WZ -> 3l + MET , l = e, mu

Contents:

WZanalysis

Datasets

• DATA

        /DoubleElectron/Run2011A-PromptReco-v1/AOD      [160329-161312]  
        /DoubleElectron/Run2011A-PromptReco-v2/AOD      [162718-164236]   
        /DoubleElectron/Run2011A-Apr22ReReco-v2/AOD    [160329-163333]  
        /DoubleMu/Run2011A-PromptReco-v1/AOD              [160329-161312 ]   
        /DoubleMu/Run2011A-PromptReco-v2/AOD              [162718-164236]    

       Certified run range from _Cert_160404-163869_7TeV_PromptReco_Collisions11_JSON.txt_    (Lumi = 5 * 10 ^{32}  / cm^2 /s)
       Integrated luminosity 190.6/pb

• MC for Spring11-PU_S1_START311_V1G1-v1/AODSIM

 
       QCD_Pt-15to20_MuPt5Enriched_TuneZ2_7TeV-pythia6       [Pt range 15to20 ... 150]
       QCD_Pt-20to30_BCtoE_TuneZ2_7TeV-pythia6                  [Pt range 20to30 ... 80to170]
       QCD_Pt-20to30_EMEnriched_TuneZ2_7TeV-pythia6           [Pt range 20to30 ... 80to170]
       TTTo2L2Nu2B_7TeV-powheg-pythia6                    
       VVJetsTo4L_TuneD6T_7TeV-madgraph-tauola                   [for WW and ZZ]    
       WZTo3LNu_TuneZ2_7TeV-pythia6
       -> no more Z0Jets_TuneZ2_7TeV-alpgen-tauola               [+1,2,3,4,5 Jets pT range 0to100 ... 800to1600]
       DYJetsToLL_TuneZ2_M-50_7TeV-madgraph-tauola
       ZBB0JetsToLNu_TuneZ2_7TeV-alpgen-tauola                           [+ 1,2,3 Jets]
       ZCC0JetsToLNu_TuneZ2_7TeV-alpgen-tauola                           [+ 1,2,3 Jets]
       ZGtoLLG_TuneZ2_7TeV-pythia6-tauola

HLT requirements

• Double leptons paths on DATA:
  • HLT_Ele17_CaloIdL_CaloIsoVL_Ele8_CaloIdL_CaloIsoVL (description in EgammaWorkingPointsv3 )
  • HLT_DoubleMu7

• HLT paths do no exist in the Spring11 MC used, they exist in Summer11 but the HLT menu will likely evolve very soon, desynchronising again the data/MC HLT menus
  • for electrons in particular algorithms used for Iso and Id at the HLT are different from those in the RECO
  • offline HLT emulation through cuts to assure being on the plateau for both DATA and MC
    • electrons - same prescription as in AN 11-123 (studied by Stephanie Baffioni):
      • 2 Ecal-Driven electrons with pT > 20, 10 and sigmaIetaIeta < 0.014(0.035), H/E < 0.15(0.10), IsoECAL/Et < 0.2(0.2), IsoHCAL/Et < 0.19(0.19)
      • despite the different HLT and offline-RECO algorithms and conditions, main changes wrt online are pT thresholds and HCAL isolation
      • efficiency measurement on DATA ongoing
    • 2 muons with pT > 20,10 and ID requirement (only globalMuons)
    • at least 2 leptons same flavor pT > 10 and Mll > 50

Cross-sections

• from MCBackgroundList
• except for
  • QCD LO
  • ZGamma (LO 14.3pb * 1.25)
  • VVJets (LO 4.8pb * k-factor 1.102)

Objects

• • |eta| < 2.5,
  • pT > 15GeV/c (for Z) , pT > 20GeV/c (for W)
  • Loose ID (for Z) = sigmaIEta IEta < 0.012(0.031) (H/E moved to HCAL Isolation) + |DEta_in| < 0.007 (0.011) + |DPhi_in| < 0.8(0.7)
  • Tight ID (for W) = sigmaIEta IEta < 0.010(0.031) (H/E moved to HCAL Isolation) + |DEta_in| < 0.005 (0.006) + |DPhi_in| < 0.027(0.021)
  • Iso = Fast Jet subtraction with Effective Areas (see Fast Jet-subtraction Effective Area)

if(elec.isEB()) {
    ( elec.dr03TkSumPt() + 
      max(0., elec.dr03EcalRecHitSumEt() - 1.) + 
      elec.dr03HcalTowerSumEt() + ( elec.hadronicOverEm() * elec.energy() * (Rt/R) ) - 
      rhoFastJet_ * 0.0997 ) / elec.p4().Pt()     
    // double R=TMath::Sqrt(sclRef->x()*sclRef->x() + sclRef->y()*sclRef->y() +sclRef->z()*sclRef->z());
    // double Rt=TMath::Sqrt(sclRef->x()*sclRef->x() + sclRef->y()*sclRef->y()); 
    // cut values: Electron is Isolated if   the computed Isolation is < 0.15 (WP 95%)   < 0.07 (WP 80%)
}

else{
    ( elec.dr03TkSumPt() + 
      elec.dr03EcalRecHitSumEt() + 
      elec.dr03HcalTowerSumEt() + ( elec.hadronicOverEm() * elec.energy() * (Rt/R) ) - 
      rhoFastJet_ * 0.1123 ) / elec.p4().Pt();
    // cut values: Electron is Isolated if   the computed Isolation is < 0.1 (WP 95%)   < 0.06 (WP 80%)
}

• • Conversion rejection

if( (fabs(convInfo.dcot() ) < 0.02 &&  fabs(convInfo.dist() ) < 0.02)  ||
    ele.gsfTrack()->trackerExpectedHitsInner().numberOfHits() > 0 )  // electron is a conversion up to a WP 80%
if( ele.gsfTrack()->trackerExpectedHitsInner().numberOfHits() > 0 )  // electron is a conversion up to a WP 95%

* muons:

• • |eta| < 2.4,
  • pT > 15GeV/c (for Z), pT > 20GeV/c (for W)
  • ID (for W and Z) = GlobalMuon + TrackerMuon + NpixelHits > 0 + NtrackerHits > 10 + normalizedChi2 < 10 + NmuonHits > 0 + Nmatches > 1 + |dxy(PV)| < 0.2
  • Iso = Fast Jet subtraction with Efefctive Areas (see Fast Jet-subtraction Effective Area)

    if( fabs(muon.Eta() ) < 1.05 ) {
        ( muon.isolationR03().sumPt +
          muon.isolationR03().emEt  +
          muon.isolationR03().hadEt  -
          rhoFastJet_ * 0.1057 ) / muon.Pt()
        // cut values: Muon is Isolated if   the computed Isolation is < 0.1 (WP 95%)   < 0.06 (WP 80%)  *BUT < 0.15 could be enought*
    }
    else{
        ( muon.isolationR03().sumPt +
          muon.isolationR03().emEt  +
          muon.isolationR03().hadEt  -
          rhoFastJet_ * 0.0769 ) / muon.Pt()
        // cut values: Muon is Isolated if   the computed Isolation is < 0.1 (WP 95%)   < 0.06 (WP 80%)  *BUT < 0.15 could be enought*
    }
    

• MET: track corrected MET (see METvsPUeffect)
  • ParticleFlowMET looked at, performances compatible with those for tcMET

Cut Based selections

Preselection

• 1) 2leptons satisfying HLT (see HLT requirements before)
• 2) 3rd lepton pT > 10
  • All leptons considered satisfy pT > 10GeV
• 3) vertex compatibility on the leptons: |dxy(PV)| < 0.5 && |dy(PV)| < 0.5

Main selections

• 4) Z candidate
  • opposite charge + same flavor
  • 60 < M < 120 GeV/c2
  • (leptons chosen for Z are excluded from further selections)

• W candidate
  • 5) 3rd lepton chosen as the highest pT one left
  • 6) veto on 2nd Z (defined with same selections used to find the first one)

• 7) |3D impact parameter| < 3 on the worse of the 3 leptons (removed from main flow, reported here as a record)

• 8) MET cut > 25 GeV (independent of step 7)

• 9) MET cut applied after step 7 (removed from main flow, reported here as a record)

Results

• table with some numbers
• For each step follow number of events in 5 rows corresponding to: 4channels combined, eee, eem, mme, mmm channels
  • The distinction between channels is made on RECO level and based on the info from the requirements up to that step
• Number of events for MC are rescaled to luminosity in DATA (190.7/pb)
• Last column is Signal+Background from MC for the only combination of the 4 channels

• Summary plot of the selections, for the 4 channels combined:

• See effect of cut-based flow:
  • After Preselection (step 3)

Show... Hide

• • Z candidate (step 4)

Show... Hide

• • W candidate (step 6)

Show... Hide

• • MET cut (step 8 - end of the analysis)

Show... Hide

Backgrounds

• DATA driven for Z+Jets, relying on MC for the others

Z+Jets with ABCD method (ONGOING)

• Study on MC using signal sample (WZ->3l) and Z+Jets background (sample from madgraph)
  • 3 different 2D distributions looked at: Combined Relative Isolation - pT, C.R.Isolation - MET (also pfMET), C.R.Isolation - DEta_in
    • distributions filled for the highest pT lepton remaining after the Z selection. Leptons satisfy Iso, Id, pT criteria required at the Preselection (steps 1-3)
    • channels looked at eee, mme for Barrel and Endcap

see Iso-pT

Show... Hide

see Iso-MET

Show... Hide

see Iso-pfMET

Show... Hide

see Iso-DEta_in

Show... Hide

• channels looked at eem, mmm for Barrel and Endcap see Iso-pT

Show... Hide

Show... Hide

Show... Hide

• First approach is to define ABCD regions following cuts to extract the signal
• Few hypothesis:
  • If one major background is surviving
  • If A is the signal region (no signal contribution in B, C, D )
  • If NA/NB = NC/ND (with NA = number of events in A region) holds for the background
• Background contribution in region A is estimated as NC*NB/ND

• Test 1 is to assess the validity of the hypothesis NA/NB = NC/ND, via the estimator [ NA - NC*NB/ND ] / NA, on the Z+Jets sample.
• Test 2 is to check how well the signal region is isolated from the background, via [NB + NC + ND ] / NTot, on the signal sample.

See some scan computed wrt a different cut applied to separate the B,C,D regions from A (for MET and DEta_in eee and mme)

Show... Hide

for MET in eem and mmm channels

Show... Hide

MET cut Inversion

• Z+Jets (madgraph sample) through reverting cut (MET tried at the moment): compare number of events in the control region MET < 25GeV

See plots

Show... Hide

• see table with numbers rescaled
• see few plots for step 4 (Z candidate)

Show... Hide

Support studies

Fast Jet-subtraction Effective Area

• most updated result of the study presented at the Electron Identification meeting 17/June/2011 seeSlides
• For electrons and muons a correction for the Isolation dependence on PU is computed.
• Combined Relative Isolation is looked at with WP95% and 80% for electrons with 2010 VBTF like settings (SimpleCutBasedEleID)
  Show... Hide
  
• For muons Combined Relative Isolation is looked at: WP95% refers to <0.15 cut, WP80% refers to <0.1cut
  Show... Hide
  

• Correction computation:
  • on leptons from signal (from WZ3L), with pT>10GeV
  • Fast Jet subtraction is applied, using Effective Areas (to account for veto-regions/Isolation thresholds not taken into account with the nominal 0.3 cone radius)
    • NewIsolation = OldIsolation - rho * EffectiveArea
    • talking about absolute isolation to subtract the PU contribution
  • All Isolation as defined in SimpleCutBasedEleID except for HCALIsolation, computed in the full cone 0.->0.3 (and consequently no H/E Id is applied as an ID cut)
  • How to:
    • rho factor event by event, as described in SimpleCutBasedEleID2011
    • New_dr03HcalTowerSumEt() (in the cone 0->0.3) a fast recipe which is a good approximation is to use
      • old_HCALIsolation (the one in [SimpleCutBasedEleID]) which is "sum of Transverse energy in HCAL 0.15->0.3"
      • H/E which is "sum of Energy in HCAL 0->0.15, over E", since H/E is in good approximation H(transverse energy)/E(transverse energy), it could be
      • new_HCALIsolation = old_HCALIsolation + (H/E * Et) where Et is the transverse energy.

• Mean-value of the single sub-detector absolute Isolation wrt the Number of vertexes

• Mean-value of the Combined Isolation before and after the correction

• Electrons: Effective Area = 0.0997 electron EB (< 1.479); EA = 0.1123 electron EE (> 1.479)
  • As Working Points I am using at the moment the same ones in [SimpleCutBasedEleID]
    Show... Hide
    
• Muons: Effective Area = 0.1057 barrel (< 1.05) EA = 0.0767 endcap (> 1.05)
  • As Working Points I am using at the moment the same ones in [SimpleCutBasedEleID] for electrons in the endcaps
    Show... Hide
    

METvsPUeffect

• Efficiency of MET > 25GeV cut wrt N.Vertexes in the event.
• Plots refer to events after Preselection (step 1-3) tcMET (left) pfMET (right)
  Show... Hide