1 / 37

Preliminary Oral Exam

Preliminary Oral Exam. Aram Avetisyan Brown University February 25, 2009. Outline. Part 1 : Validation of HCAL Software CMS in general and its hadronic calorimeter in particular Software used in CMS Details of hadronic calorimeter validation Part 2: The T 5/3 Top Partner model

Télécharger la présentation

Preliminary Oral Exam

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Preliminary Oral Exam Aram Avetisyan Brown University February 25, 2009

  2. Outline • Part 1: Validation of HCAL Software • CMS in general and its hadronic calorimeter in particular • Software used in CMS • Details of hadronic calorimeter validation • Part 2: The T5/3 Top Partner model • Description of model • Fast Simulation and the Physics Analysis Toolkit in CMS • Event selection • Charge misidentification study • Reconstruction of T5/3 invariant mass

  3. ECAL Magnet HCAL Tracker Muon chambers The CMS Experiment • Compact Muon Solenoid (CMS) Experiment • One of two general purpose experiments at the Large Hadron Collider (LHC) • Compact: 15m in diameter, 21.5m long, weighs 12500T

  4. The CMS Hadronic Calorimeter (HCAL) • HCAL is designed to measure the kinematics of jets and missing transverse energy • Brass absorber plates interspersed with plastic scintillator • The section closest to beam pipe is made of steel embedded with quartz fibers • Steel on inside and outside to provide support

  5. HCAL Subdetectors • Four subdetectors covering pseudorapidity ( ) up to |η| = 5: • Barrel (HB): |η| < 1.4 • Endcap (HE): 1.3 < |η| < 3.0 • Forward (HF): 2.9 < |η| < 5.2 • Outer (HO): |η| < 1.26 • HO is outside the magnet Inside the magnet η = 1 η = 0.5 η = 1.4 η = 2.1 η = 3 η = 2.4

  6. Beam Splash Event • Splash from 2E9 protons impacting a collimator about 150m upstream of CMS • HCAL is blue

  7. HCAL Software Objects SimHits • Simulation is done with GEANT4 • SimHit: Sum of scintillator energies for each cell • Also stores time of hit, subdetector and location of cell • Simulation only • Digi: Signal from one readout channel • Stores cell identification info and up to 10 integers corresponding to charge deposited in a 25 ns time bin • RecHit: Stores energy in GeV, time of hit and cell identification info • CaloTower: 1 HCAL cell + 5x5 ECAL crystals • Used for High Level Trigger which doesn’t have time to process all ECAL and HCAL cells Digis RecHits CaloTowers ECAL

  8. HCAL Software Validation • CMS Software Framework (CMSSW): the entire collection of software used for simulation, calibration, reconstruction, analysis, etc. • Used for both processing both Monte Carlo and real data • Constantly in development • Pre-release about once every 2 weeks • Release about once every 2 months • Validation for HCAL: compare all software objects to last good CMSSW version • Quick validation: TTbar and QCD • Full validation for known major changes, new production release or unexpected results from quick validation

  9. Monte Carlo Samples for Validation • TTbar and QCD • 50 GeV π+ shot at 3 pairs of η, φ towers • Corresponding to HB, HE and HF • 500 GeV π+ randomly distributed in |η|<1.3 and all φ • Energetic particles needed to reach HO • Neutrinos for noise validation • Generation software needs some particle to run • Simulation runs quickly • 50 GeV photons (for HF) in 3<|η|<5 and all φ • 20 and 50 GeV pions randomly distributed in |η|<5 and all φ

  10. Single Pion Events Pion first decaying in ECAL Pion decaying entirely in HCAL

  11. TTbar Events TTbar event in CMS Same event in HCAL only

  12. Full Validation Processing • Most processing is done using full chains: • SimHits, Digis, RecHits and CaloTowers produced in one instance of running CMSSW • No magnetic field for full validation • Validation is of HCAL software • Easier to hit the same tower each time • ECAL+HCAL geometry except for 500 GeV pions, photons and randomly distributed 50 GeV pions which are HCAL only • ECAL needed for CaloTowers

  13. Validation Output • Running these chains results in a large number of histograms • More than 1000 altogether • Only about 100 are ordinarily used • Histograms are analyzed using ROOT macros • Histograms are produced for all software objects and are thus quite different • Variety of dimensions, range, scaling of axis, etc. • Difficult to treat them in a unified manner

  14. Histograms: Energy Fraction vs. iEta-iPhi 50 GeV Single π full scan of CMS

  15. Histograms: Number of CaloTowers Number of CaloTowers in HB from TTbar events

  16. Histograms: HE Signal Amplitude vs. SimHits

  17. Steering File • Solution: steering file for macros that includes all of these parameters: HcalRecHitTask_energy_ecal_plus_hcal_HE 0 HcalRecHitTask_energy_ecal_plus_hcal_in_cone_HE 1 HcalRecHitTask_energy_ecal_plus_hcal_in_cone_HE.gif 100. -1 1D Statrv Chi2 NoLog 41 43 HE ecal+hcal energy in cone (GeV) HcalRecHitTask_energy_hcal_vs_ecal_HE 1 HcalRecHitTask_energy_hcal_vs_ecal_HE.gif 100. 100. 2D noStat noChi2 NoLog 41 42 HE banana plot Colors File Title x-axis range y-axis range x-axis title

  18. Automation & Web Publication • 3 separate macros: CaloTowers, Digis and RecHits • Each with its own steering file in the format from the previous slide • Shell script: • Takes CMSSW versions to be compared as arguments and passes them on to macros • Runs all three and distributes resulting histograms into the proper directories, creates index.html files for each directory • Directories are then published on the web • Static HTML formatting • Example: http://cmsdoc.cern.ch/cms/cpt/Software/html/General/Validation/SVSuite/ HCAL/2_1_0_pre9_vs_2_1_0_pre6/RecHits/Global/index.html

  19. T5/3 Top Partner Model Heavy Top Partner Heavy Bottom Partner -- Signature: -- Look for tW invariant mass peak (T5/3)

  20. Cross Sections (at 14 TeV) • Almost two orders of magnitude difference between the two mass points

  21. Fast Simulation • Monte Carlo events are generated with MadGraph • Full simulation is accurate, but slow – particularly for events with a lot of particles and jets • Fast simulation uses a simplified model of CMS and algorithms optimized for speed • No GEANT4 • 10-100 times faster for samples used in this study • Output has the same format as full simulation • Validated against full simulation at each release • It has some limitations • No charge misidentification for leptons

  22. Event Selection: Isolation • Jets: • Remove those overlapping with electrons • Leptons: • Remove duplicate electrons (same track or ECAL supercluster) • Isolation parameters for leptons: • trackIso: sum of energies of tracks within ΔR < 0.3 of lepton • caloIso: sum of energy deposits in ECAL and HCAL within ΔR < 0.3 for muons and ΔR < 0.4 for electrons • Define • relIso > 0.8

  23. Event Selection: Cuts • Cuts on electrons, muons, jets and missing transverse energy (MET) used to separate signal from background • Transverse momentum: • Due to different cross sections and masses some cuts for 1 TeV sample differ from the 500 GeV ones • Common cuts on leptons (electrons or muons): • |η| < 2.4 • First lepton with pT > 50 GeV • Second lepton with pT > 25 GeV • Lepton charges must have same sign

  24. Lepton pT Distributions Normalized to 1

  25. JetMET Cuts • Different cuts on JetMET: • M = 500 GeV • |η| < 5 • Leading jet with pT > 100 GeV • Second jet with pT > 80 GeV • 5 jets with pT > 30 GeV • M = 1 TeV • |η| < 5 • Leading jet with pT > 200 GeV • Sum of Leading and Second Jets > 300 GeV • Second jet with pT > 80 GeV • 5 jets with pT > 30 GeV • ET > 20 GeV

  26. Jet pT Distributions Normalized to 1

  27. Charge MisID • Most significant selection criterion: presence of same signed leptons • If lepton charge is misidentified, backgrounds with much larger cross-sections (e.g. ttbar) become relevant • Data-driven approach: Tag & Probe • Look at Z->l+l- events • Select pair of leptons with invariant mass near MZ at least one of which is well-measured (“tag”) • Look how often charge of the other lepton (“probe”) is the same as that of the tag • Used large, centrally produced full simulation samples

  28. Charge MisID • Muons: • Tag: Muon with ET > 20 GeV • MisID rate is slightly less than 5E-4 • Not enough events to make plots even after running over more than 1 million events • Defined as a constant 5E-4 in analysis • Electrons: • Tag: Isolated electron passing High Level Trigger with |η| < 2.5 • About 2.2% of electron pairs have the same charge so average mis-ID rate is ~1.1% • Implemented as an array in (Pt, Eta) in analysis • Tag & Probe consistent with generator level information

  29. Electron Charge MisID Plots • No relIso-based Isolation applied • MisID rate grows with |η|, but remains fairly low

  30. M = 500 GeV Results # of Events assumes 300pb-1 of integrated luminosity 289 pb-1 integrated luminosity needed to get

  31. M = 1 TeV Results # of Events assumes 70fb-1 of integrated luminosity 71 fb-1 integrated luminosity needed to get

  32. T5/3 Invariant Mass • The T5/3 can be reconstructed from the hadronic fraction branch of the decay • Require all cuts from above and that two jets be tagged as b-jets • High efficiency track counting algorithm, loose working point • b-jet closest to the lepton with the second highest pT is assumed to come from leptonic branch of decay • The remaining jets are then used to reconstruct the two W’s, the top and the T5/3

  33. T5/3 Invariant Mass • Reconstruct First W: • Pair of jets with invariant mass closest to W ( must be within 20 GeV) • Jets must have ΔR < 1.5 and are discarded if used • W must have pT > 50 GeV • Reconstruct Second W: • Pair of jets with invariant mass closest to W (must be within 25 GeV) • Jets must have ΔR < 2.0 and are also discarded if used • W must have pT > 30 GeV • Reconstruct top quark: • Consider (W, b-jet) pairs and take combination with invariant mass closest to top quark (must be within 40 GeV) • Reconstruct T5/3 from top quark and the unused W

  34. Invariant Mass W Plots

  35. Invariant Mass Plots

  36. Summary • HCAL software validation is ongoing • Validation is now close to fully automated • Changes are easy to configure via steering files • T5/3 Top Partner study at 14 TeV • Accounted for possibility of charge misidentification • Effects are small • In the case of M = 500 GeV, early discovery is feasible (300 pb-1) • M = 1 TeV needs 71 fb-1 • Reconstructed T5/3 invariant mass from hadronic decay chain • Only minor adjustments needed for 10 TeV run

  37. Acknowledgements & References • HCAL software validation: • Salavat Abdullin • https://twiki.cern.ch/twiki/bin/view/CMS/HcalValidation • T5/3 study • Tulika Bose and Meenakshi Narain • CERN-CMS-CR-2008-005

More Related