ttH (Hγγ) search and CP measurement Scott McGarvie Pedro Teixeira-Dias ATLAS-UK Higgs mtg, Durham, 18/9/06
Motivation Higgs may be CP-even (e.g. SM), CP-odd (e.g. A in MSSM), or a CP-mixed state (CP-violating 2HDM, complex MSSM) ttH channel offers possibility to measure CP of light Higgs scalar (i.e., mH<2mZ)
CP measurement in ttH: method Suggested by Gunion & He, PRL76, 24 (1996) 4468; CP-sensitive variables depend only on momenta of top quarks, not Higgs … …though in practice, accuracy of top reconstruction will depend on Higgs decay mode /A
Feasibility study at parton level Perfect top quark momentum reconstruction Establish power of method to measure/distinguish CP-even, CP-odd, mixed-CP signal No backgrounds
CP-sensitive variables (I) – tth/A a2-even a2-odd a1-even a1-odd a3-even a3-odd b1-even b1-odd
CP-sensitive variables (II) - tth/A b3-even b3-odd b2-even b2-odd b4-even b4-odd b5-even b5-odd
Maximum Likelihood Use the method of maximum likelihood to determine the mixing parameter . Where f(x) are the PDFs for the test statistic, which in our case are just the individual variables (e.g. x=b4 …)
Measurement of α αtrue = 50% < αreco > = 51.8% Distribution of αreco for 500 toy MC experiments, with s+b=100 events each (≈250 /fb)
Feasibility study result (parton level) αrecon = αtrue used b4 variable only method provides unbiased CP determination “15-20% uncertainty” (L ≈ 250 /fb)
Detector-level study (ATLFAST) tth/A bjj blυ γγ (BR (Hγγ) ≈ 10-3) Selection Procedure: 2 light jets 2 bjets 2 photons (pT 25 GeV) 1 electron or muon (pT 25 GeV)
CP sensitive variables: a1 , a2 a1-even a2-even ATLFAST LEVEL a1-odd a2-odd a1-even a2-even PARTON LEVEL a1-odd a2-odd
CP sensitive variables: b4 , b5 b4-even b5-even ATLFAST LEVEL b4-odd b5-odd b5-even b4-even PARTON LEVEL b4-odd b5-odd
Detector-level study (ATLFAST) CP-sensitive variables retain discriminant power …must include backgrounds …and improve event selection for higher background rejection TDR analysis carried out long ago and using older technology (PYTHIA 5.7 for signal and bgd; relied on PHOTOS MC to generate ISR/FSR γs… spectrum too soft) Now: use MADGRAPH ME calculation for all bgds; HERWIG 6.5 is used for CP-even/odd signals TDR bgd estimate was found to be ~5x lower than current estimates…!!
Backgrounds (MADGRAPH) Process σ (fb) Equiv. Lumi generated (fb) ttγ 3326 155 ttγγ 16 49,700 ttγj 3704 220 ttjj 268 960 bbγγ 3944 26 bbγj 348 570 bbjj 3921 25 Wγγ 123 50,700 Wγγj 134 1,540 mH=120 GeV (HERWIG) 1.5 205,000 generated event samples: ~ 6M background events; ~ 2M signal events (mH=115 – 150 GeV)
Photon pT cuts optimization
Photon pT cuts optimization pT(γ1) pT(γ2) signal signal ttγγ bgd ttγγ bgd
Mass window optimization, Δm Count events in mass window of size Δm: mh- Δm < mγγ < mh+ Δm Δm (MeV) Maximum significance for 2.3 GeV (cf 1.7 GeV in TDR)
Event selection optimization (mh=120 GeV; 100 /fb) Use the current signal and background event samples and try to improve the event selection signal-to-noise… there is room for improvement: TDR-inspired: pT(e/μ)> 25 GeV; pT(γ1), pT(γ2)> 25 GeV mh-Δm < m(γγ) < mh + Δm (Δm =1.7 GeV) Total b = 32.1 s = 19.7 Optimized event selection: pT(e/μ)> 25 GeV; pT(γ1)> 50 GeV, pT(γ2)> 30 GeV mh-Δm < m(γγ) < mh + Δm (Δm =2.3 GeV) Total b = 19.2 s = 17.8
Event selection (cont’d) Additional cuts? (...can’t afford large additional cuts in signal rate, otherwise compromise CP-even/odd discriminant power…) Requiring 1 b-tagged jet (reduces Wxx bgds) only a slight improvement in significance Top-quark reconstruction: required for CP measurement; will use bjj blv combination that minimizes
Summary & Future Plans Established power of CP-method in tth channel at LHC (parton level study) Detector level tth (hγγ) (ATLFAST): CP-sensitive variables seem to retain power much-improved bgd calculation (wrt TDR) done event selection improved still need to establish finally the precision of CP determination, for eg. L=100 /fb and 300 /fb finish writing up and get PhD. Future plans: investigate possible increase in CP-sensitivity by 1) using more than just one CP-sensitive variable 2) using a higher BR Higgs decay mode (HWW, bb) 3) combining CP-measurement from Hγγ and HWW, bb
Backup slides
Detector-level event reconstruction For events which meet these requirements: Reconstruct a hadronic W from all jet-jet pairs Assume missing momentum is due to the neutrino, use W mass constraint to fully reconstruct the neutrino momentum. Reconstruct a leptonic W in events which have 1 or more neutrino solutions Reconstruct the top quarks by combining the b-jets and Ws 4-vectors
top-quark reconstruction Reconstruct a top for all bjj and bl combinations. Four-vectors of the two tops are found by selecting the combination which minimizes the 2 mt = 175 GeV, mw = 80 GeV, from TDR The reconstructed tops are then used to determine the values of the CP-sensitive variables