Jet Identification 

Convenor: Murilo Rangel, Team: Gabriel Facini (jetid_eff, vtx. conf.), Zdenek Hubacek (jetreco).

This group is charged with reconstructing and identifying jets for Run II.

Recommended Jet-ID cuts:

• Are implemented in GoodJet.cpp in the package jet_evt and applied by D0reco.
• The jet ID selection is also implemented in the CAFe JetSelector in caf_util.
  

For Summer 2009 dataset:

• For Summer 2009 data, JetID and Vertex Confirmation Data/MC efficiencies scale factors have been derived.
  Please use the package jetid_eff v03-01-02.
• HowTo: Instructions on how to apply SF in CAFe.
• For more details see this talk .
• Taggability SF are also available in the jetid_eff package (use the version v03-01-04), but the user should consider whether or not it is useful for his/her analysis. For more details see this talk .

For p20 MC and Winter2009 dataset:

• For Winter2009 data, JetID Data/MC efficiencies scale factors have been rederived on Nov 8th 2008.
  Please use JetID efficiencies as in the package jetid_eff v03-00-02. Note that systematics have been kept from p20 preshutdown.
• HowTo: Instructions on how to apply JetID SF in CAFe.

For p20 MC and 2007-pre/post-shutdown data, and Ichep2008 dataset:

• For Ichep2008 data, JetID has been rederived (see additional material of the p20 jetid note below), with systematics kept from p20 preshutdown.
• p20 Ichep2008, Preliminary p20 preshutdown and postshutdown JetID efficiencies are in the package jetid_eff v02-00-05 (June 6th 2008).
• Instructions on how to apply JetID SF in CAFe.
• p20 Jet ID criteria remain unchenged w.r.t. p17, only the corresponding efficiencies have been remeasured.
• Documented in D0 note #5634 (April 22nd 2008).
• More information: Effect of the maximal emf fraction cut (talk at CALGO 6th Nov 2007, also includes preliminary p20 Data and MC Reco/ID efficiency plots).

Jets in-depth:

• Some Remarks on Cone Jets or: Cone jets are not circles! [Jul 16th 2008]
• Documentation

For p17:

• Documented in D0 Note #4919 (jet ID optimization)
• Documented in D0 Note #5250 (Measuring Reconstruction*Jet-ID efficiencies using the tag and probe method in p17)
• Jet ID/Reco efficiency measurements: D0 Note #5218 (Combined JetID efficiency for p17)
• IMPORTANT MESSAGE FOR p17 CAF ANALYSES:
  The default config file for JetSelector selects good jets (flag isGood) but also applies an EM veto  (flag !isEM) which removes jets whose distance in DeltaR to their closest EM cluster  is smaller than dRmin_Jet_EM  (default value: 0.4 in jetreco/rcp/JetRecoCorr5.rcp). This veto, which is NOT part of good jet selection, was originally introduced for MET calculation, where it is beneficial, but was abusively extended to other applications: for instance, it was  included for b-tagging, where it can cause 10% inefficiency in top analyses (note that the b-id group already fixed this and that version #1 of p17 b-tagging will not have it).
  
  It is thus recommended to:
  
  • keep this veto or a similar one, for MET calculation
  • improve this veto by recomputing !isEM in order to reject only jets which are close to an electron candidate having passed much tighter cuts, for jet selection.
  Note that it is also recommended to change the matching distance between electron and jet from 0.4 to 0.5, to be consistent with the cone radius (for JCCB jets).
  
  The following config files, setup by Dag Gillberg and Yann Coadou for the Single Top group, can be used as an example (see in /D0/dist/packages/top_cafe/devel/configs/ for most up-to-date versions):
  Jet.config
  EM.config
  MET.config
  Note in this example that the two EMJet removal schemes used for different purposes both happen after applyJES.
  The full flow for jets used in event selection is:
   * CAF Jet -> JES -> EMJet removal with loose electron = MatchedJet -> GoodJet in JetSelector
  while that for jets used  for MET correction is:
   * CAF jet -> JES -> EMJet removal with "d0correct" EM's = MetJet -> used in RecomputeMET.
  
  In any case, it is the responsability of each analyser:
• to choose which EMJet removal procedure(s) to apply for jet selection and/or MET calculation, either the default one (not recommended), the Single Top group one above (recommended), or another procedure more suited to his (her) analysis, if needed,
•  to calculate the efficiency of  the EMJet removal procedure chosen for jet selection, for the event topology he (she) wants to study.
  
  
• The jet ID selection is also implemented in the CAFe processor JESGoodJet in the package jes_tools, which can either use the official ID or just reapply the cuts you want. A sample config file is:
  /D0/dist/packages/jes_tools/devel/rundata/jes_tools_basic.config
  
  
• Old: Please have a look at our Projects list of ongoing projects.
  

For p14 pass II:

• CHF<0.4 & EMF>0.05: to remove instrumentation noise. Both cuts would have needed to be optimized for the ICR region (done for p17).
• EMF<0.95:  to remove electrons and photons. This cut also improves the jet energy resolution. We may consider removing it once the MET group decides on a recommended procedure to remove electron-jet and photon-jet overlaps.
• L1 confirmation: L1SET / (jet pT * (1 - CHF)) > 0.4
  
  • This is the cut to be applied when L1SET takes into account ICD remapping (default in d0correct).
    
  • Every jet with L1SET>55 GeV should also be considered L1 confirmed.
• Jets taken before the L1 readout was complete (approx. the first 65 pb-1 of data) whose pT is outside the L1 coverage should use the older cuts (which L1 confirmation replaced):  f90 < 0.5 || CHF < 0.15
• Note that top_analyze:
  
• does not use remapping and thus uses a different cut in the ICD region (0.8 < |eta| < 1.5), but which gives similar results in terms of efficiency (see Jet-ID criteria for p14 pass2 certification for details),
  
• does not confirm all jets with L1SET>55 GeV (this will falsely reject jets starting from ~150 GeV, which is not an issue for most top analyses),
• does not take partial coverage into account (this will falsely rejects jets at high |eta| in the first 65 pb-1 of data, which is again not an issue for most top analyses),
• demands the old n90>1 and HotF<10 cuts, which are no longer needed (but are not harmful either).

Useful links:

• Documentation (important)
• Certification Results (old)
• Transparencies (old)
• Software (old)
• Activities (super-old)