Presentation on theme: "Simulation Developments"— Presentation transcript:
1Simulation Developments Choice of few topics to have production version(s) for the 2010 data in autumn and longer term developmentsfocus on Generators, Geant4 PhysicsLHCb Week in St. Petersburg, 29th June 2010
2LHCb Week in St. Petersburg, 29th June 2010 Generator TuningPythia6 not expected to evolve significantly but its tuning (to LHC data) yesUse other generators for specific events, e.g. Pythia8 for diffractiveImportant to understand the MinBias (most part of the events are extracted from them)Need to have soon a proper Pythia tuning (comparison of key distributions using LHCb tune, Perugia-0 tune, Atlas tunes...)Contribution to the LPCC MB and UE working groupA first attempt to review ongoing projects for the MC tuning (from generators and validation tools to first comparisons/plans for physics results) at the simulation workshop on Jun 2nd.Report on available tools and first idea by Silvia Miglioranzi at June software week.
3Generator tuning Activity starting in LHCb LHCb Week in St. Petersburg, 29th June 2010Generator tuningActivity starting in LHCbfew people looking into Minimum Bias tuningdefining appropriate data sample and MC samples to comparegeneral reasonable agreement between data and MCDavid Voong presentationidea on b production studies and tuningRolf Oldeman presentationB/B*/B** can be tuned in Pythia, affect Flavour TaggingNot much room in Pythia to tune B spectra expect changing fragmentation function parameters
4Validation and tuning tools LHCb Week in St. Petersburg, 29th June 2010Validation and tuning toolsSupport for tuning as joint effort between LCG Generator Service and the LPCC for tuning tools supportRepository of toolsProfessor, Rivet, HepMC Analysis ToolsWeb resources for validation of different tunesIn LHCb Online tools customized to perform the MC Data Quality Monitoringby Silvia MiglioranziOnline Histogram DataBase, Online Presenter and OMAlib (Online Histogram Analysis Tools)Define and monitor key reference distribution in GaussFollow them all the way to compare with data as to feed tuning for minimum bias and beauty/charm production into generatorsDetailed instructions on how to use the Online Tools for MC Monitoring providedhttps://twiki.cern.ch/twiki/bin/view/LHCb/McMonitor
5Validation and tuning tools LHCb Week in St. Petersburg, 29th June 2010Validation and tuning toolsExample of comparison of 2 small samples - 100evts minbias each - using LHCb (ref) and Perugia-0 tune
6LHCb Week in St. Petersburg, 29th June 2010 EvtGenA merge of features developed by BaBar/Belle/CDF/LHCb since 2003 adopted last yearNeeded to re-implement some of the modified to handle the situation at LHC(b)Different description of Mixing and CP-violationHandle the fact that Pythia is used to produce the (wider variety) of b-hadronsNew models can be added at run timeE.g. new model for rare radiative leptonic B-mesons decay recently introduce by Nikolai Nikitin and Daria SavrinaDecay files and EvtGen maintenanceDECAY.DEC last updated in 2009 with PDG 2008 BFsPlenty of new decay files using available modelsEvtGen is looked after within LHCb, by Paul Harrison and Mark WhiteheadHowever, the continuing development, validation and maintenance is also the primary responsibility of the physicist-users within the collaboration.Always test the generator for your mode before using it!Let them know of problems, and they will help you fix them.
7General issues Generators statistics LHCb Week in St. Petersburg, 29th June 2010General issuesGenerators statisticsPrinting of all steps in generation (selection, cuts, flipping) in log files. A dedicated script used to extract and compile informationMigrate to FSR (File Summary Record)similar to luminosity informationmanpower need to be identifiedCentral production for generator level studies ?Easy Gauss configuration exist to have only generator content in HepMC formatPossibility to have MCParticles/MCVertices filled with generator information exist: need to introduce easy Gauss() configurable controlNeed to define file types for them and make them known to the production system
8Generators: open issues LHCb Week in St. Petersburg, 29th June 2010Generators: open issuesBeam parameters for collisionsnow can be configured via options of many toolsdefine event data class as single stored source of informationHijing for beam gas generator for luminosity studiescurrent version not working on slc5LCG will install ALICE version in near futurewill need to pick it up and do some validation in our systemMachine induced background (far away)need new files of losses at 3.5 TeV on TCTestimates to be provided by LBSevent types to be definedmerging with physics events to be provided in Gauss() configurable
9LHCb Week in St. Petersburg, 29th June 2010 Future GeneratorsFortran generators are at their end-of-life. Authors will not update them anymore, all developments will be in C++ from now on.Very productive workshop on Jun 2nd 2010Summary Report by Patrick Robbe at June software weekInvited talks from authors of C++ generators:Pythia8: Peter Skands (CERN)Sherpa: Frank Siegert (Durham)Herwig++: Mike Seymour (Manchester)Photos/Taula: Zbiegniew Was (Krakow)Presentations from LHCb (interfaces to these generators):Pythia8: Raluca Muresan (EPFL)Sherpa: Julian Wishahi, Tobias Brambach (Dortmund)
10C++ generators Amongst C++ generators, 5 are interesting for LHCb: LHCb Week in St. Petersburg, 29th June 2010C++ generatorsAmongst C++ generators, 5 are interesting for LHCb:« Production »: Pythia8, Herwig++ and Sherpa« Decay »: Sherpa, Photos and Tauola (even though Pythia8 and Herwig++ have interesting decay engines, but which will not be as detailed as EvtGen or Sherpa)Advantages of C++:Easier maintenance inside LHCb software,Modular interface (can re-use easily other parts of software – random numbers, ...)Possibility to have 2 separate instances of the same generator: for example, one for signal, the other for pile-up.
11Pythia8 Status: All « LHCb » processes available: LHCb Week in St. Petersburg, 29th June 2010Pythia8Status:Ready and tuned for Minimum Biasmultiple parton-parton interactions, with an improved model (pT ordered, similar to the one we use with Pythia6)Diffractive description better than Pythia6Improved shower modelString fragmentation: Lund fragmentation for light quarks, with Bowler modification for c/b quarks[NB: this is the part where the B spectra may be tuned]But problem with underlying event (which would not affect the majority of events generated in LHCb)All « LHCb » processes available:All 2 2 processes that we use in LHCb with Pythia6Charmonium + Bottomonium prompt productionDiffractive + elastic
12Pythia8 in LHCb Interface available and used for physics studies LHCb Week in St. Petersburg, 29th June 2010Pythia8 in LHCbInterface available and used for physics studies(P.Robbe and student)(R.Muresan)Ks0 production studiesPythia6 model (not developed anymore)Pythia8 model (actively developed)3.0 < y < 3.5
13LHCb Week in St. Petersburg, 29th June 2010 SherpaMulti-purpose generator, which can be used as a production generator, but also as a decay generator.Multiple interaction model similar to Pythia, parton showers kT orderedSome QCD NLO matrix elementsCluster fragmentation model for hadronizationNo complete generation of Minimum Bias yet:Missing elastic scattering, single diffraction, double diffraction.Under development, but soon availableImplementation will aim at complement underlying event simulation,2 2 QCD matrix elements,Goal is to have a complete description of QCD,Work partly motivated by LHCbDetailed hadron decay module, including generation of QED corrections for decayAuthors have plans to develop a large number of different decay models to describe D, B, and tau decays.
14LHCb Week in St. Petersburg, 29th June 2010 Sherpa in GaussInterface available, both for production and decay tool.A lot of interaction and support between authors and Julian Wishahi and Tobias Brambach who are integrating Sherpa inside Gauss.Some parts of Sherpa have been tuned for LHCb technical needs.To be released in near futureWorking already fine to generate inclusive b and signal eventsDecay kinematics generated according to amplitudes (and form factors)Allow spin correlations and correct angular distributionsTreatment of neutral B mixingAllow generation of signal sample by forcing particle to a given decay modeDecay tables for ~400 hadrons, ~2500 decay channels, ~400 decay modes with specific amplitudesFew remaining issues for LHCb decaysCP violation in signal B not yet implementedMass smearing problem for broad resonances,technical problem due to the separation in Gauss of production and decay
15LHCb Week in St. Petersburg, 29th June 2010 Sherpa in Gauss
16LHCb Week in St. Petersburg, 29th June 2010 Herwig++Multi-purpose event generator, successor of Fortran HERWIG generator.Fortran version development stopped, C++ version is now the main one.Current C++ version provides a full simulation of hadron-hadron collisions, with many improvements with respect to Fortran version:Angular ordered parton shower,Many processes at NLO,Gives a good description of B hadron fragmentation function.Improvements in hadronization, designed to improve the simulation of bottom and charm hadronsflavour specific parameters in hadronization (not available in Pythia)Improved simulation of hadron, tau and B decays.General properties of heavy mesons and baryons: no CP violation or mixing
17Herwig++, Event ShapesB hadron fragmentation function compared to SLD/LEP data.
18LHCb Week in St. Petersburg, 29th June 2010 Herwig++ in GaussAn old interface to Fortran Herwig, but not maintained, exists in Gauss.Interest from Cambridge group to interface Gauss to Herwig.Development will consider directly C++ Herwig++ version (and leave Fortran Herwig).Herwig++ authors are willing to help and provide support when implementing Gauss interface to Herwig.
19LHCb Week in St. Petersburg, 29th June 2010 PHOTOSUsed in LHCb to generate QED radiation from all decays to charged particles.Is called from EvtGen, we use for the moment FORTRAN version.C++ version under development, will be released end of summer.Will be able to call it from outside EvtGen:PHOTOS will read a HepMC event record with its full structure.It will then add automatically all necessary radiation photons to decay of particles.Then we will be able to have it as a « universal » tool inside LHCb generation software.
20Geant4 and Detectors Physics processes in Gauss/Geant4 LHCb Week in St. Petersburg, 29th June 2010Geant4 and DetectorsPhysics processes in Gauss/Geant4are our choices of physics list and cuts adequate?Comparison with dataonce we are comparing apples with apples (i.e. the same class of events) do we understand where discrepancies come from?
21Geant4 hadronic physics LHCb Week in St. Petersburg, 29th June 2010Geant4 hadronic physicsHadronic’s cross sections and physics listsStudy by Silvia Miglioranzi on cross sectionsall physics lists looked at have the same cross sectionsbut different multiplicities (study in progress)New physics list prepared by Geant4 for LHCbbetter cross section forreference tag last week Geant4 9.3.ref05we are currently using Geant4 9.2.p03development version of Gauss in nightlies to be based on this to allow its investigation over the summer
22Geant4 EM physics EM ‘old’ issues to be checked and revisited LHCb Week in St. Petersburg, 29th June 2010Geant4 EM physicsEM ‘old’ issues to be checked and revisitedDelta rays switched on everywhere: is OT simulation fine?Multiple scatteringlateral displacement in muon filters much better (side effect of switching on delta rays) but need to be re-checked with latest versionslopes pull issue in tracking still there? New MS model used since DC06, but effect not verifiedProduction and tracking cutsare they responsible for some of the MC/data difference in clusters multiplicities?one of the major influence on CPUReview dE/dx in thin layersit has been getting better, again check latest versionare muons still different from other particlesCalorimeters calibration done last for MC09 (G4 9.1 now at )Muon low energy backgroundnot applied: bunches now very far away but neutrons component can be very slowlast parametrization for 14 TeV collisions…
23Detector geometry and conditions LHCb Week in St. Petersburg, 29th June 2010Detector geometry and conditionsRICH refractive index calibrationfrom DDDB to SIMCOND (and LHCBCOND) to be regularly updatedas for dead strips need to define which set to use for massive productionMissing materialVelo recently updated RF foil thicknessSome parts of the cable supports in the beam pipe not yet described nor its outer support frameMaterial that should NOT be therePass detector elements one by one to take into account alignmentNeeded to VELO open/close BUT Velo-Envelope with vacuum not thereRestructuring of how geometry is passed to G4 needed to fix this
24Summary Simulation is an essential ingredient of the analysis LHCb Week in St. Petersburg, 29th June 2010SummarySimulation is an essential ingredient of the analysisMake sure it matches as much as possible what we measure and it is stabletools in place to helpgenerators tuningappropriate physics choice and cuts in G4material is well describedMake sure it is ready for the futuremake it easier to use for private studies(adopt), commission and follow evolution of new (C++) generatorsthey can already suit LHCb needsa lot of interesting work needed to validate them in LHCbfollow Geant4 and detectors evolution