Simulation Project Leaders Meeting 12-May-2006 ==== Attendees John Apostolakis (JA), Paolo Bartalini (PB), Witek Pokorski (WP), Alberto Ribon (AR), Pere Mato (PM), Paola Sala (PS), Gabriele Cosmo (GC). Excused: Rob Veenhof (RV) --------------------------------------------------------------------- ==== Minutes --------------------------------------------------------------------- ==== Simulation physics validation - Alberto - Meetings: Two meetings have taken place since the last SL meeting. o In April, there were presentations given by V.Grichine on an update concerning the developments going on in Geant4 on synchrotron radiation; by M.kossov, about an investigation done on the current default implementation of the hadronic elastic scattering process (G4LElastic) in Geant4, where problems have been identified and require improvements; an update on the CMS HCAL 2004 test-beam analysis (using Geant4 6.2.p02), where a precise study of the systematic effects (beam composition, beam profile, noise, etc.) that can affect the analysis was done, confirming the problem on the shower shapes for high energy over 150 GeV. o In May, there was a presentation on the ATLAS HEC analysis only for electrons, using 8.0.p01 and comparing with the previous Geant4 releases: the new release appears more stable, where less dependency on the production cut is confirmed, few percent more visible deposited energy is observed as expected; the energy resolution gets worse although closer to the data (a difference has been noticed between copper and lead: in the latter case, the energy resolution gets better in G4 8.0.p01; however, in both cases, the energy resolution gets closer to the experimental value); increase of CPU time (factor two) if using the same cuts, where however higher cut (up to 1mm) can be used without affecting physics performance. A.Howard gave a presentation on his ongoing analysis of TARC, a benchmark for neutron production and transportation; first preliminary results of the analysis were discussed, different physics-lists have been compared especially concering production of neutrons. M.Kossov then gave a presentation on hadronic neutron elastic-scattering, showing the current limitations in G4LElastic hadronic process; he also presented updates made in the CHIPS model for issues raised in validation. G.Folger presented the status of investigations going on for understanding the hadronic shower shapes problem (QGSP model has showers that start earlier and are shorter, whereas LHEP model is more close to the data); cross-sections, multiplicities, hadron elastic-scattering and neutron production are under investigation; currently checking variations/effects using simplified calorimeters. Effects of hadronic showers are important also for ILC and a workshop dedicated to this topic will be held in Fermilab in September. o In the meeting of June 21st a presentation is expected for the CMS HCAL analysis but using a more recent version of Geant4 (7.1.p01), where some improvements in the shower shapes are expected (7.1.p01 has cross-sections updated). - ATLAS Tilecal analysis: o There has been progress in understanding some open issues; a meeting with Fluka experts was very useful to understand Birks quenching; it was useful the suggestion to study a time cut for neutrons in order to restrict the time window according to electronic readout; this was done using a prerelease version of Fluka; the time cut in Fluka reduced the production of neutrons and the results are now more similar to what obtained with Geant4. Now currently adding time information on the hits and performing detailed studies. Still remains to understand the difference in visible energy between the two engines; when using muons as primaries, about 10% less visible energy is observed in Geant4 (version 7.1.p01) than Fluka. A first check on the material density (shoot geantinos for Geant4 and ray-particles for Fluka to check radiation-length) didn't bring to answers, the material description in the setup seems correct; as well as, when trying to use an homogeneous block of iron and shoot muons at 10 GeV, the results are fine in both MCs. The attempts to follow, as suggested also by PS is to try also with an homogeneous scintillator; to verify also the response when shooting electrons and pions; to try with a simplified sampling calorimeter and finally also verify with the latest Geant4 version 8.0.p01. The next week, the first results using version 8.0.p01 should be ready. --------------------------------------------------------------------- ==== Generator services - Paolo - Documentation: o The Les-Houches proceeding have been published; a section on LCG Generator and one on MCDB are also included. - Meetings: o There was a meeting in May, chaired by AR, where MC user support was discussed. There were reports from Pythia, Herwig++, ThePeg and Sherpa. MC projects wish to use their own bug-reporting tools and eventually have a filter for the bugs submitted by the LHC community through LCG. It was decided to provide through the LCG Generators web pages an interface to the the different MC web pages, on the style of what already done in the CEDAR project by adding also explicit links to the bug-reporting systems of the most popular MCs. It was also decided to provide a filtering mechanism involving the GENSER integrator for the actual selection and forwarding of the bug-reports. The next integrator will be trained in this new task. People in GENSER should be also involved as responsibles in the various bug-tracking systems (for issues related for example to installation problems, etc...). - Manpower: o New arrival: Oleg Zanin; new integrator from June 1st will be M.Kirsanov (i.e. the GENSER master integrator). o Resources from LCG Russia correspond to 1 FTE for MCDB and 1.1 FTE for GENSER. This is considered to be well balanced for the current needs (mostly dealing with user support). - GENSER: o The current integrator is working on user support for GENSER 1.3.0 and for the preparation of the next GENSER release 1.4.0. o The finalisation of the integration Herwig++/ThePeg will be performed by the next integrator and by one Herwig++/ThePeg developer visiting CERN in June-July. - MCDB: o Now direct access of files in CASTOR is possible. Currently concentrating on the parser and developing a query suite for accessing multiple files, and allow access other than web interface. o A.Gusev and S.Belov are the current developers. A.Gusev will leave on May 23rd. - Review of milestones: o There's the need to review the current list of open milestones and put them in sync with the recent experiment requirements. This review will take place at the end of June. o Generator Framework: it is expected a report on the recent activity for work related to the LCG Generator meeting of June or July. --------------------------------------------------------------------- ==== Fluka - Paola - The 3rd Fluka beginner course has taken place recently in Pavia. 34 students participated and the course was well received. - Currently preparing the new release (Fluka.2006) expected in June. Features included: o Improvements in the user-interface: it will be possible to input data by names (not only by numbers) for materials, regions, estimators, etc. o Various bug fixes collected since the 2005 version. o Physics: improvement in evaporation and fission, where a new theory driven model has been implemented. There will be a new mass table calculating quantities till high masses. New evaporation level densities. o Online scoring of doses, with maps of activity distribution applied everywhere, not only by regions as in the past. o Propagation of quenching also in the routines for hits/digits. - Under development: o Open source code distribution (no longer restricted only to CERN and INFN). o Nucleus-nucleus interactions: implementation of low-energy model (<100MeV) for light ions o Online quantum-molecular dynamics for nuclei up to iron. --------------------------------------------------------------------- ==== Simulation framework - Witek - GDML: o Meeting in Brussels with ESA and other collaborators in the space and astro-physics community, to discuss extensions to GDML to support tessellated solids implemented in Geant4 for CAD interface. It has been a rather productive meeting; the GDML schema was defined at the meeting itself. The work is now almost completed in GDML and will be available soon in the new release. The G4Tet solid is also being added to the GDML schema. A CAD->GDML converter is being implemented by ESA contractors and will be freely available to the Geant4 community. o Also investigating on the implementation (request from CMS) of a small subsystem to be used for reading geometry setups described as vertices to tessellated solids to be used as light interface to CAD systems. o Further improvement have been made to the GDML build system, GDML can now be installed also with static libraries. o One summer student and one technical student will join the team starting from this summer. - Persistency: o Hits persistency using Root was discussed with users in the Geant4 Hypernews forum, it is proposed to replace the existing example in Geant4 using the old persistency mechanism in Root with new one using the reflection system. This may require the implementation of the native container for hits in Geant4 to be reviewed. - MCTruth: o A tool to store Geant4 output through HepMC objects is implemented. LHCb is already starting using it. No input from CMS yet. To be decided if to distribute it as an additional extended example in Geant4 or not. --------------------------------------------------------------------- ==== Geant4 - John - Investigation on hadronic shower shapes: o Possible sources responsible for the differences between LHEP and QGSP models under active investigation: cross-sections, multiplicity and neutron production. o Currently looking also at fixed target data E > 100GeV in forward regions. Further data required (thanks to PS for reference to SPY experiment data). o Physics lists including Cascade models appear to perform better at a price of more CPU time (CPU time increase ranges in average of order 40% to 100%). Using QGSP_BERT (with Bertini), the response in two typical setups is close to the response of LHEP. o It was confirmed that using physics lists with the precise neutron transport option ("_HP"), to track neutrons precisely down to thermal energies, results in an additional factor (4-5 times) in simulation time in typical setups. This is relevant and of interest only for selected applications. o To verify also the containment of a shower in modules of a combined setup. - Hadron elastic scattering modes/processes: o The improvement of the elastic scattering was found to be significant in scintillators. Key improvements are being prepared for the June release. The full revised physics process is expected to be available in the 3rd quarter. o An alternative improved physics process for neutron-proton is also being prepared. - Performance benchmarks have been run by S.Jarp's team in IT on ATLAS and LHCb Geant4-based simulation setups. First results showed that approximately 40% of the time is spent in geometry, 25% in physics, with 15% from stdC++/cstd libraries. - Meetings: o A mini-workshop took place at CERN in April with a number of Geant4 developers to review the design and prototype implementation of multiple navigation. This new development will enable coupled transport in parallel geometries in the EM field. Geometries for physics interactions, for fast-simulation, for scoring and biasing processes will be handled in parallel. A first beta implementation is scheduled to be available with the June public release. --------------------------------------------------------------------- ==== Garfield - Rob - No report. --------------------------------------------------------------------- ==== AOB Next meeting scheduled for June 27th.