hdgeant
- comparisons of muon response in FDC between hdgeant and CPPsim
- tracking efficiency comparisons
- other comparisons of detector responses: FCAL, TOF, pions in the FDC, momentum resolutions, CPP-critical kinematic variables
Richard described his efforts in the Geant4 conversion.
- He thanked David for giving the effort a push forward.
- The original design was GEANT3 oriented, difficulties found going to GEANT4
- Some short-cuts don't work anymore, Geant4 pickier about overlaps, e. g., small overlaps of ends of CDC straws with aluminum end-plate.
- Using Boolean subtraction works in principle, but incurs unacceptably large performance hit.
- Solution from ILC fast MC effort: multiple layers in geometry definition, not all layers need to be active at all times.
- To validate new geometry, needed visualization, but G4 bugs were found and needed to be addressed.
- He plans to use CCDB/JANA-Resource scheme for magnetic field maps.
- Work done on converting/validating HDDM output stage using the new C++ API.
- Needed to expose formerly hidden methods in xstream library so that they are not masked by C++ STL routines.
- Makes things more efficient. N. b. we are now using "new" code.
- Old REST files will generate warnings. CRC checking were corrupted on write before now.
- Next step implement all features of our current event generation scheme: genr8, bggen, particle gun, E&M background with coherent bremsstrahlung spectrum (cobrems).
- Will use David's hit classes, as developed for CPPsim.
- He will set-up a git repository for the new code. We should then be able to access pre-release versions of the code for testing.
Spring 2015 Commissioning Simulations
Sean reported on preparations for simulations of Spring 2015 data.
- Many jobs have run, but conditions were not tuned properly. We are starting over with
- 3.0 GeV coherent edge
- Lower cut-off for PYTHIA simulations, use PYTHIA down to 2.5 GeV photon energy.
- Studying whether to include recent changes to mcsmear for the BCAL.
New build manager
- Template system
- Julia modules
- Tasks and Scripts
The system is well-suited for the situation where many builds are needed with slightly different versions of particular packages. All packages are built under a single directory; there is no ambiguity about which versions go together. This system was motivated by a debugging exercise Nathan went through some months ago with Paul.
There is large overlap in functionality with the version management system Mark has been working on. There is also virtually no overlap in software tools; Julia vs. Perl. It is unclear whether the two can be merged. Richard also mentioned that there is a third way to go: gridmake, which we use on for installation of the software on all OSG sites. In any case it is clear that the use-case that Nathan described needs support.