Search for the Standard Model Higgs in WW (lν)(lν) v l v l Yanyan Gao (Fermilab) February 11, 13 HEP Lunch seminar at University of Chicago
The SM Higgs Production and Decays at LHC "(pp! H+X) [pb] The SM Higgs is mainly produced in gluon fusion at LHC At 8TeV for mh=15 GeV: σ(ggh) ~ pb, σ(qqh) ~ 1.6pb Leptonic WW channel has large rate in full mass range with relatively clean signature Due to the missing ν, Higgs mass is not reconstructable large mh resolution (%) 1 pp! H (NNLO+NNLL QCD + NLO EW) pp! qqh (NNLO QCD + NLO EW) BR [pb]! 1 - + - VBF H + s # WW l qq + - WW l l + - ZZ l l qq = 8TeV LHC HIGGS XS WG 1 - pp! WH (NNLO QCD + NLO EW) pp! ZH (NNLO QCD +NLO EW) gg H s= 8 TeV LHC HIGGS XS WG 1 pp! tth (NLO QCD) qq H 8 3 4 [GeV] M H - -3-4 # WH l bb tth ttbb + - ZH l l bb + - ZZ l l ZZ l - + - l l l 3 4 + l = e, " = e, ", q = udscb M H [GeV] mh(15gev): BR(H WW lν) ~ %
Expected Sensitivity in WW Channel For am integrated luminosity of 1./fb at 8 TeV and 5/fb at 7 TeV At high mass, WW channel has the sensitivity to exclude the SM Higgs up to 6 GeV For SM Higgs with mass ~15 GeV, WW channel has expected significance is ~ 4σ 95% CL limit on!/! SM CMS Preliminary s = 7 TeV, L = 5.1 fb s = 8 TeV, L = 1. fb Expected limits 1 Combined H " bb H " ## H " $$ H " WW H " ZZ 3 4 6 m H (GeV) Local p-value -5-5 - CMS Preliminary s = 7 TeV, L % 5.1 fb s = 8 TeV, L % 1. fb 1 1!! 3 4! Combined H " bb H " ## H " $$ H " WW H " ZZ 1 115 1 15 13 135 14 145 m H (GeV) 5! 6! 7! 8! 3
Outline Overview of the analysis Main background estimations The SM Higgs search strategy Latest search results Summary and outlook 4
Analysis Challenges Higgs signal is several orders smaller than the backgrounds "! BR (fb) 7 6 5 4 3 Background Signal Reducible backgrounds W/Z+Jets Top 1 W+jets Z+jets Top WW ZZ HWW(15) (lν) (l) (lν) (lν) (lν) (lν) No mass peak: essentially counting experiments Irreducible backgrounds WW/ZZ Key in the analysis: improve signal to background ratio and develop reliable method to estimate the residual background 5
Analysis Overview Key selection falls into two steps WW selection: genuine dilepton MET events - Reject reducible backgrounds WW selection HWW selection - Distinguish the WW background using the Higgs mass and spin HWW selection Categorize final states according to number of jets (, 1, and ) lepton flavors (ee, eμ, μμ) -Jet eμ is the most sensitive channel Two complementary approaches to extract the Higgs production cross-section Cut-based Shape based (mll and transverse mass) 6 ν e W - H W + momentum direction ν e e e + spin direction HWW like e μ μ e WW like
Measuring Major Objects in CMS Charged Leptons: Si tracker, EM calorimeter (e ± ), Muon system (μ ± ) Photons: EM calorimeter Quark and gluons (collimated jets): Hadronic calorimeter Neutrinos: missing energy (MET) - negative sum of all visible energy 7
Two oppositely charged isolated leptons The WW Selection Leading lepton pt > GeV, trailing lepton pt > GeV Reducing Wjets background Large missing energy Projected MET (along one of the lepton direction) > GeV Reducing Z+jets backgrounds Use MVA discriminant in the ee/μμ to maximize DY reduction Top veto Remove events with soft muons and b-tagged jets Dilepton pt > 45 GeV Removing mostly W/Z+jets and W/Z+γ/γ* backgrounds Expected background contributions for 1/fb at 8 TeV All bkg WW Top W+jets Drell-Yan WZ/ZZ W/Z+γ(*) -Jet 433 ± 3146 ± 19 417 ± 45 334 ± 91 18 ± 118.1 ± 7.1 89 ± 1-Jet 899 ± 15 976 ± 111 1369 ± 56 88 ± 83 131 ± 8 88.6 ±5.6 46 ± 1 -Jet 39 ± 137 473 ± 1 1865 ± ± 58 579 ± 7 51. ± 3.5 41.3 ± 3.9 8
Reducible Background Estimations 9
W+jets Background Estimation W+jets events are reduced by tight requirement on lepton identification and isolation The residual background is due to jets faking the leptons The rate of jets to fake leptons (FR) is measured in data The residual background is estimated in a data-driven way ν l l + (l ) Cross-check the data-driven method in a second W+jets control region same sign events W+jets background for 1.1 fb in same-sign events Number of same-sign events in data 9 Number of non-w+jets background (WZ or Wγ) 8. ± 1.9 Number of W+jets background observed 64. ± 9.7 Data-driven method estimates 68.4 ± 4.6 The systematic uncertainty is ~36% mainly from MC closure test and the away jet pt threshold in the QCD data
Top Background Estimation The top background is reduced by vetoing events with b-quark Top-tagging: look for b-quark signature (soft muons and b-tagged jets) The residual background contributes mostly to events with jets - Top background in events with -jet is much smaller better sensitivity The residual top background is estimated using data-driven methods The systematic uncertainty is ~% (5%) in the -Jet (1-Jet) bin -Jet: theory calculations of relative tw and ttbar compositions the statistics in the top-tag control region 11
Drell-Yan Background Estimation Understanding the Drell-Yan background is crucial in the ee/μμ channels This background arises due to mis-measured MET, which is difficult to model in MC Veto the events in the Z peak and apply tighter MET cut than eμ channel worse sensitivity The residual Drell-Yan background is estimated MET E mis-measured In Z-peak N in (data) N out N in (MC) Subtract the non-dy backgrounds such as WZ/ZZ (peaking) and Top and WW (non-peaking) in counting the events in Z-peak Out Z-peak The systematic uncertainty is ~5%, main limiting factor in these channels mll modeling in MC ( Nout/Nin) statistics in the Nin(data) 1
Wγ* Background Wγ* comes from ISR and FSR The dileptons from the γ* internal conversions are generally of low pt - One of the dilepton is easily misreconstructed in the detector Low (mll, mt) signature Wγ* ISR Wγ* FSR - Reduce by pt(ll) > 45 GeV, mt >8 GeV For mγ* > 1 GeV, It is included in WZ/γ* For low mγ* < 1 GeV, we use a dedicated MC with the cross-section normalized to data in 3-lepton control region entries /.5 GeV 15 data * W+! WZ ZZ CMS preliminary top & V+jets L = 1.1 fb s = 8 TeV Wγ* l(μ + μ - ) The systematic uncertainty is ~ 3% 5 Compare e(μμ) and μ(μμ) k-factors Compare kfactors in different ranges of mll (test of the mll spectrum) 5 13 m ll [GeV]
Sifting out HWW 14
H WW versus WW Explore the kinematics differences due to Higgs spin and mass Low mass Higgs events, two leptons tend to have small opening angle and small dilepton mass entries / 5. GeV 6 4 data m H =13 WW W+jets DY top WZ/ZZ CMS, s = 7 TeV L = 4.6 fb -jet W - H W + ν e ν e e e + 5 15 m ll [GeV] momentum direction spin direction The transverse mass of the dilepton and MET carries signature of the Higgs resonance mass m T = p ll T Emiss T (1 cos φ E miss T,ll) 15
Cut based analysis Conservative approach Apply orthogonal selections on the best 5 kinematic observables Shape based analysis in eμ channel Optimistic approach Higgs Cross-section Extraction Further divide the analysis into 8 subchannels based on mt and mll M H = 15 GeV (GeV) M ll 18 16 14 1 8 6 4 CMS preliminary L = 1.1 fb cut-based signal region 8 1 14 16 18 4 6 8 (GeV) (8TeV) M T 18 16 14 1 8 6 4 16 (GeV) M ll Background 18 16 14 1 8 6 4 CMS preliminary L = 1.1 fb (8TeV) 8 1 14 16 18 4 6 8 (GeV) M T 14 1 8 6 4
WW Background Estimations For low mass Higgs, there is a signal free WW control region to estimate WW background N data S = NC data (N S /N C ) MC Theoretical uncertainties are assigned 1) PDF variations ) QCD scale variations 3) Parton showering HWW signal region (SR) WW control region (CR) The systematic uncertainty is ~%(%) for (1)-Jet bin For high mass Higgs, there is no signal free region dilepton invariant mass (GeV) We rely on MC to estimate the WW background primarily for the cut-based For the shape based analysis, the WW yield is determined from the fit 17
Systematics on Overall Yields Overall signal efficiency uncertainty is ~% dominated by theoretical uncertainties on missing higher order effects and PDF variations Overall background uncertainties are ~% in the HWW signal region W+jets: ~ 36% Top: ~% (-Jet) and ~5% (1-Jet) Wγ*: ~3% Drell-Yan: ~5% (-Jet), ~% (1-Jet) For all background estimations based on MC prediction, consider Luminosity: 4.4% for 8 TeV and % for 7 TeV Lepton selection efficiency (3-4%), momentum resolution (~%) MET resolution ~% Jet energy scale resolution ~% 18
Shape Systematic Uncertainties Instrumental Uncertainties Lepton selection efficiency and momentum scale MET and JES resolutions WW Background Difference between Madgraph and MC@NLO QCD Scale variations on the renormalization and factorization scales evaluated in MC@NLO PDF uncertainties (not included in HCP analysis, but verified afterwards) Wjets Background Fakerate derived with a different away jet pt threshold in the QCD data Top Background Difference between Powheg and Madgraph 19
Compare Different Channels Use expected limits as an example to see the relative contributions in different channels Most sensitive channels are the eμ channels in the shape analysis 95 % CL limit on!/! SM 9 8 7 6 CMS preliminary L = 1.1 fb DF jet D DF 1jet D (8TeV) DF jet cut-based SF jet cut-based SF 1jet cut-based SF jet cut-based All channels combined 5 4 3 1 1 115 1 15 13 135 14 (GeV) M H
SM Higgs Search Results 1
Events after WW Selections Events after the WW selections for all channels (ee/μμ/eμ) for 1.1/fb at 8 TeV Data All bkg WW Top W+jets Drell-Yan WZ/ZZ W/Z+γ(*) -Jet 445 433 ± 3146 ± 19 417 ± 45 334 ± 91 18 ± 118.1 ± 7.1 89 ± 1-Jet 353 899 ± 15 976 ± 111 1369 ± 56 88 ± 83 131 ± 8 88.6 ±5.6 46 ± 1 Higgs signal ~ 15 events in -Jet bin eμ Kinematic distributions in the most sensitive channels show evidence for the SM Higgs(15) o events / 3 5 data m H =15 GeV H15 W+jets VV Top Z/ #* WW stat.$syst. CMS Preliminary s = 8 TeV, L = 1.1 fb -Jet eμ o events / 5 data m H =15 GeV H15 W+jets VV Top Z/ #* WW stat.$syst. CMS Preliminary s = 8 TeV, L = 1.1 fb 1-Jet eμ 15 15 5 5 4 6 8 1 14 16 18 "! ll [ o ] 4 6 8 1 14 16 18 "! ll [ o ]
Cut-based Analysis Results (1) Number of events after applying Higgs selections for more mass points Take the most sensitive channel -jet eμ as an example mh Data H WW All bkg. WW W+jets W/Z+γ(*) Top Drell-Yan +WZ 15 349 58 ± 1 91 ± 7 3 ± 19 44 ± 16 5.6 ± 9.5 11. ±.5 6.6 ±.6 16 197 38 ± 51 16 ± 13 15 ± 1 5.9 ±.7.6 ± 1.5 13.1 ± 3.1 3.7 ±.4 38 95 ± 1 78 ± 1 4 ± 19 7.7 ± 3.5 1.3 ±.9 8.9 ± 6.4 6.3 ±.9 4 198 4 ± 11 ± 19 133 ±15 4.4 ±. 3.5 ±.1 5 ± 11 6. ±.7 At low mass region, we see clear excess ~ 15 GeV At high mass end data agrees much better with background only hypothesis 3
Cut-based Analysis Results (II) At low mass range we observe an excess consistent with SM Higgs 95% CL limit on!/! SM 1 CMS preliminary observed median expected H # WW # l" expected ± 1! expected ±! L = 1.1 fb (8 TeV) signal injection m =15 GeV ±! (stat.) H Expected Observed Significance.4 1.7 Signal Strength -.8 ±.45 1 3 4 5 6 Higgs mass [GeV] events / GeV/c data / MC 16 14 1 8 6 4.5 1.5 1.5 data m H =15 GeV H15 W+jets VV Top Z/!* WW stat."syst. CMS Preliminary s = 8 TeV, L = 1.1 fb All cut based selections are applied except mt 5 15 5 m ll-e T ll-e T m T [GeV/c ] 5 15 5 [GeV/c ] 4
Shape Analysis Results The low mass excess is enhanced as expected in the more sensitive analysis At high mass we exclude SM Higgs up to 5 GeV significant improvement compared to cut-based analysis mh = 15 Expected Observed Significance 3.7.9 Signal Strength 1.77 ±.8 M H = 15 GeV CMS preliminary s = 8TeV, L = 1.1 fb 5 GeV) Events/( GeV! Data/MC 18 16 14 1 8 6 4 1 data H(15) WW * Z/" top VV W+jets -Jet eμ 5 15 Bin index Bins index in the mll-mt D plane 95% CL limit on!/! SM 1 CMS preliminary observed median expected H # WW # l" expected ± 1! expected ±! L = 1.1 fb (8 TeV) signal injection m =15 GeV ±! (stat.) H 1 3 4 5 6 Higgs mass [GeV] 5
Post-Fit Kinematic Distributions The fit results of all nuisance parameters are consistent with pre-fit values within about ±σ Closer look at the signal region mll < 5 GeV Signal and background overall normalization is corrected by the postfit results M H = 15 GeV CMS preliminary s = 8TeV, L = 1.1 fb M H = 15 GeV CMS preliminary s = 8TeV, L = 1.1 fb Events/ GeV 3 5 data H(15) WW * Z/! top VV W+jets -Jet eμ Events/ GeV 16 14 1 data H(15) WW * Z/! top VV W+jets 1-Jet eμ 15 8 6 4 5 Data/MC 1 Data/MC 1 15 5 (GeV) M T 15 5 (GeV) M T 6
Post fit distributions on the final discriminant (GeV) M ll Background 18 16 14 1 8 6 4 CMS preliminary L = 1.1 fb (8TeV) 8 1 14 16 18 4 6 8 (GeV) M T 14 1 8 6 4 unroll M H = 15 GeV 5 GeV) Events/( GeV! Data/MC 1 1 data H(15) WW * Z/" CMS preliminary top VV s = 8TeV, L = 1.1 fb W+jets -Jet eμ 4 6 8 Bin index Bins index in the mll-mt D plane M H = 15 GeV CMS preliminary s = 8TeV, L = 1.1 fb At the signal free regions, we see very good agreement between data and the post-fit backgrounds 5 GeV) Events/( GeV! data H(15) WW * Z/" top VV W+jets 1-Jet eμ No indications of major background shape mis-modeling Data/MC 1 1 7 4 6 8 Bin index Bins index in the mll-mt D plane
Any other SM-like Higgs? Treat the SM Higgs 15 GeV as a background We repeat the analysis and see whether there is a second signal We find data consistent with the expectations of SM Higgs at 15 GeV plus background hypothesis No evidence of other SM-like Higgs 8
Vector Boson Fusion Production σ(vbf) is ~ times less than σ(gg fusion) VBF channel has a cleaner signature or 3 jets: leading two jets in Δη > 3.5 v l v l g 4 HWW Events at 1.1/fb in the eμ channel mh Data H WW All bkg. WW Top 15 1.7 ±.. ±.6.8 ±.5.9 ±.3 16 4 11.7 ± 1.5.9 ±.8 1. ±.6 1.5 ±. 8 9.3 ± 1. 9.1 ±.4.5 ± 1. 4.6 ± 1.3 4 7 3.9 ±.5 9.8 ± 3. 3.5 ±. 4.6 ± 1.3 6 3 1.4 ±. 3.7 ± 1.3 1.6 ± 1. 1.9 ±.8 No other jets between the two forward jets Both leptons within the two forward jets We perform a cut-based analysis Similar mh-dependent cuts on Higgs decays - leptons pt, transverse Higgs mass etc The S/B is ~ 1 for the low mass Higgs The measurement is statistically dominated Plenty room to improve for higher luminosity - Increasingly important in SM Higgs test by constraining qqh vs ggh 9 µ qqh+vh 8 6 4 CMS Preliminary s = 7 TeV, L $ 5.1 fb s = 8 TeV, L $ 1. fb H "!! H " WW H " ZZ H " bb H " ##. -.5..5 1. 1.5..5 3. 3.5 µ ggh+tth
Final Results (7TeV+8TeV Shape Analysis) We observe low mass excess at 3σ With signal strength consistent with SM Higgs ~15 GeV At high mass we exclude the SM Higgs up to 6 GeV mh = 15 Expected Observed Significance 4.1 3.1 Signal Strength 1.74 ±.5 95% CL limit on!/! SM 1 CMS preliminary observed median expected H # WW # l" expected ± 1! expected ±! L = 4.9 fb (7 TeV) + 1.1 fb (8 TeV) signal injection m =15 GeV ±! (stat.) H 1 3 4 5 6 Higgs mass [GeV] 3
Summary We observed an excess consistent with a SM Higgs-like particle with mass 15 GeV s = 7 TeV, L % 5.1 fb s = 8 TeV, L % 1. fb The excess is at 3σ above background only expectation CMS Preliminary m H = 15.8 GeV The signal strength (relative to the SM Higgs prediction with mass 15 GeV) is.74 ±.5 - Both systematics/statistic contribute roughly equally to the uncertainty - The measurement is so far the most precise one among all channels - Its consistency with H ZZ measurement respect the custodial symmetry We exclude the SM Higgs in the high mass range up to 6 GeV The expected sensitivity projected for the full 11+1 dataset is at the level of 5σ H " bb H " $$ H " ## H " WW H " ZZ.5 1 1.5.5 Best fit!/! SM 31
Prospects for HWW in the future (I) LS1 LS 13 15-17 19-1 phase 1 /fb @ 8 TeV 4/fb/year upgrade /fb/year@14 TeV 5/fb @ 7 TeV @13 TeV 3/fb@14 TeV At the end of the nominal LHC ~ early, we expect to collect 3/fb at 14 TeV HWW signal strength can be measured with5% precision, not enough for claiming SM EWSB The VBF production has the potential to be discovered at 5σ At the end of the HL-LHC run ~ early 3, we expect to collect 3/fb at 14 GeV Contribute to measure H VV coupling at < 5% level which poses stringent constraint on NP models LS3 phase upgrade HL-LHC 4-3 /fb/year @ 14 TeV 3/fb at 14 TeV 3
Prospects for HWW in the future (II) WW lν channel provide a testing ground for other spin resonances Such as the spin graviton like resonance with minimal couplings to W pairs Phenomenological analysis is performed with simplified backgrounds and ignoring systematics Exploit the same mll-mt shapes for different models (S. Bolognese et al, Phys. Rev. D 86 9531) At the level of the 5σ discovery with SM Higgs, we expect to rule out the m+ model at 95% level This is promising with the full 11/1 dataset in the near future SM Higgs WW X(m+) WW qq WW 33
Backup slides 34
Parton Density Function 35
Higgs Search Limits (Statistical Analysis) In the absence of discovery, we set upper-limits at 95% C.L. on signal strength μ = σ/σsm The μ 95% quantifies roughly the signal size corresponding to σbackground If μ 95% = 1 signal strength μ 95% is excluded CLs modified frequentist method Define a test statistic that quantifies the signal-ness of events Define CLs that quantifies how likely the observed data is S+B or B-only 36