Simulation Project Leaders Meeting 14-September-2004 ==== Attendees Paolo Bartalini (PB), John Harvey (JH), John Apostolakis (JA), Witek Pokorski (WP), Paola Sala (PS), Gabriele Cosmo (GC), Torre Wenaus (TW). Excused : Fabiola Gianotti --------------------------------------------------------------------- ==== Minutes Alberto Ribon has been appointed as the new coordinator for the Physics Validation sub-project. He will replace Fabiola Gianotti in this role and will join this meeting starting from the next one in October. --------------------------------------------------------------------- ==== Generator services - Paolo The Generator Library (WP1): - The new Beta version of GENSER (0.2.0) has been released yesterday (September 13th). The new release of GENSER includes new versions of Herwig and Pythia as long with new introduced packages, i.e. Jimmy and evtgenlhc. Evtgenlhc has been provided by LHCb and will be developed in GENSER. Patrick Robbe (LHCb) has been given access to the evtgenlhc package in the GENSER repository, other contributors will be guaranteed a limited access (development of new decay models, decay files etc.) - From the next version, GENSER will no longer be considered in "beta" stage and the first digit will be increased. The project is in schedule for the new version, and can be already considered ready for production. - ATLAS makes already use of GENSER in the official production (data challenge); LHCb is currently validating GENSER in order to adopt it in production. Interest has been expressed by CMS to migrate to the new framework for production in the mid-term time scale the migration should be scheduled for the data-challenge next year. ( Librarian: ) - Although during the summer-time there was not an official librarian provided by LCG-Russia, it was possible to get help from remote (M.Kirsanov, A.Sherstnev) for the preparation of a special version dedicated to ATLAS production and to finalize the introduction of Jimmy and evtgenlhcb (work started in spring by Mikhail Kirsanov). - New resources will be soon available: there will be a visit of S.Makarychev on September 20th, he will share the effort with the author of SHERPA for its integration in GENSER; SHERPA is now in a more advanced stage of development (post-alpha). JH: It's important to respond to the internal reviewers. PB: Feedbacks from the authors improved after the review (see for example the starting and promising collaboration with PHENOGRID). Andreas Pfeiffer had already interactions with A.Sherstnev for controlling the repository, his duties should mostly concentrate on the project and user support side (preparing bug fix releases for already introduced packages etc.); Andreas should not be the one taking care of introducing and verifying the new packages in the GENSER repository, which should be under the responsibility of Paolo (with possible help from A.Ribon), the authors and the russian collaborators. JH: Andreas should be heavily involved, we need to make sure between now and November that GENSER is managed as for the other LCG packages. - Concerning bug-fix releases, it was decided not to adopt a "double-versioning" strategy, but provide new versions of GENSER with limited number of packages. Storage, Event Interfaces and Particle Services (WP2) - UK collaboration for PhenoGRID is progressing. They had first postdoc working on the common interface (ThePEG) for Herwig; he should start to work next month. - ThePEG status: no much news, documentation is under improvement (A.Ribon is taking care of the documentation); the main developer for the project is Leif Lonnblad. Pythia7 is not much interested in using it; the author (Torbjon Sjostrand) says it's too complicated and has alternative solutions; Torbjon is currently visitor here in SFT; it was expressed the wish he could devote some time to support the user community and dedicate to the development of Pythia7. It is possible that for the time being Pythia7 will be developed without the framework but later on, usage of ThePEG might be recovered by some other co-author of Pythia7. Common Event Files, Event Database (WP3) - The russian visitor S.Belov concentrated on MCDB project during the summer period. Next visitor (S.Makarychev) will arrive on September 20th and will also mostly concentrate on MCDB. A meeting of the generators services has been held in August (actually, on September 1st) where a demonstration of MCDB was given. A new meeting is scheduled for September 23rd, where A.Sherstnev will give an overview of the project including the plans for future developments (integration with GENSER, integration with a simple generator level production framework, development of APIs for the integration in the LHC experiment simulation software). Tuning and validation of event generators (WP4) - No news, apart from the basic sanity checks added to some GENSER subpackages (M.Kirsanov). --------------------------------------------------------------------- ==== Fluka - Paola There's need to clarify the position of FLUKA in the generic simulation framework project. We should know if there is interest for this from the experiments. If so, the FLUKA team is very happy to collaborate and give consultancy, however a real involvement in the work is not possible, due to the lack of man-power. Need to identify the man-power to invest in the project. The FLUKA team will continue to report in this meeting about the FLUKA development progress (although the work is not done with LCG man-power), and the activity of the group should be mentioned in the quarterly report. "Free statements" about the activity or decisions in the group should not be made or taken without also a consultation with the FLUKA team. FLUKA: - Working on the public release and 'PEMF removal'. The Beta release is expected before the end of the year and will be available for internal use at CERN only, not world-wide yet. The manual is ready for publication and would like to have it published as a yellow report; there's some resistance for doing this, mainly for a question of budget (the manual is about 400 pages); the manual will also be available from the FLUKA web site. FLUGG: - Paola is working on it for the porting to Geant4-6.2. Since 6.2 is compatible with 6.X series, it should not require major interventions. Paola thanks Andreas Morch from Alice for the help. Paola is willing to act as the contact-person for FLUGG. WP: it should be first clarified what we mean by common framework. There's no need of writing something completely new from scratch but just clarify the issues that need to be resolved in order to make as much easy as possible for users in the experiments the exchange of a simulation setup between FLUKA and Geant4. From experience in the work done by Alberto Ribon and Witek, the main difficulties were coming from the lack of documentation in some cases. Proper documentation on how to use the software should be provided (have some guidelines to use MGDRAW, for example...). Writing a concrete interface for handling digitazion/hits is something specific of the kind of simulation (detector-response/ physics) done, which cannot be generalised. - It was agreed that we need to: > figure out what is missing or needs to be improved, > give a guideline for what should be written by a user in terms of code for hits and digits (provide examples) and check that the information produced is compatible between FLUKA and Geant4. Paola and Witek will meet to discuss and try to converge on these issues. --------------------------------------------------------------------- ==== Geant4 - John Quarter milestones: - Contributions to the September development release: the ability to customise the registration/de-registration of objects in the geometry stores has been coded and is now under testing. As well as models for transition-radiation and PAI have been updated. - Revised performance testing suite is in place. - Work for the regression/acceptance suite is progressing well. 1/3 of the tests for EM perfomance and MSC/shower-profile has been already released in 6.2. The tests are being progressively included in standard system testing. More developments & feedback from experiments: - The support for CLHEP-1.9.1.X is converging and most likely the September development release will support CLHEP-1.9.1.X (still coworking with CLHEP-1.8.1.0). - A patch (patch-01) to release 6.2 was released on July 27th. Among all, it includes a fix to a problem in Multiple-Scattering reported by ATLAS, causing visible deflection of direction for particles at high energies in thin volumes. CMS originally reported a loss in performance with 6.2, which apparently was due to this bug! Now fixed. - The G4Allocator class has been extended to allow for releasing on- demand of the memory allocated. The new feature has been tested by CMS and significant improvement in memory consumption was observed (jobs now not exceeding 100MB average load); CMS also reported good news concerning stability in production test with 6.2p01: almost no failures for 1 million events produced (single particle events, of which 200K minimum-bias). - ATLAS and LHCb report good reliability in production using 6.0 (+ some fixes included in 6.1) and 6.1 respectively. - Andreas Morsch reported that ALICE will move to use TGEO as their production geometry for G4. Currently testing it with FLUKA. - Concerning the storage of Geant4 geometry in POOL/ROOT, it is necessary to understand the impact of the changes (default constructor for volumes/solids) which have been asked. JA takes care to review this. Meetings: - The next Technical User Forum has been announced; it will be held in Catania during the Geant4 Collaboration Workshop (4-9 October). Platforms: - During the Geant4 Workshop there will be also a discussion on future supported platforms; the idea is to move to SLC3 as default Linux starting from the December release. - Linux/gcc-2.95.2 and Windows/VC++6 will be no longer supported. - MacOS is one of the "verified" platforms; to have it regularly tested by system testing there's the need for a stable AFS platform. No official request for support from the experiments yet. PS: also FLUKA has been requested to support MacOS with g77. SPI collaboration: - No new updates at the moment. The goal is to have the Savannah portal for bug reports put into production. GC: presenting and showing the Savannah prototype during the Geant4 workshop is a good opportunity to hear first reactions from the Geant4 collaboration. If no agreement will be found for the migration to Savannah, new resources will have to be provided in future by the collaboration for the maintenance of the existing system based on Bugzilla. - Bonsai: need to talk to Alberto Aimar to see if there has been interest expressed. Interest was expressed by Giuseppe Battistoni too, he will be put in contact with Alberto. --------------------------------------------------------------------- ==== Simulation framework - Witek GDML: - Work is progressing, although Radovan has been very busy in the past months and still has no free schedule until after CHEP. Witek is the only one working on it at the moment. - Support for parameterised volumes has been implemented; can export/import a (simple) geometry with/without parameterisation. The goal is to have it complete and fully working in a short time-scale. - Still missing support for some solids (polycone, polyhedra, ...), or exotic parameterisations (by material/solid). Preliminary tests made on more complex geometry descriptions lead to a crash due to the missing features. The objective is to progressively test it on more complex geometries (like some of the advanced examples in Geant4 or real experimental setups) and make also measurements for performance, memory consumption, file size... - Next will come the support for ROOT/VMC as well; not yet started. FLUGG: - Witek contacted A.Gheata for the VMC and FLUGG; the strategy used allows for automatic generation of the PEMF files, however, it still foresees the process of generating the files, so it is not really solving the problem. The question is how much effort should be put on this or if it is worth to wait instead for the final solution which will be anyhow provided in FLUKA. - It would be good to be contacted by people from experiments who are really interested in using FLUGG, so also feedback can be exchanged. --------------------------------------------------------------------- ==== Simulation physics validation Alberto Ribon will be the new coordinator for the Simulation physics validation sub-project. It is also hoped that Fabiola will still be involved in the project, at least as ATLAS representative. WP: pi-absorption simple benchmark note is almost completed and will be available soon. - The note on the validation of the hadronic physics-lists is now published and public (accessible from the Application-Area web-site) - Alberto Ribon and Witek need to meet and discuss how to proceed with the physics validation and for determining the next benchmarks to study. --------------------------------------------------------------------- ==== AOB JH: Recommandations from the review should be considered. November is the next review. GC: Next meeting in October, will try to find a slot in the week 11-15 (busy period: may not be necessarily on Tuesday...).