Low-pT electrons

This TWiki briefly documents the support available for and the performance of low-pT electrons.

Authors and contacts

Authors: R. Bainbridge, F. Cavallari, N. Marinelli, C. Rovelli, M. Verzetti.

Contact: ?

Quick start guide

Low-pT electrons are fully supported in ... blah ... (very short statement).

Documentation

"Recording and reconstructing 10 billion unbiased B meson decays in CMS to probe lepton flavour universality"
EPJ Web of Conferences 245, 01025 (2020), CMS-CR-2020-081

"Recording and reconstructing 10 billion unbiased b hadron decays in CMS"
CMS-DP-2019-043

Analysis note

Relevant talks:

  • Status report to XPOG, 28th April 2021 here

Etc?...

Algorithm overview

List of modules and their functions

Support for CMSSW data sets and simulation

  • All CMS data sets are supported except for Heavy Ion
  • Fast Simulation is not yet supported

Support for CMSSW releases and data tiers

Recommended

  • The recommended release version is CMSSSW_10_6_20 for studies with UltraLegacy data sets.
  • The very latest developments can be found in the development release cycle, currently 11_3_X
  • We attempt to back port useful developments to previous release cycles (currently only 10_6_X)

The table (and key/footnotes) below provides the details of status for the supported release cycles.

How to read the table? Choose the CMSSW release cycle you wish to work with and then choose a release version (i.e. the column). Look down the column to see what you get by default and by applying the available eras and modifiers (e.g. bParking, run_miniAOD_UL, run_miniAOD_devel, ...).

Release Cycle 10_6_X 11_3_X
CMSSW Version 10_6_20 10_6_26 (PR merged) 11_3_0
Recommended? Yes Yes Yes
RECO/AOD data tiers
Default behaviour Modifiers: None
Ele: Yes (Sequence,Collections)
Thr: pT > 1.0 GeV, Tight Seed
Rekey: No
EReg: No [1]
ID: Autumn18 [2]
Already used in earlier release for UL RECO of non-BParking data sets.
Sub-optimal ID.
No longer recommended.
Listed for completeness.
Modifiers: None
Ele: Yes (Sequence,Collections)
Thr: pT > 1.0 GeV, Tight Seed
Rekey: No
EReg: No [1]
ID: Autumn18 [2]
Sub-optimal ID.
Not recommended.
Listed for completeness.
Modifiers: None
Ele: Yes (Sequence,Collections)
Thr: pT > 1.0 GeV, Tight Seed
Rekey: Yes
EReg: Yes
GT(mc): xxx
GT(data): xxx
ID: 2020Nov28 [3]
Recommended for future (non-BParking) data sets.
(Only the "delta" for the bParking modifier w.r.t. default is shown below)
bParking era Modifiers: bParking
Thr: pT > 0.5 GeV, VLoose Seed
Not recommended.
Listed for completeness.
Modifiers: bParking
Thr: pT > 0.5 GeV, VLoose Seed
Not recommended.
Listed for completeness.
Modifiers: bParking
Thr: pT > 0.5 GeV, VLoose Seed
Recommended for future BParking data sets.
MINIAOD data tier
Default behaviour Modifiers: None
Pat: No
Modifiers: None
Pat: No
Modifiers: None
Pat: Yes [5]
Select: pT > 1.0 GeV, ID > -0.25 [6]
Recommended for future (non-BParking) data sets.
(Only the "delta" for various combinations of the bParking, run2_miniAOD_UL, and run2_miniAOD_devel modifiers w.r.t. default is shown below)
UL re-miniAOD
No bParking era
Modifiers: (~bParking & run2_miniAOD_UL)
Rekey: Yes
ID: 2020Sept15 [4] (Rerun on AOD)
Pat: Yes [5]
Select: pT > 1.0 GeV, ID > 1.5 [7]
Already used for UL re-miniAOD of non-BParking data sets (Winter 2020 campaign).
(Subject to no-change rule.)
No longer recommended.
Modifiers: (~bParking & run2_miniAOD_devel)
Rekey: Yes
ID: 2020Nov28 [3] (Rerun on AOD)
Pat: Yes [5]
Select: pT > 1.0 GeV, ID > -0.25 [6]
(Subject to no-change rule.)
No longer recommended.
Modifiers: (~bParking & run2_miniAOD_UL)
Rekey: Yes
EReg: Yes (Rerun on AOD)
GT(mc): xxx
GT(data): xxx
ID: 2020Nov28 [3] (Rerun on AOD)
Pat: Yes [5]
Select: pT > 1.0 GeV, ID > -0.25 [6]
Can be used for future UL re-miniAOD of non-BParking data sets.
bParking era
No UL re-miniAOD
Modifiers: (bParking & ~run2_miniAOD_UL)
Rekey: Yes
ID: Autumn18 (Rerun on AOD)
Pat: Yes [5]
Select: None (i.e. open)
Sub-optimal ID.
Not recommended on top of re-RECO of bParking data sets.
Listed for completeness.
Modifiers: (bParking & ~run2_miniAOD_UL)
Rekey: Yes
ID: 2021May17 (Rerun on AOD)
Pat: Yes [5]
Select: None (i.e. pT > 1.0 GeV, no ID req.)
Not recommended.
Modifiers: (bParking & ~run2_miniAOD_UL)
Pat: Yes
Select: None (i.e. open)
Recommended for future BParking data sets.
UL re-miniAOD
bParking era
Modifiers: (bParking & run2_miniAOD_UL)
Rekey: Yes
ID: 2020Sept15 [4] (Rerun on AOD)
Pat: Yes [5]
Select: None (i.e. open)
Not recommended on top of re-RECO of bParking data sets.
Listed for completeness.
Modifiers: (bParking & run2_miniAOD_UL)
Rekey: Yes
EReg: Yes (Rerun on AOD)
GT(mc): xxx
GT(data): xxx
ID: 2021May17 [4] (Rerun on AOD)
Pat: Yes [5]
Select: None (i.e. pT > 1.0 GeV, no ID req.)
Recommended for UL re-miniAOD of bParking data sets.
Modifiers: (bParking & run2_miniAOD_UL)
Pat: Yes
Select: None (i.e. open)
Recommended for future BParking data sets.

NANOAOD data tier
Default behaviour No No No
(Below: only "delta" w.r.t. default is shown)
To come...

Key:

  • Modifiers: indicates the logic required to obtain the desired behaviour (in certain cases, the logic for past campaigns is implicitly rather than explicitly coded).
  • Ele: indicates if reconstructed low-pT electrons are available by default in the RECO/AOD data tiers.
  • Thr: indicates thresholds applied during reconstruction:
    • ElectronSeed working point, e.g. Tight or V(ery)Loose (see Performance section below for details)
    • Minimum seed track pT, e.g. 0.5 or 1.0 GeV
  • EReg: indicates if energy regression is applied by default.
  • Rekey: indicates if the ValueMaps holding the Seed BDT scores are rekeyed by reco::GsfElectronRef
  • ID: indicates if ID is available in the form of a discriminator score from the specified BDT model. Further details can be found in the Performance section.
  • Pat: indicates if "low-pT" PAT electrons are available by default in the MINIAOD data tier
  • Select: indicates thresholds applied during the PAT select step, e.g. on the ID score and/or pT

Footnotes:

  • Current recommendations given in coloured text: recommended, possible / previously used, not recommended.
  • Embedded URLs typically point to the relevant CMSSW python configuration.
  • [1] The absence of an energy regression does not significantly affect the ID performance. A possible "patch" is to build the electron 4-momentum from GsfTrack mode momentum and electron mass.
  • [2] Recipes are available to run the latest ID algorithm as a post-processing step. in case a model is not available by default in the release or to supersede an existing model. While this is discouraged, please contact the authors for details if required.
  • [3] The "2020Nov28" ID model was trained with CMSSW_10_6_20 using 10_6_X MC B->Kee (UltraLegacy) samples.
  • [4] The "2020Sept15" ID model was trained with CMSSW_10_2_15 using 10_2_X MC B->Kee samples.
  • [5] The MINIAOD contains two collections: the "slimmedLowPtElectrons" and the "gsfTracksOpenConversions". The former embeds all useful information (BDT-based seed and ID discriminator scores, GsfTrack and cluster info, etc). The latter is used to identify low-pT electrons from photon conversions.
  • [6] The pT and ID score thresholds are tuned to give a 90% signal efficiency and add 2% to the default MINIAOD footprint

Minimal support

  • The release cycle 10_2_X is not recommended. (This release cycle is used by some of the ongoing BParking-based analyses, e.g. R(K) using 2018 data. However, the code base relies heavily on cms-merge-topic recipes supported by experts.)
  • The release cycle 11_2_X is not recommended.

The table (and key/footnotes) below provides the details of status for unsupported release cycles.

Release Cycle 10_2_X 11_2_X
CMSSW Version 10_2_15 11_2_0
Recommended? No No
Default Processing
RECO/AOD Ele: Yes
Thr: pT > 1.0 GeV, Tight
EReg: No
ID: Autumn18 Sub-optimal!
Ele: Yes
Thr: pT > 1.0 GeV, Tight
EReg: No
ID: ID: Autumn18 Sub-optimal!
MINIAOD Pat: No Pat: No
NANOAOD No No
(Re)processing with the bParking era
RECO/AOD Ele: Yes
Thr: pT > 0.5 GeV, VLoose
Ele: Yes
Thr: pT > 1.0 GeV, VLoose
MINIAOD Pat: Yes Pat: Yes

GT: MC: 106X_upgrade2018_realistic_v16_L1v1 Data: 106X_dataRun2_v33

Key:

  • Ele: indicates if reconstructed low-pT electrons are available by default in the RECO/AOD data tiers.
  • Thr: indicates thresholds applied during reconstruction ...
    • ElectronSeed working point, e.g. Tight or V(ery)Loose (see Performance section below for details)
    • Minimum seed track pT, e.g. 0.5 or 1.0 GeV
    • ... OR indicates thresholds applied during the PAT select step, e.g. on the ID score
  • ID: indicates if ID is available in the form of a discriminator score from the specified BDT model. Further details can be found in the Performance section.
  • EReg: indicates if energy regression is applied by default.
  • Pat: indicates if "low-pT" PAT electrons are available by default in the MINIAOD data tier

Recent developments

Recent PRs to CMSSW are detailed below

  • #31220
    • Merged to master on 2 Dec 2020, entered CMSSW_11_3_0
    • "Add low-pT electrons to MINIAOD, update ID, improve end user experience"

Code examples

This section provides some concrete code examples on how to use the low-pT electrons.

Performance

Shown below are some representative performance plots.

Analyses using low-pT electrons

Several analyses have expressed an interest in using low-pT electrons. Below is a (non-exhaustive) list of those we are aware of. Please volunteer to add your information here if you plan to use low-pT electrons.

  • A. N. Other, some analysis ...
Edit | Attach | Watch | Print version | History: r141 < r140 < r139 < r138 < r137 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r139 - 2021-07-06 - RobBainbridge
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Main All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2021 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback