Measurement of the Top Quark Mass with the Matrix Element Method at D0 Petra Haefner Ludwig-Maximilians-Universität München

Why measure the Top Mass? because it is SO HEAVY (~ 172 GeV) ! only (known) fermion with mass near electro-weak scale indirect constraints on mass of Higgs boson possible decay time shorter than hadronisation timescale allows direct mass measurement based on decay products Petra Haefner Scientific American, June 2003

How is Top produced? mainly in top-antitop pairs (strong interaction) 85 % annihilation ( 5 % @ LHC) 15 % gg fusion (95% @ LHC) cross section ~ 7pb (800 pb @ LHC) single top production (electroweak interaction) several production channels (s-, t-, associated t+W) larger backgrounds lower cross section (~ ½ stt) not observed yet Petra Haefner

How does Top decay? branching ratio t  Wb ~ 100 % as |Vtb| >> |Vts|,|Vtd| topology determined by W decays branching ratio W  light q ~ 2/3, W  l+n ~ 1/3 t difficult to identify consider only decays to l= e,m alljets (44%) largest branching fraction huge backgrounds lepton+jets (30%) lower backgrounds good statistics dilepton (5%) cleanest signature very low statistics Petra Haefner

The Tevatron Collider proton-antiproton collisions Run I (1992-1996) ECM = 1.8 TeV  L dt  100 pb-1 Run IIa (2002-2006) ECM = 1.96 TeV  L dt  1.2 fb-1 Run IIb (2006-?) just started in June  L dt  4 - 8 fb-1 (expected) Petra Haefner

The D0 Detector “standard“ collider detector configuration silicon microvertex & tracking detector within solenoid (2 T) LAr calorimeter high granularity excellent resolution muon chambers large coverage (h < 2.0) three trigger levels Level 1 1.5 kHz Level 2 850 Hz Level 3 50 Hz coordinate system: pseudorapidity h =  ln (tan /2) radius r polar angle j (track) distance Petra Haefner

The Tracking System two tracking subdetectors Silicon Microstrip Tracker measurement of charged particle origin distinguish primary vertices (important at high luminosity) identify b jets, t decays (secondary vertices) reject cosmic muon background Central Fiber Tracker measurement of charged particle momentum & direction dca resolution Petra Haefner

The Calorimeter sampling calorimeter energy resolution contributions uranium as absorber liquid argon as active material high granularity longitudinal shower shape divided into 3 cryostats energy resolution contributions stochastic term jet fragmentation sampling fluctuations EM fraction fluctuations noise term electronic noise multiple interactions constant term dead material magnetic field non-compensation Petra Haefner

The Muon System consists of 3 layers (A,B,C) of Proportional Drift Tubes accompanied by scintillation trigger counters (Pixel detectors) toroidal magnet between layer A and B 1.9 T magnetic field field perpendicular to beam axis muons only charged particles likely to penetrate both the tracking system and calorimeter reconstruct muon track outside calorimeter magnetic field allows stand-alone momentum measurement match muon to central track use better resolution of tracking detector for pt measurement Petra Haefner

What do we measure? lepton+jets signature 1 energetic, isolated lepton (e / m) 4 energetic jets (2 b jets, 2 light jets) missing transverse energy (1 n) event kinematics 2 solutions for n pz (along beam axis) otherwise fully determined kinematics 24 possible jet combinations (fewer with b tagging) no need to distinguish light jets from W 12 jet combinations whole detector needs to be well understood, calibrated and all subdetectors need to work for the top reconstruction! Petra Haefner

Comparison of Methods Template Method (classical approach) take all 2x12 combinations into account choose kinematic fit with smallest 2 fill histogram with reconstructed top mass compare to simulated distributions of different mtop correct combination only in ~40% of events! all events are weighted equally Matrix Element Method (developed at D0 Run I) calculate probability Pi(mtop) for every event to be a top decay take information from all 24 combinations into account combine probabilities of all combinations & events get total probability P(mtop) for the whole sample measure mtop as most likely one all combinations contribute events are weighted according to their information content Petra Haefner

Probability Calculation probability for a given combination x in one event to be a top decay parton density function (PDF) for all flavor compositions of the colliding quark and antiquark leading order matrix element used for cross section calculation leading order PDF (CTEQ5L) jet / m resolutions taken into account by transfer functions jet / lepton angles & electron energy assumed to be perfectly measured simultaneous fit to mtop & JES factor S minimisation of total uncertainty due to JES Petra Haefner

Improving the method simultaneous mtop / JES fit allows in situ calibration of jets ( constraint W mass in the matrix element) include b-tagging information obtain more pure samples use as constraint on jet combinations include muon transfer functions to describe resolutions improve agreement between MC and data improve JES / jet resolution direct improvement on largest uncertainty Petra Haefner

Jet Energy Scale top mass reconstructed from measured jet energies jet resolution dominant effect on top mass resolution need to calibrate detector to jets  Jet Energy Scale “classical approach“: calibrate EM calorimeter with Z ee assume e  g calibrate jet response with g + jet only EM calorimeter calibrated well hadronic calorimeter only indirectly calibrated complicated procedure  large systematic uncertainties Petra Haefner

Hadronic Calibration DiEM events very clean event signature easy to calibrate mass fit high statistics DiJet events hadronic W, Z decays swamped by QCD no mass constraint study single hadrons Petra Haefner Single Hadron Response need to find isolated particles backgrounds from neutral particle decays very low statistics (especially at higher E)

Eflow Algorithm improved Track-Jet- Algorithm improve jet energy resolution by improving absolute energy scale propagate tracks to CAL surface calculate h, j of path at each layer subtract expected energy deposit in CAL from Ejet (Single Hadron Response) add Etrack instead correct Ejet for in-out /out-in tracks cancels sources of fluctuations 10 % jet resolution improvement in g+jet data (before JES corrections) 20 % improvement on Z  bb mass resolution (important for Higgs search) 10 % improvement on W jj mass resolution (important for top mass) see IMPRS Seminar Talk 10.2.06 Petra Haefner

Muon Resolution Studies study differences between MC / data resolutions derive smearing factors for MC improve agreement between MC / data resolutions study effects of track quality / run periods on resolution derive muon transfer functions better description of muon resolution include transfer functions into top mass probability calculation Petra Haefner

Latest Top Mass Results CDF: l+jets, Template, 680 pb –1 l+jets, ME method, 940 pb –1 di-lep, ME method, 680 pb –1 all-had, Ideogram method, 310 pb –1 D0: l+jets, ME method, 370 pb-1 l+jets, Ideogram method, 370 pb-1 di-lep, Matrix weighting, 370 pb-1 di-lep, Neutrino weighting, 370 pb-1 Winter 2006 Summer 2005 There‘s more to come! Petra Haefner

Outlook Run IIa finished with 1.2 fb-1recorded, Run IIb just started triple the statistics of the last published analysis on tape being analysed right now first results expected soon b-tagging included in the next analyses publication in preparation new JES expected within the next months improvements on jet resolution smaller systematic errors collaboration wide effort of data reprocessing and new analysis format large improvements on all sectors from data reconstruction to MC generation, to efficiencies, scale factors, b-tagging, ... Petra Haefner