Search for a SM Higgs Boson in semi-leptonic di-tau final states with ATLAS at the LHC

Search for a SM Higgs Boson in semi-leptonic di-tau final states with ATLAS at the LHC ET miss = 9 GeV μ (pt=63 GeV) τhad (pt=96 GeV) mττ =9 GeV Nils Ruthmann mjj =65 GeV Nils Ruthmann Universität Freiburg International School of Subnuclear Physics ERICE - Italy 6 June 3

L EW = @ µh@ µ H + g 4 (v + H) (W + µ W µ + Introduction Higgs mechanism major ingredient of the SM Spontaneous breaking of the U()Y SU()L symmetry leading to massive W,Z states cos Z µ Z µ )+ W ( µ )H vh 3 4 H4 +... Lepton masses generated via independent Yukawa couplings G f v p f f G f p f fh V ( )=µ + ( ) Nils Ruthmann - ISSP 3 Erice 6.6.3

L EW = @ µh@ µ H + g 4 (v + H) (W + µ W µ + Introduction Higgs mechanism major ingredient of the SM Spontaneous breaking of the U()Y SU()L symmetry leading to massive W,Z states cos Z µ Z µ )+ W ( µ )H vh 3 4 H4 +... Lepton masses generated via,independent Yukawa couplings G f v p f f G f p f fh V ( )=µ + ( ) Higgs Boson mass hardly constrained by EW data and theory (log. loop corr., stability bounds) mh = 94 (+ 9-4) GeV - favouring low mass range LHC direct searches exclude broad range leaving a small window 95% CL Limit on / SM ATLAS Obs. Exp. ± ± Ldt = 4.6-4.9 fb s = 7 TeV Data ATLAS: 7.5-9 GeV CMS: < 7.5 GeV CLs Limits 3 4 5 6 m H [GeV] Phys. Rev. D86 () 33 @ 95% C.L. June Nils Ruthmann - ISSP 3 Erice 3 6.6.3

L EW = @ µh@ µ H + g 4 (v + H) (W + µ W µ + Introduction Higgs mechanism major ingredient of the SM Spontaneous breaking of the U()Y SU()L symmetry leading to massive W,Z states cos Z µ Z µ )+ W ( µ )H vh 3 4 H4 +... Lepton masses generated via,independent Yukawa couplings G f v p f f G f p f fh V ( )=µ + ( ) Huge progress in the past months Discovery Mass measurement Coupling measurements Spin determination ATLAS Preliminary W,Z H bb s = 7 TeV: Ldt = 4.7 fb : Ldt = 3 fb H s = 7 TeV: Ldt = 4.6 fb : Ldt = 3 fb (*) H WW H s = 7 TeV: Ldt = 4.8 fb : Ldt =.7 fb (*) H ZZ Combined ll s = 7 TeV: Ldt = 4.6 fb : Ldt =.7 fb 4l s = 7 TeV: Ldt = 4.6 fb : Ldt =.7 fb s = 7 TeV: Ldt = 4.6-4.8 fb : Ldt = 3 -.7 fb Direct observation in fermionic decays ( ) Its up to ττ / bb µ =.3 ±. m H = 5.5 GeV + Signal strength (µ) p 3 9 6-3 -6-9 5 8 - -4 ATLAS-CONF-3-34 ATLAS Preliminary Observed SM expected June 3 s = 7 TeV, Ldt = 4.6-4.8 fb, Ldt = 3-.7 fb 5 5 3 35 m H [GeV] 4 6 8 Nils Ruthmann - ISSP 3 Erice 4 6.6.3

SM Higgs branching ratios (HDECAY) bb _ WW. ττ gg ZZ cc _ - Branching ratios of a SM Higgs Boson Low mass Higgs Boson branching ratios diverse γγ : Clean signature - High mass resolution => Discovery, mass measurement ZZ: Low background - High mass resolution => Discovery, mass measurement WW: Experimentally challenging - Statistically powerful (Spin determination) bb: Challenging due to multijet bkd - Accessible in association with W/Z (tt) ττ: Large BR - Probes leptonic Yukawa coupling - Large irreducible Z bkd γγ Zγ -3 4 6 8 m (GeV/c ) H Nils Ruthmann - ISSP 3 Erice 5 6.6.3

. - SM Higgs branching ratios (HDECAY) bb _ WW ττ gg ZZ cc _ Branching ratios of a SM Higgs Boson Low mass Higgs Boson branching ratios diverse γγ : Clean signature - High mass resolution => Discovery, mass measurement ZZ: Low background - High mass resolution => Discovery, mass measurement WW: Experimentally challenging - Statistically powerful (Spin determination) bb: Challenging due to multijet bkd - Accessible in association with W/Z (tt) ττ: Large BR - Probes leptonic Yukawa coupling - Large irreducible Z bkd γγ Zγ -3 4 6 8 m (GeV/c ) H Three τ decay channels: Leptonic ( light leptons + 4 ν) ee/µµ Large bkd, statistically limited ll % lh 46% hh 4% Hadronic ( hadronic τ + ν) Large QCD background - Trigger complicated High statistics Semi-leptonic ( light lepton + 3 ν ) Presence of trigger lepton - Statistically powerful Nils Ruthmann - ISSP 3 Erice 6 6.6.3

Production mechanisms and backgrounds Mainly two production mechanisms important for H ττ searches. Gluon-Fusion Largest cross-section Jets at LO Additional partons due QCD radiation Categorisation to account for different bkd composition ( Jets low/hig MET, Jet) Main Backgrounds Z->ττ + n jets W + (n+) jets QCD.Vector-Boson-Fusion nd largest cross-section Jets at LO Additional partons due to QCD + EW radiation Topology allows strong bkd suppression Z->ττ + jets W + 3 jets tt-> bbww QCD Nils Ruthmann - ISSP 3 Erice 7 6.6.3

Hadronic τ-decay reconstruction Hadronic τ decays are identified by : low track multiplicity A narrow shower profile A specific EM/HAD energy mix Nils Ruthmann - ISSP 3 Erice 8 6.6.3

Hadronic τ-decay reconstruction Hadronic barrel/ extended barrel/ endcap (, )= (.,.) LAr EM barrel/endcap 3 () layers, middle layer: (, )= (.5,.5) Hadronic τ decays are identified by : multiplicity low track A narrow shower profile A specific EM/HAD energy mix R Reece, KG Tan, P Malecki, F Nils Ruthmann - ISSP 3 Erice Friedrich, N Dawe, K Nakamura + ; " classifier Identification based on multivariate Using pile-up corrected input variables /, # Typical background rejection of ~ at signal efficiencies of 6% & scale Energy calibrated to the visible τh!! : Typical scale uncertainties ~3% " " ' " 9 6.6.3

Preselection Events 4 8 6 4 3 ATLAS L dt =.3 fb work in progress Data SS Data t t +single-top Z ll WW/WZ/ZZ W +jets Z H(5) x 5 Events selected by: e/μ OR l+τ trigger Events 7 6 5 4 3 3 ATLAS L dt =.3 fb swork = 8 TeV in progress Data SS Data t t +single-top Z ll WW/WZ/ZZ W +jets Z H(5) x 5 Data/Bkd.5.5 4 6 8 46 8 [GeV] m T 4 6 8 4 6 8 [GeV] m T νe/μ e/μ ντ E/ T ντ τhad Jet topology separates VBF like events Data/Bkd 4 6 8.5 N Jet.5 4 6 8 N Jet Data/Bkd 3 4 8 6 4 8 Fake τ s: - W+Jets - QCD ATLAS L dt =.3 fb work in progress Data SS Data t t +single-top Z ll WW/WZ/ZZ W +jets Z H(5) x 5 6 4 4 6 8 46 8.5 MET [GeV].5 4 6 8 4 6 8 Tight lepton ID criteria Crucial to reduce multijet background before actual τ-id MET [GeV] MET resolution crucial for mass reconstruction Jet(s) ν 4-momenta fitted exploiting known τ-decay kinematic properties MET resolution dominates Real τ s: σmass - Z ττ - WW/WZ - top Poor mττ resolution Events Data/Bkd 5 5 5.5.5 3 ATLAS L dt =.3 fb work in progress Data SS Data t t +single-top Z ll WW/WZ/ZZ W +jets Z H(5) x 5 5 5 5 3 35 4 MMC m [GeV] 5 5 5 3 35 4 Nils Ruthmann - ISSP 3 Erice 6.6.3 MMC m [GeV]

Background estimation Fake τ-background needs data-driven estimation! Same sign events offer control region dominated by fake contributions QCD: Mainly gluon initiated fakes, no charge correlation OS SS W + Jets: Quark enriched jet sample with charge correlation to lepton OS>SS Fake rates depend on shower variables => Tough to simulate n Bkd OS = n All SS + n QCD OS-SS + nw+jets OS-SS + nz! OS-SS + n other OS-SS Nils Ruthmann - ISSP 3 Erice 6.6.3

Background estimation Fake τ background needs data-driven estimation Same sign events offer control region dominated by fake contributions QCD: Mainly gluon initiated fakes, no charge correlation OS SS W + Jets: Quark enriched jet sample with charge correlation to lepton OS>SS n Bkd OS = n All SS + n QCD OS-SS + nw+jets OS-SS + nz! OS-SS + n other OS-SS W normalisation measured in W control region Corrects Jet τ fake rate Depends on Quark/Gluon composition in the jet sample Charge Correlation and Jet Multiplicity dependent Additional control regions for cross-checks or normalisation corrections: tt Z->ll Quark/Gluon ratio 8 7 6 5 4 3 ATLAS Simulation work in progress OS SS Events / 5 GeV 5 5 5 3 ATLAS-CONF-6 e had Preselection Data 5 x H(5) Z (OS-SS) Others (OS-SS) Same Sign Data Stat. uncert. L dt = 3. fb ATLAS Preliminary 4 6 8 4 6 8 m T [GeV] Nils Ruthmann - ISSP 3 Erice 6.6.3

Background rejection : Real τ s Events 5 5 3 L dt =.3 fb Real τ s: - Z ττ - WW/WZ - top Data SS Data t t +single-top Z ll WW/WZ/ZZ W +jets Z H(5) x 5 5 ATLAS work in progress Data/Bkd.5.5 5 5 5 3 35 4 MMC m MMC m [GeV] 5 5 5 3 35 4 [GeV] Nils Ruthmann - ISSP 3 Erice 3 6.6.3

Background rejection : Z ττ Irreducible Z ττ background hard to separate Same final state including resonance event topology Real τ s: - Z ττ - WW/WZ - top VBF Production: Focus on specific Jet topology: Z: Mainly Drell-Yan production (+radiation) H: Jets at leading order (large angular separation, high pt) Events Data/Bkd 8 7 6 5 4 3.5 µ had H+-jet VBF + e had L dt =.3 fb ATLAS work in progress Data t t +single-top Z ll WW/WZ/ZZ Z Fake H(5) x...3.4.5.6.7.8.9 lep_centrality.5...3.4.5.6.7.8.9 Tagging jets bracket the leptons lep_centrality Events Data/Bkd 4 8 6 4.5.5 µ had H+-jet VBF + e had L dt =.3 fb ATLAS work in progress Data t t +single-top Z ll WW/WZ/ZZ Z Fake H(5) x 3 4 5 6 7 8 9 jj 3 4 5 6 7 8 9 Large rapidity separation of two tagging jets jj No additional jet activity Nils Ruthmann - ISSP 3 Erice 4 6.6.3

Background rejection : Fake τ s Events 5 5 3 L dt =.3 fb Fake τ s: - W+Jets - QCD Data SS Data t t +single-top Z ll WW/WZ/ZZ W +jets Z H(5) x 5 5 ATLAS work in progress Data/Bkd.5.5 5 5 5 3 35 4 MMC m MMC m [GeV] 5 5 5 3 35 4 [GeV] Nils Ruthmann - ISSP 3 Erice 5 6.6.3

Events Fake τ s: - W+Jets - QCD µ + e Boosted had had 6 L dt =.3 fb Data 5 4 3 ATLAS work in progress Background rejection : Fake τ s Rejection of non-resonant backgrounds: Does the event look like a resonant ττ-system? t t +single-top Z ll WW/WZ/ZZ Z Fake H(5) x MET centered between the τ s? e/μ Δ R MET τhad e/μ Δ R MET Jet faking τhad Data/Bkd Events Data/Bkd.5.5 45 4 35 3 5 5 5 µ + e Boosted had had L dt =.3 fb -.5.5 metphi_centrality -.5.5 metphi_centrality Is the : neutrino share reflected in the lepton/τ pt? ATLAS work in progress Data t t +single-top Z ll WW/WZ/ZZ Z Fake H(5) x 3 4 5 6 7 8 9.5 ptl_over_ptt.5 3 4 5 6 7 8 9 ptl_over_ptt Nils Ruthmann - ISSP 3 Erice 6 6.6.3 Jet(s) mττ : Mass consistent with 5 GeV? Events Data/Bkd 45 4 35 3 5 5 5 µ + e Boosted had had L dt =.3 fb ATLAS work in progress Data t t +single-top Z ll WW/WZ/ZZ Z Fake H(5) x 3 4 5 6.5 R.5 3 4 5 6 Small opening angle between two taus in case of high pt recoil system R

Categorisation Jets VBF like topology Not VBF Not Boosted NJet>= Not VBF Not Boosted NJet= VBF - jet + jets inclusive Boosted H Not VBF High pt ττ Good mass resolution Analyses categories aiming at: Best possible background rejection based on specific event topologies Partial enhancement of specific production modes allowing for dedicated Higgs coupling fits Split the phase space - Not restrict it Based on Jet and Resonance quantities Nils Ruthmann - ISSP 3 Erice 7 6.6.3

Results Result based on 3 fb combined with 4.6fb of 7 TeV data No excess observed => Extract upper limit on cross-section Expected exclusion limit ~.7 σsm Events / GeV 8 6 4 Boosted + e had µ had Events / GeV Data 5 x H(5) Z (OS-SS) Others (OS-SS) Same Sign Data Bkg. uncert. L dt = 3. fb ATLAS Preliminary 5 5 MMC mass m [GeV] 8 6 4 8 6 4 H+-jet VBF + e had µ had Data H(5) Z Others Fake Bkg. uncert. L dt = 3. fb ATLAS Preliminary 5 5 5 3 MMC mass m [GeV] 95% CL Limit on / SM 8 H lep had Observed CL s Expected CL s ± ± ATLAS Preliminary L dt = 4.6 fb, s = 7 TeV L dt = 3. fb, 6 4 3 4 5 m H [GeV] ATLAS-CONF-6 Nils Ruthmann - ISSP 3 Erice 8 6.6.3

Results Result based on 3 fb combined with 4.6fb of 7 TeV data No excess observed => Extract upper limit on cross-section Expected exclusion limit ~. σsm when combined with other subchannels 95% CL Limit on / SM 8 7 6 5 4 3 H Observed CL s Expected CL s ± ± ATLAS Preliminary L dt = 4.6 fb, s = 7 TeV L dt = 3. fb, µ VBF+VH B/B SM 5 4 3 ATLAS Preliminary L = 4.6 fb, L = 3. fb, s = 7 TeV best fit best fit (µ>) 95% Contour 68% Contour SM prediction Background only m H = 5 GeV 3 4 5 m H [GeV] - -3-4 6 8 ATLAS-CONF-6 µ B/B ggf SM Nils Ruthmann - ISSP 3 Erice 9 6.6.3

Outlook Searching for leptonic Higgs Boson decays now more interesting than ever H τ τ plays a crucial role in this challenge A preliminary analysis of a partial dataset shows no significant excess, but.. our collaboration is currently reviewing a improved analysis of the full fb dataset Aiming at a 3 σ sensitivity - in the presence of a signal Stay tuned! Nils Ruthmann - ISSP 3 Erice 6.6.3

Outlook Thanks for your attention Nils Ruthmann - ISSP 3 Erice 6.6.3

BACKUP Nils Ruthmann - ISSP 3 Erice 6.6.3

E/p h ±, pt < GeV Gaussian PDFs Mean: Data/MC Width: Uncertainty Testbeam h ±, pt > GeV TES Uncertainty Single particle uncertainties propagated to composite objects (Jets, Taus) Bottom-up: From single particles to a more complex object Pseudo-experiment approach Sampling over various uncertainties: Toy experiments randomly shifted energy scales Fractional uncertainty...8.6 ATLAS Preliminary prong η <.3 <E/p> corr. Data/MC.9.8.7.6.5.4.3....9 Data <.8 Simulation 3 4 5 6 7 3 4 p [GeV] Single particle resp. Underlying event Pile-Up ATLAS Preliminary Data 3 4 5 6 7 3 4 Data + Simulation Material modeling Non-closure Total uncertainty p [GeV].4. 3 4 5 6 7 8 τ p [GeV] T Nils Ruthmann 3.4.3

Multivariate analysis approach Based on the HCP results last year: continuing the same analysis just adding data would not allow for a 3 σ evidence within Run Refined analysis techniques are needed to accomplish this goal H ττ fully focussed on a BDT analysis from the beginning 95% CL Limit on / SM 8 6 had H lep Observed CL s Expected CL s ± ± ATLAS Preliminary L dt = 4.6 fb, s = 7 TeV L dt = 3. fb, Furthermore: Fully focussed on a Higgs Boson mass of 5 GeV 4 V<y V<x V5<z 3 4 5 m H [GeV] Boosted Decision Trees widely used in ATLAS Iterative training procedure on a signal and background sample Mis-Classified events in one iteration will be amplified for the next one (boosting) Final classification score averages the decision of all trees V3<v V4<k V6<q V7<p Classified as Signal Classified as Background Nils Ruthmann - ISSP 3 Erice 4 6.6.3

φ for the two methods. In contrast to the nce of long tails toward large masses in the cant improvement for low-mass Higgs boel by significantly reducing a large Z ττ Mass estimation mτ << mz => Tau decay products heavily boosted Collinear approximation assumes θ(ν,vis) = Calculate visible energy fraction: x, (,) Deviations from θ(ν,vis) = broaden resolution leading to unphysical solutions => efficiency loss Parametrized PDFs of ν-angular distributions - pt and decay dependent Fit the best solution in multi-dimensional space of opening angles and MET x-y resolution METx, METy, + angles => 5 variables floating in the fit ise Improves completely mass overwhelm resolution majorly the: Higgs search ΔRMS/RMS = 33% for Z + Jets Has high efficiency compared to collinear approximation 35 ) Arbitrary units 9 8 7 6 5 4 3 H τ τ, lep-had channel, M h =5 GeV/c Missing Mass Calculator Collinear Approximation arxiv:.4686 [hep-ex] 5 5 5 3 35 (GeV/c ) M τ τ arxiv:.4686 [hep-ex]...3.4.5.6.7.8 ΔR Nils Ruthmann - ISSP 3 Erice 5 6.6.3 Events 4 Z τ τ 3 -prong τ decays