240 likes | 375 Views
Reconstruction and Identification of hadronic -decays in ATLAS. Tau06, 9 th Workshop on Tau Lepton Physics Pisa, Italy 19-22 September 2006. Fabien Tarrade LAPP, Annecy tarrade@lapp.in2p3.fr (on behalf of the ATLAS Collaboration). Outlook. M otivation Characteristic of leptons
E N D
Reconstruction and Identification of hadronic -decays in ATLAS Tau06, 9 th Workshop on Tau Lepton Physics Pisa, Italy 19-22 September 2006 Fabien Tarrade LAPP, Annecy tarrade@lapp.in2p3.fr (on behalf of the ATLAS Collaboration)
Outlook • Motivation • Characteristic of leptons • The ATLAS detector • -jets identification • Algorithms TauRec Tau1P3P • Summary F.Tarrade
Physics processes with τ’s Higgs Processes : • Standard ModelHiggs (VBF,ttH) qqH → qqττ, ttH → ttττ • MSSM Higgs(A/H, H+) A/H→ ττ, H+→ τν • SUSY signature with τ’s in final state • Extra dimensions … new theories (?) • Z →ττ • W→τντ signal significance Exotic Processes : 5σ Standard Model : important for the commissioning mHiggs (GeV/c2) F.Tarrade
Leptonic decay modes t nt + ne + e (17.4%) t nt + nm + m (17.8%) Hadronic decay mode 1 prong t nt + p± (11.0%) t nt + p± + p0 (25.4%) t nt + p± + p0 + p0 (10.8%) t nt + p± + p0 + p0 + p0 (1.4%) t nt + K± + np0 (1.6%) 3 prong t nt + 3 p± + np0(15.2%) decays τdecay modes : ~35 % How to identify them ? 1 track, impact parameter shower shape, energy sharing ~77 % t jets 3 tracks, impact parameter secondary vertex shower shape, energy sharing ~23 % F.Tarrade
ATLAS layout Muon Detectors Electromagnetic Calorimeters Forward Calorimeters Solenoid EndCap Toroid Barrel Toroid Inner Detector Hadronic Calorimeters Shielding F.Tarrade
ATLAS tracking Inner Detector Barrel Silicon Strip Detector Forward Silicon SripsDetector Transition Radiation Tracker Pixel Detectors Inner Detectors (ID) : • Precision Tracking : • Pixel detector, Semiconductor Tracker (SCT) • Continuous Tracking for pattern • recognition and e id • Transition Radiation Tracker (TRT) • Resolution : • σ(PT)/PT = 0.05%PT (GeV) 1% • Tracking in range |η|< 2.5 • ID inside2 Tesla solenoid field Magnetic Field : F.Tarrade
ATLAS calorimetry Layer Granularity (x) Pre-sampler 0.025 x 0.1 Strips 0.003 x 0.1 Middle 0.025 x 0.025 Back 0.05 x 0.025 Back Middle φ Layer Granularity (x) Tile0 0.1 x 0.1 Tile1 0.1 x 0.1 η Tile2 0.2 x 0.1 η-Strips EM Accordion Calorimeters Calorimeter : • Electromagnetic : (in |η|<3.2) • σE/E = 10%/√E(GeV) 0.245/E(GeV) 0.7% • (low luminosity) • Hadronic :( in |η|<3) • σE/E = 50%/√E(GeV)3.0 % Hadronic Tile Calorimeter η= 2.5 η= 1.475 η= 1.8 η= 3.2 Forward LAr Calorimeters Hadronic LAr EndCap Calorimeters η-Strips F.Tarrade
-jets reconstruction π0 π+ Jets Candidate Reconstruction : • characteristics well-collimated calorimeter cluster with a small number of associated charged tracks • acceptance |η|<2.5 acceptance of the inner detector • QCD jets • electrons that shower late or with strong Bremsstrahlung • muons interacting in the calorimeter • TauRec and Tau1P3P π+ τjet π- Backgrounds misidentified as jets: τ decay jet reconstruction algorithms in ATLAS : F.Tarrade
Reconstruction TauRec (default algorithm) : • Start from different objects: clusters or isolated track • Associate tracksto the τ jet candidate • Calibrateτjet candidate energy (calorimeters) • Start from a good leading hadronic track • Create single-prong or three-prong τ jet candidate • Calibrate candidate energy (tracker+calorimeters) • Determination of the discriminantvariables (calculate likelihood, discriminant multi-variate) • Apply set of basic cuts for τ-Identification pions inner detector hadronic calorimeters resolution (%) Tau1P3P (new algorithm): pT (GeV) For TauRec and Tau1P3P : F.Tarrade
TauRec : preselection Default jets reconstruction and identification in ATLAS : • Cluster (ET>15 GeV), or track (pT> 2 GeV) • Associate tracks pointing to the objet ifΔR=√(Δη2+Δφ2)<0.3 • Select candidate with 1, 2 or 3 tracks cut number of tracks TauRec jet candidate effreconstruction= 85% efficiency of reconstruction ΔR<0.3 η F.Tarrade
TauRec : variables • Build a set of discriminant variables for jets reconstruction and identification and for the rejection of QCD jets REM ΔET12 Nstrip Strip Width Charge ET/pT(1st track) Signal Bkg Z→ττQCD jets F.Tarrade
TauRec : identification • calculate likelihood using : REM, ΔET12, Ntrack, Strip Width, Nstrip, Charge, Impact parameter, ET/pT(1sttrack) • calibrate jets candidates energy weights (Monte Carlo) applied directly to cell energies depending on their E/V content (cell energy density), ηand layer (à la H1) • Apply set of basic cutsfor τ-Identification TauRec jets σ = 10.4% = -1.8% QCD jets (ETreconstructed τ – ETtrue τ)/ ETtrue τ likelihood ε=50% J2 : R ~40 J3 : R ~100 J5 : R ~200 jet identified with TauRec rejection factor Possibility to start with different objects Good reconstruction efficiency Good energy resolution For ε(τ)=30%, 15< pT< 334.5 Rej(QCD jets) = 400 - 10 000 efficiency reconstruction and identification F.Tarrade
Tau1P3P : preselection new algorithm for jets reconstruction and identification in ATLAS : Tau1P dedicatedTau1P3P in searches for the light Higgs or soft SUSY: 's with ETvis = 20-70 GeV efficiency of reconstruction • Explores exclusive feature of the lepton hadronic does not correspond to a typical jet but : 1 track + n π0 (Tau1P) 3 tracks + n π0 (Tau3P) • Good quality hadronic track (pT> 9 GeV), find nearby good quality track (pT>1 GeV, ΔR<0.2) • 1 track of good quality + no nearby track + 2 nearby tracks ET Tau3P efficiency of reconstruction • jet candidate of Tau3P ET • jet candidate of Tau1P efficiency of reconstruction for signal limited by good quality tracking F.Tarrade
Tau1P3P : variables • Build a set of discriminant variables for jets reconstruction and identification and for the rejection of QCD jets • Energy scale for jet candidates using energy flow based on tracks instead of pure calorimeter jet techniques Tau1P Tau3P σ = 9.9% = 0.4% σ = 2.7% = 0.8% (ETreconstructed τ – ETtrue τ)/ ETtrue τ (ETreconstructed τ – ETtrue τ)/ ETtrue τ F.Tarrade
Tau1P3P : identification • Based on sampling the signal and background densities in a multi-dimensional phase-space using range-searching and probability density estimation. Combine all observables in one discriminant variable • Apply set of basic cutsfor τ-Identification various ET bins Tau1P rejection efficiency various h bins jet identified with Tau1P jet identified with Tau3P rejection Start by a good quality track jets separation with 1 or 3 tracks Good energy resolution For ε(τ)=30%, 15< pT< 60 Rej(QCD jets) = 600 - 1 000 efficiency F.Tarrade
Summary • Efficient identification is crucial for several physics studies • Good sensitivity for identifying τ’s in many physics channels, • from light Higgs to Heavy SUSY • TauRec gives good results,possibility to use different seeds : • cluster, track … • Track based Tau1P3P gives good results, separation of jet • with 1 or 3 tracks is most powerful for low PT • The tau-identification achieved will allow the study of physics • channels where the jets background is very large • two complementary algorithms have been developped in ATLAS, • so a robust reconstruction and identification should be available • to be checked with early data jets reconstruction algorithms in ATLAS : Perpectives : F.Tarrade
Trigger • Trigger in ATLAS LVL1 Calorimeter+MuonTrigger, coarse granularity ~75 kHz Level 1 2.5 μs LVL2 Region of Interest, All detectors, full granularity ~1 kHz Level 2 ~10 ms High Level Trigger Event Filter refines the selection, can perform event reconstruction using latest alignment and calibration data (full offline reconstruction) ~100 Hz Event Filter PC farms ~sec F.Tarrade
triggers lepton hadrons • Possible way of selecting taus with the ATLAS trigger • Lepton Trigger • trigger with the electron or the muon • Hadronic Tau Trigger • - LVL1 Tau Trigger ( Calo) • use EM (0.2×0.2) and hadronic (0.2×0.2) towers to define • a Region of Interest and also for the isolation • in the EM (1.2×1.2) and hadronic (1.2×1.2) calorimeter • - LVL2 Tau Trigger ( Calo+Tracking) • evaluating offline variables : em radius of the cluster, width • in energy deposition, isolation fraction, track … • - Event Filter • based on the default -jet reconstruction code PRELIMINARY t trigger efficiency : still under evaluation F.Tarrade
LVL1 Tau-Trigger ‘ • A 2x2 tower EM cluster + 2 x 2 hadronic cluster used to ID cand. ROIs • A 2x1/1x2 tower EM clusters added to the energy in the hadronic inner region (shown in red) is compared to a threshold. There are 4 in the ROI the highest ET is taken. • A ring of 12 EM towers surrounding the clusters, which is used for isolation in the EM calorimeters • 12 hadronic towers (behind the EM isolation ring) for isolation in the hadronic calo. Tau trigger at LVL1 Hadronic Cal. 2-Tower EM cluster EM Cal. F.Tarrade
Efficiencie and Rejection Definition : efficiencie and rejection effRECONSTRUCTION = N(Reco & labeled ) N(in acceptance) effIDENTIFICATION = N(Reco & labeled & Id) N(Reco & labeled ) Rejection = N(Reco & non labeled ) N(Reco & non labeled & Id) F.Tarrade
Variables for TauID Δη=0.1 granularity of the tower Δη ×Δ Δ=0.1 tau_EMRadius : REM=Σ (ETEM×ΔR) / Σ ETEM tau_IsoFrac : ΔET12=Σ (ETEM+ETHad) 0.1<ΔR<0.2 / Σ(ETEM+ETHad) tau_stripWith2 : Δη2=Σ (η2×ET)/Σ ET – ( Σ (η×ET )/ Σ ET )2 Had EM η F.Tarrade
H1 Method Weights applied directly to cell energies Better resolution and residual non linearities Cluster and Energy scale • Cone Algorithm • Highest ET tower for jet seed + cone • Iteration of cone direction, jet overlap, energy sharing, merging F.Tarrade
Energy Flow ETeflow= ETemcl +ETneuEM + S pTtrack + S resETchrgEM + resETneuEM ETeflow/ETtruth ETeflow/ETtruth 1 prong 3 prong <>=1.011 s = 0.0303 <>=1.014 s =0.0663 used only EM cells within DR < 0.2 around tau1P direction ETeflow/ETtruth ETeflow/ETtruth 1.2 1.2 0.8 0.8 Z -> tt F.Tarrade