Simulation Project Leaders Meeting 13-July-2010 ==== Attendees John Apostolakis (JA), Andrea Dotti (AD), Witold Pokorski (WP), Gabriele Cosmo (GC). Excused: Pere Mato, Rob Veenhof. --------------------------------------------------------------------- ==== Minutes - Nothing to report. --------------------------------------------------------------------- ==== Simulation physics validation & framework - Andrea - Validation studies: o The new Geant4 release 9.4.beta has been validated on the GRID; as result of this, it was found that the EM description of gamma-nuclear reactions is now different in CHIPS physics list; a first comparison has been made with FLUKA for gamma-nuclear reactions on thick target data, and the results are under evaluation. Also, the response of the Bertini (BERT) based physics lists has increased for Pb-based calorimeters; this is under study since no clear material dependence has been observed. - Meetings: o An LCG Physics Validation Meeting was held on July 7th with presentations from ALICE, LHCb, BESIII and CMS experiments. E.Sicking presented the status of the simulation of the ALICE detector using the development version of Geant4, 9.3.ref05. A comparison with the Geant3-based simulation has also been made. Main focus on the simulation of the tracking detectors, in particular Geant4 shows a higher number of clusters and more energy deposition in the TPC with respect to data and Geant3; a source of this discrepancy, to be investigated, could be the need for a better tuning of the gas gain in the Geant4-based simulation. S.Miglioranzi showed cross-sections and multiplicities studied in the context of the LHCb detector. A slab of different thicknesses of Al has been simulated with beams of pions, protons and anti-protons. The number of interacting particles and the produced secondaries were recorded. This study shows the differences between the models used in the physics lists: anti-protons have always the same cross-sections and secondaries for all physics-lists, while QGSP_BERT and FTFP_BERT agree below 4 GeV and QGSP_BERT and LHEP agree in the intermediate energy range. To be noticed that K+/K- have different cross-sections. It was also carried out a study on the impact on physics and CPU for different EM options; the highly precision 'opt3' EM option, while requiring substantial more CPU time does not bring noticeable improvement in physics simulation. The next step is to test the QGSP_BERT_CHIPS physics-list. G. Cao and H. Liu from the BESIII collaboration have shown a comparison between data and simulation for anti-protons. Given the clean environment of the collider it is possible to select a pure sample of (anti-)protons. In the past Geant4 showed difficulties in predicting the calorimetric observables studied at BESIII (total energy in cluster, shower shape) for this kind of particles. The agreement for other particles (pions, protons) has indeed been satisfactory with the use of the QGSP_BERT_HP physics list. A custom physics list has been developed by BESIII, based on QGSP_BERT and using CHIPS for the simulation of anti-protons; the results, with models from Geant4 release 9.3.p01 are encouraging and show good agreement with data. The next steps are the studies of kaons and (anti-)neutron samples. Finally S.Piperov showed comparisons of CMS test-beam data using the Geant4 development release 9.3.ref05. In particular a new comparison with kaons and anti-protons beams was done. The quality of the simulation with 9.3.ref05 using the reference CMS physics-lists (QGSP_BERT_EML, FTFP_BERT_EML and QGSP_FTFP_BERT_EML) did not change compared to the stable release, and confirmed to be good. The physics lists with CHIPS models (CHIPS, QGSP_BERT_CHIPS and QGSP_BERT_CHIPS_EML) have been studied in addition. CHIPS shows good agreement with kaon data, while response of anti-protons is overestimated; the response of kaons is under-estimated (and anti-protons over estimated) in QGSP_BERT_CHIPS and QGSP_BERT_CHIPS_EML. These two lists perform even better with respect to QGSP_BERT_EML. o Next meeting will be in September. --------------------------------------------------------------------- ==== Generator services - Witek - Meetings: o The next LCG Generator meeting is scheduled for September. - MC Generators tuning: o The web page with collection of tunes is being prepared, but no much progress is expected in the next month due to some problems on the synchronization of shifts by the Russian collaborators. - GENSER: o New generators have been installed: Pythia 6.423, LHAPDF 5.8.3, Pyquen 1.5.1, MCatNLO 1.5.1, Tauola++ 1.1. o Work has been done in order to port the latest version of Tauola++ to use the new build system; the porting has been successful and is now working; a new release of Tauola++, 1.1 includes this. o A new collaborator, V.Duk, is currently working on implementing tests for the Hijing MC generator, the only one for which tests were missing. - HepMC: o Progress has been made for including the newest version of HepMC in the nightly builds; it will be part of a new LCG release build which will then be available to experiments to try out. --------------------------------------------------------------------- ==== Fluka - Alfredo - No report. --------------------------------------------------------------------- ==== Geant4 - John - Releases: o The new Geant4 release 9.4-beta was announced on June 25th: the new release includes a new physics-list QGSP_BERT_CHIPS which introduces improvements in modeling of anti-nucleons, hyperons and kaons cross sections; similar improvements in revised QGSP_FTFP_BERT and FTFP_BERT; several obsoleted physics-lists have been removed. 9.4-beta includes a major revision of the Bertini model, where code has been re-engineered to retrofit Geant4 classes for particles and other services and improve pion cross-sections. Careful testing of the software changes was made to ensure that results are not changed by software revisions. 9.4-beta also includes changes in CHIPS for elastic and inelastic cross-sections and physics model refinements; some additional fixes have been provided by the author also after 9.4-beta and are under testing. Included in 9.4-beta also a completely revised hadronic de-excitation model, and a new geometrical shape, G4GenericTrap, implementing the 'Arb8' shape requested by ALICE. - Feedback from experiments: o ALICE: request for physics for light anti-ions (d-bar, t-bar, He-bar). o ATLAS: problems reported in energy conservation in release 9.2.p02, rare cases of 47 TeV protons produced in Bertini (problem also present in release 9.3) and 95 GeV nucleons from 2 GeV gammas in CHIPS. Both issues have been characterized and the fixes provided. Also reported an open issue regarding calibration of hits energy in CTB simulation, which appears greater than the beam energy when using 9.3.p01. The problem cannot be reproduced in bare Geant4, where energy deposited is constant between Geant4 versions. A custom algorithm is used for computing total energy, indicating that the problem may not be in the physics models but in the use of the Geant4 interfaces involved in calibration hits implementation. - Developments: o Testing for energy conservation in theoretical hadronic models is being prepared and planned to be systematically exercised in nightly builds. o The prototype of the new CMake build system has been evaluated, and refinements have been planned after a series of meetings with B.Morgan visiting CERN on July 6-9th. o Progress made on the evaluation of SVN as a possible replacement for CVS for the code repository. - Architecture review: o A Geant4 Architecture Review Mini-Workshop has been held at LAPP, Annecy, on June 29-30th: several topics for potential revision have been identified; a first wider discussion on the impact of potential changes has taken place and further topics have been identified for improvement. Issues from multi-core and multi-threading were also part of the assessment. - Meetings: o The Geant4 Collaboration Workshop in ESTEC will take place on October 4-8th October and a detailed planning is under way. o Geant4 Spokesperson Election status: the election process is ongoing and is scheduled to finish on Friday July 16th; Makoto Asai (SLAC) and Takashi Sasaki (KEK) are the nominated candidates. o G.Cooperman will be visiting CERN on July 13-16th. o The next computing performance task-force meeting, will be held on July 14th. --------------------------------------------------------------------- ==== Garfield - Rob - No report. --------------------------------------------------------------------- ==== AOB Next meeting scheduled for September 21st.