Monte Carlo tuning using ATLAS data Davide Costanzo (on behalf of the ATLAS collaboration) 1MonteCarlo tuning using ATLAS data23/08/2011
Monte Carlo simulation process The Monte Carlo simulation of the ATLAS detector is used in most ATLAS papers/results Event generators: Pythia6, Pythia8, Herwig, Herwig++, Alpgen, AcerMC Powheg, Sherpa... AtlasG4 simulation Reconstruction, Analysis, Systematics Data Reconstruction, Analysis Data/MC comparison example ATLAS-CONF Search for Supersymmetry with MET, bjets and no leptons 2MonteCarlo tuning using ATLAS data23/08/2011 Tuning a combination of event generator and simulation parameters
Event Generator tunings ATL-PHYS-PUB ATL-PHYS-PUB PYTHIA 6 is used as the main general-purpose event generator in ATLAS. The tune is performed in four stages: 1) Flavour parameters tuned to hadron multiplicities/ratios, from e+e- collisions 2) Final State Radiation (FSR) and hadronisation parameters, tuned to event shapes and jet rates from e+e- collisions 3) Initial state shower parameters and primordial kT, tuned to Tevatron and ATLAS data 4) Multiple Partonic Interaction (MPI) parameters, tuned to Tevatron and ATLAS data For HERWIG/JIMMY, only the MPI parameters were tuned: The inverse proton radius squared [PRRAD] The MPI cut-off at s = 1800 GeV [PTJIM0] The MPI cut-off evolution [EXP] Energy evolution of the MPI cut-off was added to HERWIG/JIMMY MPI model, to allow simultaneous tuning to 7 TeV and 900 GeV ATLAS UE data. 3MonteCarlo tuning using ATLAS data23/08/2011
Pythia 6 tune ATL-PHYS-PUB MonteCarlo tuning using ATLAS data23/08/2011 AUET2B tuning to ATLAS data. To be used for 2011 analyses AUET2B (CTEQ6L1) main Pythia 6 tune for ATLAS MC11
Pythia 8 tune ATL-PHYS-PUB MonteCarlo tuning using ATLAS data23/08/2011 Pythia8 has a better diffractive modelling than Pythia 6 ATLAS A1 and AU1 tunes to s=7TeV minimum bias and underlying event data For ATLAS MC11
Herwig/Jimmy tune ATL-PHYS-PUB MonteCarlo tuning using ATLAS data23/08/2011
The AtlasG4 simulation EPJC 70 (2010) 823 Over 10 9 events simulated with AtlasG4 so far A few examples of AtlasG4 tuning using data in the next few slides 7MonteCarlo tuning using ATLAS data23/08/2011
Inner Detector material validation Use photon conversions in the Inner Detector to map the material distribution: Conversion rate (in colour) for R vs z The beam pipe, the pixel barrel and part of the SCT detectors are visible 8MonteCarlo tuning using ATLAS data23/08/2011
Inner Detector material validation (2) Conversion rates for fixed eta: Small discrepancies still visible, geometry description in continuous evolution 9MonteCarlo tuning using ATLAS data23/08/2011
Inner Detector hadron-graphy Reconstruct secondary vertices inclusively and select those arising from secondary hadronic interactions MonteCarlo tuning using ATLAS data Cooling Pipe, Cables, Carbon Fibre shell Details of modules in 1 st Pixel layer. 1023/08/2011
Calorimeter response to single hadrons ATLAS-CONF Measure the calorimeter response to isolated tracks in pp collisions: E=Energy deposited in the calorimeter p=track momentum E/p for isolated tracks (2.2<p<2.8 GeV) Translates into a jet energy scale uncertainty of 1% to 3% Pythia AMBT1 tune Geant4 QGSP_BERT Physics list (Central region)
Calorimeter response to single hadrons (2) ATLAS-CONF Use Ks -> π + π -, Λ-> π - p and Λ -> π + p to measure response to pion, protons and anti-protons Response to anti-protons underestimated by MC (Only few % of the total energy in a jet) Pythia AMBT1 tune Geant4 QGSP_BERT Physics list (Central region)
Jet shape measurement PRD Differential jet shape ρ(r): Average fraction of jet PT in an annulus or radius r Jets are composed by hadrons. The distribution of the hadrons within the jet depends on the parton-jet fragmentation process ATL-PHYS-PUB
Conclusions 14MonteCarlo tuning using ATLAS data23/08/2011 Simulation is a very important component of the ATLAS physics programme Different components need to be tuned and validated. Tuning is a cyclic process : - Event generators - Geant4 MonteCarlo - ATLAS detector response A good agreement is achieved between the ATLAS data and simulation With more data available small discrepancies become visible resulting in a continuous improvement of the ATLAS Monte Carlo strategy
Response to electrons Use J/Psi and Z decay to study the response to electrons
Response to muons Use Z->μμ events to study the response to muons. The Z mass reconstructed width, and hence the muon momentum resolution is underestimated in the simulation