TF - 7/11 - Minutes, G.Cosmo
Revised, JA, 28 Nov 2007
Open requirements, JA
- Req.0103 (Input from CAD) - Proposed to be closed.
- Req.1102 (Identify a process) - Limited to which physical process it models.
Not yet expected in 9.1.
- Req. 1401 (Recoils ions) - Received contribution, being integrated.
- Req. 1501 (Categorize energy deposit by process) - no further progress.
- Req. 1602 (Primary ionisations) - in progress, sceduled for 9.1.
- Req. 1604 (Performance improvement) - Some issues addressed and included.
- Req. 1605 (Packaged phys-lists with low-en processes) - PL revisions planned for 9.1. Packaging uncertain.
- Req. 1506 (Documentation of phys-lists) - Draft version available.
- Req. 1403 (Forcing decay in tunnel) - Proposal presented in Hebden-Bridge
- Summary of requests from Underground experiments, MA
* Muon induced neutrons developments
* Radioactivity: open problem reports to be addressed
* Neutron_HP: open problem reports to be addressed, new LLBL model based on ENDL data-base to be provided.
Reports from experiments
- VRVS connection was not possible. See slide uploaded on the agenda.
- Problem with G4 user actions: moving to application bootstrap with Python classes
generated from C++: (expected) exception is thrown concerning the order of
instantiation of user-action and physics-list. It's a side-effect to be aware of.
It may be the case to throw a warning instead of an hard exception; not sure though...
Report from Geant4 Workshop, JA
- Several sessions with users, underground users, medical users, also GATE meeting
before the Workshop (topics of CPU performance in phantom geometries, low energy
physics models, verification and validation on energy deposition around point source)
- Summary by J.Perl on medical physics use of Geant4, improvement in documentation
of physics options, etc... good agreement with data in Geant4 by varying few parameters
compared to what is necessary when using EGCS-NRC instead.
- Second half was the collaboration workshop, results of improved shower-shapes with
the last two releases 8.3/9.0 series.
Supported platforms, GF
- Platforms for 9.1:
SLC4/gcc-3.4.6 32-bit/64-bit, MacOS gcc-4.01, Windows/VC++8.0, SunOS-5.8/CC-5.5
- Platforms currently having code ported but not regularly tested
- Expected needs for 2008:
* Solaris ? (yes from SLAC)
- Problem of man-power in system testing in Geant4 from 2008.
- Interest expressed in (but currently no commitments undertaken):
* MacOS 10.5 Leopard
* gcc-4.2.X interesting for bechmarking purposes (interest also in CMS)
* widespread - Consider also standard user platforms (Fedora?)
- Users are asked to respond about their needs for platforms (OS, compiler, hardware, 32/64 bit)
in 2008 by December 12th, 2007.
Geant4 CPU performance, TN
- Report from the G4 Workshop.
- Set of benchmarks, simple setups targeted to stress specific G4 areas, are currently
run manually by developers. Will be automated in near future.
- Plan defined for CPU benchmarks to be run regularly; policy defined (platform, frequency of runs,
control of the system,...); work started after the Workshop.
- Different techniques which can be used to improve CPU performance according to the use case.
- Overview of some recent CPU measurements. Improvements in latest releases came from
code reviews in physics.
Preview of 9.1 (EM and hadronics), VI
New physics-lists, GF
- Planning to drop the following physics lists in upcoming release 9.1:
- QGSP_EMV_NQE, QGSP_BERT_NQE (provided in 8.3 for backward-compatibility),
- QGSP_BERT_TRV (transition from Bertini to QGS at ~5 GeV, not ~9.5 GeV)
- New physics lists: QGSC_BERT, FTFP_BERT and/or FTFC_BERT, QBBC.
Experimental: QGS_BIC, FTF_BIC (replacing QGSP_BIC).
Neither of these are meant to replace (yet) the standard physics lists for
production use by experiments. Depending on use case and need, some will
be proposed for physics comparisons - with a view to address open issues
and improve on the existing physics lists. Will need thorough testing and
feedback from experiments and other users.
- Diffraction: plan to include projectile diffraction.
- Documentation: new draft pages for the hadronic physics lists, added to
the "User Support" official page from the Geant4 web site. Feedback requested.
Neutrino-nuclear interactions, MK
- See presentation slides on the agenda...
Wednesday, November 7, 2007