Status of Higgs and jet physics. Andrea Banfi

Transcription

1 Status of Higgs and jet physics Andrea Banfi

2 Outline Importance of jet physics in LHC Higgs analyses Zero-jet cross section NNLL+NNLO resummations Progress in Monte Carlo event generators One-jet cross section NLL+NLO resummations Progress towards NNLO Two-jet cross sections Discriminating between gluon-gluon and vector-boson fusion Further information Handbook of LHC Higgs cross sections 1,2,3 [Inclusive observables , Differential distributions , Higgs properties ] CERN workshop Jet issues in Higgs physics"

3 Higgs searches at the LHC LHC experiments will focus on determining whether the Higgs boson found in July 2012 is truly the missing piece of the Standard Model? Many analyses, both for the Higgs and for new physics searches, will involve final states containing hadronic jets

4 We wish to see... b b Higgs boson H Z e e + Black holes Dark matter At a hadron collider heavy particles are produced together with strongly interacting quarks and gluons (a.k.a. partons)

5 ... instead we might see Higgs boson Dark matter Black holes Whatever Physics shows up at the LHC, it always involves hadronic jets, the footprints of quarks and gluons in our detectors

6 QCD as a theory of jets Multi-jet cross sections Structure of jets Fixed-order QCD All-order QCD (parton shower, resummations)

7 Jet-veto cross sections Understanding jet-veto cross sections is crucial to establish if the recently found Higgs boson is compatible with that of the Standard Model -1 s = 7 TeV, L = 5.1 fb -1 s = 8 TeV, L = 12.2 fb CMS Preliminary m H = GeV H " bb H " $$ H " ## H " WW H " ZZ Best fit / SM

8 Vetoing jets: why? Example: Higgs decaying into WW suffers from a large background from top-antitop production H W + t b W + W t b W Each top quark decays into a b-jet ) veto events with jets in the final state Jet-vetoes are employed in many other Higgs analyses

9 X W + H W Zero-jet cross sections

10 Gluon fusion: Higgs to WW Divide events according to jet multiplicity: zero, one and two or more jets Zero-jet cross section, veto all jets with p t,jet >p t,veto ATLAS CMS anti-k T R =0.4 anti-k T R =0.5 0-jet selection p Tj < 25 GeV for j < 2.5 p Tj < 30 GeV, j < 4.7 p Tj < 30 GeV for 2.5 < j < 4.5 X W + H W This works well: the zero-jet cross section the huge (yellow) top-antitop background 0 jet is least contaminated by

11 Higgs plus zero-jets at fixed order The Higgs cross section in gluon fusion has been computed at very high accuracy 0 1 d 0 jet s {z} s + s A H {z} NLO NNLO finite m t,m b NLO [Spira et al. NPB 453 (1995) 17] large- large- m t m W NNLO QCD-EW [Anastasiou Melnikov Petriello NPB 724 (2005) 197] [Catani Grazzini PRL 98 (2007) ] [Anastasiou Boughezal Petriello JHEP 04 (2009) 003] Uncertainties in the Higgs total cross section are small, of order 7-8% These calculations are implemented in computer codes (FEHiP, HNNLO) producing exclusive events directly compute at NNLO First steps have been made towards NNNLO tot ) 0 jet [see e.g. Anastasiou et al and ]

12 Need for resummation At fixed-order, various ways of treating uncertainties (scale variations, Stewart-Tackmann, efficiency method) give different results pp, 8 TeV, m H = 125 GeV FIXED ORDER NNLO MSTW2008 NNLO PDFs anti-k t jets, R=0.5 with efficiency with Stewart-Tackmann scale variations 0-jet (p t,veto ) [pb] p t,veto [GeV] Origin of instability: large logarithms ln(m H /p t,veto ) at all orders in the perturbative expansion ) all order resummation needed

13 Zero-jet resummations We have performed NNLL resummation matched to NNLO, and implemented it in the code JetVHeto [AB Monni Salam Zanderighi Phys.Rev.Lett. 109 (2012) ] Our results have been independently confirmed by two different groups in the framework of Soft-Collinear Effective Theory (SCET) [Becher Neubert JHEP 07 (2012) 108 ] [Becher Neubert Rothen ] [Stewart Tackmann Walsh Zuberi ] Recent improvements: Ingredients beyond NNLL accuracy Effect of top and bottom masses in loops [Becher Neubert Rothen ] [Stewart Tackmann Walsh Zuberi ] [AB Monni Zanderighi ]

14 Zero-jet resummation summary p LHC s =8TeV MSTW2008NNLO 0 jet(25 GeV,R= 0.4) [pb] 0 jet(30 GeV,R= 0.5) [pb] BMSZ ± ± 1.47 large-m t B NR (+0.65) ( 1.15) n/a large-m t STWZ ± 1.22 pert (±0.46 clust ) ± 0.87 pert (±0.24 clust ) large- m t BMZ ± ± 1.79 exact m t,m b All results are compatible within uncertainties Theoretical uncertainties are between 10% and 15% Inclusion of mass effects increases the uncertainty

15 Resummations vs data With existing data it is already possible to have a measurement of and perform comparison with NNLL+NNLO resummations 0 jet [fb/gev] T / dp dσ fid ATLAS Preliminary data gg H gg H X H syst. unc. NNLO+NNLL' (STWZ) + X H NNLO+NNLL (JETVHETO) + X H = VBF + VH + tth H γ γ, s = 8 TeV -1 L dt = 20.3 fb 0.4 Ratio to POWHEG Particle level p [GeV] T,jet1 Good agreement with data in the zero-jet bin p t The leading-jet p t spectrum is underestimated at high p t

16 Monte Carlo vs data Monte Carlo event generators simulate soft and collinear emissions at all orders (parton shower) : they are valuable alternatives for resummation [fb/gev] T / dp dσ fid ATLAS Preliminary data gg H syst. unc. NLO+PS (POWHEG+PY8) + X H gg H+1j NLO+PS (MINLO HJ+PY8) + X H X H = VBF + VH + tth H γ γ, s = 8 TeV -1 L dt = 20.3 fb 0.4 Ratio to POWHEG Particle level p [GeV] T,jet1 Monte Carlo produce fully exclusive events at hadron level, ready to be interfaced with experimental detector simulations

17 Monte Carlo recent progress (I) State-of-the-art for Monte Carlo event generators is matching of parton shower with NLO calculations (POWHEG and m T in signal region (N jet = 0) m T in signal region (N jet = 1) dσ/dmt [fb/gev] dσ/dσ MEPS@NLO Sherpa+OpenLoops MEPS@NLO µ F,R /2 2µ F,R µ Q / 2 2µ Q Unc. quad. sum MC@NLO NLO lνlν + 0j m T [GeV] LHC HIGGS XS WG 2013 dσ/dmt [fb/gev] dσ/dσ MEPS@NLO Sherpa+OpenLoops MEPS@NLO µ F,R /2 2µ F,R µ Q / 2 2µ Q Unc. quad. sum MC@NLO NLO lνlν + 1j m T [GeV] LHC HIGGS XS WG 2013 New methods to merge different jet multiplicities ensuring NLO accuracy for each multiplicity (MEPS@NLO, HJ-MiNLO) m ll in signal region (N jet = 1) 0.45 [Hoeche Krauss Schoenherr Siegert MEPS@NLO JHEP 04 (2013) 027] 0.4 µ F,R /2 2µ F,R [Hamilton µ Q / 0.35 Nason Oleari Zanderighi 2 2µ JHEP Q 08 (2013) 082] Unc. quad. sum 0.3 MC@NLO σ/dm ll [fb/gev] IGGS XS WG 2013

18 d /dy [pb] Ratio 4 Monte Carlo recent progress (II) Improved HJ-MiNLO procedure gives the first parton shower that ensure NNLO accuracy for the Higgs total cross section y HNNLO NNLOPS [Hamilton Nason Re Zanderighi JHEP 10 (2013) 222] Overall, very good agreement between Monte Carlo and analytic resummations for the zero-jet cross section "(p T,veto ) #/" central JETVHETO NNLOPS Anti R=0.4 k T p T,veto [GeV]

19 jet X H One-jet cross sections

20 Resummed one-jet cross section The region p t,jet p t,veto is responsible for 50% of the uncertainties of the one-jet bin ) resummation of ln(p t,jet/p t,veto) needed X jet X H Resummation performed in SCET at NLL accuracy matched to NLO [Liu Petriello Phys.Rev. D87 (2013) , Phys.Rev. D87 (2013) ] Resummation gives a reduction of theoretical uncertainties NLO 1 jet(25 GeV) = % 46% NLL 0 +NLO 1 jet (25 GeV) = % 30%

21 One jet at fixed order Origin of the large (50%) uncertainty in the one-jet bin is that known at NLO only (in the large- limit) m t 1 jet is

22 One jet at fixed order Origin of the large (50%) uncertainty in the one-jet bin is that known at NLO only (in the large- limit) m t 1 jet is Recent advances in NNLO methods made it possible to obtain NNLO for the gluon-gluon channel jet [Boughezal Caola Melnikov Petriello Schulze JHEP 06 (2013) 072] σ NNLO at σhad [fb] k T algorithm R =0.5 p Tj = 30 GeV µ [GeV] σ NLO σ LO The full NNLO should be available soon

23 p t Finite masses at high pt M T R [AB Martin Sanz ] A high- gluon can resolve a loop with a top partner of mass mixing with the top through an angle (fb) T,j dσ/dp m t H M H T s = 8 TeV µ = µ F R = m H MSTW2008NNLO M T =600 GeV, sin 2 θ R 2 M T =1000 GeV, sin θ R M T =2000 GeV, sin 2 θ R (GeV) p T,j = 0.4 = 0.4 = M T =600 GeV, sin θ R = M T =1000 GeV, sin θ R = M T =2000 GeV, sin θ R no top partner = 0.1 Need perturbative control on the tail of jet- p t distribution, where ln(p t,j /m t ) is large ) case for higher order corrections

24 X X ggf H VBF H Two-jet cross sections

25 Two jets: VFB vs ggf In VBF events the Higgs tends to recoil against the two forward jets ATLAS CMS loose CMS tight anti-k T R =0.4 anti-k T R =0.5 anti-k T R =0.5 2-jet selection p Tj >25 GeV for η j <2.5 jet 1: p Tj >30 GeV, η j <4.7 p Tj >30 GeV, η j <4.7 p Tj >30 GeV for 2.5< η j <4.5 jet 2: p Tj >20 GeV, η j <4.7 η jj = η j1 η j2 > 2.8 > 3.0 > 3.0 m jj > 400 GeV > 250 GeV > 500 GeV η H (η j1 + η j2 )/2 - < 2.5 < 2.5 φ H jj > 2.6 > 2.6 > 2.6 Extra jet veto condition, cut on H jj

26 σ2(m σ 2 (ST: ρ = 0) Two-jet uncertainties 0.1 σ2(m 0.2 σ 2 (ST: ρ = 0) σ2( φ cut H jj ) [pb] A cut H jj > 2.6 is sensitive to higher order effects ) m careful estimate cut jj [GeV] m cut jj [GeV] of NLO uncertainties of gluon fusion using different methods FIG. 10: Exclusive 2-jet cross section over a range of m cut jj panel) for the ATLAS VBF selection. σ ggf 2 /σ 2 VBF [%] mπ cut jj φ [GeV] cut gg H gg+2j H (NLO +2j 8(NLO TeV) 8 TeV) m H = m125 H GeV = 125 GeV ATLAS 2-jet selection η jj > 2.8, p T Hjj < 30GeV Total uncertainty Pert. uncertainty (ST) 20% relative uncertainty ATLAS 2-jet selection m jj > 400 GeV, η jj > 2.8 µ = m H (scheme a) H jj Efficiency method ST: ρ = 0.4 ST: ρ = for fixed p THjj < 30 GeV (left panel) and fixed φ H jj > 2.6 (right [Gangal Tackmann Phys.Rev. D87 (2013) ] [GeV] Too extreme VBF cuts might increase the uncertainty in ggf ggf quantity as a function of σ ggf 2 /σ 2 VBF [%] gg H +2j (NLO 8 TeV) m H = 125 GeV m cut jj ATLAS 2-jet selection η jj > 2.8, φ H jj > 2.6 Total uncertainty Pert. uncertainty (ST) 20% relative uncertainty FIG. 11: Perturbative uncertainties of the ggf contribution relativetothevbfcrosssectionoverarangeofm cut jj p THjj < 30 GeV (left panel) and fixed φ H jj > 2.6 (rightpanel)fortheatlasvbfselection. ) for fixed check the We already saw in Sec. IV A that the perturbative uncertainties we also saw in the results in Table II. in the exclusive 2-jet cross 2 / 2 VBF section also depend In Fig. m cut 11 we show the ggf uncertainty relative to the on the chosen VBF cuts and increase with a higher cut on VBF cross jj section analogous to Fig. 9. We can clearly the dijet invariant mass, m jj. The reason for this effect is see that in this case tightening the cut on m jj does improve that at higher m jj the effective hard scale in the process the separation of the ggf and VBF contributions, is also pushed higher causing the logarithmic corrections as the perturbative ggf uncertainty relative to the VBF cut

27 VBF cuts in Monte Carlo's Accurate predictions for the two-jet cross section rely on modelling of the third jet in gluon fusion dσ/dy [pb] Rapidity of 3rd jet WBF selection Mcfm Hej PowhegBox Sherpa LHC HIGGS XS WG 2013 dσ/d φ [pb] Azumuthal separation of the two leading jets WBF selection Mcfm Hej PowhegBox Sherpa LHC HIGGS XS WG y(j 3 ) φ(j 1, j 2 ) In Monte Carlo generators the third jet is produced at LO only: discrepancies due to details of the shower and/or tree-level merging

28 Higgs plus three jets at NLO Very recently, Higgs production in gluon fusion with three additional jets has been computed at NLO with GoSam [Cullen et al. PRL 111 (2013) ] dσ/dp T,j [pb/gev] LHC 8 TeV cteq6me pdf anti-kt: R=0.5, p T > 20 GeV, η < st jet LO 1 st jet NLO nd jet LO 2 nd jet NLO 3 rd jet LO rd jet NLO p T,j [GeV] Jet- p t distributions are stable with inclusive cuts Looking forward to seeing NLO distributions with extreme VBF cuts

29 Conclusions Jets are important ingredients of many LHC Higgs analyses Uncertainties in exclusive jet cross sections are large calculations are very important, precision Study of Higgs and jets triggered many theoretical advances First steps towards NNNLO Higgs production in gluon fusion NNLL+NNLO resummation for zero jets, progress towards NNNLL NNLOPS: first Monte Carlo generator matched to NNLO Higgs plus one jet at NNLO (gg channel only) Higgs plus three jets at NLO Comparison and validation all existing theoretical tools (fixed-order calculations, resummations, Monte Carlo generators) in progress

30 Conclusions Jets are important ingredients of many LHC Higgs analyses Uncertainties in exclusive jet cross sections are large calculations are very important, precision Study of Higgs and jets triggered many theoretical advances First steps towards NNNLO Higgs production in gluon fusion NNLL+NNLO resummation for zero jets, progress towards NNNLL NNLOPS: first Monte Carlo generator matched to NNLO Higgs plus one jet at NNLO (gg channel only) Higgs plus three jets at NLO Comparison and validation all existing theoretical tools (fixed-order calculations, resummations, Monte Carlo generators) in progress Thank you for your attention

31 Extra

32 _ WH: Higgs to b-b Tricky issue: suppress background while keeping jets from b b system b b X H W + ATLAS CMS anti-k T R =0.4 anti-k T R =0.5 b- b system p Tj1 > 45 GeV, p Tj2 > 20 GeV j < 2.5 p Tj1,p Tj2 > 30 GeV allow extra jet with p Tj > 20 GeV j < 2.5 j1, j2 > 30 p TW - > 100 GeV (> 120 GeV, ) other jets p Tj < 30 GeV, j > 2.5 BDT HW - BDT WH setup is involved and will not be considered in the rest of the talk

33 Theoretical uncertainties We have combined the NNLL resummation with NNLO, using three matching schemes (a), (b) and (c) Central value: scheme (a) with µ R = µ F = Q = m H /2 Q is the resummation scale: ln(m H /p t,veto ) ln(q/p t,veto ) [AB Monni Salam Zanderighi 12] gg H, m H = 125 GeV pp, 8 TeV MATCHED NNLL+NNLO MSTW2008 NNLO PDFs anti-k t, R = 0.5 (p t,veto ) (p t,veto ) / central (p t,veto ) p t,veto [GeV]

34 Theoretical uncertainties We have combined the NNLL resummation with NNLO, using three matching schemes (a), (b) and (c) [AB Monni Salam Zanderighi 12] Central value: scheme (a) with µ R = µ F = Q = m H /2 Q is the resummation scale: ln(m H /p t,veto ) ln(q/p t,veto ) Variation of µ R,µ F with Q = m H /2 m H 4 apple µ 1 R,µ F apple m H 2 apple µ R apple 2 µ F (p t,veto ) gg H, m H = 125 GeV pp, 8 TeV MATCHED NNLL+NNLO MSTW2008 NNLO PDFs anti-k t, R = (p t,veto ) / central (p t,veto ) p t,veto [GeV]

35 Theoretical uncertainties We have combined the NNLL resummation with NNLO, using three matching schemes (a), (b) and (c) [AB Monni Salam Zanderighi 12] Central value: scheme (a) with µ R = µ F = Q = m H /2 Q is the resummation scale: ln(m H /p t,veto ) ln(q/p t,veto ) Variation of µ R,µ F with Q = m H /2 m H 4 apple µ 1 R,µ F apple m H 2 apple µ R apple 2 µ F gg H, m H = 125 GeV pp, 8 TeV MATCHED NNLL+NNLO MSTW2008 NNLO PDFs anti-k t, R = 0.5 Variation of Q with µ R,µ F = m H /2 (p t,veto ) 0.6 m H 4 apple Q apple m H 0.4 (p t,veto ) / central (p t,veto ) p t,veto [GeV]

36 Theoretical uncertainties We have combined the NNLL resummation with NNLO, using three matching schemes (a), (b) and (c) [AB Monni Salam Zanderighi 12] Central value: scheme (a) with Q is the resummation scale: µ R = µ F = Q = m H /2 ln(m H /p t,veto ) ln(q/p t,veto ) Variation of m H µ R,µ F with Q = m H /2 4 apple µ 1 R,µ F apple m H 2 apple µ R apple 2 µ F gg H, m H = 125 GeV pp, 8 TeV MATCHED NNLL+NNLO MSTW2008 NNLO PDFs anti-k t, R = 0.5 Variation of Q with µ R,µ F = m H /2 (p t,veto ) 0.6 m H 4 apple Q apple m H 0.4 Schemes (b) and (c) with µ R = µ F = Q = m H /2 (p t,veto ) / central (p t,veto ) p t,veto [GeV]

37 Theoretical uncertainties We have combined the NNLL resummation with NNLO, using three matching schemes (a), (b) and (c) [AB Monni Salam Zanderighi 12] Central value: scheme (a) with Q is the resummation scale: µ R = µ F = Q = m H /2 ln(m H /p t,veto ) ln(q/p t,veto ) Variation of m H µ R,µ F with Q = m H /2 4 apple µ 1 R,µ F apple m H 2 apple µ R apple 2 µ F gg H, m H = 125 GeV pp, 8 TeV MATCHED NNLL+NNLO MSTW2008 NNLO PDFs anti-k t, R = 0.5 Variation of Q with µ R,µ F = m H /2 (p t,veto ) 0.6 m H 4 apple Q apple m H 0.4 Schemes (b) and (c) with µ R = µ F = Q = m H /2 Total uncertainty: envelope (p t,veto ) / central (p t,veto ) p t,veto [GeV]

38 Comparison to Monte Carlo We compare the jet-veto efficiency to different Monte Carlo predictions pp, 8 TeV, m H = 125 GeV µ R = µ F = m H /2 MSTW2008 NNLO PDFs anti-k t jets, R=0.5 m t, m b corrections ε(p t, veto ) 0.6 ε(p t, veto )/ε JetVHeto (p t, veto ) JetVHeto POWHEG+Pythia MC@NLO+Herwig HJMINLO, large-m t p t,veto [GeV] All Monte Carlo results are within resummation uncertainty band In the region p t,veto = 25 agreement with MC@NLO 30 GeV NNLL+NNLO results are in better