In physics analysis with multi-electron final states high identification efficiency is needed to enhance signal selection, in
particular at low E_T where the background increases and the fake rate is much higher. In the current implementation Electron Identification is considered a separate step with respect to Electron Reconstruction and in the following can be found information on how to run the Identification algorithms in CMSSW. Brief descriptions of the algorithms and their performances are also described.
Electron ID Algorithms
The following variables are currently used to discriminate between real and fake electrons:
SuperCluster energy / track momentum at vertex
DeltaEta between SuperCluster position and track direction at vertex extrapolated to ECAL assuming no radiation
DeltaPhi between SuperCluster position and track direction at vertex extrapolated to ECAL assuming no radiation
Ratio of energy in HCAL behind SuperCluster to SuperCluster energy
Energy in 3x3 crystals / energy in 5x5 crystals
Energy of closest BasicCluster to track impact point at ECAL / outermost track momentum
Energy of closest BasicCluster to track impact point at ECAL / innermost track momentum
DeltaPhi between track impact point at ECAL and closest BasicCluster
1/E(SuperCluster) - 1/p(track at vertex)
Brem fraction = (track momentum at vertex - track momentum at ECAL)/(track momentum at vertex)
SigmaEtaEta cluster shape covariance
SigmaPhiPhi cluster shape covariance
A selected subset of the above variables can be combined to perform electron ID, using cuts, or multivariate techniques such as a likelihood, or neural net.
Description and performance of the implemented algorithms are available:
SWGuideCutBasedElectronID, "Simple Cuts", "Cuts in Categories" and "Class Based" Electron Identification
The eID algos are published in the release.
In the events the eID results are stored with value maps for the loose, tight and high energy Fixed Threshold Identification and loose and tight Category Based identification.
There is still the possibility to run on-fly all the electron identification algorithms and saved the results using both a collection of references and value maps.
An example of cfg python file for running on fly the eID algorithm is provided in
here