Electron physics object tutorial C. Charlot / LLR Automn08 tutorials, 14 oct. 2008

2  Introduce the algorithms and their parameters  Electron sequence objects data structure and content  How to read the electron collection  Electron ID modules  Disclaimer  CMSSW 21x  Isolation not covered Scope

3 Electrons in CMS  Electrons are caracterized by a high E T SuperCluster matched in position and momentum with a track  Electron object can be more complicated due to showering in the traker material  Several subclusters (grouped into a supercluster)  Eventualy several tracks from photon conversions « showering electron » brem’d  initial e-  Electron object therefore  has a particle behaviour  is a composed object  handles combined information

4 Electron reconstruction sequence Super Cluster Trajectory Seed Seeding: - Std track seeds - SC driven pixel match filter - E T, H/E Electron TrackCandidate Electron Seed Electron Gsf track GsfElectron Trajectory building: - CTF builder - Electron loss modeling - No Chi2 cut - Reduced #candidates/layer Gsf track fit: - Electron loss modeling - Mode of the gaussian mixture used for p ele - Brem fraction Preselection: - E/p - dEta, dPhi Amb. resolution

5 Electron seeding  Electron seed reconstruction starts from standard tracker seeds and superclusters  Filtered by E T and H/E  Pixel match filter applied on standard tracker seeds  Match the first hit backpropagating from supercluster E T and position to the beam spot  Large windowz (z spread)  E T dependent phi window  Match the second hit propagating from the matched first hit up to the second hit layer  Tight windows

6 Electron seed object reco:ElectronPixelSeed TrajectorySeed  A TrajectorySeed contains:  a TSOS  By convention defined on the outermost layer  a vector of RecHits  a propagation direction  An ElectronPixelSeed adds:  Ref to SuperCluster reco:SuperCluster 1  Unlike TrajectorySeeds, ElectronPixelSeeds are stored in RECO data tier

7 Electron tracking  Electron reconstruction uses Ckf trajectory building with dedicated propagators and parameters  Energy loss modeling  No Chi2 cut  Reduced number of candidates per layer  Gsf fit is used to evaluate track parameters  Energy loss modeling  Multicomponent TSOS  Further electron reconstruction makes use of mode estimate gsfElectronCkfTrackCandidateMaker_cff.py

8 Gsf track object reco:GsfTrack reco:Track  A Ttrack contains (see Tracking tutorial):  Ref. position on the track  Momentum at this position  5D curvilinear covariance matrix from the fit  Charge, Chi2 and ndof  Hit patterns (in which layers the track has hits)  Through the TrackExtra (only in RECO)  Inner and outer track parameters with covariance matrix  A reference to the TrajectorySeed  Vector of Refs to RecHits  a propagation direction  A GsfTrack adds:  Charge, momentum and momentum covariance from mode reco:GsfTrackExtra 1  GsfTrackExtra is the standard « extra » track extension for the GsfTrack  Inner and outer multicomponent states

9 Electron preselection  GsfElectrons are constructed from a reconstructed GsfTrack and a loose association with it’s corresponding supercluster  Associated at the electron seeding stage  The supercluster-gsf track preselection is loose so to keep high efficiency at the RECO stage  Compatible with affordable fake rate  Purity can be increased at a later stage applying electron ID  Preselection is based on  Supercluster E T  Delta eta between the SC position and track extrapolation  Delta phi between the SC position and track extrapolation  E/P  H/E

10 Electron preselection  Preselection parameters defined in pixelMatchgsfElectrons_cfi.py

11 Electron Object GsfElectron GsfTrackSuperCluster GsfTrackExtraBasicCluster Candidate n1 Particle behaviour (q, p4,...), PhysTools Matching and ID variables, refs to constituants

12 Electron Object  Particle properties  pdgid, 4- momentum, charge,...  References to constituant SC and Gsf track  Matching and eID variables  Track extrapolation to ECAL,..

13 Electron object  Particle interface (vertex, momentum, charge, mass,..)  Vertex position is the CPA to the beam spot  trackMomentumAtVertex from GsfElectron ≠ momentum from Gsf track  GsfElectron:: trackMomentumAtVertex() gives mode estimate, GsfTrack::momentum() provides mean estimate  momentum(), p4(), pt() uses the momentum magnitude from E-p combination  direction always given by the GsfTrack parameters  charge is given by the GsfTrack

14 Electron object  Track-Cluster match and ID variables  Track parameters are extrapolated to the ECAL in all cases  Can be extrapolated from the innermost track measurement toward the SC position  eSuperclusterOverP(), deltaEtaSuperClusterAtVtx(), deltaPhiSuperClusterAtVtx()  Or extrapolated from the outermost track measurement toward the seed BC position  eSeedOverPout(), deltaEtaSeedClusterAtCalo(), deltaPhiSeedClusterAtCalo()  HadOverEm()  Ratio of hadronic energy in tower just behind the SC over the SC energy  Shower length variable, not hadronic isolation

15 Electron object  Electron corrections  caloEnergy() returns corrected superCluster energy when electron level scale corrections are turned on  isEnergyScaleCorrected() to know if correction was applied  default is off (done at the superCluster level)  momentum() returns the momentum estimate from the E-p combination  isMomentumCorrected() to know if correction was applied  default is on

16 Electron classes & errors  Electron algorithm makes use and provides a definition of electron categories derived from the SC-track pattern and measurement quality (bremsstrahlung)  4 basic categories: golden, narrow, big-brem, showering (from the mostly non radiating to the mostly more radiating)  + basic EB/EE separation and a category for all electrons in crack (EB/EE transition region) and gaps (EB eta gaps only)  Access to electron class through int GsfElectron::classification();  0 : golden EB  10: big-brem EB  20: narrow EB  30,31,32,..: showering EB with nbrem = 0,1,2,..  40: crack and gaps  class += 100: same for EE  Errors on E measurement derived from classification + momentum error from the fit  float caloEnergyError(), trackMomentumError()

17 Electron Collections  Main collection recoGsfElectrons_pixelMatchGsfElectrons__RECO (object type) ( producer) (process)  root [0] TFile::Open(’’rfio:/castor/cern.ch/cms/store/..../reco.root’’)  root [1] TBrowser b

18 Electron Collections

19 Example reading code See  RecoEgamma/Examples/test/GsfElectronMCAnalyzer_cfg.py runs a module that reads the electron collection and produces histograms in a root file  makes use of MC truth

20 Electron ID  Configurable tool based on PhysicsTools available to perform electron ID  RecoEgamma/ElectronIdentification  Seeveral algorithms are available:  cut based (no categories), cut based (3 categories), cut based (electron classes), likelihood, neural net  For the cut based approaches, configuration specified in files  RecoEgamma/ElectronIdentification/python/cutBasedElectronId_cfi.py  RecoEgamma/ElectronIdentification/python/ptdrElectronId_cfi.py  The following can be specified in these config files:  loose or tight electron quality for the cut based eID no categories and 3 categories  Loose, medium or tight electron quality for the cut based eID using electron classes  List of discriminating variables to be used (cut based eID using electron classes)  Individual cut values  In the standard sequence, cut based (no categories) and cut based (3 categories) are run and provide a collection of references to the GsfElectron collection

21 Electron ID  runElectronID.cfg under /test runs the electronId modules and outputs a root tree:  More in (Roberto Salerno)

22 Further references  SWGuide      Workbook Electron   DataFormats  mma mma Feedback most welcome on usage & documentation