Triggering On Hadronic Tau Decays: A challenge met by ATLAS MarcusM.MorgensternOnbehalfoftheATLASCollaboration TechnicalUniversityDresden,Dresden,Germany Abstract. TheATLASexperimentattheLargeHadronCollider(LHC)hasbeenabletocollect5.25fb−1 of datain2011.FormanyphysicsanalysesbothincontextoftheStandardModel(SM)andBeyondtheStandard Model (BSM) theories such as Higgs boson searches, tau leptons play an important role. Thus, triggering on hadronic tau decays is an essential ingredient for the success of those measurements. This contribution will 2 1 summarizethedevelopedeffortstomeetthischallenge.Efficiencymeasurementsusingdatatakenin2011ata 0 center-of-massenergyof7TeVaredescribedandresultsarepresented.Anoutlookonfurtherdevelopmentsof 2 thetautriggeralgorithms,tomatchfuturerequirementsandhigherinstantaneousluminositiesaresummarisedin theend. n a J 1 Introduction canbeappliedwhichismeasuredasthesumofenergyde- 6 positsinanisolationregioninaringof4 ×4 aroundthe 2 Tauleptonsplayacrucialroleinmanyphysicsprocesses coreregionof2 ×2 towers.Aso-calledRegionofInter- ] investigatedbytheATLASexperimentattheLargeHadron est(RoI)isbuiltduringtheL1processingandispassedto x Collider (LHC), both within the Standard Model and in the Higher Level Trigger (HLT), if the event is accepted. e modelsbeyondtheSM.InparticularforHiggsbosonde- To stay below the rate limit of 75kHz a hardware based - p cays in the SM at low masses and in the Minimal Su- pre-scale(PS)factorcanbeapplied.Next,thelevel2(L2) e persymmetric extension of the SM (MSSM) the branch- systemperformstrackreconstructionwithintheRoIinad- h ingfractiontotausisexpectedtobeintheorderof10%. dition to calorimeter information (using the full detector [ For this purpose it is important to trigger on tau decays granularity in the RoI) to identify tau leptons. Identifica- 1 efficiently.Thetautriggerisaimedatefficientlyrejecting tionvariablesmeasuringtaudecayproperties,suchasthe v backgroundeventswithfaketaucandidateswhilekeeping narrownessofthetaudecayproductsorthenumberofre- 2 theacceptanceforrealtausashighaspossible.Tauleptons constructedtracks,areusedtoprovideasufficientdiscrim- 9 decayhadronically65%ofthetimeandleptonically35% ination power against QCD jets. By setting requirements 4 of the time. Only hadronic tau decays are considered in on those identification variables, several working points 5 thiscontext. Taustendtodecay intooneor threecharged corresponding to given signal efficiencies are defined. In . 1 hadrons, mainly pions and to a small amount into kaons, additionaPScanbeapplied.Figure1presentstheelectro- 0 plusadditionalneutralpionsorkaons.Thus,thetaudecay magneticradius,R ,whichmeasuresthelateralshapeof EM 2 is characterised by one or three charged tracks and nar- thetaudecay.Figure2showsthenumberoftracksforL2 1 rowenergydeposistsintheelectromagneticandhadronic tausobtainedinsignalMonteCarlo(MC),andbackground : v calorimeter.Thesedecaypatternsbecomemanifestiniso- QCDdi-jetdata. i lationcriteriaappliedtodistinguishtaudecaysfromback- X ground,mainlyarisingfromQCDmultijetcontributions. r a 2 The ATLAS Tau Trigger ents 0.12 ATLASPreliminary v d e 0.1 Signal (cid:111) MC The ATLAS trigger system [1] is aimed at reducing the se Dijet Data (2011) initial collision rate of 40MHz to a reliable data rate for mali 0.08 permanentstorageofaround200-400Hz.Thisisachivied or N byathree-levelsystem.Thefirstsystemisthelevel1(L1) 0.06 trigger system. It is a hardware based system which uses information from the electromagnetic (EM) and hadronic 0.04 calorimetersystems,providedastriggertowersofsize∆η× ∆φ = 0.1 × 0.1. At level 1 a tau lepton is identified by 0.02 applyingagiventransverseenergy,E ,cutcorresponding T 0 to the given trigger item, e.g. for L1 tau15 a ET thresh- 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 oldstrictlyabove15GeVisapplied.Theenergyisrecon- L2 R EM structedbythesumofenergydepositsinthehighest2 ×1 pairofEMtowersandthesumofenergydepositsof2 ×2 Fig. 1. Electromagnetic radius, R , for trigger taus in signal em hadronic towers behind the EM layers. In addition to the Monte Carlo and QCD di-jet data. The latter serves as back- E threshold a cut on the energy in the isolation region groundestimate[2]. T EPJWebofConferences s 0.9 y 1 ed event 00..78 ATLASPreliminaSrigynal (cid:111) MC Efficienc 00..89 s Dijet Data (2011) 0.7 ali 0.6 m 0.6 Data 2011 Nor 0.5 0.5 Zfi t t (MC) 0.4 0.4 ATLAS Preliminary 0.3 0.3 s= 7 TeV, (cid:242) dt L = 475 pb-1 0.2 0.2 EF_tau20_medium1 0.1 0.1 0 0 10 20 30 40 50 60 70 80 90 100 0 0 1 2 3 4 5 6 7 8 9 p [GeV] T L2 Track multiplicity Fig. 3. Tau trigger efficiency measured in Z → ττ decays us- Fig. 2. Number of tracks of L2 taus for trigger taus in signal ing 2011 data by applying a tag-and-probe method for the EF Monte Carlo and QCD di-jet data. The latter serves as back- tau 20 mediumitem.Theefficienyismeasuredw.r.t.offlinetau groundestimate[2]. candidatespassingmediumidentificationrequirementsasafunc- tionoftheofflinetaup [2]. T Thelasttriggerstepisaccomplishedbytheeventfilter y 1 (EF),whichrunsonfullybuiltevents..Reconstructionand c identificationalgorithmssimilartotheonesusedforoffline cien 0.9 analysisareutilised[3].Dependingonthetriggerchaina Effi 0.8 PSfactorisappliedtomatchthefinalouputrateofaround 0.7 200-400Hz. 0.6 Data 2011 0.5 Zfi t t (MC) 0.4 ATLAS Preliminary 3 Tau Trigger Efficiency 0.3 s= 7 TeV, (cid:242) dt L = 475 pb-1 0.2 EF_tau29_medium1 0.1 Forphysicsanalysesitismandatorytoknowthetriggeref- ficiency.Thisistheprobabilityforanoffline-reconstructed 00 10 20 30 40 50 60 70 80 90 100 andidentifiedtautopassthetrigger.Forthispurposesev- p [GeV] eralstudiesmeasuringthetriggerefficiencyhavebeenper- T formed using both Monte Carlo simulated and real data Fig. 4. Tau trigger efficiency measured in Z → ττ decays us- events. ing 2011 data by applying a tag-and-probe method for the EF Tomeasurethetautriggerefficiencyastudybasedon tau 29 medium.Theefficienyismeasuredw.r.t.offlinetaucan- Z → ττ MC events has been performed. The efficiency didatespassingmediumidentificationrequirementsasafunction oftheofflinetaup [2]. is measured for several trigger items as a function of the T transverse energy, E . Since MC simulation will not nec- T essarily give a reasonable description of the new energy di-jet sample is selected in collision data and the trigger frontier of the LHC collisions, data driven measurements efficiencyismeasuredbyatag-and-probemethod. ofthetriggerefficiencyareessential.Hence,astudyusing Zbosondecaysinrealdatahasbeenperformed.Hereinthe trigger efficiency is measured by a tag-and-probe method 4 Future prospects for 2012 data using data of the 2011 run at a center of mass energy of √ s = 7TeV. An event is tagged by either an electron or Afteramaintancestopinearly2012theLHCwillrestart a muon, by applying an unbiased electron or muon trig- proton-protoncollisionsinearlyApril.Todealwiththeex- ger, respectively. On the probe side tau leptons which are pectedincreaseininstantaneousluminosity,andthuswith reconstructed and identified by the offline algorithms are alargerpile-upcontributionaswellashighertriggerrates, used to measure the tau trigger efficiency. The amount of improvementsintheselectionalgorithmsofthetautrigger dataanalysedinthismeasurementcorrespondstoroughly have to be implemented. The evolution of the event rate 475pb−1.TheresultsfortheEFtau 20 mediumand as a function of the instantaneous luminosity is shown in tau 29 medium items are shown in Figure 3 and 4. For Figure5and6fordifferentL1andEFitems,respectively. theseitemsmediumtautriggeridentificationisappliedas Several options are available to match the given fixed wellasacutonE of20GeVand29GeV,respectively. outputrate.Thestraightforwardwayistoincreasethe E T T Togainknowledgeofthetautriggerefficienciesatvery of the items without prescale. Furthermore, the isolation high energies, where a SM sample of real tau leptons is criteriacanbetightened.Bothresultinaloweracceptance limited in a few pb−1 of data, QCD di-jet events can be forphysicsanalysis.Adifferentapproachtoobeythenec- used to achieve a reasonable precision. Such a measure- essaryrejectionrateistoapplymoreadvancedalgorithms, mentisbasedontheideathatajetfakingareconstructed asusedinofflineidentification.Duetothehigherrejection andidentifiedofflinetauleptonshouldfirethetautrigger power at a given signal efficiency working point, mulit- at high p at the same rate as real tau leptons. Hence, a variatetechniquescouldbeareasonablealternativetothe T HadronColliderPhysicssymposium(HCP-2011) z]106 H e [ ATLASPreliminary cal105 Data 2011, s = 7 TeV s e pr ore 104 ef L1_TAU6 b L1_TAU8 ate 103 L1_TAU11 1 r L1_TAU50 L 102 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Instantaneous luminosity [1033/cm2/s] Fig.5.Evolutionoftheeventrateasafunctionoftheinstanta- neousluminosityforL1itemsasmeasuredin2011data[2]. z] 8 H e [ 7 ATLASPreliminary at Data 2011, s = 7 TeV put r 6 EF_tau100_medium out 5 EF_tau16_loose_mu15 EF EF_tau29_medium_xe35_noMu 4 EF_2tau29_medium1 3 2 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Instantaneous luminosity [1033/cm2/s] Fig.6.Evolutionoftheeventrateasafunctionoftheinstanta- neousluminosityforEFitemsasmeasuredin2011data[2]. current cut based implementation of the tau trigger algo- rithms.Thisisplannedforthe2012datataking. References 1. ATLAS Collaboration, The Expected Perfor- mance of the ATLAS Experiment - Detector, Trigger and Physics, CERN-OPEN-2008-020, https://cdsweb.cern.ch/record/1125884 2. https://twiki.cern.ch/twiki/bin/view/AtlasPublic/TauTriggerPublicResults 3. ATLAS Collaboration, Reconstruction, Energy Cali- bration,andIdentificationofHadronicallyDecayingTau LeptonsintheATLASExperiment,ATLAS-CONF-2011- 077,https://cdsweb.cern.ch/record/1353226