S. Miglioranzi (CERN) 37th Software Week

1. Validation infrastructure and tuning of the simulation S. Miglioranzi (CERN) 37th Software Week • -rays • Multiple Scattering in MUON system • Validation tools for the Simulation • G4 Material Scans • Geometry bug fix in VELO • QMTests

2. -rays simulation status For MC09 the -rays stay turned OFF (currently only switched ON for RICH Aerogel). In future productions (from Gauss v38r0) the sub-detectors should be prepared to have the -rays turned ON -> more realistic simulation. From Simulation Applications meeting - 24.04.2009: • VELO: velo group would like to have some data to check the feasibility of -rays activation. Studies about resolution and occupancy needed. • OT:no straws simulation. -rays activation cut works on Region-basis -> OT may need to be defined inside a Region if cannot cope with -rays ON. • ST: no difficulties foreseen to have them ON. • RICH: no difficulties foreseen to have them ON. • CALO: no difficulties foreseen to have them ON (only Energy calibration will change) but check if and how the resolution will be affected and variation of CPU time needed. • MUON: no difficulties foreseen to have them ON (may have to change the low Energy background parametrization).

3. Multiple Scattering in MUON system • Muon trajectories are dominated by multiples scattering interactions in the Calorimeters and Muon Filters • The MCS in G4 is not correctly simulated in case of dense material and large step sizes (MUON Filters are a perfect example!) -> correlation between displacement and angular deviation not maintained (see details in Software week 17/03/09) • The activation of -rays forces G4 to reduce the step length -> correct description of correlation (side effect) • Include correlation plots in simulation monitoring... • need to investigate in trackers as well

4. Multiple Scattering in MUON system Simulation Time: • Default : 1 • -rays-LT.opts : 1.4 • -rays-HT.opts : 1.2 • Default + -rays only in MUON Filters: 1 S. Vecchi - Simulation Application meeting 08/05/2009 • MCS description in G49.1p03 (newG4MuMscModel) slightly improved w.r.t. G4 8.2 (G4MuMultipleScattering), without considering the -rays. • Still correlation between displacement and angle spoiled if -rays are not switched ON • Gauss v36r2 • PDG

5. Multiple Scattering in MUON system S. Vecchi - Simulation Application meeting 08/05/2009 New Multiple Scattering Simulation Checking packages: • in Muon/MuonMuoniSim (MultipleScatteringChecker.cpp(h) - 2D histos MCS) – available in v37r2 • in Sim/GaussMonitor (MonitorStepAction.cpp(h) - step length in Muon Filters histos) – not yet included in a release • RMS(dx).vs.p still slightly underestimated respect PDG by a factor ~0.5-0.6 • Default • -rays (HT/LT) • -rays in MUON Filters • PDG

6. Multiple Scattering in MUON system • Default • -rays (LT) • -rays (HT) • -rays in MUON Filters S. Vecchi - Simulation Application meeting 08/05/2009 • Switching ON the -rays only in the MUON Filters -> no variation in the number of hits. • MCHits monitoring plots need to be extended to all subdetectors

7. Development of general MC monitoring and validation tools • First version of monitoring tools already available in: http://lhcb-release-area.web.cern.ch/LHCb-release-area/DOC/gauss/generator/validation/index.html http://lhcb-release-area.web.cern.ch/LHCb-release-area/DOC/gauss/simulation/validation/index.html • Python scripts to analyse the Gauss log files and rootople • Comparisons of different Gauss versions and different cuts • web pages with monitoring plots/tables are created

8. Development of general MC monitoring and validation tools What’s next: • Set of general scripts to monitor the MC output (for Gauss, Boole, Brunel) during production. • Unique tools for monitoring production, validation and tuning • Error handling: run over the full production output to update counters and identify which and how many jobs crashed • include some info (generator statistics) in File SummaryRecord (DST) without the necessity to scan the log files (discussion this Friday at the Core Software session) • Possibility to choose from cumulative or “from a certain date” information (histo, counters, errors...) • Histogram handling:histos will mantain the Tree structure of the directories in the web page(easier visualization) • Generator statistics scriptswill be included to be run (once a week, cron jobs?) • use already existing scripts (e.g. bookeeping scripts) • investigate what can be taken from the DQM

9. LHCb material budget (G4 version) http://lhcb-release-area.web.cern.ch/LHCb-release-area/DOC/gauss/simulation/matscan/matscan_gen.html Material budget plots for geometry: • DDDB=”head-20090112" LHCbCOND="head-20090112" • DDDB="MC09-20090602" LHCbCOND="head-20090402" MC09 • Stand-alone version DDDB as in: • https://twiki.cern.ch/twiki/bin/view/LHCb/CondDBHowTo#Generate_XML_Files_from_the_Cont • PGUN (MaterialEvalGuns.opts) used: originating at (0,0,0) in LHCb ref system, generated in a -grid extending over the whole LHCb acceptance. • Fake-Detector structure-geometry created modifying the XML (Air planes positioned in subdets gaps) • Radiation length at planes summing contribution at each G4Step (StepLength/MaterialRadiationLength) Thomas’ page: http://lhcb-reconstruction.web.cern.ch/lhcb-reconstruction/Python/Radiation_and_Bfield_maps.htm

10. LHCb material budget (G4 version) http://lhcb-release-area.web.cern.ch/LHCb-release-area/DOC/gauss/simulation/matscan/matscan_gen.html G4Step does not “see” the overlapping volumes, problem avoided for the moment switching ON/OFF the simulation of specific subdets from SimGeometry.opts (still remaining issues with RICH). Future: implementation of a G4 Parallel Geometry (feasible from G49.1) in order to create manageable scorer planes. • BeforeMagnetRegion: 2710 mm (2) • Magnet Region: 7570 mm (3) • After Magnet Region: - 8250 mm (4) - 8950 mm (5) - 9450 mm (6) - 11950 mm (7) - 12320 mm (8)

11. LHCb material budget (G4 version) http://lhcb-release-area.web.cern.ch/LHCb-release-area/DOC/gauss/simulation/matscan/matscan_gen.html Radiation Length at Z = 2710 mm (full Before Magnet Region - plane 2 ) The radiation length in each eta (phi) bin is averaged in phi (eta). 0.3 X0 0.4 X0

12. LHCb material budget (G4 version) VELO From: http://lhcb-release-area.web.cern.ch/LHCb-release-area/DOC/gauss/simulation/matscan/matscan_gen.html the material scan plots (cumulative and plane2plane) for all the planes available. T3 RICH2

13. Gauss “particle stuck” bug Problems observed: • Gauss (v37r0) jobs hanging in production (up to 20%) • or n of few MeV ping-ponging between Universe and VELO tank upstream (protection has been introduced to kill this particles) • jobs stucked in production with error like: WARNING - G4PropagatorInField::ComputeStep(): Zero progress for 51 attempted steps. • charged particles of zero steps (G4 pushing them) in the same volume (protection introduced) Investigation with G4 people. Problem found: • due to precision problem there is an overlap between the volume (a Polycone) and its mother. • loss in precision of one of the planes of the Polycone Problem solved: • geometry modified to introduce a small gap (few m) between the Polycone and its mother • corrected plane parameter to avoid precision loss

14. Gauss “particle stuck” bug • likelysimilar problem in Velo2Rich (the whole event is aborted as the problem is caught by a different G4 class) • This problem was not detected by the overlaps checking tool https://twiki.cern.ch/twiki/bin/view/LHCb/GeomOverlap because dependent on optimizer. Moreover overlap-checking not working for polycones volumes (?). • G4 crewdetected the overlap with the overlap check in G4PVPlacement (... , G4bool pSurfChk=false) constructor. Optional boundary check to activate when new volumes introduced. • Could be set as option when new LHCb->G4 geometry will be put in place

15. Gauss QMTests Gauss (from v37r0) has introduced a set of QMTests to be run in the nightlies: • Default QMTest running over few MinBias events • SAM test for checking the installation on remote sites, running one event as the G4 geometry (hence the access) is done during the first event. • A generator-only test running over 1000 events (B0->KS0J/psi(1S)). • in preparation: more generator-only tests (to test different signals) running on few events

16. Summary • Sub-detectors need to be prepared to turn the simulation of -raysON (not for MC09 but sometimes this year, before the data taking) • MCS: updated correlation studies using recent Gauss version have been produced (S. Vecchi) and set as part of the routine simulation validation • New Simulation monitoring scripts are foreseen to improve the monitoring of MC production. • New Material Scans using G4 have been produced (code will be included in next Gauss releases) • Gauss production bug (geometry overlap) found and solved. • QMTests for Gauss now available to be run in nightlies. • Geant4 9.2p01 is going to be built by the end of June (now included in slot lhcb2 in the nightly builds) and a Gauss development version (v38r0) prepared.

17. MCS simulation in G4 • The activation of -rays forces G4 to reduce the step length -> correct description of correlation (side effect) • for the MSC starting from g4 9.0 or 9.1 possibility to control the step size but tests needed to understand how to use them inside Gauss/GiGa • -rays will be turned ON. In the future the “activation cut” on the Electron range (5 mm, see previous slides) can be optimized (larger cut) in order to save CPU time. • Include correlation plots in simulation monitoring M3 M3 S. Vecchi -rays OFF -rays ON