Cut Based Electron ID

See the following presentation for the description of the algorithm.

Starting from CMSSW 3_1_X has been implemented a traceability system for the different identification algorithms. It allows to decouple the versions of the eID from the CMSSW releases and helps to keep track of the various selection working points evolving in time. Basically the electronIdCutBased_cfi.py file is used like a DB and the PSet where the cuts are defined has a meaningful name:

electronIDType+electronQuality+"EleIDCut"+electronVersion
So for example the set of cuts for Loose Robust selection will be identified as robustlooseEleIDCutsV00. See the following presentation for the description of eID traceability

"Simple Cuts"

Most recent information can be found at SimpleCutBasedEleID

This kind of selection is aimed for early data taking. It has been designed to be as simple, efficient and robust as possible. No electron classification is involved here. Cuts are applied on the following 4 variables (different value for barrel and endcaps):

  • Ratio of energy in HCAL behind SuperCluster to SuperCluster energy
  • 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
  • SigmaIEtaIEta cluster shape covariance (SigmaEtaEta was used in 1_6_X)

Note that sigmaIEtaIEta is designed to be almost identical to an eta corrected sigmaEtaEta except that it is more stable vs eta.

The Fixed Threshold method has three levels of identification: a loose set of cuts (sometimes is referred to as "Robust'') to deal with the identification of electrons coming from Z decays, a second one for W decay where a tighter identification is needed and a third one for high energy electrons.

Loose

The set of thresholds (barrel and endcap) used in the loose Fixed Threshold Identification (previously referred like "Robust") for barrel and endcap

Version 00

Optimization of the cut values using CMSSW 16X samples

Variable barrel endcap
H/E 0.115 0.150
|Δηin| 0.0090 0.0105
|Δφin| 0.090 0.092
σηη 0.0140 0.0275

Version 01
Optimization of the cut values using CMSSW 22X samples

Variable barrel endcap
H/E 0.075 0.083
|Δηin| 0.0077 0.0100
|Δφin| 0.058 0.042
σiηiη 0.0132 0.027

Tight

The set of thresholds (barrel and endcap) used in the tight Fixed Threshold Identification for barrel and endcap

Version 00

Optimization of the cut values using CMSSW 16X samples

Variable barrel endcap
H/E 0.015 0.018
|Δηin| 0.0025 0.0040
|Δφin| 0.020 0.020
σηη 0.0092 0.025

Version 01
Optimization of the cut values using CMSSW 22X samples

Variable barrel endcap
H/E 0.01 0.01
|Δηin| 0.0040 0.0066
|Δφin| 0.025 0.020
σiηiη 0.0099 0.028

High Energy

The set of thresholds (barrel and endcap) used in the Fixed Threshold Identification for High Energy electron for barrel and endcap

Version 00

Optimization of the cut values using CMSSW 16X samples

Variable barrel endcap
H/E 0.050 0.100
|Δηin| 0.005 0.007
|Δφin| 0.09 0.09
σηη 0.011 0.0275

Version 01
Optimization of the cut values using CMSSW 22X samples

Variable barrel endcap
H/E 0.05 0.05
|Δηin| 0.005 0.007
|Δφin| 0.09 0.09
σiηiη n/a 0.0275
E2x5/E5x5 >0.94 OR E1x5/E5x5 > 0.83 n/a

"Cuts in Categories"

A detailed description of this identification can be found at the following link.

Version 00

Optimization of the cut values using CMSSW 16X samples

Version 01

Optimization of the cut values using CMSSW 22X samples

Version 02

Optimization of the cut values using CMSSW 33X samples

"Class Based"

This eID is devote to the HZZ/MultiLepton analysis, that are the most demanding analysis for low pT electrons (down to pT of 5 GeV/c) and they require the highest possible efficiencies for eID (epsilon^4). It is especially important to optimize eID and eIsolation against the specific background (e.g. Zbb at low M_H).

See the following presentation for finding more details and more references.

The following variables are currently used to discriminate between real and fake electrons:

  • Energy of closest BasicCluster to track impact point at ECAL / outermost track momentum
  • Ratio of energy in HCAL behind SuperCluster to SuperCluster energy
  • 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 x electron charge
  • SigmaiEtaiEta cluster shape covariance

The optimization of this eID was performed on MC HZZ events with at least 3 electrons (see slides 5 of the linked presentation), it means that a set of electrons, that are not necessary firing the trigger, are used. The working point obtained is a high efficiency one: 95% for all the classes and for electron with = 10 GeV/c

Cut Based Electron ID Performance/Validation

The follow page show the electron ID performance, in terms of efficiency in signal samples vs eta and pT for several CMS.RelVal samples:

Edit | Attach | Watch | Print version | History: r14 < r13 < r12 < r11 < r10 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r14 - 2010-10-07 - RobertoSalerno
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    CMSPublic All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback