Simulation Project Leaders Meeting 1-June-2010 ==== Attendees John Apostolakis (JA), Alberto Ribon (AR), Witold Pokorski (WP), Pere Mato (PM), Gabriele Cosmo (GC). Excused: Andrea Dotti, Alfredo Ferrari, Rob Veenhof. --------------------------------------------------------------------- ==== Minutes - Andrea Dotti replaces Alberto Ribon in the leadership of the Physics Validation project, as of June 2010. All thanked Alberto for the excellent progress and collaboration in the Simulation Project, which benefitted greatly from his contributions during the past five years. All hoped to have occasions to communicate and collaborate again. - V.Uzhinskiy will join soon the Geant4 SFT group as project associate. He is expected to start by mid-June. --------------------------------------------------------------------- ==== Simulation physics validation & framework - Alberto, Andrea - Validation studies: o A new LCG note (CERN-LCGAPP-2010-01) on the status of hadronic physics in Geant4 at the start of the LHC physics program will be soon released. - Meetings: o At the Physics Validation meeting of May 19th, M.Simonyan from ATLAS reported on the validation study on response of single isolated hadrons in ATLAS with collision data at 900 GeV energy; the simulation was carried out using Pythia and Geant4 with QGSP_BERT physics list. Reported good agreement within 5%, which shows a good description of dead material and noise; only in the transition region between the barrel and the end-cap the agreement between data and simulation is less good; here, the presence of cables and dead material not well described in the simulation can be likely the cause of the observed discrepancy. Another presentation was given by S.Banerjee on the comparisons in CMS of the simulation response for single isolated hadrons with data at 7 TeV; good agreement of the simulation (Pythia + Geant4/QGSP_BERT_EML) within 3% in the barrel region; some larger discrepancy in the end-cap region, where calibration studies are undergoing. Sunanda also presented comparisons with the HCal test-beam data, using the latest Geant4 release 9.3.p01, and comparing two physics lists: FTFP_BERT and QGSP_FTFP_BERT (which use Fritiof model instead of LEP); these lists show results as good as QGSP_BERT or better and a smoother response as a function of the beam energy. CMS is likely to move their simulation to Geant4 9.3.p01, considering also to adopt in production the new list QGSP_FTFP_BERT_EML, which gives slight faster performance compared to FTFP_BERT_EML. A third presentation was given by M.Kossov on the treatment of low-energy neutrons as an alternative to LEP and the HP models already available in Geant4; the new approach uses parameterisation of fission and fusion (capture); the development is not yet included in the main development trunk. o Next meeting will be on July 7th, when presentations from LHCb and ATLAS HEC are expected. --------------------------------------------------------------------- ==== Generator services - Witek - Meetings: o Planning meeting held on May 28th, where a status report of the project was given and the new activities for the next months discussed. - MC Generators tuning: o New activity involving Generator Services on tuning of MC generators in the context of LPCC. Some roles have been assigned to the project, among these, the creation and maintenance of a new repository for the tuning tools, as it is done for the MC generator packages, and the creation of web tools to present the validation of the different tunes (a 'static' version providing with plots against data and a 'dynamic' version, based on a data-base where queries can be made for the on-the-fly generation of plots based on tuning parameters of the MC). The manpower for this new activity will mainly come from the GENSER team, with additional resources from LPCC. - GENSER: o New generators installed: NLOJet++, MCFM. New versions: Pythia 8.135, Sherpa 1.2.1, LHAPDF 5.8.2, Pyquen 1.5.1, Herwig++ 2.4.2. o Received requests from ATLAS for the installation of new generators and different versions of old generators. - Validation: o Valuable contribution from DESY for the HepMC analysis tool, which now want to automatize for use also in nightly builds. o Implemented a number of validation tests using Rivet, allowing the comparison of results between different versions of the generators. - HepMC: o New 2.06 release announced in May; encouraged experiments (still using 2.03) to try it out! Received a request to provide a consistent build on top of the latest version of HepMC, Pythia8, and the other packages in the nighties, so that the process for testing the current development can be facilitated. o Work is going on for the implementation of a Root-I/O library for HepMC. --------------------------------------------------------------------- ==== Fluka - Alfredo - No report. --------------------------------------------------------------------- ==== Geant4 - John - Releases: o Preparing for 9.4-beta, scheduled for end of June; the deadline for the development is Friday, June 4th. Improvements in different areas are expected, together with also removal of some old obsoleted classes (including the old G4MultipleScattering class) and obsoleted physics lists. Prepared for the regression testing of the candidate releases on the WLCG and OSG Grids. o The development tag of May was made available Friday 28 May; among the various fixes and developments, it includes a major review of the de- excitation/evaporation model, which was carried out in the past months by V.Ivantchenko and Jose-Manuel Quesada Molina (Sevilla Univ.). - Feedback from experiments: o Received request from ALICE for improved cross section and interactions of anti-proton and kaons. A new physics-list has been put in place, in which CHIPS cross-sections and modeling are used for anti-baryons and hyperons (replacing the ones from the parameterised LEP/HEP models derived from Geant3 GHEISHA) and CHIPS cross-sections are used for kaons. The new physics-list is included in the reference tag of May. It has been sent to both ALICE and LHCb, which requested improvement in the treatment of kaons at the last Technical Forum (March 30th). o Interest also expresses by ALICE for the treatment of energy loss in gases for TPC; interaction on this is ongoing. - Developments: o CPU improvements: good feedback and interaction with CMS and ATLAS. Two sets of improvements were created by Geant4: the first in the geometry by SFT team; the second in Bertini by SLAC colleagues. The improved allocation of navigation-history objects in geometry was measured. It was confirmed to drastically reduce memory churn. The revised implementation of the Bertini model contributes to decrease significantly allocations on the heap and memory churn in general. These developments introduced since the Geant4 development release of April, combined to bring almost 5% overall CPU time improvement in CMS tests with QGSP_BERT. o Interaction with D.Levinthal from Intel, looking at the results with PTU profiling; first results do not identify a clear area for improvement in the code. Tests on more realistic experiments setups are planned. Currently using the Intel compiler for inter-procedural optimization studies. o Working on preparing a proposal for common activities within the SFT group, spanning prototype of Root vectors, configuration management with Cmake, and potential consolidation in the geometry. o A prototype Savannah portal is under preparation for forwarding bug-reports to the Geant4/SFT team from experiments at CERN. o Indico site established for the SFT/simulation weekly meetings (Internal). - Architecture review: o In the process of finalizing the document of the first assessment; it will be circulated to Geant4 Collaboration members this week. - Meetings: o Mini-Workshop at LAPP on 29-30 June on Geant4 architecture. o 15th Geant4 Collaboration workshop at ESTEC on October 4-8th. o CALOR conference held in Bejing: results on comparison with data presented by ATLAS and CMS; identified data on anti-proton from CMS which could be used; reports from CALICE/DREAM projects. - Collaborators visits: o O.Kadri visit during the past fours weeks included work on an improved Multiple Scattering model (WentzelVI) for muons and hadrons. This separates hard scattering (large angle) and soft scattering. Also it includes step limitation in order to obtain improved estimates of the correlation of deflection and lateral discplacement (this will address LHCb request for full correlation of angle deflection and exit point even when 'infinite' cuts are used). o A.Bagulya developments included revised ionization for negatively charged particles. This fixes issue reported by ATLAS with high negative- charge exotic particles. --------------------------------------------------------------------- ==== Garfield - Rob (offline report) - Meetings: o RD51 meeting in Freiburg im Breisgau. - Garfield: o Most development is going on in the photon transfer term for which microscopic and macroscopic models give similar values, lending quite some credence to the approach. The photon feedback in avalanches is not yet understood at this level; interesting presentation by Paulo Fonte on the relevance of electron diffusion achieving the same effect. --------------------------------------------------------------------- ==== AOB Next meeting scheduled for July 6th.