Simulation Project Leaders Meeting 25-September-2007 ==== Attendees John Apostolakis (JA), Witek Pokorski (WP), Rob Veenhof (RV), Gabriele Cosmo (GC). Excused: Alfredo Ferrari (AF), Pere Mato (PM), Alberto Ribon (AR). --------------------------------------------------------------------- ==== Minutes - The next LHCC Comprehensive review is scheduled for November 19th. - Quarterly Reports: the 3rd quarter 2007 activity reports must be prepared by end of September. --------------------------------------------------------------------- ==== Simulation physics validation - Alberto (offline report) - Meetings: o In the Physics Validation meeting on June 20th, there was a presention from V.Uzhinskiy on the recent improvement on the Fritiof model in Geant4, which has improved significantly the hadronic shower shape description when using the FTFP Physics List. CMS presented an update on the combined ECAL/HCAL for 2004/06 test beam; it was confirmed that there is a difference in response for low energy pions between data and Geant4 simulation, even when Birks effects are taken into account, and the size of the crystal matrix is correctly described in the simulation. As far as the hadronic shower shapes are concerned, Bertini cascade improves the description. o In the Physics Validation meeting on July 25th there were two presentations from ATLAS. The first was on the new analysis of the TileCal 2002 test-beam data, including Fluka (see below). The second was an update of the HEC test-beam, with Geant4 8.2.p01, including a study of the effect of Birks quenching in liquid argon. The main conclusions we can draw is that Bertini cascade improves the longitudinal and lateral hadronic shower profile, whereas its effect on the energy response and energy resolution is opposite for the two analyses: in the case of TileCal, it improves the agreement with the data, whereas for HEC it is farther from the data. Work is in progress to understand this difference. o The next Physics Validation meeting is scheduled for October 17th. - Simple benchmarks: o First preliminary results of the fourth simple benchmark, based on the Helios data for diffraction in proton-nucleus collisions, has been discussed with Geant4 people. Extension to Fluka is undergoing. - Extension of ATLAS TileCal 2002 analysis with Fluka: o Results for 20 GeV beam energy has been presented at the LCG Physics Validation meeting on July 25th. Geant4 QGSP_BERT and Fluka provide a good description of the hadronic showers. It has been confirmed, not only for recent Geant4 versions (7.1 and 8.1), but also for the original one (5.2), that Geant4 QGSP shower shapes are shorter and narrower than data, in agreement with the other test-beam analyses. o Work is still undergoing to extend the analysis to the other beam energies, where the evaluation of the background due to beam contamination with other particle species is more difficult. --------------------------------------------------------------------- ==== Generator services - Witek - Manpower: o Challenging situation since beginning of the summer; three out of four developers/integrators have become unavailable since moving to other duties in the experiments or respective home Institutions; currently O.Zanin is carrying on the effort, contributing remotely from Russia. Three new contributors have been proposed by M.Kirsanov, but they will have to go through a training period before becoming fully effective. o The summer student M.Montull has left and has been working on the validation of Pythia8. - Meetings: o In the last monthly meeting which took place last week there has been a report about the status of GENSER by M.Kirsanov; also a presentation by Z.Was on Photos and event records; it was discussed the issue of whether to store the number of parton interactions as an additional information (which can be deduced from the event anyway), no decision has been taken. A discussion also took place concerning handling of the status code for particles in HepMC; in ATLAS the status has to be changed in order to keep the information coming from Geant4. - GENSER: o New generators: Charybdis and Isajet and new versions of several others. o O.Zenin made some work to integrate Rivet in GENSER for validation; some new validation tests have been implemented using Rivet. The idea is to extend further the validation suite with Rivet in GENSER and further foster cooperation with the Rivet team. o Following a request from LHCb, WP has built three generators (Pythia, Photos and LHAPDF) on the Windows platform. People are happy with it and request for more... It should be noted though that unfortunately building on Windows is very unconvenient given the unavailability of a publicly accessible Windows machine to be used and the missing expertise and manpower required for providing 'official' support on Windows for all the generators. --------------------------------------------------------------------- ==== Fluka - Alfredo - No report. --------------------------------------------------------------------- ==== Simulation framework - Witek - Flugg: o The Flugg package has been now included in the LCG application area CVS and the Simulation Project is now providing effort for the maintenance. A new version of the package has been released, compatible with the new versions of Fluka and Geant4. o To do: review the code and resolve some known issues affecting the ATLAS TileCal setup at 90 degrees. - GDML: o The new student Z.Torzsok is getting now deep in the matter; improvements have been implemented in the GDML processor, and new solids have been added to the schema (extruded-solid). A new version of GDML (2.10.0) has been released last week. o Now starting working on the integration of the Geant4 plugin in Geant4. The idea is to move away from the Saxana processor and move to a DOM-based parser; this will most likely imply developing from scratch the processing code. The objective is to reach a better maintainable package, easier to install and more user-friendly for the Geant4 users, maintaining full compatibility with the existing GDML schema and GDML files. It is foreseen that a complete functional system will be available in the next couple of months. - Persistency: o Root I/O binding is now completed; features have been included in Root (support for C arrays of pointers...), and some necessary fixes in Geant4 now available in the last patched release 9.0.p01. o Still missing an example to fully demonstrate the functionality to be eventually added in Geant4. o To be discussed also the addition of an optional UI command in Geant4 for storing/loading geometry. --------------------------------------------------------------------- ==== Geant4 - John - Meetings (12th Geant4 Workshop): o The 12th Geant4 Workshop took place September 13th-19th at Hebden Bridge, (West Yorkshire) hosted by the University of Manchester, UK. o Before this workshop, on Wednesday 12th there was the first Geant4-GATE mini-workshop. GATE is the Geant4 Application for Tomographic Emission created and maintained by the OpenGATE Collaboration. It is a tool for simulating PET and SPECT detectors, based on Geant4 and including extensions that enable users to define detectors and make measurements. Users can do this without utilising C++, simply by issuing UI commands. A number of topics and issues were presented and discussed: biasing, fast navigation for voxel phantoms, profiling of their application. A number of hot-spots were identified when using low-energy EM processes (10-30% of the application time). Avenues to satisfy the urgent need of GATE users and developers to obtain improved CPU performance and potential methods for obtaining it were discussed. o The first three days of the Geant4 Workshop was a Users workshop. Presentatations were made by HEP, medical and space users. There was much feedback from the HEP underground community: issues with neutrons production from muons and new propagation were communicated and discussed. New CHIPS models for muon-nuclear interactions are now available in Geant4, and these were presented and recommended. A presentation from the CALICE group summarised existing results. When new results are available, they now have a standing invitation to present them at the Physics Validation meeting. o The new prototype for regular/lattice-navigation was presented to the users with measurements showing comparable speed as for the voxelised navigation, but with much reduced memory consumption when applied to size-able phantoms of the order of several millions voxels. o Results of two comprehensive studies of Geant4 for medical applications were presented: against Bremmshtralung data on which EGSnrc was validated (where Geant4 is seen to perform as well) and an extended study of electron dose point kernels in water. Their significance is their validation of the electron transport of Geant4, relevant to many application areas including HEP. o HEP: CMS still reports issues regarding the multiplicity of particles produced in shower from few GeV; the origin for this seems to be understood and could be related to EM energy deposition in crystal for hadrons... ATLAS reported on the CPU performance monitoring and studies. CMS also reported that when enabling field value caching, a significant gain in CPU performance is seen [there was an observation made during the SL meeting regarding reproducibility of events which could be affected if the cached values are not flushed at the start or end of each event!]. o There have been presentations by ESA concerning recent progress made on the interface to CAD systems. Two solutions are near to completion: OpenFrontier, a CAD-like 3D editor for geometrical models based on OpenCASCADE with STEP import/export facilities and GDML export module; and FASTRAD, a geometry editor with STEP import/export facilities, translation of STEP descriptions to tessellated solids/GDML. These will become available as open source software (OpenFrontier) or without a cost (the FASTRAD module). An additional module of FASTRAD that provides generation of C++ code is expected to require a paid license. o The second part of the meeting was the Collaboration Workshop. Among the topics discussed, the improvements in the way fluctuations are handled depending on step size. It was also reported about a project for development of EM ions-ions interactions, a prototype with full functionalities may be available by December 2008. o The Geant4 Steering Board meeting and the 20th Technical Forum have been also held during the Workshop. o The next Geant4 Collaboration Workhop will be hosted in Kobe (Japan) on October 8-12th 2008. o Independently, there has been interest from people who wish to use Geant4 for the detailed simulation of signal propagation in gaseous detectors. Exploratory discussions are underway with the Garfield and interested parties. - Releases: o Two public patches of Geant4 were released in August: 8.3.p01 and 9.0.p01. 8.3.p01 included in particular a fix in the magnetic field module for exceptions which used to be thrown in rare cases of occurrence of zero steps; other fixes are reported in the release notes. o The next Geant4 release 9.1 is scheduled for December 14th 2007. --------------------------------------------------------------------- ==== Garfield - Rob - Gas-based detectors o A Micropattern workshop was held at CERN Sep 10-11th. It featured a review of the various gas-based techniques under study for future detectors, such as Micromegas, Pixel and GEM. A proposal for an R&D collaboration is to be submitted later. In this context, the need is recognised for an update of the simulation software. RV has been asked to come forward with a proposal, which will consist of 2 parts: the areas where new algorithms are needed and the user interface. o New algorithms are needed for: - field calculations for configurations where neither analytic nor finite element methods are applicable: investigation into boundary element methods and integral equation techniques are needed; - detectors are reaching the scale where electrodes are of the same size as the mean free path and the statistical transport methods used up to now will no longer be applicable: one solution would be to use Magboltz with the real fields; - gas gain: secondary induction process to be simulated. The process is normally known as "Penning transfer"; this is a mechanism under which energy is transferred from the excited states of one species of molecules to molecules of the other, which in the process are ionised; - dynamic effects are not currently included in any calculation, yet they do matter in e.g. GEM operation; some exploratory studies have been done in this domain, and further work awaits progress in field calculations as such. o User interfaces: - the most convenient user interface for development (plotting/graphics) would use Root, this would also be appreciated by several major users; - in parallel, it is expressed a clear interest for an integration with Geant4, possibility currently under study (Alain Bellerive, John Apostolakis). - Activation o The Fluka team has ironed out a number of issues with neutron counting (related to biasing) in the CMS model (geometry, production/interaction files) and new tests have been run to compare with earlier calculations. A report on this remains to be written. o The model is gradually being used inside the CMS collaboration, in particular for the beam conditions monitor. - NA60 o No activity since the last meeting. --------------------------------------------------------------------- ==== AOB Next meeting scheduled for November 6th.