Simulation Project Leaders Meeting 13-March-2007 ==== Attendees John Apostolakis (JA), Alfredo Ferrari (AF), Witek Pokorski (WP), Alberto Ribon (AR), Rob Veenhof (RV), Pere Mato (PM), Gabriele Cosmo (GC). --------------------------------------------------------------------- ==== Minutes - Participation to the CHEP07 conference in Canada must be notified as soon as possible. April 1st is the deadline for the abstracts submission, April 16th is the deadline for providing the complete list of names of candidates to the PH secretariat. - The first 2007 quarterly report is expected by end of March. - Summer students: one student has been selected for Generator Services. Unsuccessful selection for Simulation Framework and Geant4 so far. --------------------------------------------------------------------- ==== Simulation physics validation - Alberto - Meetings: o At the meeting held on February 28th there has been a presentation by P.Strizenek (MPI) on the ATLAS/HEC, where monitoring of the various Geant4 versions is done in their analysis; extended comparison were made using 8.1.p02 also with the QGSP_BERT physics list; the adoption of Bertini Cascade improves as expected the shower shapes profile, on the other the energy resolution for charged pions unexpectedly is reported to be in worse agreement with the data; reasons need to be understood. There was a presentation also from AR on the final results of the 3rd simple benchmark. M.Kossov presented latest results on precise simulation of muon-capture in CHIPS. Finally, a presentation by V.Grichine on a study made for comparisons of nucleon-nucleus cross-sections. o The next meeting is scheduled for March 28th, where a presentation on the ATLAS combined test beam simulation is expected. - Simple benchmarks: o Final results for the 3rd simple benchmark presented at the last Physics Validation meeting and LCG note released (CERN-LCGAPP-2007-01), see also: http://lcgapp.cern.ch/project/mgmt/doc.html. o For future benchmarks there are three possible choices under consideration: setups based on SPY, NA49 or HELIOS experiments. Evaluation is undergoing and priority is given to studies that could help to understand and improve the shower shapes description in Geant4. - Extension of ATLAS TileCal 2002 analysis with Fluka: o The bi-weekly meetings with the Pisa group are now established. The analysis has now reached an advanced state and comparison with data is going on. The results will be discussed soon with Geant4 and Fluka experts. --------------------------------------------------------------------- ==== Generator services - Witek - Meetings: o There're been two monthly meetings, were the status of the project and GENSER restructuring has been presented, together with validation studies. - GENSER: o Version 2.0.0 of GENSER has been released beginning of February, the new version follows the new scheme and structure of the package, with the new build system and configuration. o Since then two generators asked by CMS have been added: Pyquen and Hydjet. o Experiments (Atlas, CMS, LHCb) are using it and no major problems have been reported so far. - Manpower: o There's currently a gap of two weeks where no Russian collaborators are present. A new collaborator, P.Polarush from LCG/Russia, is joining the GENSER group. o L.Sonnenschein from MC-Net is now at CERN working on HepMC and development on the Rivet tool. The idea is to make use of Rivet for validation of the generators. - On request of T.Sjostrand an 'interactive' manual of Pythia (interactive choice of parameters for generation of input cards) was put in place. Ben Lloyd has contributed in doing this. - On suggestion of M.Mangano, Ben has also contributed to create a FAQ page to be used for the Generators. The page is based on Twiki: https://twiki.cern.ch/twiki/bin/view/GeneratorServices/WebHome. --------------------------------------------------------------------- ==== Fluka - Alfredo - Meetings: o There has been a Collaboration meeting in mid-December, where discussions about the reorganisation of Collaboration took place; it was also decided to establish a 'shift' procedure on weekly-base for the mailing list, in order to provide more support to posted requests; also a FAQ page has been added on the web: http://www.fluka.org/Faq.html. - Fluka: o Work done particularly on graphical interface. New Fluka backend GUI to inspect and manipulate standard FLUKA output (http://www.fluka.org/tools/FlukaGUI.html), and new Python user interface FLAIR (http://www.fluka.org/flair/) for graphical input, run control and analysis of results. A presentation of the FLAIR utility will be planned for an Application Area meeting. o An interim new release Fluka-2006.3b will be made avaiable beginning of next week which will mainly include new usability features and some bug fixes not affecting physics performance, included a fix for a bug found in the quenching implementation on MGDRAW. The new release will be also installed in the common AFS LCG area. - Tutorials: o Two courses are planned: a course in Houston, US (May) and another in Legnaro, Italy (October). Pilot courses will be also taken at CERN along with the traditional users meeting occuring every 3 weeks. --------------------------------------------------------------------- ==== Simulation framework - Witek - GDML: o B.Lloyd has finished the implementation of the installation script. It will be part of the next release which will be announced soon. o With help of J.McCormick (SLAC) it has been added a new example showing how to add user extensions to GDML. A possible enhancement for this is to show extensions of the schema for customised solids. o Ben has started investigations for possible integration of GDML binding in Geant4 for the Geant4 specific components. - Manpower: o B.Lloyd will leave end of April. --------------------------------------------------------------------- ==== Geant4 - John - Releases: o A pulic patch to release 8.2 was made available end of February. The patch included several fixes, particularly addressing a problem of event failure rate on 8.2 reported by ATLAS when using the QGSP physics-list with magnetic field. Test cases reproduce the problem only when significant overlaps in the geometry are present. Such overlaps were also discovered in a number of problem cases in the experiment detector description. Using QGSP_EMV the problem is not present, and the simulation is faster due to the simpler multiple-scattering model that delivers results which are less precise and less stable with variation of cuts. The more precise multiple-scattering algorithm is used in most physics lists and is the default for this process. The workaround consisted of changes in the treatment of the geometrical safety in the field-propagation and removing an unnecessary call to point location by the multiple-scattering process. ATLAS report a reduction of the event failure rates to 10% with these fixes. In ATLAS some of the issues affecting the geometry setup have been fixed. Other fixes included in the patch address some problems of NaN detected and a fix for a long-standing bug in G4Polycone reported by CMS. - CPU performance: o According to reports from ATLAS, in most physics channels the performance using QGSP_EMV is near to that of Geant4 7.1-patch2, except in one type of events (dijets) where an increase by a factor 2 in time was observed. Increased neutron production and 5x transport steps were found, but the primary source of increased CPU time appears to be photon transport in the EM endcap calorimeter. The issue is perplexing and under investigation. - Shower shapes: o The LCG note on the shower profile studies is in an advanced state. Preliminary conclusions are now made. The note will be made available in the next days. A. Ribon has been driving the effort and acted as editor. o The combination of diffraction and quasi-elastic seems to be the culprit of the problem, it was con-stated that improvements in this area are necessary. A first implementation of a revised FTF model by V.Uzhinsky, which includes improvements in diffraction, was included in the last development release at the end of February. First studies using the new model in place of QGS have shown significant improvement in the shower shape profile. Additional thin-target validations are ongoing. We are communicating with experiments to try out the latest development release (reference tag) and in particular the revised FTF model. - Tutorials: o There will be a Geant4 tutorial at SLAC in May, and another one at Ecole Polytechnique in Paris in June. The latter will include participation of SFT Geant4 team members, and will include lectures in French and English. --------------------------------------------------------------------- ==== Garfield - Rob - Gas detectors: o Further refinements in the simulation of stopping ions in gas mixtures, in particular regarding the energy loss fluctuations. Shulek straggling functions need adjustments so as to ensure negative energy losses are avoided, while at the same time maintaining the correct mean and width. This work is now sufficiently complete to be made available. A report has been written on the subject and feedback is awaited. o A prototype Grid implementation of Garfield by the Grid team is complete and is being tested. It is anticipated that this will be used to generate transport tables for Atlas and Alice, varying the gas mixture and the magnetic field, both in strength and in orientation. o Although retired, Steve Biagi is taking an active interest again in the calculation of transport properties. - Activation studies: o The conversion to Fluka format of the CMS silicon detector is essentially complete. The description of the central part of the beam line is in progress. o Before updating the CMS Fluka file with these new elements, Fluka is to be run with the existing geometry so as to verify the neutron rates in the cavern. These can be compared with calculations performed earlier. Work on this has been on hold for the last month pending the above activities and is expected to resume shortly. --------------------------------------------------------------------- ==== AOB Next meeting scheduled for May 8th.