2010 HI Dilepton Trigger Study with Tag and Probe Method

Basement of Study

• General Tag and Probe Method : link
• Muon Tag and Probe Method : link
• Tag and Probe Software : link
• Quarkonia Task Force T & P : link
• Tutorial : link

Motivation

• Compare to the Trigger Efficiency between using MC and Tag & Probe method for the November HI data

Selection Condition

• HI Dilepton Muon Sellection for Global Muon reconstruction Efficiency (20101103 updated) Example for Z0
  • Definition of Tag muons : SinlgeMu3 (hltL1Mu3L1Filtered0)

         process.tagMuons = cms.EDFilter("PATMuonSelector",
                 src = cms.InputTag("patMuonsWithTrigger"),
                 cut = cms.string("isGlobalMuon && !triggerObjectMatchesByFilter('hltL1Mu3L1Filtered0').empty()"),
         )

• • Definition of Probe muons

         process.tkTracks  = cms.EDProducer("ConcreteChargedCandidateProducer", 
                 src  = cms.InputTag("hiGlobalPrimTracks"),     # L4 
                 particleType = cms.string("mu+"),
         ) 
         process.tkProbes = cms.EDFilter("CandViewRefSelector",
                 src = cms.InputTag("tkTracks"),
                 cut = cms.string("pt > 0 && abs(eta) < 2.4"),
         )

• • Passing probe condition : inside of mass cut region and triggered events

         process.tpPairs = cms.EDProducer("CandViewShallowCloneCombiner",
                 #cut = cms.string('2.6 < mass < 3.6'),
                 cut = cms.string(''),
                 decay = cms.string('tagMuons@+ probeMuons@-')
         )
         process.tkToGlbMatch = cms.EDProducer("MatcherUsingTracks",
                 src     = cms.InputTag("tkTracks"), # all tracks are available for matching
                 matched = cms.InputTag("glbMuons"), # to all global muons
                 algorithm = cms.string("byDirectComparison"), # check that they
                 srcTrack = cms.string("tracker"),             # have the same 
                 srcState = cms.string("atVertex"),            # tracker track
                 matchedTrack = cms.string("tracker"),         # can't check ref
                 matchedState = cms.string("atVertex"),        # because of the
                 maxDeltaR        = cms.double(0.01),          # embedding.
                 maxDeltaLocalPos = cms.double(0.01),
                 maxDeltaPtRel    = cms.double(0.01),
                 sortBy           = cms.string("deltaR"),
         )
         process.tkPassingGlb = cms.EDProducer("MatchedCandidateSelector",
                 src   = cms.InputTag("tkProbes"),
                 match = cms.InputTag("tkToGlbMatch"),
         )

• HI Dilepton Muon Sellection for Trigger Efficiency (20101103 updated): Example for Z0
  • Definition of Tag muons : SinlgeMu3 (hltL1Mu3L1Filtered0), SingleMu5 (hltL1Mu5L1Filtered0) .. etc

         process.tagMuons = cms.EDFilter("PATMuonSelector",
                 src = cms.InputTag("patMuonsWithTrigger"),
                 cut = cms.string("isGlobalMuon && !triggerObjectMatchesByFilter('hltL1Mu3L1Filtered0').empty()"),
         )

• • Definition of Probe muons

         process.probeMuons = cms.EDFilter("PATMuonSelector",
                 src = cms.InputTag("patMuonsWithTrigger"),
                 cut = cms.string("isGlobalMuon"),
         )

• • Passing probe condition : inside of mass cut region and triggered events

         process.tpPairs = cms.EDProducer("CandViewShallowCloneCombiner",
                 #cut = cms.string('2.6 < mass < 3.6'),
                 cut = cms.string(''),
                 decay = cms.string('tagMuons@+ probeMuons@-')
         )

• Default of QTF TnP :
  • Muon Collection : mergedCaloMuons

         # Merge calomuons into the collection of muons
         from RecoMuon.MuonIdentification.calomuons_cfi import calomuons; 
         process.mergedMuons = cms.EDProducer("CaloMuonMerger",
             muons     = cms.InputTag("muons"), 
             caloMuons = cms.InputTag("calomuons"),
             tracks    = cms.InputTag("hiSelectedTracks"),
             minCaloCompatibility = calomuons.minCaloCompatibility,
             muonsCut  = cms.string("track.isNonnull"),
             caloMuonsCut = cms.string(""),
             tracksCut    = cms.string("pt > 2 || (abs(eta) > 1 && p > 2)"),
         )

• • TagMuons : globalMuons

         process.tagMuons = cms.EDFilter("PATMuonSelector",
                 src = cms.InputTag("patMuonsWithTrigger"),
                 cut = cms.string("isGlobalMuon"),
         )

• • ProbeMuons : allMuons (all tracks among the mergedCaloMuons)

         process.probeMuons = cms.EDFilter("PATMuonSelector",
                 src = cms.InputTag("patMuonsWithTrigger"),
                 cut = cms.string("track.isNonnull"),
         )

• Total Efficiency : e_trk * e_reco * e_trg
  • Definition of each of Efficiency = (Passing Probe)/(All of pairs)

Recipe with HI MC

• Option changed accroding to HI Dilepton group selection (2010/10/22)
  • Prod cfg : tp_hi_Prod_mc_SM3.py (the case of matched with SingleMu3)
  • Ana cfg : tp_hi_Ana_mc_SM3.py (the case of matched with SingleMu3)
• Only making PAT Muons : hiPatMuonsProducer_cfg.py
• Setup update (2010/10/15)
  • Make UserCode/DonghoMoon/HiTagAndProbe package
  • script : script_pre5
  • hiPatMuonsWithTrigger_cff.py : setup with Heavy Ion environment to produce HI PATMuons
• How to use
  • QTF method
    • tp_hi_mc.py : modified from QTF method to produce Fit Tree (input RecoSamples)
    • hiFitTrigger.py : modified from QTF method to analyze Tree information of output file of tp_hi_mc.py
  • General TnP
    • tp_hi_Prod_mc.py : modified from testTagAndProbeFitTreeProducer_JpsiMuMu.py to produce Fit Tree (input RecoSamples)
    • tp_hi_Ana_mc.py : modified from testTagAndProbeFitTreeProducer_JpsiMuMu.py to analyze Tree information of output file of tp_hi_Prod_mc.py

• Setup at 390pre4 and pre5 (2010/10/11)

     addpkg PhysicsTools/TagAndProbe V02-06-11 (or later)
     addpkg MuonAnalysis/MuonAssociators V01-10-00
     addpkg MuonAnalysis/TagAndProbe V07-01-00

     addpkg PhysicsTools/PatAlgos
     cvs update -rV08-01-02 PhysicsTools/PatAlgos/python/tools/helpers.py

• 1. Following QTF Method : MuonAnalysis/TagAndProbe/ (390pre4)
  • remove MuonAnalysis/TagAndProbe/plugin/MuonSelectorJPsi.cc, MuonSelectorJPsiPlusMu.cc, MuonSelectorWmunu.cc (not build at 390pre4, but don't need for HI PAT)
  • in the /afs/cern.ch/user/d/dmoon/public/html/TagAndProb
    • move hiPatMuonsWithTrigger_cff.py into your MuonAnalysis/MuonAssociators/python
    • with tp_hi_mc.py which use reco input file (not pat muons), run process make pat and TagAndProbeTreeProducer for Analysis
    • input outfile of tp_hi_mc.py into hiFitTrigger.py and run it
  • tp_hi_mc.py : modified from tp_from_aod_simple_MC.py - process order : recoMuFilter >> Make PAT Muons with Trigger Match >> TagAndProbeTreeProducer
  • hiFitTirgger.py : modified from fitTrigger.py - TagAndProbeTreeAnalyzer

• 2. Following the Method to use "Reco Sample" directly (not using PAT) : PhysicsTools/TagAndProbe/ (390pre5)
  • in the PhysicsTools/TagAndProbe/test/
    • TagAndProbeFitTreeProducer_JpsiMuMu.py : input sample (Reco)
    • TagAndProbeFitTreeAnalyzer_JpsiMuMu.py : input file (output file of FitTreeProducer)

• 3. Changing 2nd method with PAT muons (ongoing)
  • Making PAT muons >> TagAndProbeFitTreeProducer : tp_hi_Prod_mc.py
  • TagAndProbeFitTreeAnalyzer : tp_hi_Ana_mc.py

Result

• Samples : 5k Z0 Embedded on MB Hydjet, /castor/cern.ch/user/d/dmoon/cms390pre5/Hydjet_MinBias_2.76TeV_Z0_Flat_Emb_HLT_SMX_Reco/
• tp_hi_Prod_mc_SM3_Z0.py
• TrgResultZ0

Global Muon Reconstruction Efficiency for Z0 (20101103 updated)

• Samples : 5k Z0 Embedded on MB Hydjet, /castor/cern.ch/user/d/dmoon/cms390pre5/Hydjet_MinBias_2.76TeV_Z0_Flat_Emb_HLT_SMX_Reco/
• tp_hi_Prod_mc_GLB_Z0.py
• GlbResultZ0

L1DoubleMuOpen Trigger Efficiency

• Samples : 3000 J/psis Embedded on MB Hydjet, /castor/cern.ch/user/d/dmoon/cms390pre5/Hydjet_MinBias_2.76TeV_Jpsi_Flat_Emb_HLT_Reco/
• flag setup in FitTreeProducer : tp_hi_Prod_mc.py

      flags = cms.PSet(
            Glb = cms.string("isGlobalMuon"),
            TM = cms.string("isTrackerMuon"),
            L1DoubleMuOpen = cms.string("!triggerObjectMatchesByFilter('hltDoubleMuLevel1PathL1OpenFiltered').empty()"),
      ),

• eta and pt bin size modified and L1DoubleMuOpen Trigger added in FitTreeAnalyzer : tp_hi_Ana_mc.py

        L1DoubleMuOpen_pt_eta = cms.PSet(
            EfficiencyCategoryAndState = cms.vstring("L1DoubleMuOpen","true"),
            UnbinnedVariables = cms.vstring("mass"),
            BinnedVariables = cms.PSet(
                #pt = cms.vdouble(3.0, 6.0, 20.0),
                #pt = cms.vdouble(0.0, 3.0, 6.0, 9.0, 12.0, 15.0, 18.0, 20.0),
                pt = cms.vdouble(0.0, 2.0, 4.0, 6.0, 8.0, 10.0),
                #abseta = cms.vdouble(0.0, 1.2, 2.4),
                #abseta = cms.vdouble(0.0, 0.6, 1.2, 1.6, 2.4),
                eta = cms.vdouble(-2.4, -1.2, 0.0, 1.2, 2.4),
            ),  
            BinToPDFmap = cms.vstring("gaussPlusLinear", "*pt_bin0*", "gaussPlusQuadratic")
        ),

• Trigger Efficiency pt and eta depending on each pt and eta bins

eta eff pt_bin0	eta eff pt_bin1	eta eff pt_bin2	eta eff pt_bin3	eta eff pt_bin4

pt eff eta_bin0	pt eff eta_bin1	pt eff eta_bin2	pt eff eta_bin3

• One of Example for Mass distribution and fitting as the passing/failing/all probes

Mass distribution Passing/Failing/All probes	Mass fitting Passing/Failing/All probes

• L1DoubleMuOpen? Trigger Efficiency pt vs eta

Trigger Efficiency L1DoubleMuOpen pt vs eta

Bug report

• 2010/10/15 : Error message from Making PAT Muons

      %MSG
      %MSG-e DataNotAvailable:  PATMuonProducer:patMuonsWithoutTrigger  15-Oct-2010 12:53:13 CEST Run: 1 Event: 3
      No primary vertex available from EventSetup, not adding high level selection     

• • it came from that patMuons dosen't match Heavy Ion collections, it should be changed as below
  • fixed version of hiPatMuonsWithTrigger_cff.py

      beamLineSrc = cms.InputTag("hltOfflineBeamSpot"), # Heavy Ion BeamSpot
      pvSrc = cms.InputTag("hiSelectedVertex") # Heavy Ion vertex collection

• Trigger info problem

      %MSG
      %MSG-e errorTriggerResultsValid:  PATTriggerProducer:patTrigger  15-Oct-2010 12:53:12 CEST Run: 1 Event: 2
      TriggerResults product with InputTag TriggerResults::HLT not in event  
      No HLT information produced. 

• • This is coming from that can't get trigger result correctly due to different process name
  • solved, to add 2 lines in the configuration file as below

      from MuonAnalysis.MuonAssociators.hiPatMuonsWithTrigger_cff import changeTriggerProcessName
      changeTriggerProcessName(process, "HISIGNAL")   # Custom re-run HLT

Todo list

• tuning the environmental setup of Analyzer
• bug fixing trigger info and vertex problem in making PATMuons

-- DongHoMoon - 11 Oct 2010