1. LHCb Simulation LHCC Computing Manpower Review 3 September 2003 F.Ranjard / CERN

2. 03/09/2003LHCb Simulation2 Contents Introduction First studies using Geant4 The Gauss Project Description Manpower Status & Validation Gauss in 2004 Participation to LCG Application projects Manpower needs 2003-2007

3. 03/09/2003LHCb Simulation3 Introduction Currently the LHCb Simulation program SICBMC is a Fortran Geant3-based program. DC 2003 has been done with SICBMC. SICBMC will be retired end of 2003. DC2004 will be done with Gauss a C++ Geant4- based program.

4. 03/09/2003LHCb Simulation4 First studies using Geant4 In 2000 Sajan Easo/RAL started a standalone project to study the feasibility to use Geant4 to simulate RICH setup. In addition to existing Geant4 processes he had to develop specific processes: Photoelectric process: creation of photoelectrons in HPDs. Energy loss: in the silicon of HPDs. 2001Test beam setup was simulated with this program and results presented at Geant4 workshop in 2002. In 2000 D.Rusinov/ITEP made some comparison of ECAL test beam data with Geant3 and Geant4.

5. 03/09/2003LHCb Simulation5 Gauss project The Gauss project started end of 2001, the goal was: replace the Fortran SICBMC program be in time for the Computing TDR Gauss is a Gaudi application: It uses all services provided by Gaudi: ROOT I/O, Histogram facility, Detector Description facility, Message service, etc… It is flexible to accommodate the full apparatus or a test beam setup.

6. 03/09/2003LHCb Simulation6 Gauss: 2 steps Event Generation Algorithms Detector Simulation Algorithms (Geant4) Digitizing Algorithms Generator Event Record (HepMC) MCParticles MCVertices MCHits Digits MCDigits ROOT W.Pokorski

7. 03/09/2003LHCb Simulation7 Event Generation for primary generation: interface to Pythia 6.x and SingleParticleGun (for tests) for B-decays: interface to EvtGen (from lange@slac.stanford.edu & ryd@hep.caltech.edu ) lange@slac.stanford.eduryd@hep.caltech.edu NOTE: all of the algorithms are using RndmGenSvc and ParticlePropertySvc from Gaudi spill-over handling at the level of digitization. Pileup in the same bunch crossing is handled at generation time. structure easy to extend in case other generators (e.g. Herwig) were needed.

8. 03/09/2003LHCb Simulation8 Detector simulation Sensitive Detectors G4 Physics Lists LHCb Event Model Digitization Algorithms Detector Data XmlDDDB GiGa HepMC Pythia6.205 EvtGen Magnetic Field Svc W.Pokorski

9. 03/09/2003LHCb Simulation9 GiGa Development of a Geant4 interface to Gaudi started in 1999: a first prototype was available in 2000. It converts the geometry available through DetectorDataSvc to Geant4. It interfaces the Magnetic Field available through MagneticFieldSvc to Geant4. It converts HepMC data to Geant4. It converts Geant4 hits to LHCb Event data model. It allows modular physics lists It calls specific sub-detector code when traversing a sensitive detector. It is part of the Gauss framework.

10. 03/09/2003LHCb Simulation10 Gauss framework: manpower in 2003 Tasks to cover: Overall framework and coordination Event Generator interface GiGa interface Physics lists handling Sensitive detector handling Tracker hit converter 1.3FTE provided by: 1 CERN fellow (contract ends Dec.2003) 1 project associate

11. 03/09/2003LHCb Simulation11 Subdetector participation in 2003 Subdetectors are responsible for: Detector geometry description in XML Geant4 specific actions to the detector. Specific Converter from G4Hits to LHCb MCHits. Validation. 4.9 FTEs Detector:Contributor: VeloLiverpool RICH 1RAL Inner TrackerZurich Outer TrackerNikhef RICH 2RAL ECAL, HCAL, Spd, PrsLAL, ITEP, INP, INR MUONRome

12. 03/09/2003LHCb Simulation12 Gauss status : August 2003 All detectors are included in the setup. Full Pythia events are simulated and G4 Hits converted to LHCb event data model. Output data files are stored on CASTOR for future used by BOOLE the digitization application. Summer test-production, with DIRAC at CERN, just finished: 120000 minimum bias events produced in 30 jobs (6 crashes due to a known problem being discussed with the G4 team). Performance: Gauss is about a factor 2 slower than SICBMC but tracking of photoelectrons was not included in SICBMC.

13. 03/09/2003LHCb Simulation13 Gauss validation We should distinguish between “validating” Gauss (i.e. proving that it can replace SICBMC) and “tuning” Geant4 physics settings. Validating Gauss each subdetector should come up with a set of “benchmarks” validating Gauss velo, trackers, muon – comparison with SICBMC multiplicities energy deposition ToF, etc Done : in good agreement. HCal – some study of different physics list done in 2002. ECal/Spd/Prs – comparison with SICBMC ongoing

14. 03/09/2003LHCb Simulation14 Gauss validation (2) Tuning Geant4 physics settings It requires test beam data. So far only the RICH has made such analysis validation of different RICH-specific process (Cherenkov, Rayleigh scattering, etc) using test beam data

15. 03/09/2003LHCb Simulation15 Red lines: Charged particle Green lines : Photons. Transmission in Aerogel, Cherenkov radiation, Rayleigh scattering, etc, simulated properly S. Easo A Typical event in the RICH Testbeam

16. 03/09/2003LHCb Simulation16 Single Particle Gun: 80 GeV/c Pions. Selected Tracks with Beta > 0.99 Mean in SICBMC= 23.2 for Beta ~ 1.0 GAUSS uses segmented mirrors, causing a small loss wrt. SICBMC. Mean= 23.0 Hits from CF 4 in GAUSS

17. 03/09/2003LHCb Simulation17 Hits From RICH1 Charged Tracks: 7 GeV/c Pion And its secondaries. Rayleigh Scattering On. C=0.00719  m 4 /cm A=0.9368 for 4 cm. Ionization And Multiple Scattering on.

18. 03/09/2003LHCb Simulation18 Forthcoming validation RICH – new test beam in October (new gas radiator, new version of photodet.) Calorimeters (Spd/Prs/Ecal/HCal) - test beam periods this summer between the middle of June and end of August Results expected late Autumn 2003

19. 03/09/2003LHCb Simulation19 Gauss in 2004: Necessary Improvements Event generation Pythia tuning and EvtGen decay files Framework strategy of saving trajectories (MCParticles) a lot of influence on CPU time performance, disk space. production of useless particles ( ionization of air ) tracking of low energy particles looping in magnetic field conversion of useless G4 tracks to MCParticles Keep vertex types. Physics lists - selection of right processes with right cuts for each region Subdetectors Physics validation

20. 03/09/2003LHCb Simulation20 Framework : manpower in 2004 Tasks: Coordination Development: GiGa, Physics lists, tuning Documentation Job profile: A physicist with good knowledge of C++ (estimate 1.6FTE) Post occupied by a CERN fellow (contract ends Dec 2003) Will be replaced by existing manpower (find a replacement) Need a second person (CERN fellow?)

21. 03/09/2003LHCb Simulation21 Event Generation: manpower in 2004 Tasks: Coordination, Studies and tuning of different event generators. Coordination, tuning, maintenance of EvtGen, writing of B-decays files Job profile: Physicists with great experience (estimate 0.3 FTE) Post-graduate physicists (estimate 0.3FTE) Bristol and LAL-Orsay will contribute to fulfill these posts.

22. 03/09/2003LHCb Simulation22 LCG Simulation projects Physics validation lead by F. Gianotti http://lcgapp.cern.ch/project/simu/validation/ Participation to regular meetings Simulation framework lead by A.Dell’Acqua http://lcgapp.cern.ch/project/simu/framework LHCb contribution In 2003: 0.2 FTE from CERN fellow (contract ends Dec. 2003)

23. 03/09/2003LHCb Simulation23 Manpower needs: 2003-2007 20032004200520062007 Framework1.6 (1.1)1.61.1 Event generation0.20.60.3 Subdetectors5.4 (4.9)7.25.0 6.5 Velo0.8 Silicon Trackers0.3 Outer Trackers0.6 0.4 RICHs1.5 Calorimeters1.22.00.5 Muon Chambers1.0 (0.5)2.01.5 2.0