Simulation Project Leaders Meeting 24-October-2006 ==== Attendees John Apostolakis (JA), Alfredo Ferrari (AF), Witek Pokorski (WP), Alberto Ribon (AR), Pere Mato (PM), Rob Veenhof (RV), Gabriele Cosmo (GC). Excused: John Harvey (JH). --------------------------------------------------------------------- ==== Minutes - LHCC Comprehensive Review outcome: no major notes to be reported for the Simulation Project; acknowledged the need of improving the integration of the various MC packages in GENSER. --------------------------------------------------------------------- ==== Simulation physics validation - Alberto - Meetings: o In the Simulation Validation meeting on September 20th there were several interesting presentations made: A presentation on an ATLAS TileCal using a beam at 90 degrees was made by M.Simonyan, studying pions and protons; the problems affecting the longitudinal shower shape simulation in Geant4 were confirmed. What was interesting was that the problem is also observed at low beam energies, whereas in CMS the problem was seen only at high beam energies. In this analysis it was also measured the transverse/lateral profile by adopting LHEP and QGSP physics lists and their extensions including Bertini intra-nuclear cascade. The results show that LHEP shape is narrower than data, and QGSP is even narrower; the inclusion of Bertini model makes the hadronic showers wider, in better agreement with data. This can be probably explained by the fact that neutrons play an important role in the lateral shower profiles. It must also be noticed that lateral profiles are also sensitive to the experimental analysis (time window, quenching, etc...). Another presentation on neutron background studies in CMS with Geant4, aimed to evaluate the impact on physics signals, was made by N.Srimanobhas, showing reasonable agreement using the G4NDL3.8 (or 3.9) data sets. Other presentations on Geant4 developments were also made. See: http://indico.cern.ch/conferenceDisplay.py?confId=a063186. - 3rd simple benchmark: o A meeting with Fluka experts to discuss the preliminary results of the analysis happened recently; it was suggested the analysis should be done using the new high energy model in Fluka, for consistency with all recent results. It was also suggested to add some plots of the ratio of the simulations over the data to make easier the evaluation of the comparisons. A presentation of the final results will be done in November. - Extension of ATLAS TileCal 2002 analysis with Fluka: o Everything is now ready to start the production and then the comparison with the data. Some technical problems need to be fixed in the setup upgrade. See Simulation Framework report below. --------------------------------------------------------------------- ==== Generator services - Witek - Meetings: o The announced planning meeting for Generator Services was hold last Friday, October 20th; the representatives from all stake-holders (experiments and authors) were present. The status and plan were presented. The outcome was positive and an agreement was reached on the main items discussed. It was decided to move away from SCRAM in GENSER and use the original build systems provided in each generators package. It was agreed to reduce the man-power effort currently devoted to MCDB and move it to areas like GENSER and validation where there's more urgent need. It was discussed how to improve first level support, by improving the existing Savannah portal, restructuring the relevant web pages, and advertising GENSER to the experiments. The possibility of holding tutorials on the generators by the authors to individual experiments was also discussed. New prioritised milestones have been set. o The next monthly Generator Services meeting will be in November 22nd. - Man power: o Increase man power for GENSER and validation. --------------------------------------------------------------------- ==== Fluka - Alfredo - New Fluka release: o No big issues were raised since the announcement last September. o Further changes are expected concerning ionisation on TPC for ALICE; it was required to provide the number and positions where primary ionisations are created, as well as the energy of each of them. o The new release of Fluka will be installed shortly in LCG AFS area as planned. - Activities: o Working on heavy ion direct pair production (no major impact on LHC calculation foreseen). o New neutron cross-section library: not yet found a person for helping in finalising the work. o Need to better understand nuclear data formatting and distribution within codes to be compliant with the Nuclear European Agency (usage of NJOY application for data generation and reprocessing). NEA is funding the evaluation of cross-sections at cryogenic temperature, which can be interesting for LHC and useful for activation studies. - Flugg: o Discussions have been taken place with WP concerning the future maintenance and support of Flugg. Paola Sala can share some of her time to it, at least at the beginning, but a more stable solution is necessary for the future; resources are required in this area to assure proper support. o License issues need to be clarified now that Geant4 runs with a license. --------------------------------------------------------------------- ==== Simulation framework - Witek - FLUGG: o Following the presentation by ATLAS on the TileCal simulation at the last Physics Validation meeting, where a setup using a beam at 90 degrees was used, it was decided to upgrade the current setup to also upgrade to the latest Geant4 version. Some work is required to bring the setup up to date. o A trial with the full CMS geometry in Flugg for the activation studies was made. A problem concerning the size of a common-block in Fluka was fixed and it is now possible to import the whole CMS geometry through Flugg in Fluka. Some other run-time problems need still to be understood and fixed. Also, some geometrical elements (like the cavern and beam details) need to be added to the setup. - GDML: o B.Lloyd is actively working on the re-implementation in C++ of the GDML->Root converter. First positive results achieved, the complete LHCb geometry can be read with a speed a factor twice faster than using the Python reader. --------------------------------------------------------------------- ==== Geant4 - John - Geant4 Workshop in Lisbon, October 4th-10th: o Many topics discussed: hadronic validation, CPU performance, developments and refinements to multiple-scattering. o There has been a report from the INCL team about the implementation of a prototype for interfacing INCL with Geant4 and a plan for a complete rewrite in C++ to be included in future in Geant4. o A presentation was made from remote (VRVS) by CERN OpenLab about a study on CPU perfomance using the ATLAS simulation. No hotspots were found; time spent in the geometry is the leading contribution, which generally is much smaller in simplified setups instead; it was found that the contribution from the hadronic processes is not negligible. 20% average gain by upgrading the compiler to gcc-4.X compared to gcc-3.X; worst performance using -O3 compared to -O2. Further investigations are now going on in SFT by Rafi. A Study in ATLAS for fast simulation was presented, showing interesting gain when turn off neutrinos generation (up to 8% time gain). o An Oversight Board (OB) meeting was taken at the Workshop where an internal review of Geant4 has been planned for the period March-June 2007. o Also the first official Steering Board (SB) meeting has been taken, where mainly internal communication and organisational issues were discussed. It was also planned the new December release of Geant4 which will be a 'minor' release, 8.2. - The next Technical Forum meeting has been scheduled for November 21st at CERN. There will be also one focused on Space Applications at the Geant4-Spenvis Space Users' Workshop to be held at NASA JPL in Pasadena on November 6-10th. - Some expected developments in release 8.2: o Refinements to the navigator for the usage of parallel geometries; it will be possible to shoot charge particles on parallel geometries with field activated. o Expected developments in multiple-scattering near boundaries: new optional way to adopt single scattering. - Validation: o Comparison with Agababyan data: results on high energy proton interactions will be shown at the Physics Validation meeting, October 25th. o It was found by AR that the Whitmore data used for the 3rd simple benchmark don't include diffraction; good agreement with Geant4 QGS model and data is reached with and without diffraction. A review of the diffraction in the QGS model is planned. Further validation studies may include SPY data and NA49 pion production data. - Experiments feedback: o Requests from ALICE for number of primary ionisations in gaseous materials and simulation of polarised lambas decay. o Issues raised in ATLAS: in upgrading to Geant4 7.x release series a change documented in the migration was missed regarding the revised procedure to activate user step limits. During Grid productions the set of low-energy photon evaporation data was not found. An estimate of the effects was undertaken by the Geant4 team, and it is assessed to be very small for all observables tested. Currently working on establishing a better process for verifying the presense of required data. --------------------------------------------------------------------- ==== Garfield - Rob - Garfield activities: o Rare decay experiments have very low counting rates and to reduce the cost of read-out electronics, they sometimes connect each n'th channel. This leads to the need for a truely periodic weighting field when computing induced signals, as opposed to the more usual need for a non-periodic weighting field with a periodic electric field. Analytic expressions for weighting fields with true periodicity have been added. o SNO, T2K and rare decay experiments wish to separate alpha, neutron and nuclear recoil signals. The simulation of the ionisation patters for particles which stop in the gas is not taken care of by Heed, which concentrates on particles of much higher energies. Ionisation simulation procedures were added which produce patterns that respect the statistical properties computed by the SRIM program (www.srim.org). Tests are in progress. - CMS activation studies with Fluka: o RV currently on the learning phase to use Fluka with a CMS crystal as example. o WP is trying to import the full CMS geometry description through Flugg. This should help in keeping the CMS model up to date - it is not expected to be a fully automated approach since it is likely that some materials will have to be added which are important for activation calculations but not for tracking. Further, for activation calculations, details of the read-out channels are irrelevant. A much simplified setup could therefore provide accurate results in far less time. --------------------------------------------------------------------- ==== AOB Next meeting scheduled for December 12th.