Simulation Project Leaders Meeting 23-June-2005 ==== Attendees John Apostolakis (JA), Alfredo Ferrari (AF), Witek Pokorski (WP), Alberto Ribon (AR), John Harvey (JH), Pere Mato (PM), Gabriele Cosmo (GC). Excused: Paolo Bartalini, Rob Veenhof --------------------------------------------------------------------- ==== Minutes The workplan for LCG Phase II of the simulation project is now available from the Simulation Project web site: http://lcgapp.cern.ch/project/simu/AASimuPlansPII.pdf A short discussion took place concerning the support for the future versions of the Fortran compiler: - AF reported about his experience with the Fortran compiler now supported by the GNU-gcc team from gcc-4.0.0 (gfortran): the team supporting gfortran is not particularly open to interaction, many bug-fixes are left unresolved for long time (more than one year). One of the main problems is the lack of support for ENTRY statements, these are eavily used in FLUKA and CERNLIB; an official bug-report was open by IT to RedHat and was fixed quickly in the RedHat gfortran, still the patch has not been applied to the official gfortran distribution. - AR confirms that other problems were observed in the first evaluation for the migration of the code in Herwig: return arguments from subroutines, BLOCK DATA and no support for DOUBLE COMPLEX. - g95 seems to be more compliant with F77 and also providing better support, it is however exposing to risks since not officially supported by the GNU-gcc team. --------------------------------------------------------------------- ==== Simulation physics validation - Alberto - Combined-calorimeter: there has been intense activity in ATLAS recently; a report will be given at the next Physics Validation meeting in September. The setup will be very useful as validation benchmark test, since fairly complete in terms of geometry (represents a slide of a full detector setup) and physics simulated. - A lot of progress has been made in extending the Tile calorimeter test-beam setup to be used also with Fluka . The validation will be re-run also with the most recent version of Geant4, to see if there are also improvements in the shower-shape measurements. The exercise is useful since it can be applied also for the combined test-beam study and future benchmarks. - The work for the 3rd simple benchmark still has to start, AR is expected to tackle this in August, when he will be most of time available. - There has been a presentation by Vicente Lara at the Physics Validation meeting on June 22nd, on the validation and comparisons of hadronic models in Geant4 (binary-cascade, evaporation, precompound, ...). The physics-lists are used to assemble together the different models, and understanding the transition from a model to the other is something that now can be done in an accurate way. It is thought extremely useful to reiterate these studies with the current versions of the software, and eventually integrate them with the validation suite which is regularly run at every new release of the software. - There will be a meeting in Genova in mid-July, hosted by INFN-Genova, to discuss among all topics also physics validation. --------------------------------------------------------------------- ==== Generator services - Paolo Extract of the report provided by PB offline: - Manpower: Current shifter are: M.Kirsanov (GENSER), L.Dudko (MCDB). - Events: + Workshop in Les Houches (attended by P.Bartalini, L.Dudko). Reports on Project, in particular MCDB. Usage of MCDB advertised. + Workshop in Perugia (Beauty 2005, attended by P.Bartalini). Report on the project. Enphasis on EvtGen. + Special Application Area Meeting. Presentation of the Project. Report on the Simple Generator Level Production Framework by J.C.Maestro, "beta version" (achieved milestone 05.14). Very first presentation of Pythia 8 by T.Sjostrand. Report on Herwig++ by P.Richardson, "Tests with THEPEG" achieved milestone 05.13). Report from EvtGen by A.Ryd. Feedbacks from experiments. Discussions have also taken place: about GENSER (requests for new sub-packages and versions; quarterly release scheme and lighter modular releases); transition of Fortran packages to gcc-4.X; support of LCG Generator to the development of the new object oriented Monte Carlo Packages (EvtGen, Pythia 8, ThePEG/Herwig++) and possible support to HepMC; collaborations between LCG Generator, experiments and Monte Carlo projects on specific issues (Simple Production Framework, Monte Carlo implementation of NRQCD models for prompt J/? production); interface of the LCG MCDB with generator framework; validation & Tuning (from basic sanity checks to physics validation and Monte Carlo Tuning); future role and perspectives of the LCG Generator meetings as a permanent forum on Physics Generator issues. - GENSER: + New release 1_1_0 scheduled for mid July 2005. + A survey from Refael Yaari about future support for Fortran was received. + LCG Generator has now direct collaboration with Herwig for the transition to gcc-4.X. First working meeting organised by AR. A list of actions was prepared. + GENSER configuration: migration to the LCG TOOLBox version LCG_35; Sherpa 1.0.5 should be added in LCG_34. + New generators: CASCADE 1.2.09, CHARYBDIS 1.001, FEYNHIGGS 2.2.9. TESTS 1_00 (test package), HEPMC (started to look), DPMJET (author contacted, waiting for code), TRUENOIR (author contacted, waiting for code). + New versions of installed generators: PYTHIA (6.320, 6.321), LHAPDF (4.0, 4.1b), ALPGEN (2.0, in process), MADGRAPH (3.1.0, in process), TAUOLA (2.7.1 with CMS and ATLAS interfaces and good examples), HERWIG (6.510). + Bug fixes: revision of examples, many did not work; SAVE statements added in HERWIG. + New examples in PHOTOS, PYTHIA and ISAJET. + Documentation: end-user manual and librarian instructions updated. + Known problems: with shared libraries and on *_dbg platforms on slc3.0.5 with debugger's parsing of variables. - MCDB: + http://mcdb.cern.ch is now ready to be used and server upgraded. + Work going on Authorization Block (95% done), Search Engine (30% done) and Documentation (70% done). - Production Framework (LCG/CMS): + CMKINInterface has been ported to the GENSER framework; now the code dependencies rely on CLHEP/HepMC and is fully portable to other frameworks. + Pythia and Herwig interfaces build and run seamless (inside Pythia source tree) using the same datacards that are used in CMKIN, and with a ntuple file as output that could be converted to a root file via h2root. Still needed a local copy of some libraries (libkine.a and some other), and there are some problems with the Pythia and Herwig libraries from GENSER framework because some symbols are missing and there are problems at linking time, so the CMKIN copies are used instead (as is done in the CMS framework). + There are still some undesired dependencies on the older CMS framework (wrappings etc.). Working to get rid of them. Can now be considered at the beta stage (achieved milestone 05.14). --------------------------------------------------------------------- ==== Fluka - Alfredo - Now very close to the release of the new Beta version of Fluka. No outstanding problems left; most things are ready. Some slight political issues need to be sorted out still. - The license is approved. Download of the source-code will require signature of the license. The license foresees free use of the software, while restrictions apply in case of: + changes to the code, which should be authorised; + publications, which should come from official versions of the software. The Fluka coordination committee decided to put in place a Scientific Committee for day-to-day management of the project, for deciding about possible external contributions, and for providing technical advice on license issues. - Physics: on line radio-active decay is ready; cross-check from the radio-protection people will be completed in two weeks. - Probably still some bugs that will require beta-testing. Many common-blocks have been changed, rationalised, so user migration to the new release will be required (alpha version already available to beta testers since quite some time). - It will be observed a penalty in initialisation due to PEMF removal; the entity of the penalty mainly depends on the number of materials used; it is in the order of 18 secs in average. It is possible that in future an option to still use file storage will be added. - Linux/gcc and Compaq/dxx will be supported. In future also MacOSX and Windows with CygWin/gcc. - Documentation (manual and report) should be out already by end of June, with on-line HTML version as well; a second report focussed on the physics will be preared starting after the release. - Date: 26 July. - The new Fluka will be installed and distributed also in the LCG area as soon as it will be released. --------------------------------------------------------------------- ==== Simulation framework - Witek - FLUGG: + Started to work on the TileCal test-beam, currently concentrating on the technical part. Manuel Gallas managed to produce a GDML file of the geometry setup which has been imported using the GDML-FLUGG interface. A simple setup for producing hits a-la Geant4 has been implemented and committed in the repository. The alpha-version of new Fluka has been used, and the whole setup is working fine as first attempt. AF noted that in the past tests, effects of quenching with the scintillators were a problem; it was noted that this will require investigation but will mainly affect the final study of the detector response therefore not directly related to the hits production from the simulation engines. + Having Fluka centrally installed in the LCG area would help a lot. AF confirmed that there will be no problem to provide that. - Geant4 geometry object persistency: + The implementation of 2 plugins (a writer for Root and a detector construction class for Geant4) have been completed. Reflex is used. + The current prototype required adaptations in Geant4, mainly to deal with the need for Root to require default contructors for the classes whose objects should be made persistent. + Some outstanding problems need to be resolved, mainly concerning the presence of pointer to arrays or, more important, in the case of pointers to non-native types (code was instrumented in the current prototype...). A solution in future versions of Root may be provided for this. + The prototype have been tested with the LHCb geometry successfully; the whole geometry can be saved in roughly 180 Kb and retrieved. + A document summarising the work done will be prepared soon. + It was noted that the requirement for forcing the presence of default constructors should be investigated and understood since by design there're constraints in Geant4 that need to be taken into account. Alternative solutions should be investigated as well (private default constructors, alternative persistency techniques...). - GDML: + A new build system based on autoconf will be included in the new release 2.1.1 which is scheduled by end of June. + There has been request for support of default materials in GDML. + Work is progressing in the implementation of the Root->GDML writer using PyRoot. + Problem of manpower, Radovan Chytracek can no longer contribute. A request for a technical student will be reiterated at the next board of September. --------------------------------------------------------------------- ==== Geant4 - John - The public minor release 7.1 is planned on June 30th, 2005. + The testing phase (4 weeks period) has been completed and now working on packaging and assembling the documentation. The validation using the GRID is ongoing, with the kind help of the IT Grid contacts, the ALICE Grid team (whose lent us their virtual organisation for June) and the 'host' sites. Overall several technical issues (related to new aspects and this being only the second exercise) are causing it to be behind schedule. First results are available and positive. + No revolutionary changes, as a minor release of Geant4 is concerned. In the physics, the most important updates are concerning a refactoring of the photon processes to follow a model approach as it is already done for the energy-loss processes; the update will be completely transparent to the users. + The complete list of changes and features can be accessed from the release notes: http://geant4.cern.ch/geant4/source/ReleaseNotes4.7.1.html . - A mini-workshop of Geant4 developers was held at CERN in late May to study and prepare a first design iteration for allowing parallel navigation on real or parallel geometries. The aim is to extend and rationalise the existing framework to allow also treatment of charged particles (with or without field) and provide a coherent approach for uses in scoring, biasing and fast-simulation. This development is of particular interest for LHCb for their radiation studies. It is expected to have a prototype implementation ready by the end of the year. - The user tutorial at CERN was very successful: 46 people attended, lectures and hands-on sessions were given using the CERN Training Center facilities. Nearly 30 attendees were from CERN and other LHC related activities. Several people who did not make it were informed about the later workshop/tutorial in Finland. - A vote on the proposed new Collaboration Agreement for the Geant4 Collaboration is ongoing. The vote will potentially extend to Thursday, June 30th and a potential positive outcome will move the Agreement to the Institutes and Laboratories for their final consideration and signing. - License on data: touched on the question of using of external data libraries in creating 'tailored' libraries for simulation. It was noted the commonality of addressing this issue and agreed to communicate regarding understanding and experiences for data libraries used for the simulation. AF offered to inform us on his discussions with data distributors of the neutron xsection libraries, with whom he is communicating - to help improve understanding of specific requirements of the providers or distributors or any constraints in their use. - Manpower: the candidate for the fellowship on radiation-studies has a choice between CERN projects. He has been contacted with information about the work, and agreed to respond regarding his choice by mid-July. --------------------------------------------------------------------- ==== Garfield - Rob No report. --------------------------------------------------------------------- ==== AOB Next meeting in July or September.