April 27 2004 Simulation project leaders meeting minutes ======================================================== ==== Attendees Fabiola, Paolo, Witek, Torre, Alfredo, John, Andrea ==== Minutes The next meeting will be Tuesday June 8 10:00. Putting in place a simulation project leader soon is being addressed, principally by John Harvey in the context of PH/SFT group planning. PEB and SC2 presentations of the simulation project went well. The PEB raised no objections to the plans presented. == Generator services - Paolo GENSER 0.1.0 was just released. In addition, a special version for ATLAS, 0.1.1, containing a modification needed by them for DC2 production, was made available. Paolo has received information from Slava on plans for CERN visits by the Russian team. There will be no librarian from mid-April to mid-May, then again no librarian during July and August. The person assigned as librarian for May/June is a new person. It was agreed that this plan does not at all meet the needs of the project for librarian services, particularly with an experiment now using GENSER in data challenge production. We must find a solution quickly. Torre will take this up with John Harvey and Les. The Russian group plans to increase their effort on MCDB. An updated version of the MCDB plan with short- and long-term plans will be ready soon. MCDB developers currently use their own dedicated MySQL server but plans are under way with Alberto to move to an SPI server (that can accommodate the instabilities of MCDB development). Paolo has established comprehensive milestones for the project -- see the plan on the web. The generator services project was reviewed in March. Manpower concerns were identified as a major concern (including the librarian coverage issue). There is possible interest from generator providers in project participation in ThePeg (Lief Lonnblad) support and development. We expect to receive the review report at the end of April. == Simulation physics validation - Fabiola The project is very short of manpower. More people are needed and Fabiola wants to retain the people she has. The note on Geant4 hadronic physics validation is being completed. There should be a complete draft ready for circulation by the end of the week. Fabiola expects to have the note out in mid-May. Witek is almost finished the pion absorption simple benchmark but requires some help from Alfredo with FLUKA questions. He will then write a summary note. In the second half of May there will be a meeting to address the next simple benchmark priorities. Both FLUKA and Geant4 will be asked to come with the list of benchmarks they see as important and relevant for the LHC. Alberto is finishing the G4-FLUKA comparison in ATLAS pixels. Final report in physics validation meeting next week. Then will document work in note as first example of resurrecting FLUGG for an application related to LHC testbeam. The plan is that Alberto will then take up a new physics validation study using either ALICE or LHCb testbeam data, and spend part of his time developing the Geant4 acceptance suite. In em physics studies, Geant4 validation with LHCb ecal is getting underway. Also Fabiola is arranging with ATLAS a presentation on an ECAL study addressing shower shapes and response linearity at the per mil level. John notes it would be good to discuss the concerns with 6.0 performance in this area; Fabiola says they have not yet moved to 6.x and they will contact Michel to be sure they use the right version. == Generic framework - Andrea, Witek Andrea verifies absolute lack of interest in the experiments for a common generic simulation framework - everyone seems happy with what they have. There is interest from ATLAS, LHCb and possibly G4 in using SEAL components -- python tools and the plugin manager -- to support the dynamic creation of a custom simulation configuration in the python environment. CMS has expressed interest in python scripting support but not beyond that. Andrea is also interested in making use of GeoModel as a transient generic geometry model and gateway not only to G4 but to VMC/ROOT -- a transient alternative to GDML. Either 'ATLAS' work or 'LCG' work depending on how wide the interest is. If LCG work is proposed in these areas a concrete proposal should be written to take to the AF and the PEB. Following up on Alberto's pixel testbeam work with FLUGG etc to produce a generic infrastructure usable for other testbeam studies still has to be done. Witek will look at Alberto's work and determine what generalization is needed. We also have to proceed with GDML plans. The strategy was OK with the PEB, we now need a specific workplan. Witek and Andrea will talk to Radovan and other interested parties and write down a plan consistent with Radovan's schedule (assuming he does the work, and he can do it most efficiently). It was agreed to have a separate meeting to discuss GDML plans in more detail. Witek and Andrea are asked to organize the meeting. == FLUKA - Alfredo Recent focus has been on heavy ions, addressing open issues with the LHC machine as well as other areas (eg space science). Very busy period for LHC machine calculations as machine components approach the build stage. There is growing need and pressure to build as much as possible of the machine into the simulation geometry. Not enough manpower to do the whole thing; doing the most critical parts. Four full-time fellows are now working on LHC calculations. DPMJET version 3 has now been integrated and is under test; support for version 2.5 will be maintained as well. Other recent achievement is a first working model for em dissociation of heavy ions, important for addressing LHC machine issues with ions. Cross section is large for 1-2 neutrons being stripped and the machine impact of this needs study. Some improvements in geometry input, supporting booleans in complicated expressions. Bug fix release on web contains this. New release is still planned for end of year. Schedule looks OK at this point if no big problems or new workloads arise, but LHC machine work will have the priority. The next FLUKA course is scheduled for 2nd week Jan 2005 in Houston. == Geant4 - John In the first quarter the team contributed to a patch to Geant4 6.0, two monthly development tags and a minor release, Geant4 6.1. This release was focussed on aiding experiment simulation productions. It added features that help to identify problems in key areas quickly. Also replaced old mu absorption process with new version in 6.1, as agreed with FLUKA. Issues in muon multiple scattering were identified in the first quarter. They were fixed as they were found, and all fixes were included in 6.1. Initial release notes for 6.1 were not clear enough on what had been fixed, so a follow-up announcement with more detail was made. This turned out to be an over-correction! It unduly alarmed people. Finding the right balance in the information provided is tricky. A more regular flow of notifications of fixes is planned, so that people get used to the pattern. Vladimir Ivantchenko has come to the end of his Project Associateship term. He will still be available for consultation, but his main focus will shift to other Geant4 applications. The next Geant4 technical forum will be at Vanderbilt U in Nashville on May 12. It will be a space users forum. A problem has been reported by Atlas in test productions with the latest Geant4 release, 6.1 of end-March. Crashes occur in mysterious circumstances, with 20-30% of jobs crashing in runs with the same initial conditions. It is under intense joint investigation. Its intermittency makes it hard to find. The crash seems to disappear if preceding events are skipped. Andrea confirmed that they are investigating why the particular condition happens. Witek reports that LHCb is happy with G4 6.1. Failure rate is the lowest ever seen, about one job per mil. Only remaining crashes are in hadronic physics (the ATLAS crash seems to come in em physics).