Study of tt-Photon Events with the CMS-detector Bad Honnef 27. August 2007 Thomas Hermanns III. Physikalisches Institut B tt-Photon Events at the LHC Cross section for tt-pair production at s= 14 TeV: 830 pb (NLO) About 1 tt-pair per second (L=1033cm-2s-1) Consider top-decays only via Vtb1 Consecutive decays of the W-bosons W+/-: electron or muon channel W-/+: two (light) quarks “Semileptonic or lepton+jets channel” Branching fraction: 29.6% High statistics appropriate for rare events in the realm of top-physics Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 2 Motivation of the Analysis Determination of photon spectra at the CMS-experiment Separation of photons radiated off top-quarks Distinction of various QED-coupling scenarios Lorentz-invariant vertex parameterisation tt q, q, k 2 ie F1V k 2 5 F1A k2 2 2 q q iF k F k 2V 5 2A 2mt SM prediction at Born level F1V Thomas Hermanns 2 3 F2V Qt gt 2 0 2 F1A F2A 0 U. Baur, A. Juste, L.H. Orr, D. Rainwater „Probing electroweak top quark couplings at hadron colliders“ Physical Review D 71, 054013 (2005) Bad Honnef, 27. September 2007 III. Physikalisches Institut B 3 The Feynman-Diagrams Three classes for the hard process g + g t + t + Photon (8 diagrams) q + q t + t + Photon (8 diagrams for q=u,d) q + q + Photon t + t (8 diagrams for q=u,d) Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 4 The Dataset Cross section for 2 3 process (TopRex generator) E,min> 5 GeV: 16.8pb E,min> 100 GeV: 0.2pb Compare to: 830pb for top-quark-pair production (NLO) Indistinguishable processes for photons radiated off incoming quarks and top-quarks in the case of annihilation Study of a Pythia tt-dataset, to get photons radiated off top-quarks (final state radiation) Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 5 Photons in Signal Events Signal event Semileptonically (electron and muon) decaying tt-pair Photon radiated off a top-quark (top photon) CMS tt-inclusive dataset Total number of events: 3,900,000 events About 3,800 potential signal events (generator filter) Less than 1,000 events remaining after preselection cuts Need for a private dataset Filtering tt-semileptonic signal events on generator level E >10 GeV ||< 2.5 in events Cross-section: 0.1 pb 19,950 signal events Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 6 The Event Selection Selection criteria applied on reconstructed objects 1 electron or muon candidate pT>20 GeV/c Isolated in tracker and calorimeter 2 jets candidates from b-quarks Iterative cone algorithm (R=0.5) pT>20 GeV/c 2 jets candidates from light quarks Iterative cone algorithm (R=0.5) pT>20 GeV/c 1 photon candidate E >20 GeV/c Isolated in tracker and calorimeter Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 7 Photon Classes Try to match of every preselected photon candidates to a MC-photon Distance in (,)-plane: (Rgen-reco) < 0.05 Deviation of energy |Egen-reco| < 0.1 Egen Signal Photons Match of reconstructed photon candidate to generator top-photon Background Photons Match of reconstructed photon candidate to any generator photon but the top-photon Fake Photons No match of reconstructed photon candidate to any generator photon Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 8 Calorimeter Energy Deposits (Signal Events) Hadronic over electromagnetic energy ratio E(had)= E(i) Energy deposited in HCAL behind ECAL-supercluster (R=0.3 around line through supercluster midpoint) with i R-cone Implement cut at R=0.2 Remove large tails of mainly fake photon candidates Signal Photons Fake Photons Bkg. Photons HCAL Supercluster ECAL Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 9 Shower Shape Variables (Signal Events) Spread of electromagnetic energy in calorimeter cells Energy of clusters (squares and rectangles) ECAL cells Compare ratio of different shape variables Cut on E(3x3) divided by E(5x5) Lowest overlap between signal and background/fake distributions (normalized) Signal Photons Fake Photons Bkg. Photons Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 10 Isolation in Silicon Tracker (Signal Events) Veto against tracks in the vicinity of the photon direction Number of tracks in a cone around photon candidate Varying cone size as well as ratio of track momentum and photon energy No tracks within R=0.2 with pTrack > 0.1 EPhoton Signal Photons Thomas Hermanns Fake Photons Bad Honnef, 27. September 2007 III. Physikalisches Institut B 11 Numerical Results of the Photon Identification 10,022 out of 19,950 events fulfil minimum tt-photon requirements Results of the photon identification S/B improved by a factor of about 26 less than 25% of all signal photons lost Preselection Ehad/EEM, Shower Shape Tracker Isolation Efficiency Signal 6,027 5,107 4,539 75,3% Background 971 344 195 20,1% Fake 40,934 6,741 1,046 2,6% S/B 1/7.0 1/1.4 3.7/1 Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 12 Photon Spectrum (Signal Events) Energy spectrum of signal, background and fake photons Spectrum of signal photons slig htly harder Increase photon energy cut from 20 GeV to a higher value Signal Photons Fake Photons Bkg. Photons Thomas Hermanns Bad Honnef, 27. September 2007 Signal Photons Fake Photons Bkg. Photons III. Physikalisches Institut B 13 Background to tt-Events Events with a similar decay structure W/Z+Jets Vector-Boson pair production W+Photon tt-dileptonic ... Signal-Background Separation Strategy Reject background events via an appropriate tt-event solution (using Top Quark Analysis Framework) Remove photons as efficient as background and fake photons in signal event Incorporation of background events recently started First Overview for WZ- and Z+4Jet Events Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 14 Number of Photons Signal Events WZ-Events Z+4 Jets-Events After Preselection Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 15 Numerical Results Demand for a high energetic photon essential already at preselection step WZ: 1,446 out of 190,000 events Z+4Jets: 3,735 out of 21,402 event Rejection of background and fake photons comparable to signal events Background Photons Fake Photons Thomas Hermanns Preselection Calorimeter Isolation Tracker Isolation Efficiency WZ 140 56 35 25.0% Z+4 Jets 450 210 119 26.4% WZ 4,754 1,073 124 2.7% Z+4 Jets 15,171 4,528 354 2.3% Bad Honnef, 27. September 2007 III. Physikalisches Institut B 16 Conclusions Signal Photon Identification using Calorimeter Isolation and Tracker Isolation Robust criteria to reject background and fake photons Comparable efficiencies for various input dataset tt-background demanding a proper tt-reconstruction Preselection already gives a reasonable separation Using TQAF should reduce that component further tt-Photon analysis should ... consider angular relations between photon and objects of the tt-decay (probably correlated to tracker isolation) respect kinematic constraints if the tt-photon decay chain Compare current results with TopRex-Dataset More realistic description of physics Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 17 Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 18 Efficiency Purity Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 19 TQAF Top Quark Analysis Framework (TQAF) effort of the CMS-top-group to establish a common framework for top-quark analysis integration of CMS-standard tools (electron-ID, kinematic fit, ...) analysis code used and debugged by many people benefit from work one has to do but it was already done in the past Three-Layered-Structure production of top-objects independent of final state and analysis goal • Building of event solutions assuming a certain event hypothesis • lepton identification, calibration of jets, ... combing jets to build a W-Boson, top-quark, ... Actual analysis • • Thomas Hermanns direct access to objects according to a certain event solution criterion to select best solution: MC-matching solution Bad Honnef, 27. September 2007 III. Physikalisches Institut B 20 R-Cuts (Generator Particles) Photon radiated off top-quark x-axis: R(photon - top-quark) y-axis: R(photon - elektron/muon) Photon radiated off elektron/muon x-axis: R(photon - elektron/muon) y-axis: R(photon - top-Quark) Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 21 Identification of top-Photons Energy (reco./gen.) Pseudorapidity (reco./gen.) -Angel (reco./gen.) Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 22 Efficiency of Signal Photon Identification High efficiency in identifying the signal photons Tight cuts on energy and distance in (,)-plane Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 23 Identifikation der top-Photonen Energie der rekonstruierte Photonen Energie der generierten top-Photonen Kriterien zur Identifikation E(reco-gen) (reco-gen) (reco-gen) Thomas Hermanns Bad Honnef, 27. September 2007 III. Physikalisches Institut B 24
© Copyright 2025 Paperzz