Minutes2008 02 26

Minutes of 26 Feb 2008 meeting of Geant4 Technical Forum

Editor: G.Cosmo, 26 Feb 2008
Revisions: J.Apostolakis, 6 May 2008

- 16:30 - First time ever all telephone/VRVS connections work !

1. Open Issues
 - R1102: revised in the present development version;
 implementation planned for next release.
 - R1401: feedback requested. To be closed.
 - R1501: in development
 - R1602: postponed to next release
 - R1604: to be closed. Ongoing process
 - R1502: prototype under study or feasibility
 - R1503: not delivered in 9.1. No updates.
 - R1504: no updates
 - Composite external fields: provided examples in development release
 Should be reviewed if actions are part of the 2008 development plan.

2. Plans for developments in 2008
 - Planned release in December; no decision yet on potential release in June.
 - Major revision of integration testing to improve tests and checking of
 - Concern on resources available, much reduced in hadronics and in testing.
 Affecting planning.
 - Feedback is asked on the relative priority of the proposed developments
 for 2008.
 - Details: see list of planned developments
 - Request (G4beamline): documentation of physics-lists, what it does and which
 application each one is good for... current detailed pages are outdated.
 G.Folger replies that a draft of the new pages are in place and on the web:
 + http://geant4.web.cern.ch/node/73
 but no progress was made in the last few months. The request is accepted.
 - Issues with hadronic shower shapes: several areas in the planned
 developments are affected.
3. Matters arising
 1. Report from HARP on hadronics, sent to the experts to be discussed;
 2. How Value of proton mass in Geant4 reportedly does not agree with PDG 2006
 value (at 7th digit). How to obtain the values of masses of particles
 before initialization? 
 - First response: An update of particle properties to the values given by
 PDG-2006 was done for Geant4 release 8.2. Open question to be verified
 whether the EM physics module uses these values, and if mismatch exists
 with values of electron, proton and neutron mass imported as constants
 from CLHEP.
 Note: Confirmed that all particle properties except proton, neutron and
 electron masses were updated for PDG 2006. Those masses were kept
 equal to values in CLHEP. Action to communicate request for update
 of these values in CLHEP. 

4. Plans for Geant4 in experiments
 4.1 CMS
 In production 8.3.p01 until January, moving to 9.1.p01. Using QGSP_EMV
 physics list. Production made on SLC4/gcc-3.4.5, 32-bits. 
 Feedback: removal of std::deque in CHIPS brought a reduction of up to 25%
 in total dynamic memory allocated; still problems in reproducibility.
 Plans: new version 2.0 beginning of March, planning to use 9.1.p01 for
 massive production ~O(300 Mevents); open to include fixes and performance
 improvements, impact to physics to be quantified.
 Geant4 is the basic choice, keeping GFlash fast parameterization as an
 Calorimeters: formalized task-force, provide by the end of Spring (version
 2.1) the best version of simulation available in comparison with test-beam
 data. Some open problems are presented in the slides, see slides for
 details: energy deposits in crystals and scintillators, resolution at
 high-energy... good discussions and exchange of feedback undertaken.
 Requested support for 9.1.p01 for whole 2008 for fixes and improvements in
 performance. Required developments in physics separated from maintenance.

 4.2 ATLAS 
 In continuous production since several months, using 8.3.p02.
 Only maintenance mode changes (fixes) allowed. Currently undergoing major
 efforts to understand which physics-list to use in production and to assess
 shower parameterization. Until now used QGSP_EMV, but use of Bertini Cascade
 in the physics-list in use seems required (QGSP_BERT).
 Currently production version running smoothly, with no crash reports, very
 reliable simulation suite.
 New features required, parallel-geometries/BREPS/... plans to branch out to
 9.1.p01 for release 14, to be used (also later versions) in release 15.
 Required traceability of changes introduced, especially when concerning
 physics performance. Tracking in field becoming a concern. Any improvement
 in CPU is welcome.

 4.3 LHCb: 
 Last version 26r0 using Geant4 7.1.p01. Requirements when switching to a new
 version is the compatibility, completeness of features included, stability
 and validated physics.
 Version 8.3.p01 used in last development version of Gauss >30r0; preparing
 to retire 26r0. Trying to make the cycle shorter to put latest Geant4 release
 in the development version.
 Some concerns related to dE/dx evolving among the various Geant4 versions,
 differences observed in energy-loss when switching on/off delta-rays,
 different for different particle types.
 Problems with multiple-scattering in the muon system have been fixes in 8.X
 series of Geant4, still when switching on delta-rays.
 Concerns in the performance obtained by various physics-lists with/without
 tracking/range cuts.
 Need to find problems as early as possible; agreed with Geant4 on the
 observables to monitor at every new Geant4 release.
 Proposed a month as time delay for testing a new release before release date.
 Plan to move to Geant4 8.3.p02 / 9.1.p01 for releases 31r0 / 33r0 of Gauss
 respectively. Ready-for-data version (of Gauss) to be ready by end of May.
5. Converging needs for experiments production
 - Goal: Reduce number of Geant4 versions used in mass production in LHC
 experiments. Ultimately move all experiments to single Geant4 release for
 agreed period of time. Currently (2008) ensure that no more than different
 versions are used in production from all experiments.
 Two versions (8.3.p02 and 9.1.p01) is the current status. Agreed that for
 2008, experiments will target Geant4 release 9.1 as the target version for
 future development, and the target for migrating production. 
 No request for routine back-porting of fixes to old versions. The effort
 required means that this is not possible. Only problems significantly
 affecting physics results will be considered for back-porting during interim
 period. Maintenance and support efforts are concentrated on release 9.1. 
 The Geant4 collaboration undertakes to provide patches with fixes for
 Geant4 Release 9.1 for the full calendar year 2008.

6. Platforms supported
 - Continued support for SUN as a production platform is expected to be dropped
 during 2008. Plans are to drop it during spring 2008, unless need and
 resources (human and computer) are identified.
 - The potential to provide some support for an additional Linux flavour,
 potentially SuSe, was discussed briefly. A call for volunteers to
 contribute to this is being considered.
Meeting closed.

Meeting Date: 

Tuesday, February 26, 2008