Top and W Polarizations in Single top t-channel (2011 Data)

Introduction

This page presents the tools developed and the datasets produced by the LPSC Grenoble group for the measurements of the top-quark and W-boson polarization observables in the single top t-channel production: degree of polarisation of the top-quark, W-boson helicity and polarization fractions.

Details about the different polarization observables and their measurements from angular correlations of the charged lepton with appropriatly chosen spin axes are available in the talk given in the Single-top workshop 6 December 2011 ("Measurement of W and top-quark polarizations").

A - Analysis Software

The analysis software is saved in the Root svn repository of the $SVNGRP/Institutes/Grenoble package (to check out: svn co $SVNGRP/Institutes/Grenoble/trunk Grenoble). This software has been developed for the analyses of the 2011 data and MonteCarlo datasets centrally produced by the Top Working Group with releases AtlasPhysics-17.0.X.Y.Z.

An overview of the software is available in the talk given in the Single-top meeting 13 September 2011 ("Single-top D3PD status").

1 - Object and Event Calibration Tools

Script for checking out the Combined Performances and Top Groups packages and associated root and data files (list of current tags): runPackages

Packages for event and object corrections and estimations of associated systematic errors (with links to the documentation pages):

2 - Analysis Steps

Programs for data sample reductions, event and object selections, object and event calibrations and derived variables calculations:

  • Filtering.cxx: main code for event filtering ("skimming") and branch selection ("slimming") of the TopD3PD datasets. This code also includes period and pile-up re-weighting of Monte Carlo events.
  • PreAnalysis.cxx: main code for object energy/momentum resolution smearing, object energy/momentum re-scaling and MissingEt re-computing from the reduced D3PDs produced by the filtering step or from the TopD3PDs (in this case the job done in the filtering step is also executed by the PreAnalysis code). The nominal and up/down shifted energies and momenta are calculated and recorded.
  • PreSelection.cxx: main code for object and event pre-selections for the nominal and up/down energies and momenta (one output file per correction type for baseline analysis and systematic estimates). The object scale factors (nominal and up/down values) are calculated and recorded. Different pre-selection options are possible for the topology of the selected events:
    • "SingleLepton": 1 "tight" lepton+jets
    • "DiLeptons": 2 "tight" leptons+jets
    • "LooseSingleLepton": 1 "loose" lepton+jets
    • "LooseDiLeptons": 2 "loose" leptons+jets
    • "JetElectron": 1 jet-electron+jets
  • Selection.cxx: main code for single top t-channel selection (electron and muon flavours are treated separatly), recontructions of the neutrino/W-boson/top and derivation of the angular correlations required for the polarization measurements. The QCD jet-electron normalizations and the W+jets overall normalizations and flavour fractions are calculated and recorded. The Monte Carlo events are also re-weigthed taking into account the integrated luminosity and the corresponding cross sections. The Selection.cxx program can also be used for selection of s-channel events.
  • TruthSelection.cxx: main code for selection of the truth particles (hard process and final particles after radiation) corresponding to the t-channel (AcerMC generator).
  • TruthPolarization.cxx: main code for extraction of the angular correlations from the truth information (hard process and final particles selected with the TruthSelection program)
  • Analysis.cxx: main code for creation of histograms (control histograms and angular distributions) with all normalizations, event weights and object scale factors applied
  • TruthAnalysis.cxx: main code for creation of histograms from truth information (angular distributions at parton level) with generator weighting

3 - Object and Event Selections and Calibrations

Classes for object and event selections and calibrations:

  • EventFilter.cpp: class for the first analysis step ("Filtering" or "PreAnalysis")
  • TriggerSelector.cpp: class for trigger selection and for trigger/reco object matching
  • PeriodScaler.cpp: class for Monte Carlo pile-up and period re-weighting (this class also includes the integrated luminosity per data taking period)
  • ElectronEnergyScaler.cpp: class for electron energy resolution smearing and energy scaling correction
  • MuonEnergyScaler.cpp: class for muon momentum resolution smearing and momentum scaling correction
  • JetEnergyScaler.cpp: class for jet momentum energy resolution smearing, energy scaling correction, energy scaling uncertainty calculation and efficiency reconstruction estimate
  • MissingEtScaler.cpp: class to re-compute the MissingEt with object energy/momentum nominal corrections, with object energy/momentum uncertainty shifts and with MissingEt systematics
  • ElectronSelector.cpp: class for electron selection
  • MuonSelector.cpp: class for muon selection
  • JetSelector.cpp: class for jet selection
  • JetElectronSelector.cpp: class for jet-electron selections (with electron or muon pT and eta acceptances)
  • EventSelector.cpp: class for selection of lepton(s)+jets events in the second analysis step ("PreSelection")
  • EventScaler.cpp: class for derivation of the object (electrons, muons, jets) scale factors
  • ChannelSelector.cpp: class for selection of t-channel events (PreTag and 1Tag selections in the two jets bin) in the third analysis step ("Selection")

4 - Neutrino, W-boson and top-quark Reconstructions

Classes and functions for neutrino, W-boson and top reconstructions and selections:

  • NeutrinoTools.cpp: functions for reconstruction of the neutrino from the MissingEtx and MissingEty values and from the lepton Lorentz vector (the W-boson mass is assumed to be 80.42 GeV). The solution with the lowest pz value is chosen; in case of a negative discriminant, the solution corresponding to a null discriminant is considered (real part of the complex solutions).
  • TopSelectionWriter.cpp: class for neutrino, W-boson and top reconstructions from the reconstructed objects and the generator particles (GetEvent and GetTruthEvent methods).
  • TruthSelector.cpp: class for selection of the truth particles associated with single top t-channel production: top-quark, spectator light-quark and b-quark from the production vertex, W-boson and b-quark from the top decay vertex, lepton (e,mu,tau) and neutrino from the W-boson decay vertex and lepton (e,mu) from the tau decay vertex (hard process and final particles generated by the AcerMC code).

5 - Polarization Observables

Class for calculation of the angular correlations:

  • TopSelectionWriter.cpp: class for calculation of the angular correlations from the reconstructed objects and the generator particles (CalculateCorrelations method).
  • HistogramMaker.cpp: class for histogram making (control histograms and angular distributions)

The angular correlations and polarization observables extracted from the AcerMC generator particles and reconstructed objetcs are presented in the talk given in the Single-top meeting 14 February 2012 ("Top and W polarization in AcerMC"). Note that the AcerMC event weight was not considered in the distributions presented in this talk.

B - Data Sample Production

Production on the grid of "Filtering", "PreAnalysis" and "PreSelection" D3PDs from the TopD3PDs: tag p834 for mc11c TopD3PDs and tag p822 for TopD3PD data skims. Scripts for production on the grid are available in the Grid svn repository of the Grenoble package: submitXXX scripts in the scripts directory and runYYY interfaces in the datasets directory.

The "Selection" Nuples are produced and stored locally. Scripts for local monitoring, transfer and merging of the "PreSelection" datasets produced on the grid are available in the Local svn repository of the Grenoble package: core scripts in the scripts directory and execution interfaces in the datasets directory.

1 - Data (data11_7TeV)

  • Integrated luminosity = 4713.11 +- 183.81 pb-1 (3.9% error)

  • Total numbers of processed TopD3PD events:
    • Egamma: 241049741
    • Muons: 229319086
    • JetTauEtmiss: 733960

  • Sizes of the "PreSelection" D3PDs:
    • Egamma:
      • "SingleLepton": 200MB
    • Muons:
      • "SingleLepton": 230 MB
    • JetTauEtmiss: 6 MB

2 - Monte Carlo (mc11c_7TeV)

  • List of "PreSelection" and "Selection" file labelling (suffixes) with nominal corrections and for estimates of their systematic errors:
    • Nominal corrections: no suffix
    • Electron systematics:
      • el_scaled_down and el_scaled_up
      • el_smeared_down and el_smeared_up
    • Muon systematics:
      • mu_smeared_IDdown and mu_smeared_IDup
      • mu_smeared_MSdown and mu_smeared_MSup
      • mu_smeared_noscale
    • Jet systematics:
      • jet_goodReco
      • jet_scaled_down and jet_scaled_up
      • jet_smeared_syst
    • MissingEt systematics:
      • outsoft_down and outsoft_up
      • pileup_down and pileup_up

  • Values of the reference cross sections: TopMC11
  • List of cross-sections used in the "Selection" step: CrossSections.dat - a correction factor equal to 1.52/1.20 is applied on the Wc K-factor when reading this file as coded in the ReadCrossSections method of the AnalysisReader.cpp class.

C - Object and Event Selections

Events and objects are selected in a three-steps procedure

1 - Filtering

Event filters applied in the first data reduction step:

  • Good run list selection (data only): rejection of events with RunNumber and lbn values not listed in the Top GRL file

  • Heavy flavor overlap removal and weighting cleaning (Monte Carlo samples only): rejection of events with top_hfor_type >= 4 and rejection of events with a generator weight equal to zero

  • Lepton filter:
    • Monte Carlo samples: at least one mediumPP electron or at least one tight muon
    • Egamma stream: at least one mediumPP electron
    • Muons stream: at least one tight muon
    • JetTauEtmiss stream: no lepton filtering

  • Trigger filter:
    • Monte Carlo samples: electron or muon trigger passed
    • Egamma stream: electron trigger passed
    • Muons stream: muon trigger passed
    • JetTauEtmiss: no trigger filtering (trigger filtering applied at the dAOD level)

  • Trigger chains:
    • Electrons: EF_e20_medium (periods B-J), EF_e22_medium (period K) and EF_e22vh_medium1 or EF_e45_medium1 (periods L-M)
    • Muons: EF_mu18 (periods B-I) and EF_mu18_medium (periods J-M)

2 - PreSelection

Reconstructed objects and lepton+jets events are selected in the second step. The different lepton flavours of the selected events are recorded in the same output D3PD. For single lepton selections, the two lepton flavours are tagged with the ElectronPreTag and MuonPreTag flags.

a - Electron Selection

  • el_author = 1 or 3
  • quality: el_OQ & 1446 = 0
  • (corrected) el_cl_track_Et > 25 GeV
  • abs(el_cl_eta) < 2.47 excluding 1.37 < abs(el_cl_eta) < 1.52
  • "tight" selections for "SingleLepton" and "DiLeptons" options:
    • el_tightPP = 1
    • isolation: Eiso90Etcone20 and Eiso90Ptcone30
  • "loose" selections for "JetElectron" option:
    • el_loosePP = 1
    • no isolation

b - Muon Selection

  • mu_author = 12
  • mu_tight = 1
  • inner detector hit requirements:
    • mu_expectBLayerHit = 0 or mu_nBLHits > 0
    • mu_nPixHits+mu_nPixelDeadSensors > 1
    • mu_nSCTHits+mu_nSCTDeadSensors >= 6
    • mu_nPixHoles+mu_nSCTHoles < 3
    • n = mu_nTRTHits+mu_nTRTOutliers
      • for fabs(mu_eta) < 1.9: require n > 5 and nTRTOutliers/n < 0.9
      • for fabs(mu_eta) >= 1.9: if n > 5 then require nTRToutliers/n < 0.9
  • (corrected) mu_pt > 20 GeV
  • abs(mu_eta) < 2.5
  • "tight" selections for "SingleLepton", "DiLeptons" and "JetElectron" options:
    • isolation: mu_etcone20 < 4 GeV and mu_ptcone30 < 2.5 GeV
    • isolation vs jets: removal of muons in dR(mu_eta,mu_phi,jet_eta,jet_phi) < 0.4 from any jet (before overlap removal with electrons) with (corrected) jet_pt > 25 GeV and abs(jet_jvtxf) > 0.75

c - Jet Selection

  • Removal of overlaps with the selected electrons within dR(jet_eta,jet_phi,el_tracketa,el_trackphi) < 0.2 (only the closest jet to each electron and with jet_E >= 0 is checked)
  • (uncorrected) jet_E >= 0
  • (corrected) jet_pt > 25 GeV
  • abs(jet_emscale_eta+jet_EMJES_EtaCorr) < 4.5
  • abs(jet_jvtxf) > 0.75

d - Jet-Electron Selection

  • eljet_emscale_pt > 25 GeV
  • Electron-like selections:
    • abs(eljet_emscale_eta) < 2.47 excluding 1.37 < abs(eljet_emscale_eta) < 1.52
  • Muon-like selections:
    • abs(eljet_emscale_eta) < 2.5
  • The electron-jet is removed from the container of selected jets if there is an overlap

e - Event Cleaning

  • Vertex selection: selection of events with the first primary vertex reconstructed from at least 5 tracks (vxp_type[0] = 1 or 3 and vxp_nTracks[0] >= 5)
  • LAr cleaning: selection of events without LAr noise bursts (larError = 0 - only for data samples) - This cut is applied after all the cuts listed below (recommendation of the Egamma group).

f - SingleLepton Event Selection

  • Lepton selection: exactly one electron (e channel) or exactly one muon (mu channel)
  • Lepton final pT selection: (corrected) el_cl_track_Et > 25 GeV e channel) or (corrected) mu_pt > 25 GeV (mu channel)
  • Veto lepton: no muons (e channel) or no electrons (mu channel)
  • Trigger passed: electron (e channel) or muon (mu channel) trigger chain passed
  • Trigger matching: the selected lepton matches the trigger object within dR(trigger,reco) < 0.15
  • W transverse mass selection:
    • "SingleLepton": (corrected) WtransMass > 30 GeV
  • MissingEt selection:
    • "SingleLepton": (corrected) MET_RefFinal_em_tight_et > 30 GeV
  • Jet cleaning: rejection of events with at least one jet remaining after removal of electron duplicates having jet_isBadLoose = 0 and (corrected) jet_pt > 20 GeV
  • Jet selection: at least two jets

h - JetElectron Event Selection

  • Jet-electron selection: exactly one jet-electron with electron acceptance cuts (e channel) or muon acceptance cuts (mu channel)
  • Veto lepton: no muons (e channel) or no "loose" electrons and no muons (mu channel)
  • Trigger: no Trigger selection and no trigger matching
  • W transverse mass selection: WtransMass > 30 GeV
  • MissingEt selection: (corrected) MET_RefFinal_em_tight_et > 30 GeV
  • Jet cleaning: rejection of events with at least one jet remaining after removal of electron duplicates having jet_isBadLoose = 0 and (corrected) jet_pt > 20 GeV
  • Jet selection: at least two jets

3 - Selection

Single top t-channel events are selected in the third step and the "Electron" or "Muon" lepton flavours are treated separatly (one output Ntuple per lepton flavour):

  • Selection 2 jets pretag:
    • Lepton flavour: selection of electron+jets or muon+jets events
    • QCD rejection: no selection
    • Final jet selection: exactly two jets

  • Selection 2 jets 1tag:
    • Selection 2 jets pretag
    • Tagged jet selection: exactly one jet tagged defined as:
      • abs(jet_emscale_eta+jet_EMJES_EtaCorr) < 2.5
      • MV1 0_905 working point

D - Event Weights and Object Scale Factors

Event weights and object scale factors are calculated in the different analysis steps (Monte Carlo samples only)

1 - PreAnalysis Weights

Event weights are calculated by the PeriodScaler class in the "PreAnalysis" step:

  • PeriodWeight: re-weighting factor of the Monte Carlo events to match the luminosities of the different data taking configurations (periods)
  • pileup_weight: re-weighting factor of the Monte Carlo events to match the real distribution of the average number of interactions per crossing
  • TotalWeight = MC weight*PeriodWeight*pileup_weight: global re-weighting factor of the Monte Carlo events (MC weight is the generator event weight)

The total weighted input numbers of events are recorded in the "AODEvents" histogram of the "AODBookkeeping" folder:

  • Bin 1: No weight
  • Bin 2: MC weight
  • Bin 3: MC*Period weight
  • Bin 4: MC*Pileup weight
  • Bin 5: MC*Period*Pileup weight (TotalWeight)

2 - PreSelection Scale Factors

Object scale factors are calculated by the EventScaler class in the "PreSelection" step:

  • Electrons:
    • SF_recoId and SF_recoId_error: reconstruction and identification scale factors and associated errors returned by the electron_SF_R17.h header file
    • SF_trigger and SF_trigger_error: trigger scale factors and associated errors returned by the electron_SF_R17.h header file
    • SF = SF_recoId*SF_trigger, SF_up = SF*(1+SF_total_error) and SF_down = SF*(1-SF_total_error): global scale factors and associated up/down errors (SF_total_error is the quadratic sum of the relative recoId and trigger errors)

  • Muons:
    • SF_reco and SF_reco_error: reconstruction scale factors and associated errors returned by the AnalysisMuonEfficiencyScaleFactors.cxx class (SF_reco_error is the linear sum of the statistical and systematic errors)
    • SF_Id and SF_Id_error: identification scale factors and associated errors returned by the MuonSF.cxx class
    • SF_trigger and SF_trigger_error: trigger scale factors and associated errors returned by the MuonSF.cxx class
    • SF = SF_reco*SF_Id*SF_trigger, SF_up = SF*(1+SF_total_error) and SF_down = SF*(1-SF_total_error): global scale factors and associated up/down errors (SF_total_error is the quadratic sum of the relative reco, Id and trigger errors)

  • Jets:
    • MV1_0_905_SF: b-tag (tagged jets) or inefficiency (untagged jets) scale factors returned by the CalibrationDataInterfaceROOT.cxx class. The tagging criterion is: jet_flavor_weight_MV1 > 0.905363 (60% efficiency working point). The combined MV1 weight is calculated by the mv1Eval method of the MV1.cxx class.
    • MV1_0_905_SF_btag_up:
    • MV1_0_905_SF_btag_down:
    • MV1_0_905_SF_mistag_up:
    • MV1_0_905_SF_mistag_down:

3 - Selection Weights

Additional event weights and scale factors are calculated in the "Selection" step (coded in the Selection.cxx main code):

  • DataWeight = CrossSection*Luminosity/InputWeight: weight to re-normalize the Monte Carlo events to the real data (InputWeight is the total input number of events weighted with TotalWeight)
  • EventWeight = DataWeight*TotalWeight: total weight for comparison of the Monte Carlo observables distributions to the measured distributions and for calculation of the expected numbers of events after correction with object scale factors

4 - W+jets Normalizations

For the W+jets Monte Carlo samples, overall normalization factors and flavour fractions are calculated in the Selection.cxx code and recorded via histograms in the WjetsBookkeeping folder:

  • WjetsEvents histogram: overall normalization weights and associated errors calculated with the Wasymmetry_rel17.h header file; these weights are calculated for the selected lepton flavour:
    • Bin 1: Wjets_pretag_SF: overall normalization for the 2 jets pretag selection
    • Bin 2: Wjets_pretag_SF_up: overall normalization with error up for the 2 jets pretag selection
    • Bin 3: Wjets_pretag_SF_down: overall normalization with error down for the 2 jets pretag selection
    • Bin 4: Wjets_1tag_SF: overall normalization for the 2 jets 1tag selection
    • Bin 5: Wjets_1tag_SF_up: overall normalization with error up for the 2 jets 1tag selection
    • Bin 6: Wjets_1tag_SF_down: overall normalization with error down for the 2 jets 1tag selection

  • Wbb, Wcc, Wc and Wlight histograms: flavour fractions and associated errors (scale factors) calculated with the HFsys_factor_ttbar_emu.h header file; the fractions and their systematic errors are extracted for each W+jets flavour (given by top_hfor_type: 0=Wbb, 1=Wcc, 2=Wc, 3=Wlight) from the numbers of selected events (2 jets pretag cuts). For each W+jets flavour, the corresponding histogram contains the following values:
    • Bin 1: Wjets_fraction_nominal
    • Bin 2: Wjets_fraction_Fbb_Fc_up
    • Bin 3: Wjets_fraction_Fbb_Fc_down
    • Bin 4: Wjets_fraction_HF_LF_up
    • Bin 5: Wjets_fraction_HF_LF_down
    • Bin 6: Wjets_fraction_Fbb3_Fbb4_up
    • Bin 7: Wjets_fraction_Fbb3_Fbb4_down
    • Bin 8: Wjets_fraction_Fc3_Fc4_up
    • Bin 9: Wjets_fraction_Fc3_Fc4_down

5 - QCD Normalizations

The QCD overall normalizations extracted with the Jet-Electron method are recorded in the JetTauEtmiss data samples. They are calculated for the selected lepton flavour by the Selection.cxx code and are recorded in a histogram stored in the QCDBookkeeping folder:

  • QCDEvents histogram:
    • Bin 1: QCD_pretag: overall normalization for the 2 jets pretag selection
    • Bin 2: QCD_pretag_up: 2 jets pretag normalization with error up (50%)
    • Bin 3: QCD_pretag_down: 2 jets pretag normalization with error down (50%)
    • Bin 4: QCD_1tag: overall normalization for the 2 jets 1tag selection
    • Bin 5: QCD_1tag_up: 2 jets 1tag normalization with error up (50%)
    • Bin 6: QCD_1tag_down: 2 jets 1tag normalization with error down (50%)

E - Selection Cut Flows

The selection cut flows are recorded in histograms stored in the D3PDBookkeeping folder:

  • Filtering or PreAnalysis: D3PDEvents histogram:
    • Bin 1: D3PD Events
    • Bin 2: Good LumiBlock (data) or Good Truth HF (Monte Carlo)
    • Bin 3: Lepton Filter
    • Bin 4: Trigger Filter

  • PreSelection: ElectronPreTag and MuonPreTag histograms:
    • Bins 1 to 4: Filtering/PreAnalysis cut flow
    • Bin 5: Vertex Selection
    • Bin 6: Lepton Selection
    • Bin 7: Final Lepton Selection
    • Bin 8: Veto Lepton Selection
    • Bin 9: Trigger Passed
    • Bin 10: Trigger Matched
    • Bin 11: Electron-Muon Overlap (not used)
    • Bin 12: Wtrans Mass Selection
    • Bin 13: MissingEt Selection
    • Bin 14: Jet Cleaning
    • Bin 15: Jet Selection
    • Bin 16: LAr Cleaning

  • Selection: Electron or Muon histogram:
    • Bins 1 to 16: PreSelection cut flow
    • Bin 17: Lepton Flavour
    • Bin 18: QCD Rejection (not used)
    • Bin 19: Final Jet Selection
    • Bin 20: Tagged Jet Selection (1tag selection only)

F - Histograms

Histograms are produced by the Analysis.cxx and TruthAnalysis.cxx codes.

The histograms are filled by taking into account all event weights and object scale factors. The corresponding systematic errors are propagated according to the analysis type:

  • default: all nominal normalizations and scale factors
  • event_SF_up: global object scale factor with errors up (lepton_SF_up*jets_SF_up)
  • event_SF_down: global object scale factor with errors down (lepton_SF_down*jets_SF_down)
  • QCDEvents_up:
  • QCDEvents_down:
  • Wjets_SF_up:
  • Wjets_SF_down:
  • Wjets_Fbb_Fc_up:
  • Wjets_Fbb_Fc_down:
  • Wjets_HF_LF_up:
  • Wjets_Fbb3_Fbb4_up:
  • Wjets_Fbb3_Fbb4_down:
  • Wjets_Fc3_Fc4_up:
  • Wjets_Fc3_Fc4_down:

1 - Control Distributions

2 - Angular Distributions

Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r2 - 2012-03-10 - AnnickLleres
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Sandbox All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback