-- ChengguangZhu - 01 Dec 2007

Electron

e Reconstruction

hard electrons

• electromagnetic sliding-window clusters: 1. tower building 2. pre-cluster(seed) 3. cluster filling, used for electron and photon identification
  • each tower size is is 0.025*0.025
  • 5*5 window is used to scan [-2.5,2.5] and [-3.14,3.14] space
  • taken the window as seed, if >3GeV transverse energy found,
  • position is computed with 3*3 window to decrease sensitivity to noise.
  • If seeds are close as 2*2, the one with higher transverse energy is kept only
  • Cluster filling: different sizes (3x5, 3x7 and 5x5, release 12) of clusters are filled.
  • Cluster calibration
  • egamma identification ( in release 13, this step is done before cluster filling and calibration)
• combined sliding-window clusters: similar procedure, used for tau identification

soft electrons

Reconstructed electron parameters in AOD EgammaAOD

e Isolation

12.0.x

• The selection cut is egammaParameters::etcone20 < 6GeV :
  • Etcone20 means, in EM calorimeter, the transverse energy (ref) deposited in a cone of half-opening angle 0.2 around electron minus transverse energy representing electron (5*7 cells). This redundant transverse energy comes from detector noise or jets if electron is not isolated, also called excess transverse energy.
• This is a strange cut, as it uses ET instead of E, and also because it does NOT cut on the excess energy as a function of the cluster energy (the excess energy has cluster energy dependence). D0 used isolation=[Etotal(0.4)-Eem(0.2)]/Eem(0.2).
  • But it works when one doesn't ask very precise on electron ID; X-axis is the etcone20, Z(→ee) + jets MC events, good match means that the reconstructed electron matches a truth electron from Z-decay, within ΔR=0.1. (plot from R. Madaras's slide)
    
• bug: when subtracting the electron's energy, the used size is twice larger than preori one (5*7 cells)
  • This looser the isolation power, because the excess energy is smaller than expected

13.0.x

• use the ratio ET,Cone/ET,Cluster, and you can set different cut depending on ET (to be detailed).
  • better fake electron rejection

4momentum in AOD

• The 4 momentum of electron are saved in AOD as vector(eta, phi, e, m) and can be measured with information from either calorimeter or inner detector
  • e is more precisely measured with calirimeter
  • eta and phi are more precisely measured with inner detector.
  • in AOD (release 10), the default 4 momentum are measured with calorimeter
  • in ADO (release 12), the default e are measured with calorimeter, but eta and phi are measured with inner detector.
• in release 12, you can access non-default eta and phi measured with calorimeter with the following C++

for ( elecItr=elecTES->begin(); elecItr != elecItrE; ++elecItr) {
      const CaloCluster* cluster = 0;
      cluster = (*elecItr)->cluster();
      if(cluster) {
        (*elecItr)->setE  ( cluster->e()  );
        (*elecItr)->setEta( cluster->eta() );
        (*elecItr)->setPhi( cluster->phi() );
        (*elecItr)->setM  ( cluster->m() );
    }
}

e PID

• function isem() and isemse()
  • low- isolated electrons: (whatever author), it is best to use isem() (or the corresponding likelihood, like ElectronWeight and BgWeight but they might not have been tuned at low ET/pT). isem() will give good answer for electrons with author=3
  • low- non-isolated electrons: use preferably SofteElectronWeight, SofteBgWeight - it has been tuned on WH sample but can fit also for example electrons in dijets
  • Typical AOD code for low pT electrons can be found in

example code for low pT isolated electrons
  /afs/cern.ch/atlas/software/builds/AtlasAnalysis/14.0.0/PhysicsAnalysis/BPhys/BPhysExamples/src/GetElectron.cxx
  /afs/cern.ch/atlas/software/builds/AtlasAnalysis/14.0.0/PhysicsAnalysis/BPhys/BPhysExamples/src/Jpsiee.cxx
or example for electrons in jets
  /afs/cern.ch/atlas/software/builds/AtlasAnalysis/14.0.0/PhysicsAnalysis/JetTagging/JetTagTools/src/SoftElectronTag.cxx

• Definition of isEM

Show Hide

enum   BitDef { 
  ClusterEtaRange = 0, ClusterHadronicLeakage = 1, ClusterMiddleSampling = 2, ClusterFirstSampling = 3, 
  TrackEtaRange = 8, TrackHitsA0 = 9, TrackMatchAndEoP = 10, TrackTRT = 11 
} 
enum   PID { 
  IsEM = 0, ElectronWeight = 1, BgWeight = 2, NeuralNet = 3, 
  Hmatrix = 4, SofteIsEM = 100, SofteElectronWeight = 101, SofteBgWeight = 102, 
  SofteNeuralNet = 103 
} 

enum BitDef {
    // Cluster based egamma
    ClusterEtaRange        =  0,
    ClusterHadronicLeakage =  1,
    ClusterMiddleEnergy    =  4, 
    ClusterMiddleEratio37  =  5,
    ClusterMiddleEratio33  =  6,
    ClusterMiddleWidth     =  7,
    ClusterStripsEratio    =  8,
    ClusterStripsDeltaEmax2=  9,
    ClusterStripsDeltaE    = 10,
    ClusterStripsWtot      = 11,
    ClusterStripsFracm     = 12,
    ClusterStripsWeta1c    = 13,
    ClusterIsolation       = 14,
    //Track based egamma
    TrackBlayer            = 16,
    TrackPixel             = 17,
    TrackSi                = 18,
    TrackA0                = 19,
    TrackMatchEta          = 20,
    TrackMatchPhi          = 21,
    TrackMatchEoverP       = 22,
    TrackTRThits           = 24,
    TrackTRTratio          = 25,
    TrackTRTratio90        = 26
  };

  enum PID {
    IsEM                = 0,
    ElectronWeight,
    BgWeight,
    NeuralNet,
    Hmatrix,
    Hmatrix5,
    SofteIsEM,
    SofteElectronWeight,
    SofteBgWeight,
    SofteNeuralNet,
    IsolationLikelihood,
    AdaBoost,
    PhotonWeight,
    BgPhotonWeight,
    LastEgammaPID
  };

Muon

Muon Isolation

• The selection cut is CombinedMuonParameters::etcone20 < 6GeV :
  • The meaning is deposited in cone20 around muon minus the loss of the muon in (Had+EM) calorimeter.
  • deposited in cone20 is the sum of energy of all relative (EM+Tile) calo cells which stored in container named "AllCalo" (ESD level)
  • The etloss of muon is obtained by measurement or parametrization according to initial setting.

Jet

Etmiss