Download presentation
Presentation is loading. Please wait.
Published byKellie Small Modified over 8 years ago
1
G.Corti LHCb Software Week - 6 October 2009 Gauss Status MC09 addition over the summer and what is still expected development for first data generators status validation infrastructure and tuning of the simulation short summary by Patrick Robbe Silvia Miglioranzi
2
LHCb Software Week - 6 October 2009G.Corti Page 2 Gauss and MC09 Gauss v37r* series and DecFiles v18r*-v19r* Gauss: v37r0 [2009-04-08] v37r2 [2009-06-03] v37r5 [2009-10-06] DecFiles: v18r0 [2009-05-27] v18r1/4 [2009-06-10] v19r1 [2009-10-06] Huge amount of events produced 10 9 (+) minimum bias few hundred millions signal events (b, c, Z, Higgs. etc…) Some problems in productions few starts and stops at the beginning –and sometime for a new event type… : must check before request problem still exists: jobs stalling or failing –few events lost (G4Exception) or particle lost (zero step and ping pong) –many jobs in v37r3p1 failing at finalization likely memory corruption; to be investigated with valgrind Infrastructure in place to report (un)known problems and summarize them for production –basic elements in place and under test in production (S.Miglioranzi) –merging for production under development (V.Romanovski)
3
LHCb Software Week - 6 October 2009G.Corti Page 3 Checking errors and stability At job level production run script to look for Gauss specific errors and report where they occurred At production level merge and provided statistics
4
LHCb Software Week - 6 October 2009G.Corti Page 4 Gauss v37r*: summary Generators HepMC 2.03.09 CLHEP free Pythia 4.18.2 tuned to reproduce particle multiplicities higher c and b cross sections charmonia and bottomonia production Photos 215.2 LHAPDF 5.3.1 no differences for PDF used, but more available EvtGen (April 2009) merging of changes/developments since 2003 from BaBar, CDF, D0, CLEO, LHCb internal CP violation under study LHCb models validated Hiijng 1.383bs.2 some issues with 64-bit and support Particle Properties and Decay tables updated to PDG2008 Geant4 – 9.1.0p03 EM physics completely re-written Hadronic physics extended Physics lists provided as integral part Framework Pass from generator to Geant4 only particles which will interact with detector, that is to say particles with non-zero travel lenght. only special particles (quarks, strings…) are not saved in MCTruth make known to Geant4 only those not yet known and that travel in LHCb Possible to save all particles as MCParticles and by-pass Simulation phase Simulation Conditions and other controls Pass arguments separately to gaudirun.py Gauss sets appropriate tags in DDDB and SIMCOND for a production Choice of contents/Deployment for production in AppConfig Beam parameters sets as properties of algorithms/tools EventType and generator specified see Patrick’s talk
5
LHCb Software Week - 6 October 2009G.Corti Page 5 Gauss v37r3(p1), v37r4 and v37r5: newest features Introduced Gauss() configurable evolving, not all foreseen controls yet there kept two separate Generation and Simulation phases find(ing) optimal way of configuring it for various cases Generator stand alone vs. skipG4 vs. only reading files not clear where to put the monitoring sub-phases for Generation and Simulation –in top level Gauss sequence best for reading –in Generation and Simulation sub-sequence best when its sub-sequence is executed have segmented control of geometry and simulation response of the subdetectors –work in progress by V.Coco uses LHCbApp() in production set directly some options –LHCbApp().EvtMax documentation for existing control parameters and some of foreseen new one in a dedicated web page – https://twiki.cern.ch/twiki/bin/view/LHCb/GaussConf
6
LHCb Software Week - 6 October 2009G.Corti Page 6 Gauss() configurable options Try to make it easier for anyone to configure safely Gauss() for common or production needs and many others already there or to come
7
LHCb Software Week - 6 October 2009G.Corti Page 7 Gauss v37r3(p1), v37r4 and v37r5: newest features New or re-commissioned features in generation part for physics Continuous release of event types Fixes or adaptation to new EvtGen SUSY production with Pythia BcVegPy back in production for 3.5 TeV, 5 TeV and 7 TeV beam energy Fix bugs observed in production Spill-over available and working Generate spill-over in a single file and a single job Multiple generator sequences ( one per bunch crossing ) followed by corresponding simulation sequences freezes the sequencing, more difficult to execute other algorithms if one event is aborted in any of the bunch crossing the whole ‘event’ is discarded if event in a bunch crossing is empty only headers are written out Works for any simulation settings (beam, geometry, phase) Cannot be used for signals not extracted from min bias (eg. Higgs, W, Z, SUSY) would requite non trivial changes in the code Used for upgrade studies At the moment specify the list of spill-over paths may want to move to bunch spacing with intelligent choice of spill-over paths see Patrick’s talk
8
LHCb Software Week - 6 October 2009G.Corti Page 8 Gauss v37r3(p1), v37r4 and v37r5: newest features Short-cuts to help in switching off detectors RICHesOff.py, CaloOnly.py Simulation for Machine Induced Background Geometry of BLS and of RB84 beam line can be added LbMIB adapted to new sources format with time correlation new sources to be released for latest layout for 7 TeV ParticleGuns Repackaging of particle guns for production ParticleGuns LbPGuns all generator packages need options/Generator.py to have consistency for production system New features in particle guns for cosmics Simulation for Upgrade Gauss().Upgrade = true will set the appropriate options togheter with choice of database geometry and spillover Documentation Generators statistics available for MC09 and scripts also for private productions DecFiles packages now has ‘automatic’ release documentation including table with decfiles and if/when tested –http://lhcb-release-area.web.cern.ch/LHCb-release-area/DOC/decfiles/
9
LHCb Software Week - 6 October 2009G.Corti Page 9 AppConfig and simulation conditions for Gauss Simulation Conditions and Gauss configurations for production in AppConfig While they start with MC09 they are not specific to the MC09 production nor the detector described there. For the MC09 detector use with Conditions/MC09-....py For the Upgrade detector with Conditions/Upgrade-....py Web page with avalaible steering options available with explanations –https://twiki.cern.ch/twiki/bin/view/LHCb/GAppConfhttps://twiki.cern.ch/twiki/bin/view/LHCb/GAppConf etc…
10
LHCb Software Week - 6 October 2009G.Corti Page 10 Gauss development (post-)MC09 Put back or validate additional generators or features to the main stream only those that are (will be) requested or that have clear support EvtGen internal CP violation Hijing HiddenValley AlpGen Pythia8 Herwig (with Jimmy and MCatNLO) SHERPA New event types a continous effort as many new requests now Further features in Gauss() configurable flexibility with consistency checks eventtype and generator detectors on/off execution of different phases modify script to generate python options for event types in DecFiles - C++ generator to eventually replace Pythia6, move to latest version - C++ generator alternative to Pythia OR EvtGen, huge work by T.and J.Wisabi with help from authors For Higgs and exotica: requested for MC09 - In the past by Cambridge, still wanted ? - fix 64 bits issues see Patrick’s talk Provide integration and support but decrease active development for MC09
11
LHCb Software Week - 6 October 2009G.Corti Page 11 Gauss development post-MC09 Will become the main development from mid-October Move (again) to latest versions of external software Pythia 6.4.20.2, LHAPDF 5.7.0 GEANT4 9.2.p02 supported until end of 2010 9.3 beta release is out but… will only be released end of December HepMC 2.05 ? to be discussed/synchronized with other LHC experiments –likely not this year but is what authors of C++ Move to latest geometry Fix geometry problems observed in production VELO is looking into this Magnetic field latest maps for both polarity Beam pipe supports as in upgrade DDDB Change settings in simulations delta-rays will be on expect some modification may become necessary
12
LHCb Software Week - 6 October 2009G.Corti Page 12 Gauss development post-MC09 Move to latest Gaudi and LHCb should not be a major problem but requires testing and validation Move to new platform and compiler slc5 and gcc43 some problems still in the nightlies (not necessarly in Gauss…) MacOS Finish porting to Windows (at least partial?) FORTRAN issues and Geant4 9.3 only on Visual Studio 9
13
LHCb Software Week - 6 October 2009G.Corti Page 13 Gauss development post-MC09 Setting and access of beam parameters At the moment only small number of parameters available in data (related to collision/bunch) and parameters are set as options of various tools New class will be introduced BeamParameters with all basic data set by a unique algorithm and derived methods to be used by tools. Migrate to new ParticlePropertySvc Generators statistics and FSR Move counters used in the generation phase of Gauss to obtain the efficiency of the Generator Level Cuts applied on the generated events necessary to compute event yields: Also other counters used to monitor generation quantities: – Cross-sections (bb, cc,...) –« Luminosity » (i.e. number of interactions per event) –Fraction of B*(*), of B0/B+/Bs/b-baryons Propagate the objects to later processing steps It would give the exact numbers for the subset one may be looking at Remove post-processing of log files with python scripts could just use it for web publication after a production No manpower identified so far
14
LHCb Software Week - 6 October 2009G.Corti Page 14 Gauss development post-MC09 Modify how to pass geometry from LHCb TDS to Geant4 geometry To take into account information from “alignment” in SIMCOND for all detectors To allow use of parallel geometry for scoring To enable possibility to check overlaps at construction Major work, complete re- engineering of GiGa geometry conversion Should restart in November when Marco Pappagallo is back
15
LHCb Software Week - 6 October 2009G.Corti Page 15 Gauss validation and tuning Many tools introduced to check, test, validate and tune the simulation Essential at all stages regression tests checking production porting to a new platform validating a new version (or a new) generator soon to compare with data and make sure it reproduces what observed and if not why Technical and regression QMTests in nightly builds and SAM tests to check installation Geometry Visualization of geometry and overlaps as seen by Geant4 –geometry not quite the same as what in DDDB+SIMCOND –dedicated Gauss job with different visualization options Scans of radiation length as seen by Geant4 with publication on web page Generator and physics lists (or in other words models for particle transport) Some reference samples indentified –particle guns, min bias and b inclusive –review “signal” reference samples Specialized Data Quality infrastructure (Online presenter and OMAlib) for Gauss –Need to identify relevant distributions for generator and simulation –Web report see Silvia’s talk
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.