Simulation Project Leaders Meeting 11-December-2012 ==== Attendees (virtual meeting) John Apostolakis (JA), Andrea Dotti (AD), Alfredo Ferrari (AF), Witold Pokorski (WP), Rob Veenhof (RV), Gabriele Cosmo (GC). --------------------------------------------------------------------- ==== Minutes --------------------------------------------------------------------- ==== Simulation physics validation & framework - Andrea, Witek - Validation: o Results for the new release 9.6 are comparable to those obtained with the previous release series. The main change is observed in the hadronic showers for their lateral profile: showers in 9.6 are wider in Iron and Copper, and narrower in heavier absorbers, like Tungsten and Lead. o A prototype of the tool to make on the fly plots from FNAL DB has been created, providing multiple file upload, release highlights for upcoming Geant4 releases, expert login to view/edit/delete tests, extended data base schema. It will be soon rolled over to production. o The SimplifiedCalorimeter suite has been sucessfully ported on Geant4-MT. Reproducibility of events has been verified over the sequential version. --------------------------------------------------------------------- ==== Generator services - Witek - GENSER: o The project is moving to a new approach where the MCGenerators trees follows strictly the LCGCMT releases. This assures that the versions of all the external software packages used by the generators (like HepMC, FastJet, etc.) is compatible with those used by the experiments. The new GENSER directory name schema is now MCGenerators_lcgcmtXX, where XX stands for the lcgcmt versions. This new schema has been adapted since LCGMT release 64, and therefore we have now the following releases available: MCGenerators_lcgcmt64, MCGenerators_lcgcmt64a, MCGenerators_lcgcmt64b. In addition, we also have MCGenerators_lcgcmt_preview which is used for testing new version of the external software which has not yet been made available in the LCGCMT release. All the GENSER machinery is now being migrated to the new use CMT as the underlying configuration management system, following the approach adapted by LCG externals. In future, further integration of GENSER and LCG external software service should be envisaged. o New generators installed: amcatnlo-1.1, hepmcanalysis-3.4.13, photos++-3.52, herwig++-2.6.1a, thepeg-1.8.1, pythia6-426p12, pythia8-170 --------------------------------------------------------------------- ==== Fluka - Alfredo - No report. --------------------------------------------------------------------- ==== Geant4 - John, Gabriele - Release 9.6: o The new release 9.6 was announced on schedule on November 30th, 2012. o Release 9.6 features full event reproducibility in all physics models, with two exceptions for CHIPS stopping and neutron HP; so, event reproducibility is valid for the majority of the physics lists, including FTFP_BERT, QGSP_BERT and QGSP_FTFP_BERT. o WentzelVI multiple-scattering model is used by default in all EM physics lists; for electrons and positrons the WentzelVI model is used only above 100 MeV. This was done to provide more accurate simulation for both small and large angles of high energy charged particles. o An universal interface to angular generators is established for all EM models (Bremsstrahlung, Rayleigh and PhotoElectric effect). The Seltzer- Berger model for Bremsstrahlung has been improved to adopt a more detailed grid of differential cross-sections and updated screening functions. o Hadronic processes now trigger re-sampling of the interaction if energy/momentum non-conservation limits are exceeded. A warning message is printed if this occurs. o Diffraction dissociation in Fritiof has been improved for protons, pions and kaons. o The Bertini model has been extended to handle nuclear capture at-rest of pi-, K-, Sigma-, Xi- and Omega- and direct photo-nuclear interactions. o New total cross-section data set based on SAID data-base is introduced and now required; data is used for proton, neutron and pion cross-sections on hydrogen in majority of physics-lists. o A new option in the nuclear de-excitation module allows to ignore Doppler broadening for long lived meta-stable isotopes and provide instead decay at rest. o The INCL cascade model (INCL++) was updated and can now handle heavy-ion collisions. o Barashenkov-Glauber-Gribov nucleon-nucleus inelastic cross sections are now used in most physics lists instead of Axen-Wellisch proton-nucleus and Laidlaw-Wellisch neutron-nucleus inelastic cross sections. o A new physics-list FTFP_BERT_HP is added: it is similar to FTFP_BERT, but using HP for low-energy neutrons. o Removed direct dependency on the CHIPS model from most physics-lists. o For all elements, a vector of isotopes is always created using abundance values from NIST (in previous Geant4 versions, the vector of isotopes was created only for NIST materials); as a result it is no longer needed to define natural composition of isotopes in the user code. o A new optimized implementation of G4TessellatedSolid provides substantial CPU performance boost and reduced memory footprint, making it now suitable for import or reasonably complex structures with hundreds thousands facets. o Enhanced the G4RunManager class to allow for better integration with external frameworks and avoid need for code duplication. o Optimized tuning for parameters in G4SmartTrack, now used by default results in 4-5% reduction in total execution time in complex setups. o Improved OpenGL rendering speed through major re-design of scene and transient processing; improved Qt interface and viewer; extended OpenInventor visualization driver. o Deprecation warnings for methods and classes planned for removal in the next major release have been added to the code. o Major cleanup of the code for cases of variable shadowing. o Updated data sets: G4EMLOW6.32, G4NDL4.2, G4NEUTRONXS1.2, RadioactiveDecay3.6, G4SAIDDATA1.1 and PhotonEvaporation.2.3. o Reviewed all extended examples and provided new documentation based on DoxyGen. - Development: o A new development release is going to be made before the end of the year and will include most changes for deprecated classes and methods planned for removal for the next major release. o Porting of Geant4-MT on MacOSX has been made with little modifications on the current avialable prototype. A first simple application implementing event-level parallelism in Geant4-MT using the Intel Thread-Building-Blocks (TBB) library has been realized, demonstrating that Geant4-MT can be used within an external framework based on TBB. No changes were required in Geant4-MT itself, except for the addition of few simple classes in the application code. - Experiments feedback: o Measured discrepancy in the Z->ee channel in ATLAS, with a shift in distribution and decrease of the low tails by ~10%. Effect likely due to the protection introduced in the multiple-scattering in latest patches to fix problems of wide scattering. o Validation of 9.5.p02 going on in CMS. So far, results are compatible with the older release used in production. First porting on release 9.6 reveals some new warnings at run-time, currently under investigation. o Following the last Geant4 Technical Forum, held on December 6th, ATLAS proposes to include a new random engine and Gaussian distribution in CLHEP for general use, providing CPU improvement of the order of 2-4% combined in ATLAS simulation. Also, requested in CMS to envisage optional adoption of an alternative fast mathematical library, which could bring ~9% speedup in CMS simulation at the price of reduced precision. - Meetings: o New tutorials will be held in Belfast (Northern Ireland) on January 21-25th and at the Universisty of Wollongong (Australia) on April 19-14th. --------------------------------------------------------------------- ==== Garfield - Rob - Activities: o Waveform fits of simulated PPC avalanches have clarified several discrepancies between measurements and calculations of the Townsend coefficient, the diffusion and the velocity of freons used in RPCs. o Charging-up studies of GEMs have meanwhile been complemented with space charge studies with a view to better understand ion back flow. o Steve Biagi about to release a program to simulate photon-induced charge clouds, focus on gases used in radiation detection. o Decorrelation distance for heavy noble gases have been increased. This leads to a noticeably better estimate of the longitudinal diffusion. o AGH (Krakow) has upgraded its gain measurement laboratory with a further reduction of the dark current by means of guard rings - this laboratory is noted for the excellent calibration of its measurements. o Bursa gas lab on track. Building to be handed over in Spring. - Meetings: o Collaboration meeting in Stony Brook (Oct 1-5). o "Mini week" (Dec 3-5) and (Jan 30-Feb 1). o Collaboration meeting 2013 to be combined with the MPGD conference will be held in Zaragoza. --------------------------------------------------------------------- ==== AOB Next meeting scheduled for February 26th.