Observation of H→bb¯ decays and VH production with the ATLAS detector

Contents lists available atScienceDirect

Physics

Letters

B

www.elsevier.com/locate/physletb

Observation

of

H

→

b

b decays

¯

and

V H production

with

the

ATLAS

detector

.

The

ATLAS

Collaboration



a

r

t

i

c

l

e

i

n

f

o

a

b

s

t

r

a

c

t

Article history:

Received25August2018

Accepted6September2018

Availableonline14September2018 Editor:M.Doser

AsearchforthedecayoftheStandardModelHiggsbosonintoabb pair¯ whenproducedinassociation withaW or Z bosonisperformedwiththeATLASdetector.The data,corresponding toanintegrated luminosity of 79.8fb−1 werecollected inproton–proton collisions duringRun 2 of the LargeHadron Collideratacentre-of-massenergyof13TeV.ForaHiggsbosonmassof125GeV,anexcessofevents overtheexpectedbackgroundfromotherStandardModelprocessesisfoundwithanobserved(expected) significance of 4.9 (4.3) standard deviations. A combination with the results from other searches in Run 1andinRun2fortheHiggsbosoninthebb decay¯ modeisperformed,whichyieldsanobserved (expected)significance of5.4(5.5)standard deviations,thusprovidingdirectobservation oftheHiggs bosondecayintob-quarks.The ratioofthemeasuredevent yield foraHiggsbosondecayingintobb¯ totheStandardModelexpectationis1.01±0.12(stat.)+₋₀0._.16₁₅(syst.).Additionally,acombinationofRun2 results searchingforthe Higgsbosonproducedinassociationwithavectorbosonyieldsan observed (expected)significanceof5.3(4.8)standarddeviations.

©2018TheAuthor(s).PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense

(http://creativecommons.org/licenses/by/4.0/).FundedbySCOAP3_.

1. Introduction

The Higgs boson [1–4] was discovered in 2012by the ATLAS and CMS Collaborations [5,6] with a mass of approximately 125 GeV from the analysis of proton–proton (pp) collisions pro-duced by the Large Hadron Collider (LHC) [7]. Since then, the analysisofdatacollectedatcentre-of-massenergiesof7 TeV,8 TeV and13 TeV inRuns1and2oftheLHChasledtotheobservation of manyof the production modes and decay channels predicted by the Standard Model (SM). The bosonic decay channels are well established and have entered an era of precision measure-ments [8–18]. The decay into

τ

-lepton pairs was first observed in the combinationof the ATLAS andCMS analyses [19,20]. The mainHiggsbosonproductionmodes,gluon–gluonfusion(ggF)and vector-boson fusion (VBF), were already measured following the analysisoftheRun 1data,andrecentlythecouplingoftheHiggs bosontotopquarkswasdirectlyobservedbytheATLASandCMS Collaborations [21,22] through the observation of the associated productionofaHiggsbosonandatop-quarkpair(t

¯

t H ).

The dominant decay of the SM Higgs boson is into pairs of

b-quarks, with an expected branching fraction of approximately

58%foramassofmH

=

125 GeV [23].However,largebackgrounds

frommulti-jet productionmakea search inthedominantgluon–

 _{E-mail address:atlas.publications@cern.ch.}

gluonfusionproductionmodeverychallengingathadroncolliders. The most sensitive production modes for detecting H

→

bb de-

¯

cays are theassociated productionofa Higgsboson anda W or Z boson [24] (V H ), where the leptonic decay of the vector bo-son enablesefficient triggering anda significantreduction of the multi-jet background. As well asprobing the dominant decay of the Higgs boson, this measurement allows the overall Higgs bo-son decaywidth [25,26] to be constrainedandprovides the best sensitivityto the Z H and W H productionmodes, whichare (for instance)importantelementsintheinterpretationofHiggsboson measurementsineffectivefieldtheories [27].

SearchesinthischannelattheTevatronbytheCDFandD0 Col-laborationsshowedan excessofeventswitha significanceof2.8 standarddeviationsforaHiggsbosonwithamassof125 GeV [28]. Analysingthe2015and2016dataandcombiningwiththeRun 1 results [29,30], both the ATLAS and CMS Collaborations reported evidence for Higgs boson production and decay in this channel, with observed (expected) significances of 3.6 (4.0) and 3.8 (3.8) standarddeviations,respectively [32,33].Searchesfor H

→

bb de-

¯

cayshavealsobeenconductedintheVBF [34–36] andt

¯

t H [37–42] channels,andwithhightransversemomentumHiggsbosons [43], butwithmarkedlylowersensitivities.

This Letter reports an update to the search forthe SM Higgs boson decaying intoa bb pair

¯

in the V H production mode with theATLASdetectorinRun2oftheLHCpresentedinRef. [32].This update uses 79.8 fb−1 of pp collision data collected at a

centre-https://doi.org/10.1016/j.physletb.2018.09.013

0370-2693/©2018TheAuthor(s).PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).Fundedby SCOAP3_.

of-massenergy of13 TeV, tobe comparedwith 36.1 fb−1 forthe previous result. In addition,an updated version of theATLAS re-construction codeand improvedobject calibrations are used,the impact of the luminosity andmodelling systematic uncertainties arereducedfromupdatedmeasurementsandestimates,andlarger samples of simulated events are used to model the background processes. Events are selected in 0-, 1- and 2-lepton channels, basedonthe numberofchargedleptons,



(electrons ormuons), to explorethe Z H

→

νν

bb,

¯

W H

→ 

ν

bb and

¯

Z H

→ 

bb signa-

¯

tures, respectively. The dominant background processesafter the event selection are V

+

jets, tt,

¯

single-top and diboson process. Multivariate discriminants, builtfrom variables that describe the kinematicsof theselectedevents,are usedto maximisethe sen-sitivity to the Higgs boson signal. Their output distributions are combined using a binned maximum-likelihood fit, referred to as the global likelihood fit, which allows the signal yield and the background normalisations to be extracted. The signal extraction method iscross-checked withthe dijet-mass analysis, wherethe signal yield is extracted using the mass of the dijet system as themainfit observable,andvalidatedusingthedibosonanalysis, wherethenominalmultivariateanalysisismodifiedtoextractthe

V Z , Z

→

bb diboson

¯

process.Theresultofthemultivariate analy-sisisthencombinedwiththatofthepreviouslypublishedanalysis ofRun 1data [30],withothersearchesforbb decays

¯

oftheHiggs bosonand withother searchesin the V H production mode.The lattertwocombinationsleadtotheobservationofboththebb de-

¯

cayoftheHiggsbosonandV H production.Anobservationofthe

bb decay

¯

ofthe Higgs boson by the CMS Collaboration[31] was submittedforpublicationatthesametimeasthisLetter.

2. The ATLAS detector

ATLAS [44] is a general-purpose particle detector covering nearly the entire solid angle1 around the collision point. An in-ner tracking detector, located within a 2 T axial magnetic field generatedbyathinsuperconductingsolenoid,isusedtomeasure thetrajectoriesandmomentaofchargedparticles.Theinnerlayers consistofhigh-granularitysiliconpixeldetectorscoveringa pseu-dorapidityrange

|

η

|

<

2

.

5,andincludeaninnermostlayer [45,46] that was added to the detector between Run 1 and Run 2. Sil-icon microstrip detectors covering

|

η

|

<

2

.

5 are located beyond thepixeldetectors.Outside themicrostripdetectors andcovering

|

η

|

<

2

.

0,therearestraw-tubetrackingdetectors,whichalso pro-videmeasurementsoftransitionradiationthatareusedinelectron identification. A calorimeter systemsurrounds the inner tracking detector,covering the pseudorapidity range

|

η

|

<

4

.

9.Withinthe region

|

η

|

<

3

.

2, electromagnetic calorimetry is provided by bar-rel(

|

η

|

<

1

.

475) andendcap (1

.

375

<

|

η

|

<

3

.

2) high-granularity lead/liquid-argon (LAr) sampling calorimeters,with an additional thinLArpresamplercovering

|

η

|

<

1

.

8 tocorrectforenergylossin materialupstreamofthecalorimeters.Hadroniccalorimetryis pro-videdby a steel/scintillator-tile calorimeterwithin

|

η

|

<

1

.

7, and copper/LArendcap calorimetersextendthe coverage to

|

η

|

=

3

.

2. Thesolidanglecoveragefor

|

η

|

between3.2and4.9iscompleted withcopper/LArandtungsten/LAr calorimetermodules optimised forelectromagneticandhadronicmeasurements,respectively.The

1 _{ATLASusesaright-handed coordinatesystemwith itsoriginat thenominal} interactionpoint(IP)inthecentreofthedetectorandthe

z-axis

coincidingwiththe axisofthebeampipe.The

x-axis

pointsfromtheIPtowardsthecentreoftheLHC ring,andthe y-axis pointsupward.Cylindricalcoordinates(r,φ)areusedinthe transverseplane,φbeingtheazimuthalanglearoundthe

z-axis.

Thepseudorapidity isdefinedintermsofthepolarangleθasη= −ln tan(θ/2).Thedistancein(η,φ) coordinates, R =( φ)2_{+ ( η)}2_{,isalsousedtodefineconesizes.Transverse} momentumandenergyaredefinedas

p

T=p sinθand

E

T=E sinθ,respectively.

outermost part of the detectoris the muon spectrometer, which measures the curvedtrajectories of muons inthe magnetic field of three large air-core superconducting toroidal magnets. High-precision tracking is performed within the range

|

η

|

<

2

.

7 and there arechambers forfasttriggering withinthe range

|

η

|

<

2

.

4. A two-level trigger system [47] is used to reduce the recorded data rate. The first level is a hardware implementation aiming to reduce the rateto around 100 kHz, while the software-based high-leveltriggerprovidestheremainingratereductionto approx-imately1 kHz.

3. Object and event selection

The eventtopologies characteristic of V H , H

→

bb processes

¯

considered contain zero, one or two charged leptons, and two ‘b-jets’ containingparticles fromb-hadron decays.Theobjectand eventselectionsfollowthoseofRef. [32] toalargeextent.

3.1. Objectreconstruction

Tracks measuredin theinner detectorareused toreconstruct interactionvertices [48],ofwhichtheonewiththehighestsumof squaredtransversemomentaofassociatedtracksisselectedasthe primaryvertex.

Electronsare reconstructedfromtopologicalclustersofenergy depositsinthecalorimeter [49] andmatchedtoatrackinthe in-ner detector. FollowingRefs. [32,50], loose electrons are required to have pT

>

7GeV and

|

η

|

<

2

.

47,to havesmall impact

param-eters,2 to fulfila loosetrackisolation requirement, andtomeet a ‘LooseLH’qualitycriterioncomputedfromshowershapeandtrack quality variables [51].In the1-lepton channel,tight electrons are selected using a ‘TightLH’ likelihood requirement and a stricter calorimeter-basedisolation.

Muons arerequired to be within theacceptance ofthe muon spectrometer

|

η

|

<

2

.

7,tohave pT

>

7GeV,andtohavesmall

im-pactparameters. Loose muons are selectedusing a ‘loose’ quality criterion [52] andaloosetrackisolation.Inthe1-lepton channel,

tight muonsfulfilthe‘medium’qualitycriterionandastrictertrack isolation.

Hadronically decaying

τ

-leptons [53,54] are required to have

pT

>

20GeV and

|

η

|

<

2

.

5, to be outside of the transition

re-gion between the barrel and end-cap electromagnetic calorime-ters 1

.

37

<

|

η

|

<

1

.

52, and to meet a ‘medium’ quality crite-rion [54]. They are only used in the analysis to avoid

τ

-leptons beingmisidentifiedasjets.

Jets arereconstructed fromtopological clusters [55] using the anti-kt algorithm [56] with radius parameter R

=

0

.

4. A jet

ver-tex tagger [57] is used to remove jets associated with vertices other than the primary one for jet pT

<

60GeV and

|

η

|

<

2

.

4.

Jet cleaning criteria are used to identify jets arising from non-collisionbackgroundsornoiseinthecalorimeters [58] andevents containing such jets areremoved. Jets are requiredto have pT

>

20GeV in thecentral region(

|

η

|

<

2

.

5),and pT

>

30GeV outside

(2

.

5

<

|

η

|

<

4

.

5) of the trackeracceptance.In the central region, they aretagged ascontaining b-hadrons usingamultivariate dis-criminant [59] (MV2),withtheselection tunedtoproducean av-erage efficiency of 70% for b-jets in simulated t

¯

t events, which corresponds to light-flavour(u-, d-, s-quarkandgluon)and c-jet

misidentificationefficienciesof0.3%and12.5%respectively. Simulatedjetsarelabelledasb-,c- orlight-flavourjets accord-ing to which hadrons with pT

>

5GeV are found within a cone

2 _{Transverseandlongitudinalimpactparametersaredefinedrelativetothe}

pri-mary vertexposition, wherethebeam lineisusedtoapproximatethe primary

ofsize

R

=

0

.

3 around theiraxis.SimulatedV

+

jets eventsare categoriseddependingonthelabelsofthejetsthatformtheHiggs bosoncandidate:V

+

ll whentheyarebothlight-flavourjets,V

+

cl

when there is one c-jet and one light-flavour jet, and V

+

HF (heavyflavour)inallothercases,mainlytwob-jets.Owingtothe largerejectionoflight-flavourjetsachievedbytheMV2 discrimi-nant,simulated V

+

ll, V

+

cl and W W eventsare notsubjected

totheb-taggingrequirementduetothe resultinglownumberof

simulatedevents,butinsteadtheyareweightedbytheprobability thattheirjetspasstheb-taggingselection [32].

In addition to the standard jet energy scale calibration [60],

b-taggedjetsreceive additionalflavour-specificcorrectionsto im-prove their energy measurement (scale and resolution): if any muons are found within

R

=

0

.

4, the four-momentum of the closest muon is added to that of the jet, and a residual correc-tion is applied to equalise the response to jets with leptonic or hadronicdecays of heavy-flavourhadrons. In the 2-lepton chan-nel,a per-event kinematic likelihood uses the full reconstruction oftheeventkinematicstoimprovetheestimate oftheenergyof theb-jets.Thecorrectionsimprovetheresolutionofthedijetmass byupto40% [32].

The missing transverse momentum Emiss_T is reconstructed as thenegative vector sumof themomenta ofleptons, hadronically decaying

τ

-leptons andjets, andofa‘soft term’ builtfrom addi-tionaltracks matchedto the primary vertex [61]. The magnitude of Emiss_T is referred to as Emiss_T .An overlap removalprocedure is applied to avoidany double-counting between the reconstructed leptons,includingthehadronicallydecaying

τ

-leptons,andjets.

3.2.Eventselectionandcategorisation

Eventsarecategorisedintothe0-,1- and2-leptonchannels de-pendingonthenumberofselectedelectronsandmuons,totarget the Z H

→

v vbb,

¯

W H

→ 

ν

bb and

¯

Z H

→ 

bb signatures,

¯

respec-tively. In all channels, events are required to have exactly two

b-taggedjets,whichformtheHiggsbosoncandidate.Atleastone

b-taggedjetisrequiredtohavepT greaterthan45 GeV.Eventsare

furthersplit into 2-jet or 3-jet categoriesdepending onwhether additional,untaggedjetsarepresent.Inthe0- and1-lepton chan-nels,only one such jet is allowed, asthet

¯

t backgroundis much larger in events withfour jets or more. In the 2-lepton channel anynumberofjetsisacceptedinthe3-jetcategory.

Thereconstructedtransversemomentum pV_T ofthevector bo-soncorresponds to Emiss

T inthe0-leptonchannel,tothevectorial

sumofEmiss_T andthecharged-leptontransversemomentuminthe 1-leptonchannel,andtothetransversemomentumofthe2-lepton systeminthe2-leptonchannel. Asthesignal-to-backgroundratio increasesfor large Higgsboson transverse momenta [62,63], the analysisfocusesonahigh-pV_T region definedas pV_T

>

150GeV.In the2-leptonchannel,thesensitivityisincreasedbytheadditionof amedium-pV

T regionwith75GeV

<

pTV

<

150GeV.

Twoversionsoftheanalysisarecarriedout, oneusinga mul-tivariateapproach andthe otherusingthe dijetmassasthefinal discriminant.The event selection shown inTable 1 is applied to both versions, with further selections applied for the dijet-mass analysis.Thetwoversionsoftheanalysisalsohavedifferentevent categorisations,withfurtherdetailsoutlinedbelow.

0-lepton channel The online selection uses E_Tmiss triggers with thresholdsthatvaried from70 GeV to110 GeV betweenthe2015 and 2017 data-taking periods. Their efficiency was measured in

W

+

jets, Z

+

jets andt

¯

t eventsindatausingsingle-muontriggers, resulting in correction factors that are applied to the simulated events,rangingfrom1.05attheoffline Emiss_T thresholdof150 GeV toanegligibledeviationfromunityatan Emiss

T above 200 GeV.A

requirementon thescalarsumof thetransverse momenta HT of

thejetsremovesa smallpartof thephase spacewherethe trig-gerefficiencydependsmildly onthenumberofjetsintheevent. Events withanyloose leptonare rejected. High Emiss

T in multi-jet

eventstypicallyarisesfrommismeasuredjetsinthecalorimeters. Sucheventsareefficientlyremovedbyrequirementsonthe angu-larseparationofthe Emiss

T ,jets,and pmissT (themissingtransverse

momentumcalculatedusingonlytracksreconstructedintheinner trackingdetectorandmatchedtotheprimaryvertex).

1-lepton channel In theelectronsub-channel,eventsarerequired tosatisfyalogicalORofsingle-electrontriggerswithidentification andisolationcriterialooserthanthoseusedintheofflineanalysis, andpT thresholdsthatstartedat24 GeV in2015andincreasedto

26 GeV in 2016and 2017. The muon sub-channel uses thesame

Emiss_T triggers and correction factors as the 0-lepton channel, as these triggers effectively selecton pV

T given that muons are not

includedintheonlineEmiss_T calculationandtheyperformmore ef-ficientlythanthesingle-muontriggersintheanalysisphasespace. Events are requiredtohave exactlyone high-pT tight electronor

muon,andnoadditionalloose leptons.Intheelectronsub-channel an additionalselectionof Emiss_T

>

30GeV isapplied toreduce the backgroundfrommulti-jetproduction.Eventsarecategorisedinto thesignalregion(SR)orintoacontrolregionenrichedinW

+

HF events (W

+

HF CR) using selections on the invariant mass of the two b-taggedjets(mbb), andon the reconstructedmass ofa

semi-leptonically decaying top-quark candidate(mtop). The latter

iscalculatedastheinvariantmassofthelepton,thereconstructed neutrino3 andtheb-taggedjetthat yieldsthe lowestmassvalue. TheresultingpurityoftheW

+

HF controlregionisaround75%.

2-lepton channel The onlineselectionintheelectronsub-channel isthesameasinthe1-leptonchannel. Inthemuonsub-channel, a similar OR of single-muon triggers is used, with lowest pT

thresholdsincreasingwithluminosity andrangingfrom20 GeV to 26 GeV.Events musthaveexactlytwo loose leptons,oneofwhich musthave pT

>

27GeV,andtheinvariantmassoftheleptonpair

mustbe compatiblewiththatofthe Z boson.Eventswith same-flavour leptons enter the signal region, while events with one muonandone electrondefine ane

μ

controlregionwhichisover 99%pureint

¯

t andsingle-top-quarkevents.

Theacceptancesinthethreechannelsaftertheeventselection, as well as thepredicted cross-sections timesbranching fractions for

(

W

/

Z

)

H with W

→ 

ν

, Z

→ 

, Z

→

νν

, and H

→

bb are

¯

giveninTable2.Thenon-negligibleacceptancefortheqq

→

W H

process in the0-lepton channel is mostly dueto events withan unidentified hadronically decaying

τ

-lepton produced in the W

decay,whilethelargeracceptanceforthegg

→

Z H process com-pared with qq

→

Z H is due to the harder pV

T spectrum of the

gluon-inducedprocess.

3.3. Multivariateanalysis

Boosteddecisiontrees(BDT)aretrainedineightsignalregions, corresponding totwo jet categoriesforthethreelepton channels inthehigh-p_TV region,inadditiontothetwojetcategoriesforthe 2-leptonmedium-pV_T region.TheBDToutputsareusedasthefinal discriminatingvariablesintheanalysis.TwosetsofBDTsare con-structedwiththesameinputvariablesandparameters.The nomi-nalone(BDTV H)isdesignedtoseparate Higgsbosoneventsfrom

thesumofexpectedbackgrounds,whilethesecond one (BDTV Z)

3 _{ThetransversecomponentoftheneutrinomomentumisidentifiedwithE}miss T , and thelongitudinalcomponentis obtainedbyconstrainingthelepton–neutrino systemtothe

W mass.

Table 1

Summaryoftheeventselectionandcategorisationinthe0-,1- and2-leptonchannels.

Selection 0-lepton 1-lepton 2-lepton

e sub-channel μsub-channel

Trigger Emiss

T Single lepton EmissT Single lepton

Leptons 0 loose leptons 1 tight electron 1 tight muon 2 loose leptons with pT>7 GeV

with pT>7 GeV pT>27 GeV pT>25 GeV ≥1 lepton with pT>27 GeV

Emiss

T >150 GeV >30 GeV – –

m – – 81 GeV<m<101 GeV

Jets Exactly 2 / Exactly 3 jets Exactly 2 /≥3 jets

Jet pT >20 GeV for|η| <2.5

>30 GeV for 2.5<|η| <4.5

b-jets Exactly 2 b-tagged jets

Leading b-tagged jet pT >45 GeV

HT 120 GeV (2 jets),>150 GeV (3 jets) – –

min[ φ(Emiss

T ,jets)] >20◦(2 jets),>30◦(3 jets) – – φ(Emiss T ,bb) >120◦ – – φ(b1,b2) <140◦ – – φ(Emiss T ,pmissT ) <90◦ – – pV

T regions >150 GeV 75 GeV<pTV <150 GeV,>150 GeV

Signal regions – mbb≥75 GeV or mtop≤225 GeV Same-flavour leptons

Opposite-sign charges (μμsub-channel)

Control regions – mbb<75 GeV and mtop>225 GeV Different-flavour leptons

Opposite-sign charges

Table 2

Thecross-section(σ)timesbranchingfraction(B)andacceptance forthethree channelsat√s=13 TeV.The

qq- and gg-initiated Z H processes

areshown sep-arately.Thebranchingfractionsarecalculatedconsideringonlydecaysintomuons andelectronsfor Z→ ,decaysinto allthreeleptonflavoursfor W→ ν and decaysintoallneutrinoflavoursfor

Z

→νν.Theacceptanceiscalculatedasthe fractionofeventsremaininginthecombinedsignalandcontrolregionsafterthe fulleventselection.

Process σ× B[fb] Acceptance [%]

0-lepton 1-lepton 2-lepton

qq→Z H→ bb¯ 29.9 <0.1 0.1 6.0

gg→Z H→ bb¯ 4.8 <0.1 0.2 13.5

qq→W H→ νbb¯ 269.0 0.2 1.0 –

qq→Z H→ννbb¯ 89.1 1.9 – –

gg→Z H→ννbb¯ 14.3 3.5 – –

is used to validate the analysis by the extraction of the diboson

V Z , Z

→

bb process

¯

fromthesumofallotherSMprocesses. Thesameinputvariables,BDTsettingsandBDToutputbinning transformationasthosedetailedinRef. [32] areused,withone ex-ceptioninthe 2-leptonchannel wherethe E_Tmiss isreplaced with

Emiss

T

/

√

ST (where ST isthescalarsumoftransverse momentaof

thechargedleptons andjetsintheevent).Eighttothirteeninput variablesdescribingthekinematicsoftheeventsareused depend-ingonthechannels, ofwhichmbb, pVT and

R

(

b1

,

b2

)

(where b1

andb2refertothetwob-taggedjets)arethemostdiscriminating.

3.4. Dijet-massanalysis

Across-checkofthemainmultivariateanalysisisperformedby usingtheinvariant massofthetwob-taggedjetsasthe discrimi-natingvariable.AdditionalselectionsdisplayedinTable3increase thepurityofthesignal regionsandarenecessaryto improvethe sensitivityofthismethod.

The high-pV

T region is split into two regions 150GeV

<

pTV

<

200GeV andpV_T

>

200GeV,withfurtherrequirementsplacedupon

R

(

b1

,

b2

)

. Selections on the transverse mass of the W boson

(mW

T ) andon EmissT

/

√

ST reduce thet

¯

t background inthe 1- and

2-leptonchannels,respectively.

Table 3

Summaryoftheevent selectioncriteriainthe0-,1- and2-lepton channelsfor thedijet-massanalysis,appliedinadditiontothosedescribedinTable1for the multivariateanalysis.

Channel

Selection 0-lepton 1-lepton 2-lepton

mW T – <120 GeV – Emiss T / √ ST – – <3.5 √ GeV pV T regions pV

T 75–150 GeV 150–200 GeV >200 GeV

(2-lepton only)

R(b1,b2) <3.0 <1.8 <1.2

In the 1-lepton channel the mbb distribution is able to

suffi-cientlyconstrainthe W

+

HF background,thusitisnotnecessary toseparateeventsintoadedicatedW

+

HF CR.

4. Data, simulated samples and multi-jet background

The data used in this analysis were collected at a centre-of-mass energy of 13 TeV during the 2015–2017 running periods. Eventsareselectedforanalysisonlyiftheyareofgoodqualityand ifalltherelevantdetectorcomponentsareknowntohavebeenin goodoperatingcondition, whichcorrespondsto atotalintegrated luminosityof79

.

8

±

1

.

6 fb−1 [64,65].Therecordedeventscontain anaverageof32inelastic pp collisions.

MonteCarlo(MC)simulatedeventsareusedtomodelthe back-groundsfromSMprocessesandV H , H

→

bb signal

¯

processes.All simulatedprocessesarenormalisedusingthemostaccurate theo-retical cross-sectionpredictions currentlyavailable andwere gen-eratedatleasttonext-to-leading-order(NLO)accuracy.Allsamples ofsimulatedeventswerepassed throughtheATLAS detector sim-ulation [66] basedon GEANT 4 [67] and werereconstructedwith the standardATLAS reconstruction software.Theeffects of multi-pleinteractions inthesameandnearbybunch crossings(pile-up) were modelledbyoverlaying minimum-biasevents,simulated us-ingthesoftQCDprocessesof Pythia8.186[68] withtheA2 [69] set oftunedparameters(tune)and MSTW2008LO [70] parton distribu-tionfunctions(PDF).Forallsamplesofsimulatedevents,exceptfor

Table 4

Thegeneratorsusedforthesimulationofthesignalandbackgroundprocesses.Ifnotspecified,theorderofthecross-sectioncalculationreferstotheexpansioninthestrong couplingconstant(αS).TheacronymsME,PSandUEstandformatrixelement,partonshowerandunderlyingevent,respectively.()Theeventsweregeneratedusingthe firstPDFintheNNPDF3.0NLOsetandsubsequentlyreweightedtothePDF4LHC15NLOset [73] usingtheinternalalgorithmin Powheg-Boxv2.(†)TheNNLO(QCD)+NLO(EW) cross-sectioncalculationforthe

pp

→Z H process alreadyincludesthe

gg

→Z H contribution. The

qq

→Z H process isnormalisedusingthecross-sectionforthe

pp

→Z H process,aftersubtractingthe

gg

→Z H contribution. Anadditionalscalefactorisappliedtothe

qq

→V H processes asafunctionofthetransversemomentumofthevector boson,toaccountforelectroweak(EW)correctionsatNLO.Thismakesuseofthe

V H differential

cross-sectioncomputedwith Hawk [74,75].

Process ME generator ME PDF PSand

hadronisation

UEmodel

tune

Cross-section order Signal, mass set to 125 GeV and bb branching fraction to 58%¯

qq →W H → νbb¯

Powheg-Boxv2[76]+ GoSam[79]+MiNLO[80,81]

NNPDF3.0NLO()_{[77] Pythia8.212 [68] AZNLO [78] NNLO(QCD)+}

NLO(EW) [82–88] qq →Z H

→ννbb¯/bb¯

Powheg-Boxv2+

GoSam+MiNLO

NNPDF3.0NLO() _{Pythia8.212 AZNLO NNLO(QCD)}(†)₊

NLO(EW) gg →Z H

→ννbb¯/bb¯

Powheg-Box v2 NNPDF3.0NLO() _{Pythia8.212 AZNLO NLO+}

NLL [89–93] Top quark, mass set to 172.5 GeV

tt¯ Powheg-Box v2[94] NNPDF3.0NLO Pythia8.230 A14 [95] NNLO+NNLL [96]

s-channel Powheg-Box v2[97] NNPDF3.0NLO Pythia8.230 A14 NLO [98]

t-channel Powheg-Box v2[97] NNPDF3.0NLO Pythia8.230 A14 NLO [99]

W t Powheg-Box v2[100] NNPDF3.0NLO Pythia8.230 A14 Approximate NNLO [101]

Vector boson+jets

W→ ν Sherpa 2.2.1[71,102,103] NNPDF3.0NNLO Sherpa 2.2.1[104,105] Default NNLO [106]

Z/γ∗→  Sherpa 2.2.1 NNPDF3.0NNLO Sherpa 2.2.1 Default NNLO

Z→νν Sherpa 2.2.1 NNPDF3.0NNLO Sherpa 2.2.1 Default NNLO

Diboson

qq→W W Sherpa 2.2.1 NNPDF3.0NNLO Sherpa 2.2.1 Default NLO

qq→W Z Sherpa 2.2.1 NNPDF3.0NNLO Sherpa 2.2.1 Default NLO

qq→Z Z Sherpa 2.2.1 NNPDF3.0NNLO Sherpa 2.2.1 Default NLO

gg→V V Sherpa 2.2.2 NNPDF3.0NNLO Sherpa 2.2.2 Default NLO

thosegeneratedusing Sherpa [71],the EvtGenv1.2.0program [72] was used to describe the decays of bottom and charm hadrons. A summary of all the generators used for the simulation of the signalandbackgroundprocessesisshowninTable4.Samples pro-ducedwithalternativegeneratorsareusedtoestimate systematic uncertaintiesintheeventmodelling,asdescribedinSection5.

ThebackgroundprocessesinvolvingW or Z bosondecaysinto leptons(includingthoseinwhichtheW bosonarisesfroma top-quark decay)are collectivelyreferred to inthe followingas elec-troweak(EW)backgroundsandweresimulatedasdescribedabove. In contrast, the multi-jet background is estimated in all three channelsusingdata-driven methods.Inboth the 0- and2-lepton channels, the multi-jet contribution is estimated from template fitsto data,using thesimulated samplesto modelthe EW back-groundsandafunctionalformtomodelthemulti-jetbackground. The template fitis performedusing a variable that provides sig-nificant discrimination between the multi-jet and EW processes, withanyselectiononthatvariableremoved.Inthe0-lepton chan-nel,min

[ φ(

Emiss

T

,

jets

)

]

isused,andinthe2-leptonchannel,the

dileptonmassdistributionisusedforthe casewherethecharges ofthelepton candidateshavethesamesign, assumingthe multi-jet contributionis symmetricfor opposite- andsame-sign lepton charges.Inbothcases,itisfoundthatthemulti-jetcontributionis sufficientlysmallthatitcan beneglected inthegloballikelihood fitwithouthavinganyimpactontheextractedsignal.

Themulti-jet backgroundis foundto be non-negligible inthe 1-leptonchannel and isestimatedseparately inthe electron and muon sub-channels, and in the 2- and 3-jet categories. In each category,a templatefittothetransverse massdistributionofthe

W boson candidate is performed, which offers the clearest

dis-crimination between the multi-jet and EW processes, to extract themulti-jet yield.The template usedforthe multi-jet contribu-tion is obtained from data in a control region after subtraction ofthe residual EW contribution,based on MC predictions, while thetemplate for theEW contribution in the signal region is ob-taineddirectlyfromMCpredictions.Thecontrolregionisenriched

inmulti-jeteventsthatarekinematicallyclosetothe correspond-ing signal region butnot overlapping withit, andis defined by applying the nominal selection but inverting the stricter lepton isolation requirements.To increase thestatisticalprecision of the data-driven estimate,the numberofrequired b-taggedjetsis re-duced fromtwo to one in the multi-jet enriched control region. The templatefit appliedinthe signal regiondetermines the nor-malisation of the multi-jet contribution, while the shape of the BDTdiscriminant(orofotherrelevantobservables)isobtained us-ing a control region analogously to the m_TW template. Both the normalisationandshapederivedfortheBDTdiscriminantarethen usedinthegloballikelihoodfit.Themulti-jetcontributioninthe 2-jet categoryis foundto be 1

.

9% (2

.

8%) ofthetotal background contributionintheelectron(muon)sub-channel,whileinthe3-jet categoryitisfoundtobe0

.

2% (0

.

4%).Theseestimatesaresubject to sizeable systematic uncertainties, which are described in Sec-tion5.

5. Systematic uncertainties

The sources of systematicuncertainty can be broadlydivided into four groups: those of experimental nature, those related to themodelling ofthesimulatedbackgrounds,those relatedtothe multi-jet background estimation, and those associated with the Higgsbosonsignal simulation.Theestimationoftheuncertainties closelyfollowsthemethodologyoutlinedinRef. [32] andisbriefly summarisedbelow.

5.1. Experimentaluncertainties

The dominant experimental uncertainties originate from the

b-tagging correction factors, determined from the difference be-tweentheefficiencymeasuredindataandsimulation,fromthejet energyscalecorrectionsandfromthemodellingofthejetenergy resolution.Theb-taggingcorrection factorsare derivedseparately

forb-jets, c-jetsandlight-flavourjets [107–109].Allthree correc-tionfactors haveuncertainties estimated frommultiple measure-ments,whichare decomposedintouncorrelatedcomponentsthat arethentreatedindependently,resultinginthreeuncertaintiesfor

b-jets and c-jets, and five forlight-flavour jets. The approximate size of the uncertainty in the tagging efficiency is 2% for b-jets,

10%forc-jetsand40%forlight-flavourjets.Additional uncertain-tiesare consideredintheextrapolationoftheb-jetefficiency cal-ibration to jets with pT

>

300 GeV and in the misidentification

ofhadronically decaying

τ

-leptons asb-jets. The uncertainties in the jet energyscale andresolution are basedon their respective measurements [60,110]. The many sources of uncertainty in the correction ofthejet energyscaleare decomposedinto23 uncor-relatedcomponentsthataretreatedasindependent.Anadditional specific uncertainty in the energy calibration of b- and c-jets is considered.

Uncertaintiesinthereconstruction,identification, isolationand trigger efficiencies of muons [52] and electrons [50], along with theuncertaintyintheirenergyscaleandresolution,areestimated using13 TeV data. Theseare found to have only a smallimpact ontheresult.Theuncertainties intheenergyscaleandresolution ofthejetsandleptons arepropagatedto thecalculationof E_Tmiss, whichalsohas additionaluncertainties fromthescale,resolution and reconstruction efficiency of the tracks used to compute the softterm [61],along withthemodellingoftheunderlyingevent. An uncertaintyisassigned tothe Emiss

T trigger correction factors,

determined fromthe difference between thetrigger efficiency in data andsimulation, to account for the statistical uncertaintyin the measured correction factors andfor differences betweenthe correctionfactorsdeterminedfromW

+

jets, Z

+

jets andt

¯

t events.

Theuncertaintyinthecombined2015–2017integratedluminosity is2.0%.It isderived,following amethodologysimilar tothat de-tailedinRef. [64],andusingtheLUCID-2detectorforthebaseline luminosity measurements [65]. The average number of interac-tionsperbunchcrossingisrescaledby1.03toimproveagreement betweensimulation and data, basedon the measurement ofthe visiblecross-sectioninminimum-biasevents [111],andan uncer-tainty,aslargeasthecorrection,isincluded.

5.2. Simulatedsampleuncertainties

Modellinguncertainties are derived forthesimulatedsamples andbroadly cover three areas: normalisations, acceptance differ-encesthat affectthe relativenormalisations betweenanalysis re-gionswithacommonnormalisation,andtheshapesofthe differ-entialdistributionsofthemostimportantkinematicvariables.The overall normalisations and associated uncertainties for the back-groundprocessesaretakenfromthecurrentlymostaccurate cal-culationsasdetailedinTable4,apartfromthemainbackgrounds whosenormalisationsareleftunconstrained(floated)intheglobal likelihood fit. The additional systematic uncertainties in the ac-ceptance differences and in the shapes are derived either from particle-levelcomparisonsbetweennominalandalternative simu-latedsamples,orfrom comparisonswithdata incontrol regions. The particle-level comparisons are cross-checked with detector-level simulations whenever these are available, and good agree-ment is found. The alternative samples were either produced by othergeneratorsorbyalteringthenominalvaluesofgenerator pa-rameters.Whenacceptanceuncertaintiesareestimated,the nomi-nalandalternativesamplesarenormalisedusingthesame produc-tion cross-section. Shape uncertainties are considered in each of theanalysisregionsseparately,withthesamplesscaledtohavethe samenormalisationineachregion.Inthiscase,theuncertaintyis takenfromthealternativesample thatdiffersmostinshapefrom thenominalsample. Shape uncertaintiesare onlyderived forthe

Table 5

Summaryofthesystematicuncertaintiesinthebackgroundmodellingfor

Z

+jets, W+jets,

tt,

singletop-quarkandmulti-jetproduction.An‘S’symbolisusedwhen onlyashapeuncertaintyisassessed.Theregionsforwhichthenormalisationsfloat independentlyarelistedinbrackets.Wherethesizeofanacceptancesystematic uncertaintyvariesbetweenregions,arangeisdisplayed.

Z+jets

Z+ll normalisation 18% Z+cl normalisation 23%

Z+HF normalisation Floating(2-jet,3-jet)

Z+bc-to-Z+bb ratio 30–40% Z+cc-to-Z+bb ratio 13–15% Z+bl-to-Z+bb ratio 20–25%

0-to-2leptonratio 7%

mbb,

p

VT S

W+jets

W+ll normalisation 32% W+cl normalisation 37%

W+HF normalisation Floating(2-jet,3-jet)

W+bl-to-W+bb ratio 26%(0-lepton)and23%(1-lepton) W+bc-to-W+bb ratio 15%(0-lepton)and30%(1-lepton) W+cc-to-W+bb ratio 10%(0-lepton)and30%(1-lepton)

0-to-1leptonratio 5%

W+HF CRtoSRratio 10%(1-lepton)

mbb,

p

VT S

tt (all are uncorrelated between the 0+1- and 2-lepton channels) tt normalisation Floating(0+1-lepton)

Floating(2-lepton2-jet,2-lepton3-jet)

0-to-1leptonratio 8%

2-to-3-jetratio 9%(0+1-leptononly)

W+HF CRtoSRratio 25%

mbb,

p

VT S

Single top-quark

Cross-section 4.6%(s-channel),4.4%(t-channel),6.2%(W t)

Acceptance2-jet 17%(t-channel),55%(W t(bb)),24%(W t(other))

Acceptance3-jet 20%(t-channel),51%(W t(bb)),21%(W t(other))

mbb,

p

VT S(t-channel,

W t

(bb),

W t

(other))

Multi-jet (1-lepton)

Normalisation 60–100%(2-jet),90–140%(3-jet)

BDTtemplate S

mbb and pTV variables, asit was foundsufficient toonly consider

the changesinducedin thesevariables tocovertheoverall shape variation of the BDTV H discriminant. Full details are provided in

Ref. [32].

5.2.1. Backgrounduncertainties

The systematic uncertainties affecting the modelling of the background samples are summarised in Tables 5 and 6 and key detailsofthetreatmentofthebackgroundsarereportedbelow.

V

+

jets production The V

+

jets backgroundsare subdivided intothree differentcomponentsbaseduponthe jetflavour labels ofthetwob-taggedjetsintheevent.Themainbackground contri-butions(V

+

bb,V

+

bc,V

+

bl andV

+

cc)arejointlyconsideredas the V

+

HF background.Theiroverall normalisation,separately in the2- and3-jetcategories,isfreetofloatinthegloballikelihood fit. Theremaining flavour components, V

+

cl and V

+

ll, consti-tute less than

∼

1% of the background in each analysis region, so only uncertainties in the normalisation of these backgrounds are included.Acceptance uncertaintiesare estimatedforthe rela-tive normalisations ofthe differentregions that sharea common floatingnormalisationparameter.Inthecaseofthe W

+

HF back-ground, this includes the uncertainties in the ratio of the event yieldinthe0-leptonchanneltothat inthe1-leptonchanneland, in the 1-lepton channel, in the ratio of the event yield in the

W

+

HF controlregiontothatinthesignalregion.Forthe Z

+

HF background,thereisanuncertaintyintheratiooftheeventyield inthe0-leptonchanneltothatinthe2-leptonchannel.

Uncertain-Table 6

Summaryofthesystematicuncertaintiesinthebackgroundmodellingfordibosonproduction.An‘S’symbolis usedwhenonlyashapeuncertaintyisassessedand‘PS/UE’ indicatespartonshower/underlyingevent.When extractingthe(W/Z)Z diboson productionsignalyield,asthenormalisationsareunconstrained,the normalisa-tionuncertaintiesareremoved.Wherethesizeofanacceptancesystematicuncertaintyvariesbetweenregions, arangeisdisplayed.

Z Z

Normalisation 20%

0-to-2leptonratio 6%

Acceptancefromscalevariations 10–18%

AcceptancefromPS/UEvariationsfor2ormorejets 6%

AcceptancefromPS/UEvariationsfor3jets 7%(0-lepton),3%(2-lepton)

mbb,

p

VT,fromscalevariations S(correlatedwith

W Z uncertainties)

mbb,

p

VT,fromPS/UEvariations S(correlatedwith

W Z uncertainties)

mbb,frommatrix-elementvariations S(correlatedwith

W Z uncertainties)

W Z

Normalisation 26%

0-to-1leptonratio 11%

Acceptancefromscalevariations 13–21%

AcceptancefromPS/UEvariationsfor2ormorejets 4%

AcceptancefromPS/UEvariationsfor3jets 11%

mbb,

p

VT,fromscalevariations S(correlatedwith

Z Z uncertainties)

mbb,

p

VT,fromPS/UEvariations S(correlatedwith

Z Z uncertainties)

mbb,frommatrix-elementvariations S(correlatedwith

Z Z uncertainties)

W W

Normalisation 25%

ties are also estimated in the relative normalisation of the four heavy-flavourcomponentsthatconstitutetheV

+

HF background. Thesearetaken asuncertainties in thebc,cc andbl yields com-paredwiththedominantbb yieldandareestimatedseparatelyin eachchannelinamannersimilartotheacceptancesystematic un-certainties.Uncertaintiesarealsoderivedfortheshapesofthembb

andp_TV distributions,whichareevaluated forW

+

HF from com-parisonswithalternativesamplesandforZ

+

HF fromcomparisons withdatainmbb sidebands.

tt production Duetothesignificantlydifferentregionsofphase space probed, the tt background in the 0- and 1-lepton chan-nels(jointlyreferredtoas0+1-leptonchannelinthefollowing)is consideredindependentlyfromthett backgroundinthe2-lepton channel; different overall floating normalisation factors are con-sidered, and acceptance uncertainties are derived separately and takenasuncorrelatedbetweenthe0+1- and2-leptonchannels.For the0+1- leptonchannels,uncertainties areconsideredinthe nor-malisationratiosofthe2-jet and3-jetcategories, ofthe W

+

HF controlandsignalregions,andofthe1-leptonand0-lepton chan-nels. For the 2-lepton channel, the normalisations in the 2- and 3-jet categories are both left floating, and are effectively deter-minedintheir respectivee

μ

controlregions.Uncertaintiesinthe shapesofthe pV

T andmbb distributionsareestimatedinthe

0+1-and2-leptonchannelsseparately fromcomparisonswith alterna-tive samples. In addition, the modelling of the tt background is validatedin the 2-lepton channel by using the data events from thee

μ

controlregion tomodel thisbackgroundinthesignal re-gion,withgoodagreementfound.

Single top-quark production In theW t- andt-channels, uncer-taintiesare derived forthe normalisation,acceptance andshapes of the mbb and pTV distributions. For the W t-channel, the

esti-matedmodellinguncertaintiesarebasedontheflavourofthetwo

b-tagged jets, due to the different regions of phase space being probed when there are two b-jets (bb) present compared with eventswhere there are fewer b-jets present(other). Only a nor-malisationuncertaintyisderived forthe s-channel,since its con-tributionisnegligibleoverall.

Diboson production The dibosonbackgroundsarecomposedof threedistinctprocesses: W Z ,W W andZ Z production.Giventhe

smallcontributionfromW W production(

<

0

.

1% ofthetotal back-ground) only a normalisation uncertainty is assigned. The more importantcontributions fromthe W Z and Z Z backgrounds have uncertaintiesderivedfortheoverallnormalisation,therelative ac-ceptancebetweenregions andforthe mbb and pVT shapes. These

arederivedfollowingtheproceduredescribedinRef. [32] andare outlinedinTable6.

5.2.2. Signaluncertainties

The systematic uncertainties that affect the modelling of the signal aresummarised inTable 7.Theyarederived following the procedureoutlinedinRef. [32],butwithupdatedalternative sam-ples generated witha larger number of events,and using a pa-rameter tune optimized more recently for the evaluation of the parton shower uncertainty. Thissubstantially reduces the parton shower and underlying event (PS/UE) uncertainties. The system-aticuncertainties inthecalculationsofthe V H production cross-sectionsandthe H

→

bb branching

¯

fraction4 _{are assigned}

follow-ingtherecommendationsoftheLHCHiggsCrossSectionWorking Group [26,92,93,112,113].

5.3. Multi-jetbackgrounduncertainties

Systematicuncertaintiescanhaveanimpactonthedata-driven multi-jetestimateusedinthe1-leptonchannelintwoways:either changingthemW_T distributionsusedinthemulti-jettemplatefits, thus impacting the extractedmulti-jet normalisations,or directly changing the multi-jet BDT distributions used in the global like-lihood fit. Several uncertainties are considered, uncorrelated be-tween theelectronandmuon sub-channels.Therespective varia-tionsareaddedinquadratureforthenormalisations,orconsidered asseparate shapeuncertainties.Variationsareobtainedby chang-ing the definition ofthe multi-jet control region (more stringent isolation requirements,adifferentsingle-electrontriggerto probe a potential trigger bias in the isolation requirements), and vary-ing the normalisation ofthe contamination fromthe top (t

¯

t and

4 _{Suchsystematicuncertaintiesfullydegeneratewiththesignalyielddonot} af-fectthecalculationofthesignificancerelativetothebackground-onlyprediction.

Table 7

Summaryofthesystematicuncertaintiesinthesignalmodelling.An‘S’symbolisusedwhenonlyashape un-certaintyisassessedand‘PS/UE’indicatespartonshower /underlyingevent.Wherethesizeofanacceptance systematicuncertaintyvariesbetweenregions,arangeisdisplayed.

Signal

Cross-section (scale) 0.7% (qq), 27% (gg)

Cross-section (PDF) 1.9% (qq→W H), 1.6% (qq→Z H), 5% (gg)

H→bb branching fraction¯ 1.7%

Acceptance from scale variations 2.5–8.8%

Acceptance from PS/UE variations for 2 or more jets 2.9–6.2% (depending on lepton channel)

Acceptance from PS/UE variations for 3 jets 1.8–11%

Acceptance from PDF+αSvariations 0.5–1.3%

mbb, pVT, from scale variations S

mbb, pVT, from PS/UE variations S

mbb, pVT, from PDF+αSvariations S

pV

T from NLO EW correction S

W t)and V

+

jets processesinthemulti-jetcontrol region.In ad-dition,thefollowingsystematicuncertaintieshaveanimpactonly on themulti-jet normalisation: use ofanother discriminant vari-ableinsteadofmW

T forthetemplate fit(the azimuthalseparation

between the directions of the lepton transverse momentum and thevectorialsumofthemomentaofthetwoorthreejets)and,for theelectronsub-channelonly,theinclusionofthe Emiss_T

<

30 GeV region,which significantly enhancesthemulti-jet contributionin thetemplatefit.

6. Statistical analysis

The statistical procedure is based on a likelihood function

L(

μ

,

θ )

, constructed asthe product of Poisson probability terms over the bins of the input distributions. The parameter of inter-est,

μ

,is thesignal strength that multipliesthe SM Higgsboson productioncross-sectiontimesthebranchingfractionintobb and

¯

isextractedbymaximisingthelikelihood.Systematicuncertainties enter thelikelihood as nuisanceparameters (NP),

θ

.Most ofthe uncertaintiesdiscussedinSection5areconstrainedwithGaussian orlog-normalprobabilitydensityfunctions.Thenormalisationsof the largestbackgrounds, tt,

¯

W

+

HF and Z

+

HF,can be reliably determined by thefit,so they are left unconstrained inthe like-lihood.The uncertainties dueto thelimited numberofeventsin thesimulatedsamplesusedforthebackgroundpredictionsare in-cluded using the Beeston–Barlowtechnique [114]. As detailed in Ref. [30],systematicvariations thatare subjecttolarge statistical fluctuationsaresmoothed,andsystematicuncertainties thathave anegligibleimpactonthefinalresultsareprunedaway region-by-region.

Theprobabilitythatthebackground-onlyhypothesisis compat-iblewiththeobserveddataisdeterminedusingtheq0teststatistic

constructed fromthe profile-likelihood ratiowith the asymptotic approximation [115].

6.1. Multivariateanalysis

Asdiscussed inSection3.3,thegloballikelihoodfitcomprises eightsignal regions, definedasthe 2- and3-jetcategoriesinthe high-pV_T region for the three channels, and in the medium-pV_T regionforthe2-leptonchannel.TheBDTV H multivariate

discrimi-nantoutputdistributionsintheseregionsareinputtothefit.The eventyields are used inthe two W

+

HF control regions ofthe 1-lepton channel. In the four e

μ

control regions ofthe 2-lepton channel, thembb distributions are inputto the fit,exceptforthe

2-jet category of the high-pV

T region, where the event yield is

used.Thepost-fitnormalisationfactorsoftheunconstrained back-groundsinthe globallikelihoodfitto the13 TeV data are shown inTable8.

Table 8

Factorsappliedtothenominalnormalisationsofthe

tt,

W+HF and

Z

+HF backgrounds,asobtainedfromthe globallikelihoodfit tothe 13 TeV datafor the nomi-nalmultivariateanalysis,usedtoextracttheHiggsboson signal.Theerrorsrepresentthecombinedstatisticaland systematicuncertainties.

Process Normalisation factor

tt 0- and 1-lepton 0.98±0.08 tt 2-lepton 2-jet 1.06±0.09 tt 2-lepton 3-jet 0.95±0.06 W+HF 2-jet 1.19±0.12 W+HF 3-jet 1.05±0.12 Z+HF 2-jet 1.37±0.11 Z+HF 3-jet 1.09±0.09

The effectsofsystematicuncertaintiesonthe measurementof the signalstrength aredisplayed inTable 9.The impactofa cat-egory of systematic uncertainties is defined as the difference in quadraturebetweentheuncertaintyin

μ

computedwhenallNPs arefittedandthatwhentheNPsinthecategoryarefixedtotheir best-fitvalues.Thetotalstatisticaluncertaintyisdefinedasthe un-certainty in

μ

whenall theNPsarefixed totheir best-fitvalues. The totalsystematicuncertaintyis thendefinedasthedifference in quadrature between the total uncertainty in

μ

and the total statisticaluncertainty.Asshowninthetable,thesystematic uncer-tainties dueto themodelling ofthe signal play adominant role, followed by theuncertainty dueto thelimited size ofthe simu-latedsamples,themodellingofthebackgroundsandtheb-tagging

uncertainty.

6.2. Dijet-massanalysis

In thedijet-mass analysis, the numberof signal regions is in-creasedtofourteenasaconsequenceofsplittingtheeventregions withp_TV

>

150 GeV intwo,whiletheW

+

HF CRsaremergedinto the corresponding SR, asoutlined in Section 3.4. The mbb

distri-butions are input to thefit in all categories, except forthe 2-jet medium- and high-p_TV categories of the 2-lepton e

μ

control re-gion,wheretheeventyieldisused.

6.3. Dibosonanalysis

In thediboson analysis,a measurement ofthe signal strength

of the Z Z and W Z processes is conductedto validate the main

multivariateanalysis.Themethoddiffersfromthegloballikelihood fit only by the use of the BDTV Z output distributions as inputs,

instead of BDTV H. The parameter of interest,

μV Z

, is the signal

strengthofthecombinedW Z and Z Z dibosonprocesses,andthe SMHiggsbosonisincludedasabackgroundprocessnormalisedto

Table 9

Breakdownofthecontributionstotheuncertaintyinμ. The suminquadratureofthe systematicuncertainties attachedtothecategoriesdiffersfromthetotal system-aticuncertaintyduetocorrelations.

Source of uncertainty σμ Total 0.259 Statistical 0.161 Systematic 0.203 Experimental uncertainties Jets 0.035 Emiss T 0.014 Leptons 0.009 b-tagging b-jets 0.061 c-jets 0.042 light-flavour jets 0.009 extrapolation 0.008 Pile-up 0.007 Luminosity 0.023

Theoretical and modelling uncertainties

Signal 0.094

Floating normalisations 0.035

Z+jets 0.055

W+jets 0.060

tt 0.050

Single top quark 0.028

Diboson 0.054

Multi-jet 0.005

MC statistical 0.070

thepredictedSMcross-section withan uncertaintyof50%,which conservativelyencompassesthepreviousmeasurementand uncer-tainty [32].

6.4.Combinations 6.4.1. Run1

The results of the statistical analysis of the 13 TeV data are combined with those from the data recorded at 7 TeV and 8 TeV [30] toimprovetheprecisionofthemeasurement.Detailed studies of the impact of the correlation of systematic uncertain-tiesbetweenthe two analyses are reported in Ref. [32]. In most cases,the impactofcorrelationswas foundto benegligible.Only a b-jet-specific jet energy scale, and theory uncertainties in the Higgs boson signal (overall cross-section, branching fraction and

pV

T-dependent NLOEW corrections)are correlatedacrossthe

dif-ferentcentre-of-massenergies.

6.4.2. H

→

bb

¯

A second combination is performed with the results of the searches for the H

→

bb decay

¯

in the tt H [

¯

37,39] and VBF [34,

36] productionmodescarriedoutwiththeRun1andRun2data. As the analysis targeting the VBF production mode has a size-ablecontributionfromgluon–gluon fusionevents,itisreferredto asthe VBF+ggFanalysis inthe following. Constrainingthe cross-sectionsof the production modesto be as predictedby the SM, the combinationmeasures the ratio of the branching fraction of theHiggsbosonintob-quarks totheSM prediction.Theonly NP correlated across thesix analyses is the H

→

bb branching

¯

frac-tionthataffectstheSMprediction.AfewotherNPsarecorrelated across some ofthe analyses, following the studies conductedfor thecombinationsofRun1results [19],ofanalysesofthett H pro-

¯

ductionmode [21],andofRun 2results.

6.4.3. V H

A third combination is also performed combining the Run 2

V H ,H

→

bb result

¯

withotherresultsintheV H productionmode, butforthecaseoftheHiggsbosondecayingintotwophotonsor via Z Z∗intofourleptons.

The measurement of V H production in the H

→

γ γ

channel, which uses five reconstruction-level categories to target leptonic decaysofthevectorboson,andtwo categoriestargetinghadronic decaysofthevectorboson,asdescribedinRef. [9],isupdated us-ing79.8 fb−1 ofdata.Photonsarereconstructedfromcalorimeter energyclustersformed usingan enhanced dynamical, topological cell-clustering-basedalgorithm [49].Thesignalyieldisextractedin each categoryusingafitto thediphotoninvariant mass distribu-tionintherange105–160 GeV.Contaminationinthesecategories from non-V H Higgs boson production is constrained using sep-arate categoriesdesigned to measure thett H [

¯

21], VBF, and ggF productionmodes.

The measurement of V H production in the four-lepton final state, H

→

Z Z∗

→

4



, where



=

e or

μ

, was performed with 36.1 fb−1 [10] andhasnowbeenextendedto79.8 fb−1.Themain enhancements are: improved electron reconstruction [49] andan additional event category targeting vector-boson decays that in-clude missingtransverse momentum dueto the presence ofone ortwoneutrinosinthefinalstate.Thisresultsinthree V H

cate-gories,targetingthehadronicdecaysofthevectorboson,charged leptonicdecaysofthevectorbosonanddecaysofthevectorboson containingoneormoreneutrinos.

The combinationisundertaken asoutlined inRef. [116]. Con-straining the branching fractions for the Z Z∗, diphoton and bb

¯

decays tobe as predictedby theSM, this combinationmeasures thesignalstrengthoftheV H productionmode.

7. Results

7.1. ResultsoftheSMHiggsbosonsearchat

√

s

=

13 TeV

Fig.1showstheBDToutputdistributionsinthemostsensitive, high-pV_T,region.Thebackgroundpredictioninallpost-fit distribu-tions isobtainedby normalisingthebackgrounds andsettingthe nuisanceparametersaccordingtotheresultsofthesignal extrac-tion fit. The post-fit signal and background yields are shown in Table10forallsignalregions.

For a Higgs boson mass of 125 GeV, when all lepton chan-nelsarecombined,theprobability p0 ofobtainingasignalatleast

asstrongastheobservationfrombackgroundaloneis 5

.

3

·

10−7_,

whilst the expected value is 7

.

3

·

10−6_{. The observation}

corre-spondstoanexcesswithasignificanceof4.9standarddeviations, to be compared with an expectation of 4.3 standard deviations. Thefittedvalueofthesignalstrengthis:

μ

bb

V H

=

1.16+₋00..2725

=

1.16

±

0.16(stat.)+ 0.21 −0.19

(syst.).

Fig. 2 shows the data, background and signal yields, where final-discriminant bins in all regions are combined into bins of log₁₀

(

S

/

B

)

. Here, S and B are the fitted signal and background yieldsineachanalysisbin,respectively.

Table11showsthesignalstrengths, p0 andsignificancevalues

fromthecombinedfitwithasinglesignalstrength,andfromafit where the lepton channels each have their own signal strength. The probability that the signal strengths measured in the three leptonchannels5 arecompatibleis80%.

5 _{Theprobabilityofcompatibilitybetweenfitsdifferingonlyintheirnumberof} parametersofinterestisevaluatedintheasymptoticsregime,wherethedifference

Fig. 1. TheBDTV H outputpost-fitdistributionsinthe0-lepton (top),1-lepton (middle)and2-lepton (bottom)channelsfor2-b-tagevents,inthe2-jet (left)andexactly

3-jet(or≥3 jetsforthe2-leptoncase)(right)categoriesinthehigh-pV

T region.Thebackgroundcontributionsafterthegloballikelihoodfitareshownasfilledhistograms. TheHiggsbosonsignal(mH=125 GeV)isshownasafilledhistogramontopofthefittedbackgroundsnormalisedtothesignalyieldextractedfromdata(μ=1.16),and

unstackedasanunfilledhistogram,scaledbythefactorindicatedinthelegend.Thedashedhistogramshowsthetotalpre-fitbackground.Thesizeofthecombinedstatistical andsystematicuncertaintyforthesumofthefittedsignalandbackgroundisindicatedbythehatchedband.Theratioofthedatatothesumofthefittedsignal(μ=1.16) andbackgroundisshowninthelowerpanel.TheBDTV Houtputdistributionsareshownwiththebinningusedinthegloballikelihoodfit.

Table 10

TheHiggsbosonsignal,backgroundanddatayieldsforeachsignalregioncategoryineachchannelafterthefullselectionofthe multivariateanalysis.Thesignaland backgroundyieldsarenormalisedtotheresultsofthegloballikelihoodfit.Allsystematicuncertaintiesareincludedintheindicateduncertainties.Anentryof“–”indicates thataspecificbackgroundcomponentisnegligibleinacertainregion,orthatnosimulatedeventsareleftaftertheanalysisselection.

Process 0-lepton 1-lepton 2-lepton

pV

T>150 GeV, 2-b-tag pTV>150 GeV, 2-b-tag 75 GeV<pVT<150 GeV, 2-b-tag pTV>150 GeV, 2-b-tag

2-jet 3-jet 2-jet 3-jet 2-jet ≥3-jet 2-jet ≥3-jet

Z+ll 17±11 27±18 2±1 3±2 14±9 49±32 4±3 30±19 Z+cl 45±18 76±30 3±1 7±3 43±17 170±67 12±5 88±35 Z+HF 4770±140 5940±300 180±9 348±21 7400±120 14160±220 1421±34 5370±100 W+ll 20±13 32±22 31±23 65±48 <1 <1 <1 <1 W+cl 43±20 83±38 139±67 250±120 <1 <1 <1 <1 W+HF 1000±87 1990±200 2660±270 5400±670 2±0 13±2 1±0 4±1

Single top quark 368±53 1410±210 2080±290 9400±1400 188±89 440±200 23±7 93±26

tt¯ 1333±82 9150±400 6600±320 50200±1400 3170±100 8880±220 104±6 839±40 Diboson 254±49 318±90 178±47 330±110 152±32 355±68 52±11 196±35 Multi-jet e sub-ch. – – 100±100 41±35 – – – – Multi-jetμsub-ch. – – 138±92 260±270 – – – – Total bkg. 7850±90 19020±140 12110±120 66230±270 10960±100 24070±150 1620±30 6620±80 Signal (post-fit) 128±28 128±29 131±30 125±30 51±11 86±22 28±6 67±17 Data 8003 19143 12242 66348 11014 24197 1626 6686 Table 11

Measuredsignalstrengthswiththeircombinedstatisticalandsystematicuncertainties,expectedand observed

p

0andsignificancevalues(instandarddeviations)fromthecombinedfitwithasinglesignal strength,andfromacombinedfitwhereeachoftheleptonchannelshasitsownsignalstrength,using 13TeV data.

Signal strength Signal strength p0 Significance

Exp. Obs. Exp. Obs.

0-lepton 1.04+₋0.340.32 9.5·10−4 5.1·10−4 3.1 3.3 1-lepton 1.09+₋0.460.42 8.7·10−3 4.9·10−3 2.4 2.6 2-lepton 1.38+0.46 −0.42 4.0·10− 3 ₃ .3·10−4 ₂ .6 3.4 V H, H→bb combination¯ 1.16+₋0.270.25 7.3·10−6 5.3·10−7 4.3 4.9

Fig. 2. Eventyieldsasafunctionoflog10(S/B)fordata,backgroundandaHiggs bo-sonsignalwith

m

H=125 GeV.Final-discriminantbinsinallregionsarecombined

intobinsoflog10(S/B),with

S being

thefittedsignaland

B the

fittedbackground yields.TheHiggsbosonsignalcontributionisshownafterrescalingtheSM cross-sectionaccordingtothevalueofthesignalstrengthextractedfromdata(μ=1.16). Inthelowerpanel,thepullofthedatarelativetothebackground(thestatistical significanceofthedifferencebetweendataandfittedbackground)isshownwith statisticaluncertaintiesonly.Thefulllineindicatesthepullexpectedfromthesum offittedsignalandbackgroundrelativetothefittedbackground.

Fig. 3. ThefittedvaluesoftheHiggsbosonsignalstrengthμbb

V Hfor

m

H=125 GeV

forthe

W H and Z H processes

andtheircombination.Theindividualμbb V H values

forthe (W/Z)H processes areobtainedfrom asimultaneousfitwiththe signal strengthforeachofthe

W H and Z H processes

floatingindependently.The proba-bilityofcompatibilityoftheindividualsignalstrengthsis84%.

Acombinedfitisalsoperformedwithfloatingsignalstrengths separately forthe W H and Z H productionprocesses.The results ofthisfitareshowninFig.3.TheW H and Z H productionmodes

betweentheirmaximumlikelihoodsfollowsaχ2 _{distributionwith anumberof} degreesoffreedomequaltothedifferencebetweenthenumbersofparametersof interest.

Fig. 4. Thedistributionof

m

bbindataaftersubtractionofallbackgroundsexceptfor

the

W Z and Z Z diboson

processes,asobtainedwiththedijet-massanalysis.The contributionsfromallleptonchannels, pV

T regionsandnumber-of-jetscategories aresummedandweightedbytheirrespective

S

/B, with S being thetotalfitted signaland

B the

totalfittedbackgroundineachregion.Theexpectedcontributionof theassociated

W H and Z H production

ofaSMHiggsbosonwith

m

H=125 GeV is

shownscaledbythemeasuredsignalstrength(μ=1.06).Thesizeofthecombined statisticalandsystematicuncertaintyforthefittedbackgroundisindicatedbythe hatchedband.

have observed (expected) significances of 2.5 (2.3) and 4.0 (3.5) standarddeviations,respectively,withalinearcorrelationbetween thetwosignalstrengthsof

−

1%.

7.2. Resultsofthedijet-massanalysis

Forallchannelscombinedthefittedvalueofthesignalstrength is

μ

bb

V H

=

1.06+₋00..3633

=

1.06

±

0.20(stat.)+ 0.30 −0.26

(syst.),

ingoodagreementwiththeresultofthemultivariateanalysis.The observedexcesshasasignificanceof3.6standarddeviations, com-pared to an expectation of 3.5 standard deviations. Good agree-mentisalsofoundwhencomparingthevaluesofsignalstrengths intheindividualchannelsfromthedijet-massanalysiswiththose fromthemultivariateanalysis.

Thembb distributionisshowninFig.4summedoverall

chan-nels andregions,weighted by their respectivevaluesofthe ratio offittedHiggsboson signalandbackgroundyieldsandafter sub-traction of all backgrounds except for the W Z and Z Z diboson

processes.

7.3. Resultsofthedibosonanalysis

As a validation of the Higgs boson search analysis, the mea-surement of V Z production based on the multivariate analysis describedinSection6.3returnsavalueofsignalstrength

μ

bb_{V Z}

=

1.20+₋0₀._.20₁₈

=

1.20

±

0.08(stat.)+₋0₀._.19₁₆

(syst.),

in good agreement with the Standard Model prediction. Analo-gously to the V H signal, fits are also performed with separate signal strengths for the W Z and Z Z production modes, andthe resultsareshowninFig.5.

7.4. Resultsofcombinations

7.4.1. Run1andRun2combinationforV H ,H

→

bb

¯

The resultof the Run 2 analysisis combined withthe Run 1

V H , H

→

bb result

¯

followingthe methodologydescribed in

Sec-Fig. 5. Thefittedvaluesofthe

V Z signal

strengthμbb

V Z forthe

W Z and Z Z

pro-cessesandtheircombination.Theindividualμbb

V Z valuesforthe(W/Z)Z processes

areobtainedfromasimultaneousfitwiththesignalstrengthsforeachofthe

W Z

and

Z Z processes

floatingindependently.Theprobabilityofcompatibilityofthe in-dividualsignalstrengthsis47%.

Fig. 6. ThefittedvaluesoftheHiggsbosonsignalstrengthμbb

V Hfor

m

H=125 GeV

forthe

W H and Z H processes

andtheircombination,usingthe7 TeV,8 TeV and 13 TeV data.The individualμbb

V H valuesfor the(W/Z)H processes areobtained

fromasimultaneousfitwiththesignalstrengthsforeachofthe

W H and Z H

pro-cessesfloatingindependently.

tion 6.4. The observed p0 value is 5

.

5

·

10−7, corresponding to

anexcesswithasignificanceof4.9standarddeviations,compared withanexpectationof5.1standarddeviations.Themeasured sig-nalstrengthis:

μ

bb_{V H}

=

0.98+₋0₀._.22₂₁

=

0.98

±

0.14(stat.)+₋0₀._.17₁₆

(syst.).

Fits arealsoperformedwiththesignal strengthsfloated indepen-dentlyforthe W H and Z H productionprocesses.Theprobability ofcompatibilityofthesignal strengthsforthe W H and Z H

pro-duction processesis 72%,andtheresults ofthisfitare shownin Fig.6.

7.4.2. ObservationofH

→

bb decays

¯

The V H resultisfurthercombinedwithresultsofthesearches forthe StandardModelHiggsboson decayingintoa bb pair

¯

pro-ducedinassociationwithatt pair

¯

andinvector-bosonfusionfor bothRun 1andRun 2,toperformasearchforthe H

→

bb decay.

¯

ForaHiggsbosonmassof125GeV,andassumingtherelative pro-ductioncross-sectionsarethosepredictedbytheSM,theobserved significancefortheH

→

bb decay

¯

is5.4standarddeviations,tobe comparedwithanexpectationof5.5standarddeviations.Withthe additionalassumptionthattheproductioncross-sectionsarethose