Semileptonic Boosted Tops Brock Tweedie Johns Hopkins University
Semileptonic Boosted Tops Brock Tweedie Johns Hopkins University 10 July 09 K. Rehermann & B. T. , To Appear
The Problem b-jet l ~ mt / p t n Isolation probability (DRbl > 0. 4) 1 Te. V 2 Te. V R 0. 34 0. 07 L 0. 56 0. 14
Our Philosophy • Try to use these nonisolated leptons • Avoid using MET for discrimination • Do not b-tag
Leptons Inside of Jets • Physics backgrounds – Heavy flavor (prompt and radiative) – p+ decays in flight • Instrumental backgrounds – p 0 C “e” – p+ C “m” Save for real experimentalists!
Event Generation • PYTHIA and HERWIG ttbar continuum and generic dijet – Includes prompt heavy flavor, light meson decays-in-flight in LHC-like detector volume • Basic requirement: muon with Pt > 30 Ge. V – ~4% pass rate for dijet…per-jet probability ~2% I will exclusively use PYTHIA plots / #s. HERWIG is practically identical.
m+jets Event Reconstruction • Set aside leading muon • Put remaining particles into perfect 0. 1 x 0. 1 calorimeter • Cluster with C/A – Set R according to Ht in hemisphere opposite the muon – Jet Pt > 50 Ge. V • Leading jet == hadronic top • Jet closest to muon == b-jet from semilep top
Semi-Leptonic Tops vs Light Jets b n m b m • • n b m K m + jet + MET hard m and MET m. T(m+MET) ~ m. W mass = mt p • m + jet + MET + JUNK • Soft/collinear singularities • Splittings more common late in the shower (more gluons!)
Mini-Isolation DR ~ mt / Ptt DR ~ ? DR ~ mb / Ptb m m n n B B B W t
Mini-Isolation • Many options for cone definition: – R ~ 1/Ptb ~ 1/Ptm – R ~ 1/Ptt – R = fixed # –… • They all perform comparably • R ~ 1/Ptm convolves additional discriminating power from muon Pt distributions
Mini-Isolation cone DR = (15 Ge. V) / Ptm top light jet W+jets Demand >90% isolated
(Thaler & Wang) xm = 1 - mb 2/mbm 2 top light jet W+jets Demand xm > 0. 5
Mini-Isolation After xm Cut top light jet W+jets
xm After Mini-Isolation Cut top light jet W+jets
Efficiencies of Leptonic Cuts (Pt ~ 1 Te. V) top light jet W+jets Mini-iso 0. 91 0. 0038 0. 95 xm 0. 89 0. 0373 0. 96 Combined 0. 86 0. 0014 0. 93
Efficiencies of Leptonic Cuts (Pt ~ 2 Te. V) top light jet W+jets Mini-iso 0. 89 0. 0026 0. 95 xm 0. 84 0. 0405 0. 95 Combined 0. 81 0. 0012 0. 92
Semi-Leptonic Tops vs W-strahlung b n n m b q’ q W n m m • • m + jet + MET hard m and MET m. T(m+MET) ~ m. W mass = mt • m + jet + MET • hard m and MET • m. T(m+MET) ~ m. W • m. W < mass < sqrt(s-hat)
Ideal Top Mass Distributions top light jet W+jets
DRbm top light jet W+jets • Wjj Mad. Graph 2 C 4
Ideal-Mass vs DRbm Discrimination
Efficiencies of Leptonic Cuts (Pt ~ 1 Te. V) top light jet W+jets DRbm 0. 97 0. 9970 0. 45 Combined leptonic cuts 0. 84 0. 0013 0. 39
Efficiencies of Leptonic Cuts (Pt ~ 2 Te. V) top light jet W+jets DRbm 0. 95 0. 9936 0. 27 Combined leptonic cuts 0. 76 0. 0011 0. 21
Backgrounds with Top-Mass Cut Now use MET for global even reco. Define hn == hm. Hadronic top-mass cut efficiencies: et ~ 85% / eq/g ~ 25%
Backgrounds with Top-Tag
Resonance Efficiencies (incorporates m+jets BR) top-mass cut top-tag
Summary • Assuming this all works in the detector, light QCD can be made negligible practically “for free” – In principle, rejection factors at the ~50, 000 level – Allows for a comfortable margin of theory error • W-strahlung is still non-negligible – O(1) rejection “for free” by exploiting geometry
Summary • Still various additional discriminators after incorporating MET – m. T(top), m(top) – Internal angular variables • Possibilities for improvements in high-Pt t -tagging and b-tagging
Summary • We will be seeing how these perform in full CMS simulation in the coming months
Extras
Resonances
Discovery Reach Estimates S/sqrt(B) > 5 & S > 15 G = 0% G = 15%
Subjet Rates * old PYTHIA results
- Slides: 31