Physics with Tau Lepton Final States in ATLAS on behalf of the ATLAS Collaboration HEP 2012, Valparaiso (Chile), 06.01.2012
The Tau Lepton m τ = 1.8 GeV, heaviest lepton cτ = 87 μm, short lifetime hadronic decay: 65% leptonic decay modes are hard to distinguish from prompt electrons or muons only hadronic tau decay modes are considered decay products are well collimated mostly 1 or 3 charged tracks (prongs) non-negligible amount of neutral pions secondary vertex-reconstruction possible (3-prong taus) 3π ± ν 1π ± 2π 0 ν 3π ± 1π 0 ν 14% 14% 39% 1π ± 1π 0 ν others 7% 8% hadronic mode 65% τ decay leptonic mode 35% 17% 1π ± ν 2/18
Tau Reconstruction Two steps to find tau candidates: Reconstruction build a tau candidate starts from anti-k T jets (R=0.4) with p T > 10 GeV and η < 2.5 no separation attempt between tau leptons and fake sources (jets, e ±, μ ± ) Identification aimed to distinguish between QCD jets and tau leptons variables used are based on tau decay properties 3/18
Tau Identification Variables (assortment) Ref.: ATLAS-CONF-2011-152 collimated energy deposits red: tau signal (MC) Z ττ W τν black: dijet background (data) energy weighted shower width in the calorimeter R Cal = E T R cells cells E T 4/18
Tau Identification Variables (assortment) Ref.: ATLAS-CONF-2011-152 narrow cone red: tau signal (MC) Z ττ W τν black: dijet background (data) Maximal distance (ΔR) between a core track and tau axis 5/18
Tau Identification 3 different methods are used: Cut-based selection Likelihood identification (LLH) Boosted-Decision-Tree (BDT) identification different sets of identification variables separately trained for single- and multi-prong tau candidates training of LLH/BDT binned in p T LLH/BDT trained for different numbers of reconstructed vertices Ref.: ATLAS-CONF-2011-152 BDT Output Score LLH Output Score https://twiki.cern.ch/twiki/pub/atlaspublic/eventdisplaypublicresults/zpileup_alltracks.png 6/18
Tau Identification Ref.: ATLAS-CONF-2011-152 tau ID training signal: MC Z ττ, W τν blue: Cuts background: QCD dijet data 2011, L = 130 pb -1 red: BDT LLH/BDT: efficiency flat in p T black: LLH every ID available with predefined cuts: loose, medium, tight (~60%, ~45%, ~30% signal efficiency) 3-prong, 20 < p T < 40 GeV 3-prong, 40 < p T < 100 GeV 7/18
Z ττ Cross-Section Measurement https://twiki.cern.ch/twiki/pub/atlaspublic/eventdisplaypublicresults/ztautau_event_display_run160613_event9209492_ed.png 8/18
Z ττ Cross-Section Measurement measured with 2010 data, L = 36 pb -1 4 final states: eτ h, μτ h, eμ, μμ Ref.: arxiv:1108.2016v1 muon isolation concentrate on channels with τ h event selection: single lepton trigger (e/μ) one e/μ with p T > 15/16 GeV τ h with p T > 20 GeV opposite charge lepton isolation (against QCD background) transverse mass m T < 50 GeV (against W lv) E T miss vector between lepton and tau (against W lv) visible mass between 35 75 GeV 9/18
Z ττ Cross-Section Measurement - Backgrounds QCD multijet: low Monte-Carlo statistics data-driven method (ABCD) measured in same-sign control region (B) OS/SS ratio measured in non-isolated regions (C/D) ratio corrected by electro-weak contribution from MC N A = N B N C N D electro-weak (EW): estimated from MC (signal) (OS) W lv MC normalized to data using a control region (high m T, low cos ϕ) (SS) 10/18
Z ττ Cross-Section Measurement - Results systematic uncertainty tau energy scale 11% affected channel eh,μh tau efficiency 8.6% eh,μh μ efficiency 3.8%, 8.6% μh, μμ e efficiency 9.6%, 5.9% eh, eμ acceptance 3% all luminosity 3.4% all total measured cross-section: σ = 0.97 ± 0.07 (stat) ± 0.06 (syst) ± 0.03 (lumi) nb theory: σ = 0.96 ± 0.05 nb (NNLO) 11/18
W τv Cross-Section Measurement https://twiki.cern.ch/twiki/pub/atlaspublic/eventdisplaypublicresults/wtaunucand-155697-6769403.png 12/18
W τv Cross-Section Measurement Ref.: arxiv:1108.4101v2 measured with 2010 data, L = 34 pb -1 only hadronic tau decay channel event selection: combined tau missing energy trigger reject events with electrons/muons with p T > 15 GeV (against W/Z decays with e/μ in final states) leading τ h within 20 < p T < 60 GeV E T miss > 30 GeV missing E T significance (S Etmiss ) > 6 QCD background: data-driven method (ABCD) defined by missing E T significance and tau identification (signal) EW background: obtained from simulation 13/18
W τv Cross-Section Measurement - Results systematic uncertainty trigger efficiency 7% tau ID efficiency 10.3% energy scale 8% acceptance 1.9% luminosity 3.4% total measured cross-section: σ = 11.1 ± 0.3 (stat) ± 1.7 (syst) ± 0.4 (lumi) nb theory: σ = 10.46 ± 0.52 nb (NNLO) 14/18
Tau ID Efficiency Measurement Ref.: ATLAS-CONF-2011-152 Z ττ μτ h obtained in 2011 data, L = 800 pb -1 tag-and-probe method following closely the Z ττ cross-section analysis 8-12% uncertainty on tau ID efficiency measurement dominated by QCD multijet estimation μ (tag) τ h (probe) 15/18
Tau ID Efficiency Measurement Ref.: ATLAS-CONF-2011-152 W τν obtained in 2011 data, L = 1.37 fb -1 event selected by missing energy (tag) number of τ h estimated by template fit on the number of tracks distribution real hadronic tau decays (MC) misidentified electrons (MC) misidentified QCD jets (data) 5-7% uncertainty on tau ID efficiency measurement 16/18
Summary tau reconstruction and identification well understood ID algorithms are continuously optimized LLH/BDT methods will be used in tau trigger as well in 2012 measured cross-section of W/Z in good agreement with theory NNLO predictions (update with more data will come soon) Standard Model channels provide feedback for tau ID key channels to improve tau identification ATLAS has an extensive physics program with taus e.g. see in this session: Higgs Boson Searches with Tau Leptons in the Final State at ATLAS (Guilherme Nunes Hanninger) Top quark production at ATLAS (includes tt μτ h + X) (Maria Moreno Llacer) 17/18
References ATLAS-CONF-2011-152 Performance of the Reconstruction and Identification of Hadronic Tau Decays with ATLAS, ATLAS Collaboration https://cdsweb.cern.ch/record/139819 arxiv:1108.4101v2 Measurement of the W to tau nu Cross Section in pp Collisions at sqrt(s) = 7 TeV with the ATLAS experiment, ATLAS Collaboration, submitted to Physics Letters B http://arxiv.org/abs/1108.4101v2 arxiv:1108.2016v1 Measurement of the Z to tau tau Cross Section with the ATLAS Detector, ATLAS Collaboration, submitted to Phys. Rev. D http://arxiv.org/abs/1108.2016v1 https://twiki.cern.ch/twiki/bin/view/atlaspublic/tautriggerpublicresults https://twiki.cern.ch/twiki/bin/view/atlaspublic/eventdisplaypublicresults 18/18
Backup Backup slides 19/18
CERN / ATLAS geometrical distance between two objects: with pseudorapidity: Width: Diameter: Weight: 44 m 25 m 7000 t 20/18
Tau Reconstruction 1-prong tau decay https://twiki.cern.ch/twiki/pub/atlaspublic/eventdisplaypublicresults/wtaunucand-155697-6769403.png 21/18
Tau Identification Performance Ref.: ATLAS-CONF-2011-152 blue: Cuts red: BDT black: LLH 1-prong, 20 < p T < 40 GeV 1-prong, 40 < p T < 100 GeV 3-prong, 20 < p T < 40 GeV 3-prong, 40 < p T < 100 GeV every ID available with predefined cuts: loose, medium, tight (~60%, ~45%, ~30% signal efficiency) 22/18
Tau Electron Veto Ref.: ATLAS-CONF-2011-152 Electrons can be misidentified as taus dedicated vetoes Boosted-Decision-Tree (BDT) Cut-based predefined scores available: loose, medium, tight (~95%, ~85%, ~75% signal efficiency) 23/18
Tau Energy Scale Uncertainty using Z ττ, W τν, Z ττ MC simulation evaluated from sources: MC generator, underlying event model amount of detector material EM energy scale topological clustering noise threshold pileup hadronic shower model non-closure Ref.: ATLAS-CONF-2011-152 tau energy scale uncertainty: 4 7% in progress: energy scale uncertainty directly from data 24/18
Tau Trigger Efficiency Ref.: TauTriggerPublicResults ATLAS trigger: 3 steps Level 1, Level 2, Event Filter Tau Trigger: L1,L2: calorimeter + track information EF: full tau reconstruction + dedicated tau ID methods tau trigger efficiency measured with 2011 data, L = 475 pb -1 using Z ττ data (eτ h, μτ h channels) tag: unbiased lepton (e, μ) trigger probe: well identified tau 25/18