Simulation Project Leaders Meeting 6-February-2008 ==== Attendees John Apostolakis (JA), Witold Pokorski (WP), Alberto Ribon (AR), Gabriele Cosmo (GC). Excused: Alfredo Ferrari (AF), Pere Mato (PM), Rob Veenhof (RV). --------------------------------------------------------------------- ==== Minutes - 2007/Q4 quarterly reports are under preparation --------------------------------------------------------------------- ==== Simulation physics validation & framework - Alberto - Meetings: o In the Physics Validation meeting of January 30th there have been three presentations, two by ATLAS on time structure of hadronic showers and one by CMS on the follow-up of the CMS 2006 test-beam studies. ATLAS combined test-beam: performance shown by using the QGSP_BERT and the QGSP_BERT_HP physics-list variant are similar; discrepancies are seen at larger time (1 microsecond or later) and understood to be due to an artifact of the rough model of the neutron capture currently implemented, which is known requiring improvement. ATLAS HEC: by including Birks quenching for the liquid-argon, and also by taking into account the time window (up to few tens of nanoseconds), Geant4 with QGSP_BERT physics-list (as in release 9.0) becomes now closer to the data, not only for the shower shapes, but also for the energy response and energy resolution, consistently with what is seen in the other ATLAS test-beam setups (TileCal and Combined). CMS test-beam 2006: by adding Birks quenching in the PbWO4 crystals with the same parameters as for the BGO crystals, by taking into account the time window, and by including the Cherenkov contribution to the light output of the PbWO4 crystals, it has been reported that there is now a better agreement with data, demonstrating that most of the discrepancies previously observed were due to instrumental effects and not to the actual simulation itself. - Extension of ATLAS TileCal 2002 analysis with Fluka: o The analysis is now completed. There has been a meeting last week with P.Sala to discuss the results, some questions regarding the energy resolution have been advanced. For the shower shapes, Fluka shows the best agreement, but Geant4 with QGSP_BERT describes also reasonably well the shower profile, and an improved agreement is expected with more recent Geant4 releases (the old release 8.1 was used in the analysis); for the energy response both MCs do well; for the energy resolution, Geant4 with QGSP_BERT seems in good agreement while Fluka is a bit lower than data. A presentation of the results is expected this week in an internal ATLAS meeting (TileCal week), and then at the next LCG Physics Validation meeting on February 27th. - Extension of ATLAS HEC analysis with Fluka: o A first setup using Flugg has been put in place by WP; the setup includes the full geometry and first modifications to the original stand-alone code provided by the ATLAS HEC group; the feasibility study has been done with positive results. It is now required further changes (e.g. beam direction and sensitivity of volumes of interest) and the work is handed over to AR. - Simple benchmarks: o The activity for the 4th benchmark has been postponed pending assignment of a student to complete the analysis of the data. - Flugg: o An important bug fix triggered by feedback received from a user in ATLAS has been provided (initial location in the geometry was systematically missing in the initialisation of the tracking in Flugg). This has allowed to use Flugg with the most recent version of Geant4. o A setup based on Flugg has been created for the feasibility study of the ATLAS HEC analysis. The first runs have been successful. o The code review of the package which is in the plan of work still remains pending due to lack of manpower. - GDML: o Z.Torszok is working on the implementation of the GDML writer in the Geant4 plugin. The reader has been released with Geant4 9.1. Some feedback from users has been received and few fixes have been introduced in the latest public patch 9.1.p01 of Geant4. o A proposal for implementing specific UI commands in Geant4 for importing and exporting GDML geometries is under evaluation. --------------------------------------------------------------------- ==== Generator services - Alberto - Meetings: o The next monthly meeting will take place on February 6th and will be mainly dedicated to MadGraph. - GENSER: o Few new version of generators have been installed in GENSER. o A request from Rivet to move to autotools for building the libraries (for those packages not providing a build-system their own) is under evaluation. Users should not be affected by that. - HepMC: o A new release (2.03.04) has been announced by L.Garren, including a bug-fix requested by ATLAS to IO_HERWIG; it also resolves the issue of the four-vectors implementation, by removing the simple math methods accidentally introduced in a previous release. o AR has planned a meeting on March 5th dedicated to HepMC for discussing technical issues and have more open discussions on needs and release planning. Such meetings will be taken regularly from now on. o The status of the migration to HepMC-2 by CMS and ATLAS will be verified at today's Generator Services meeting. --------------------------------------------------------------------- ==== Fluka - Alfredo - No report. --------------------------------------------------------------------- ==== Geant4 - John - Releases: o The Geant4 9.1 release has been made on schedule on December 14th. Among the features contributed by LCG/SFT members are the extension of Binary Cascade to enable its use in re-scattering (the potential re-interaction inside the nucleus of the products of a high energy interaction), improved straggling for ions (at low energies) and revised high-energy gamma tail for muon Bremsstrahlung, needed for NA49. Other new features include the alpha version of the Liege cascade (INCL), who translated the Fortran code of the Liege/Saclay group's INCL version 4.2, and the ABLA evaporation code; these were contributed by the HIP team. Among the fixes included in 9.1: a problem reported by ATLAS in high-energy collision of sigma-plus, which used to create a high-energy backward pion; in pre-compound and evaporation models; in the angular distribution of Bertini at high-energy; in quasi-elastic scattering. New physics-lists QGSC-BERT and FTFP-BERT avoid using the parameterised modeling for nucleon, pions and kaons. For additional details and issues see the release notes at: http://geant4.cern.ch/support/ReleaseNotes4.9.1.html o A patch release to 9.1 has been released on February 5th; several of the fixes have been back-ported to 9.0 and 8.3 as well (as patch-2). The fixes to 9.1 result in improved reproducibility; while partial improvement for this is seen for 9.0p2 and 8.3p2. Details are available at: http://geant4.cern.ch/support/Patch4.9.1-1.txt http://geant4.cern.ch/support/Patch4.9.0-2.txt http://geant4.cern.ch/support/Patch4.8.3-2.txt - Meetings: o A meeting with CMS HCAL experts in December, discussed in depth the apparent significant differences seen in the 2006 test-beam analysis. After this, the HCAL team identified a difference in the simulation and carried out an improvement in it. This greatly reduced the apparent significant disagreement, in particular for energy response, between data and simulation. The improvements were presented at the LCG Physics Validation meeting of January 30th. o Started a dialog in December with experiments on migration to new Geant4 releases, and support of existing ones. CMS undertook to start testing releases 9.0 and 9.1 early, with a view towards putting a more recent release (instead of 8.3) into production if no obstacle is encountered. o A meeting with ATLAS, CMS and LHCb is being organized to discuss about converging on a single version of Geant4 to be used in production by all experiments within 2008, and the length of time for supporting the releases used in production. Currently ATLAS utilize 8.3; LHCb is continuing a long term production using 7.1. CMS's baseline has been 8.3, and it is evaluating release 9.1 (and 9.0). o The Steering Board meeting took place on February 5th. JA reported the expected reduction in manpower affecting the system testing team, and first actions to streamline regression tests have been defined. The plans of work of the working-groups were presented and a release plan discussed, as well as support for releases in production by experiments. A release is scheduled for December as usual; a decision whether a scheduled June release will occur will be taken in April. o There will be a Technical Forum meeting planned for February 26th. --------------------------------------------------------------------- ==== Garfield - Rob (offline report) - Activation: o PH department and CMS have reached an agreement on the use of Fluka. CMS is amongst others prepared to state that the CERN-CMS technical groups will use exclusively the official Fluka release for future FLUKA work. o Work seems to be going on effectively, in particular by S.Mueller on the CMS side. CMS hopes that he can count on the continued valuable support of people like S.Roesler and M.Brugger. - Garfield: o RD51 proposal writing is in progress - the simulation part is ready. To prepare the way for the major upgrades which are planned, work is in progress to make the framework benefit of Root functionality. Next will be the full microscopic tracking of electrons in inhomogeneous fields based on Magboltz. This work is expected to be carried out largely by a doctoral student. Next in line are new ionisation simulation programs. --------------------------------------------------------------------- ==== AOB Next meeting scheduled for March 18th.