Bounding the Higgs boson width from off-shell production and decay to ZZ 4l and ZZ 2l2ν Mykhailo Dalchenko on behalf of the CMS Collaboration LLR, Ecole Polytechnique 21 July, 214 1 / 12
The Higgs boson width The total decay width is a fundamental parameter Relates to the couplings to all massive particles sensitive to the B physics Hard to measure: Γ H 4 MeV @ m H 126 GeV Γ H [GeV] 3 2 1 1 2 LHCXSWG LHC HIGGS XS WG 2 Direct constraints Current (excellent) experimental resolution is too large for direct measurements: σ m O(1) GeV Constraint from the resonance width in H ZZ 4l: Γ H < 3.4 GeV @ 95% CL Constraint from the resonance width in H γγ: Γ H < 2.4 GeV @ 95% CL 2 3 5 M H [GeV] PRD 89 (214) 927 (see S. Chhibra and C. Martin talks) 2 / 12
Width constraints from off-shell Higgs Off-shell Higgs boson production is small but the BR to 2 real Z is large above 2m Z Under change of width, peak yield remains constant if we scale the couplings in appropriate way σ pp H ZZ g 2 Hgg g 2 HZZ Γ Off-shell Higgs yield is proportional to this scale σ pp H ZZ g 2 Hgg g 2 HZZ Interference between the signal and background is sizeable and must be accounted for (see D. de Florian talk) 3 / 12
Analysis overview Use gg 4l (GG2VV and MCFM for ZZ 4l, GG2VV only for ZZ 2l2ν) and VV 4l, 2l2ν (Phantom) MC samples Build probability templates for signal, background and interference using m 4l and a kinematic discriminant D gg in case of ZZ 4l P total = µrp gg H 4l + µrp interference + P gg 4l r = Γ/Γ, µ - signal strength Build similar probability templates for VBF Build similar probability templates for signal, background and interference using mll T in case of ZZ 2l2ν Perform a combined fit of the off-shell and on-shell regions 4 / 12
4l analysis Selection settings and reconstruction are identical to PRD 89 (214) 927 Two pairs of OS/SF leptons. Z 1 closest to the Z mass, Z 2 the remaining with highest sum of p t 4 < m Z1 < 12GeV 12 < m Z2 < 12GeV One lepton with p t > 2 GeV/c, another with p t > GeV/c. m 4l > GeV, m l + l > 4 GeV PLB 736 (214) 64 Reducible background ( Z + X ) estimated from data in control regions with Z 1 + at least 1 loose lepton. Irreducible bkg calculated from MC, q q annihilation from POWHEG Phenomenological model for the qq ZZ shape Events / GeV 6 5 4 3 2 CMS 19.7 fb (8 TeV) + 5.1 fb (7 TeV) Events / 3 GeV 16 14 12 8 6 4 2 gg+vv ZZ qq ZZ kin D bkg >.5 1 12 13 14 15 m 4l (GeV) 2 3 4 5 6 7 8 m 4l (GeV) 5 / 12
ggmela Built from signal and background probabilities: D gg,a = P gg,a = a P gg sig + a P gg int + Pgg bkg and P q q,a = P q q bkg ggmela discriminant was developed in the context of the PRD 89 (214) 927 High performances for separating gg ZZ from qq ZZ where gg ZZ includes signal, continuum and their interference for any relative signal strength a. Signal strength a must be chosen when building the discriminant. Pgg,a P gg,a+p q q,a, where From preliminary studies we expected sensitivity for run 1 data to be around, so we chose D gg,. About 3% improvement when including it in the fit procedure 6 / 12
Distributions of selected events H ZZ 4l Events / bin 8 PLB 736 (214) 64 CMS 19.7 fb (8 TeV) + 5.1 fb (7 TeV) All contributions (Γ H = Γ H, µ gg+vv ZZ (Γ = Γ, µ = 1) qq ZZ MELA D gg = 1) >.65 Events /.5 15 PLB 736 (214) 64 CMS 19.7 fb (8 TeV) + 5.1 fb (7 TeV) 2 All contributions (Γ H = Γ H, µ = 1) gg+vv ZZ (Γ = Γ, µ = 1) qq ZZ m 4l >33 GeV 6 4 2 5 3 4 5 6 7 8 m 4l (GeV).2.4.6.8 1 MELA D gg signal-enriched region m 4l > 33 GeV D gg >.65 7 / 12
2l2ν analysis 6 times higher branching ratio w.r.t. 4l final state Use 4l results in the on-shell region High Z + jets background (fake ET miss from hadronic energy mismeasurement) Other backgrounds: - Irreducible: ZZ, WZ (from MC) - Non-resonant: top, WW Analysis variable: transverse mass ] 2 MT 2 = [ p 2T,ll + m2ll + Emiss,T 2 + m2 ll [ p T,ll + E ] 2 miss,t 8 / 12
Distributions of selected events H ZZ 2`2ν Events / bin PLB 736 (214) 64 CMS 3 2 19.7 fb (8 TeV) All contributions (Γ H = Γ H, µ = 1) gg+vv ZZ (Γ = Γ H, µ = 1) H qq ZZ Z+jets top/w+jets/ww WZ signal-enriched region mt > 35 GeV ETmiss > GeV 1 2 3 4 5 6 7 8 9 mt (GeV) 9 / 12
Systematic uncertainties gg ZZ - Part of cross section uncertainties cancel in the ratio between off-shell and on-shell - Shape uncertainties obtained varying PDFs: CT, MSTW and NNPDF - Correlated shape-yield uncertainties produced varying the scales and applying corresponding K-factor - K bkg = K sig (1. ±.1) q q ZZ - QCD scale: correlate shape and yield uncertainties - PDFs: constant 4% - q q ZZ EWK corrections (2-7%) Lepton efficiency for the trigger, reconstruction and selection Background estimation from data (mostly 2l2ν, up to 25%) All systematic uncertainties correlated between the on-shell and off-shell regions affect µ but not Γ H in the combined measurement / 12
Results Joint unbinned likelyhood fit Perform fit simultaneously on-shell (4l) and off-shell (4l + 2l2ν) Fitted parameters: µ F, µ V, Γ H /Γ H = 4.15 MeV) (Γ H 95% CL exclusion limit Best fit Γ H Observed: 22 MeV Expected: 33 MeV Observed: 1.8 +7.7 1.8 MeV Expected: 4.2 +13.5 4.2 MeV -2 lnl 8 6 4 2 CMS 4l observed 4l expected H ZZ PLB 736 (214) 64 s = 7 TeV, L = 5.1 fb 2l2ν + 4l on-peak observed 2l2ν + 4l on-peak expected Combined ZZ observed Combined ZZ expected s = 8 TeV, L = 19.7 fb 95% CL 68% CL 2 3 4 5 6 Γ H (MeV) 11 / 12
Conclusions First experimental constraint on the Higgs width using H ZZ events Γ H < 22 MeV @95% CL More than 2 order of magnitude improvement w.r.t. direct on-peak measurement Mild model dependency Assume essentially no new particles in the gluon fusion loop Assume no contribution from B physics in the background Perspectives The expected tail/peak cross-sections ratio is 2 times higher @13 TeV than @8 TeV The measurement may become limited by systematic uncertainties Improved determination of theory cross sections, in particular for the gg background is needed 12 / 12
Cross sections and events generation Gluon fusion, GG2VV and MCFM Latest versions of generators include signal, background and interference Assume m H = 125.6 and Γ = 4.15 MeV for generations as in PRD 89 (214) 927 Set factorization and renormalization running scale of m 4l /2 Apply m 4l -dependent K-factor (NNLO/LO) for signal from G. Passarino (Eur. Phys. J. C74 (214) 2866) Use K background = K signal assigning a % uncertainty on this assumption following M. Bonvini et al. (Phys. Rev. D88 (213) 3432) Eur. Phys. J. C74 (214) 2866 Vector boson fusion, Phantom VBF production is 7% of the total at H(125.6) peak Fraction of VBF increases with m 4l Same settings, similar lineshape except second enhancement due to t t in the gg H 12 / 12
H ZZ 4l 1D distributions in the full analysis range Events / bin 6 5 4 CMS (unpublished) 19.7 fb (8 TeV) + 5.1 fb (7 TeV) gg+vv ZZ (Γ = Γ H, µ =1) H V,F gg+vv ZZ qq ZZ Events /.5 6 4 CMS (unpublished) 19.7 fb (8 TeV) + 5.1 fb (7 TeV) All contributions (Γ H = Γ H, µ = 1) gg+vv ZZ (Γ = Γ, µ = 1) qq ZZ m 4l >22 GeV 3 2 2 3 4 5 6 7 8 m 4l (GeV).2.4.6.8 1 MELA D gg 12 / 12
H ZZ 4l ggmela inputs, signal-enriched Events / 2.5 GeV 2 CMS (unpublished) 19.7 fb (8 TeV) + 5.1 fb (7 TeV) 8 6 4 2 All contributions (Γ H = Γ H, µ = 1) gg+vv ZZ (Γ = Γ, µ = 1) qq ZZ m 4l >33 GeV Events /.1 2 CMS (unpublished) 19.7 fb (8 TeV) + 5.1 fb (7 TeV) 3 2 All contributions (Γ H = Γ H, µ = 1) gg+vv ZZ (Γ = Γ, µ = 1) qq ZZ m 4l >33 GeV 5 m 1,2 (GeV) -.5.5 1 cos θ 1,2 Events /.314 15 CMS (unpublished) 19.7 fb (8 TeV) + 5.1 fb (7 TeV) All contributions (Γ H = Γ H, µ = 1) gg+vv ZZ (Γ = Γ, µ = 1) qq ZZ m 4l >33 GeV Events /.314 15 CMS (unpublished) 19.7 fb (8 TeV) + 5.1 fb (7 TeV) All contributions (Γ H = Γ H, µ = 1) gg+vv ZZ (Γ = Γ, µ = 1) qq ZZ m 4l >33 GeV Events /.1 15 CMS (unpublished) 19.7 fb (8 TeV) + 5.1 fb (7 TeV) 2 All contributions (Γ H = Γ H, µ = 1) gg+vv ZZ (Γ = Γ, µ = 1) qq ZZ m 4l >33 GeV 5 5 5-2 2 Φ -2 2 Φ 1 -.5.5 1 cos θ* 12 / 12
H ZZ 4l 2D templates Signal-enriched All analysis range D gg 1 CMS (unpublished) 19.7 fb (8 TeV) + 5.1 fb (7 TeV) D gg 1 CMS (unpublished) 19.7 fb (8 TeV) + 5.1 fb (7 TeV).9.4.8.3.6.8.2.4 5.7.1.2 4 5 6 7 8 m 4l (GeV) 4 6 8 m 4l (GeV) 12 / 12
H ZZ 2l2ν m T ll and E miss T Events / bin 3 2 jets CMS (unpublished) 19.7 fb (8 TeV) = jets All contributions (Γ H = Γ H, µ gg+vv ZZ (Γ =Γ, µ = 1) qq ZZ Z+jets top/w+jets/ww WZ = 1) Events / bin 3 2 1 jets 1 jet CMS (unpublished) 19.7 fb (8 TeV) All contributions (Γ H = Γ H, µ gg+vv ZZ (Γ =Γ, µ = 1) qq ZZ Z+jets top/w+jets/ww WZ = 1) Events / bin VBF-type CMS (unpublished) 19.7 fb (8 TeV) VBF All contributions (Γ H = Γ H, µ gg+vv ZZ (Γ =Γ, µ = 1) qq ZZ Z+jets top/w+jet/ww WZ = 1) 1 1 1 2 3 4 5 6 7 8 9 m T (GeV) 2 3 4 5 6 7 8 9 m T (GeV) 15 2 25 3 35 4 45 5 miss E T (GeV) 12 / 12
H ZZ 4l 1D likelihood scans 1D m 4l 1D D gg -2 lnl 8 CMS (unpublished) 19.7 fb (8 TeV) + 5.1 fb (7 TeV) 4l(m ) observed 4l 4l(m ) expected 4l -2 lnl 8 CMS (unpublished) 19.7 fb (8 TeV) + 5.1 fb (7 TeV) 4l(D ) observed gg 4l(D ) expected gg 6 H ZZ 6 H ZZ 4 95% CL 4 95% CL 2 68% CL 2 68% CL 2 4 6 8 12 Γ H (MeV) 2 4 6 8 12 Γ H (MeV) 12 / 12
H ZZ 2l2ν 1D likelihood scan, on-shell measurements from 4l 1D m T ll -2 lnl 8 CMS (unpublished) 19.7 fb (8 TeV) + 5.1 fb (7 TeV) 2l2ν + 4l on-peak observed 2l2ν + 4l on-peak expected 6 H ZZ 4 95% CL 2 68% CL 2 4 6 8 12 Γ H (MeV) 12 / 12