Measurement of photon–jet transverse momentum correlations in 5.02 TeV Pb + Pb and pp collisions with ATLAS

(1)

Article

Reference

Measurement of photon–jet transverse momentum correlations in 5.02 TeV Pb + Pb and pp collisions with ATLAS

ATLAS Collaboration AKILLI, Ece (Collab.), et al.

Abstract

Jets created in association with a photon can be used as a calibrated probe to study energy loss in the medium created in nuclear collisions. Measurements of the transverse momentum balance between isolated photons and inclusive jets are presented using integrated luminosities of 0.49 nb −1 of Pb + Pb collision data at sNN=5.02 TeV and 25 pb −1 of pp collision data at s=5.02 TeV recorded with the ATLAS detector at the LHC. Photons with transverse momentum 63.1

ATLAS Collaboration, AKILLI, Ece (Collab.), et al . Measurement of photon–jet transverse

momentum correlations in 5.02 TeV Pb + Pb and pp collisions with ATLAS. Physics Letters. B , 2019, vol. 789, p. 167-190

DOI : 10.1016/j.physletb.2018.12.023

Available at:

http://archive-ouverte.unige.ch/unige:112525

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

.The ATLAS Collaboration

a r t i c l e i n f o a b s t ra c t

Articlehistory:

Received19September2018

Receivedinrevisedform19November2018 Accepted10December2018

Availableonline13December2018 Editor:W.-D.Schlatter

Jets createdin associationwith a photon can beused as acalibrated probe tostudyenergy loss in themediumcreatedinnuclearcollisions.Measurementsofthetransversemomentumbalancebetween isolatedphotonsandinclusivejetsarepresentedusingintegratedluminositiesof0.49 nb⁻¹ofPb+^Pb collisiondataat√_s

NN=⁵.02 TeV and25 pb⁻¹ofppcollisiondataat√

s=⁵.02 TeV recordedwiththe ATLASdetectorattheLHC.Photonswithtransverse momentum63.1<p^γ_T<200 GeV andη^γ<2.37 are pairedwith all jets inthe event that have p^jet_T >31.6 GeV and pseudorapidity η^jet<2.8. The transverse momentum balance given by the jet-to-photon pT ratio, xJγ, is measured for pairs with azimuthal opening angle φ >7π/8. Distributions ofthe per-photonjet yield as a functionof xJγ, (1/N_γ)(dN/dxJγ), are corrected for detector effects via a two-dimensional unfolding procedure and reportedattheparticlelevel.Inppcollisions,thedistributionsarewelldescribedbyMonteCarloevent generators.InPb+^Pbcollisions,thexJγ distributionismodiﬁedfromthatobservedinppcollisionswith increasingcentrality,consistentwiththepictureofpartonenergylossinthehotnuclearmedium. The dataarecomparedwithasuiteofenergy-lossmodelsandcalculations.

©²⁰¹⁸^TheÂuthor(s).^Published^byÊlsevier^B.V.^Thisîsânôpenâccessârticleûnder^the^CC^BY^license (http://creativecommons.org/licenses/by/4.0/).FundedbySCOAP³.

1. Introduction

The energy loss of fast partons traversing the hot, decon- finedmediumcreatedinnucleus–nucleuscollisionscanbestudied in a controlled and systematic way through the analysis of jets produced in association with a high transverse momentum (pT) promptphoton [1–7].Atleadingorderinquantumchromodynam- ics, the photon andleading jet are produced back-to-back inthe azimuthal plane, with equal transverse momenta. Measurements ofprompt photon productionin Au + Âu ^collisions ât ^the ^Rela- tivisticHeavyIonCollider(RHIC) [8] andPb+^Pb^collisionsât^the LargeHadronCollider(LHC) [9] haveconfirmedthat,sincephotons donot participateinthestronginteraction,their productionrates are not modified by the medium [10]. Thus, photons provide an estimateofthepTanddirectionofthepartonproducedintheini- tialhard-scatteringbefore it haslost energythrough interactions withthemedium. Measurementsof jetproductionwithdifferent requirements on the photon kinematics can therefore shed light onhowtheabsoluteamountofpartonenergylossdependsonthe initialpartonpT.

Furthermore,photon–jet events offeraparticularly usefulway toprobethedistributionofenergylostbyjetsinindividualevents,

E-mailaddress:atlas.publications@cern.ch.

andarecomplementarytomeasurementssuchasthedijetpTbal- ance [11–13].Whereasthosemeasurementsreporttheratioofthe transverse momenta of two ﬁnal-state jets, both of which may have lost energy, photon–jet events provide an alternative sys- temin whichone high-p_T objectis certain toremain unaffected by the hot nuclear medium. Finally, jetsproduced in association withaphotonaremorelikelytooriginatefromquarksthanthose produced in dijetevents atthe same p_T.Thus, when considered together withmeasurements of dijets or ofinclusive jet [14–16]

andhadron [17–19] productionratesinPb+ ^Pbcollisions,analysis ofphoton–jet eventscan helpto furtherconstrain theﬂavour (i.e.quarkversusgluon)dependenceofpartonenergyloss.

Studies of photon–hadron correlations, in which high-pT hadrons are usedas a proxyfor the jet, were ﬁrst performedat RHIC [20–22], and measurements using fully reconstructed jets havesincebegunattheLHC [23,24].IntheLHCstudies,thedistri- bution ofthephoton–jet azimuthal separation,φ,was foundto beconsistentwiththatinsimulatedphoton–jeteventsembedded intoaheavy-ionbackground,andthejet-to-photontransversemo- mentumratio,x_Jγ=^p^jet_T /pγ

T,was studiedforinclusivephoton–jet pairs. Theper-photon jetyield (1/Nγ)(dN/dxJγ) distributionwas shiftedtosigniﬁcantlysmallervaluesinPb+^Pb^data.

In these previous measurements, the x_Jγ distributions in Pb+^Pb^events^were ^not^corrected^for^detector^resolution^effects, which led to a substantial broadening of the reported distribu- https://doi.org/10.1016/j.physletb.2018.12.023

0370-2693/©²⁰¹⁸^TheÂuthor(s).^Published^byÊlsevier^B.V.^Thisîsânôpenâccessârticleûnder^the^CC^BY^license(http://creativecommons.org/licenses/by/4.0/).Fundedby SCOAP³.

(3)

168 The ATLAS Collaboration / Physics Letters B 789 (2019) 167–190

tionsindata.As aresult, qualitativecomparisonswithmodelsor evenwiththeanalogousdistributionsinproton–proton(pp)data couldonlybeaccomplishedbyapplyinganadditionalsmearingto thecomparisondistributions tointroduce detectoreffects.Recent measurements ofdijet pT correlations [12] and inclusivejetfrag- mentationfunctionsatlargelongitudinalmomentumfraction [25]

inPb+^Pb^collisions^used^unfolding^procedures^to^correct^for^bin- migrationeffectsandreturnthedistributionstotheparticlelevel, i.e.freefromdetectoreffects.

ThisLetterreportsastudyofphoton–jetcorrelationsinPb+^Pb collisions at a nucleon–nucleon centre-of-mass energy √

sNN = 5.02 TeV and pp collisions at the same centre-of-mass energy

√s=⁵.02 TeV. The data were recorded in 2015 with the AT- LAS detector at the LHC and correspond to integrated luminosities of 0.49 nb⁻¹ and 25 pb⁻¹, respectively. Events containing a prompt photon with 63.1<pγ

T <200 GeV and pseudorapidity

η

^γ<2.37 (excludingtheregion1.37<

η

^γ<1.52) arestudied.

The p_T balanceof photon–jet pairs forjetswith p^jet_T >31.6 GeV and

η

^jet<2.8 which are approximately back-to-back with the photon in the transverse plane, φ >7

π

/8,is analysed through theper-photonyield ofjetsasafunctionofx_Jγ,withall jetsthat meetthisselectionrequirementcountedseparately.InMonteCarlo simulations,thefractionofphotonspairedwithmorethanonejet rises from1%to ≈^15% ^over^the ^reported^photon ^pT ranges.The particularphotonandjet pT rangesusedinthemeasurementare chosen tobe evenlyspacedonlogarithmic scales tofacilitatethe unfoldingproceduredescribedbelow.

The yields are corrected via data-driven techniques for background arising from combinatoric pairings of each photon with unrelatedjetsin Pb+ ^Pbêvents ând^from^thecontamination by neutralmesons inthe photonsample. The resulting x_Jγ distributions are corrected forthe effects ofthe experimental resolution onthephotonandjet pT viaatwo-dimensional unfoldingproce- duresimilar tothatusedinRef. [12]. Duetohigher-ordereffects, photon–jet eventsdonotgenerallyhavethe back-to-backleading ordertopologymentionedabove.Thustheppdata,whichincludes theseeffects,providesthereferencedistributionsagainstwhichto interpretthe resultsinPb + ^Pbêvents.^This^Letter ^directly^com- paresphoton–jet datainPb+^Pbând ^ppêvents,ând^with^Monte Carloeventgeneratorsandanalyticcalculations [26–29].

2. Experimentalset-up

The ATLAS experiment [30] is a multipurpose particle detec- torwitha forward–backwardsymmetriccylindricalgeometryand nearly4

π

coverage.¹Thisanalysisreliesontheinnerdetector,the calorimeterandthedataacquisitionandtriggersystem.

Theinnerdetectorcomprisesthreemajorsubsystems:thepixel detector and the silicon microstrip tracker, which extend out to

|

η

| =².5, and the transition radiation tracker which extends to

|

η

_{| =}2.0. The inner detector covers the full azimuth and is im- mersed ina 2 T axial magneticﬁeld. The pixeldetector consists of four cylindrical layers in the barrel region and three disks in each endcapregion.Thesilicon microstriptrackercomprises four cylindricallayers(ninedisks)ofsiliconstripdetectorsinthebarrel (endcap)region.

1 ATLASusesaright-handed coordinatesystemwith itsoriginat thenominal interactionpoint(IP)inthecentreofthedetectorandthez-axisalongthebeam pipe.Thex-axispointsfromtheIPtothecentreoftheLHCring,andthe y-axis pointsupward.Cylindricalcoordinates (r,φ)areusedinthe transverseplane,φ beingtheazimuthalanglearoundthez-axis.Thepseudorapidityisdefinedinterms ofthe polarangleθas η= −ln tan(θ/2). Transversemomentumandtransverse energyaredefinedas pT=^p^sinθandET=Ê^sinθ,respectively.Risdefinedas (η)²+(φ)².

Thecalorimeterisalarge-acceptance,longitudinally-segmented sampling detector covering |

η

_|<4.9 with electromagnetic (EM) and hadronicsections.The EM calorimeteris a lead/liquid–argon samplingcalorimeterwithanaccordion-shapedgeometry.Itisdi- vided intoa barrel region,covering |

η

_|<1.475, andtwo endcap regions, covering1.375<|

η

|<3.2.TheEMcalorimeterhasthree primary sections,longitudinalin showerdepth,called“layers”, in the barrel region andup to |

η

_{| =}2.5 in the end cap regions. In the barreland ﬁrst partof theend cap(|

η

|<2.4), with theex- ception of the regions 1.4<|

η

|<1.5, the ﬁrst layer has a ﬁne segmentation in

η

(

η

=⁰.003–0.006) to allow the discrimina- tionofphotonsfromthetwo-photondecaysof

π

⁰ and

η

mesons.

Overmostoftheacceptance,thetotalmaterialupstreamoftheEM calorimeterrangesfrom2.5 to6 radiationlengths.Inthetransition regionbetweenthebarrelandendcapregions(1.37<|

η

_|<1.52), the amount of material rises to 11.5 radiation lengths, and thus thisregionisnotusedforthedetectionofphotons. Thehadronic calorimeterislocated outsidetheEMcalorimeter.It consistsof a steel/scintillator-tile sampling calorimeter covering |

η

|<1.7 and a liquid–argon calorimeter with copper absorber covering 1.5<

|

η

_|<3.2.

The forward calorimeter (FCal) is a liquid–argon sampling calorimeterlocatedoneitherside oftheinteractionpoint.Itcov- ers 3.1<|

η

_|<4.9 and each half is composed of one EM and two hadronic sections, with copper and tungsten serving as the absorber material, respectively. The FCal is used to characterise the centrality of Pb + ^Pb ^collisions ^as ^described ^below. ^Finally, zero-degree calorimeters(ZDC)are situatedatlargepseudorapid- ity,|

η

|>8.3,andareprimarilysensitivetospectatorneutrons.

Atwo-leveltriggersystemisusedtoselectevents,withaﬁrst- level trigger implemented in hardware followed by a software- based (high-level) trigger. Data for this measurement were ac- quiredusinga high-level photontrigger [31] coveringthe central region (|

η

_|<2.5). At the ﬁrst-level trigger stage, the transverse energyofEMshowersiscomputedwithinregions ofφ×

η

₌

0.1×⁰.1, and those showers which satisfy an E_T threshold are usedtoseedthehigh-leveltriggerstage.Atthisnextstage,recon- structionalgorithmssimilartothoseappliedintheoﬄineanalysis usethefulldetectorgranularitytoformtheﬁnaltriggerdecision.

Thetriggerwasconfiguredwithanonlinephoton-pT thresholdof 30 GeV (20 GeV)inthepp(Pb+^Pb)^running^periodând^required thecandidatephotontosatisfyasetofloosecriteriafortheelec- tromagnetic showershape [31]. ForthePb+ ^Pbdata-taking, the high-level trigger included a procedure to estimate and subtract theunderlyingevent(UE)contributiontothe ET measuredinthe calorimeter [9],ensuring highefficiency inhigh-activity Pb+ ^Pb events.

Inadditionto thephotontrigger,Pb+ ^Pb^data^were^recorded withminimum-biastriggers;theseeventsareusedtocharacterise thecentralityofPb+^Pb^collisions^as^described ⁱⁿ^Section^3.^The minimum-bias triggers arebased on thepresence ofa minimum amount ofapproximately50 GeV of transverseenergy inall sections of the calorimetersystem(|

η

|<3.2) or, forevents that do not meet this condition, on substantial energy deposits in both ZDC modules andan inner-detector trackidentiﬁed by the high- leveltriggersystem.

3. DataselectionandMonteCarlosamples

Photon–jeteventsin ppandPb+^Pb^collisions^are^initially^selected foranalysisby thehigh-leveltriggers describedabove. The typical number ofinteractions per bunch crossing in the pp and Pb+^Pbdata-takingwereoneandsmallerthan10⁻⁴,respectively.

Events are required to satisfy detector and data-quality requirements, andto contain a vertex reconstructed from tracks in the

(4)

that represent0–10%(largest ET values anddegree ofnuclear overlap),10–20%,20–30%,30–50%and50–80%(smallest

E_Tval- uesand degree of nuclear overlap) of the population. The mean numberofparticipatingnucleonsinminimum-biasPb+ ^Pb^collisions,Npart,rangesfrom33.3±¹.5 in50–80%eventsto358.8±².3 in0–10%events.

MonteCarlosimulationsof√

s=⁵.02 TeVppphoton–jetevents are used to correct the data for bin migration and ineﬃciency effects,andforcomparisonwithdistributionsmeasuredinppcol- lision data. For all the samples described below, the generated eventswere passed through a full Geant4 simulation [36,37] of theATLASdetectorunderthesameconditionspresentduringdata- takingandwere digitised andreconstructed in thesame wayas thedata.

Fortheprimarysimulationsamples,thePythia8.186 [38] gen- eratorwasusedwiththeNNPDF23LOpartondistributionfunction (PDF)set [39],andgeneratorparameterswhichweretunedtore- producea set of minimum-biasdata (the “A14” tune) [40]. Both the direct and fragmentation photon contributions are included in the simulation. Six million pp events were generated with a generator-level photon in the pT range 50 GeV to 280 GeV. Ad- ditionally,a sampleof18million eventswere producedwiththe samegenerator,tune andPDF, andwere overlaidatthedetector- hit level with minimum-bias Pb + ^Pb ^events ^recorded ^during the 2015 run. The relative contribution of events in this “data- overlay” sample were reweighted on an event-by-event basis to matchthe

ETdistributionobservedinthephoton–jet eventsin Pb+ ^Pb^data^selected ^for^analysis. ^Thus ^the^Pb + ^Pb^simulation samplescontainunderlying-eventactivitylevelsandkinematicdis- tributionsofjets(usedinthecombinatoricphoton–jetbackground estimation)identicaltothoseindata.

Additional samples of 0.3 million pp events and 6 million events overlaid with Pb + ^Pb ^data ^were ^produced ^with ^the Sherpa2.1.1 [41] generatorusing theCT10PDF set [42], aswere 0.6 million ppHerwig7 [43] eventswiththeMMHTUEtuneand PDF set [44]. The Sherpa samples were generated with leading- ordermatrixelementsforphoton–jetfinalstateswithuptothree additional partons, which were merged with the Sherpa parton shower. The Herwig events were generated in a way that in- cludesthedirectandfragmentationphotoncontributions.Boththe Sherpaand Herwig samples were filtered for the presence of a photon in the required kinematic region, and are used because theycontain differentphoton + ^multijettopological distributions andjet-flavourcompositions.

At generatorlevel, photons are required tobe isolated by re- quiringthesumofthetransverseenergycarriedby primary par- ticles² ina coneofsize R=⁰.3 aroundthephoton, E^iso_T ,to be

2 Primaryparticlesaredefinedasthosewithapropermeanlifetime,τ,exceeding cτ=^{10 mm.}^For^the^jetândîsolationÊTmeasurements,muonsandneutrinosare excludedfromthedefinition.

posited in EM calorimeter cells, following a procedure used for previous measurements of isolated prompt photon production in Pb + ^Pbcollisions [9].The procedure is similar tothat used ex- tensively in pp collisions [48,49], but isapplied to the calorimeter cells after an event-by-event estimation and subtraction of the pile-upand UE contributionto the depositedenergy ineach cell [14]. InPb+ ^Pbcollisions, all photon candidatesare treated asiftheywereunconvertedphotons.Photonidentiﬁcationisbased primarilyonshowershapesinthecalorimeter [50],selectingthose candidates which are compatible with originating from a single photon impacting the calorimeter. The measurement of thephoton energyis based onthe energycollected in a smallregion of calorimetercellscentredonthephoton(

η

×φ=⁰.075×⁰.175 inthe barreland

η

×φ=⁰.125×⁰.125 inthe endcaps),and is corrected via a dedicated calibration [51], which accounts for upstream losses and both lateral and longitudinal leakage. The sum of transverse energy in calorimeter cells inside a cone size of R=⁰.3 centred on thephoton candidate, excluding a small centralareaofsize

η

×φ=⁰.125×⁰.175,isusedtocompute the isolation energy E^iso_T .It is correctedforthe expectedleakage ofthephotonenergyintotheisolationcone.

Reconstructed photon candidates are required to satisfy identification and isolation criteria. The identification working point (called “tight”)includes requirementson each of severalshower- shape variables [50]. These criteria reject two-photon decays of neutral mesons using information in the finely segmented first calorimeterlayers, and rejecthadrons whichbegan showering in the EM section using informationfrom thehadronic calorimeter.

The isolation energy is required to be Eîso_T <3 GeV in pp collisions. In Pb+ ^Pbcollisions, where UE fluctuationssignificantly broaden thedistributionof Eîso_T values,thisrequirementissetto approximatelyonestandarddeviationoftheGaussian-likepartof thedistributioncentredatzero,Eîso_T <8 GeV.

In simulation, prompt photons in pp collisions have a total reconstruction and selection efficiency greater than 90%. At low pT≈^{60 GeV in}^the^most^central^Pb+^Pbcollisions,thisefficiency is≈^60%,^rising^withîncreasing ^pT andinlesscentralcollisions.In all events, the pT scale, defined as the mean ratio of measured photon pT to the generator-level pT, for photons which satisfy thesecriteriais within 0.5%(1%)of unityin thebarrel(endcap).

The p_T resolutiondecreasesfrom3%to2%overthemeasured p_T range.

4.2. Jetreconstruction

Jetsarereconstructedfollowing theprocedurepreviously used in 2.76 TeV and 5.02 TeV pp and Pb + ^Pbcollisions [14,15,52], whichisbrieﬂysummarisedhere.Theanti-kt algorithm [46] with R=⁰.4 is appliedto energydeposits inthe calorimetergrouped into towers ofsize

η

×φ=⁰.1×⁰.1. An iterative procedure,

(5)

based entirely on data, is used to obtain an event-by-event estimate of the average

η

-dependent UE energy density, including thatfrompile-up,whileexcludingfromtheestimatethecontribu- tionfromjetsarisingfromahardscattering.An updatedestimate ofthe jet four-momentum isobtainedby subtracting the UE energy from the constituent towers of the jet. This procedure is also applied to pp data. The pT values of the resulting jets are corrected for the average calorimeter response using an

η

- and pT-dependent calibration derived from simulation. An additional correction,derivedfrominsitu studiesofeventswitha jetrecoil- ingagainstaphotonorZ bosonandfromthedifferencesbetween theheavy-ionreconstructionalgorithmandthat normallyusedin the13 TeVppdata [53],isapplied.Afinalcorrectionattheanal- ysis level is applied to correctfor a deficiency in jet calibration dueto itbeingderived fromaneventsamplewithadifferentjet flavourcomposition.

The distribution of reconstructed jet pT values was studied in simulation asa function of generator-level jet pT. In pp and Pb+^Pbcollisions,thejetpTscaleiswithin1%ofunity.Inppcol- lisions,thejet p_Tresolutiondecreasesfrom15%atp_T≈^{30 GeV to} 10%atp_T≈^{200 GeV.}În^Pb+^Pbcollisions,theresolutionatfixed jet pT becomes worseinmore centralcollisions inawayconsis- tent with the increasing magnitude ofUE fluctuationsin the jet cone.In themostcentral eventsandat thelowest jet-pT values, theresolution reaches50%. Athigh p_T, theresolution asymptoti- cally becomes centrality-independent and, at200 GeV,consistent withthat in ppcollisions. Moreinformation aboutthejet reconstruction and jet performance in this dataset may be found in Ref. [54].

5. Dataanalysis

5.1. Photonpurityandyield

After applying the identiﬁcation and isolation selection criteria in pp collisions, approximately 19500, 7800, 4100 and 400 photons are selected with pγ

T =⁶³.1–79.6 GeV, 79.6–100 GeV, 100–158 GeV and 158–200 GeV, respectively. In Pb + ^Pb ^collisions,theanalogousyieldsare15400,6300,3500 and300.These rawyields are determined asafunction of pγ

T andare then cor- rectedforbackgroundandfortheeffectsof pTbinmigration.

First, the selected photon sample is corrected for the back- groundcontribution,primarilyfrommisidentiﬁedneutralhadrons.

For each pγ

T and centrality range,the purity of prompt photons withinthisrangeisestimatedwithadouble-sidebandapproach [9, 48,49],whichissummarisedinthefollowing.

In addition to the nominal selection, background-enhanced samples of photon candidates are defined by selecting photons failing at least one of four specific shower-shape requirements (referred to as the “non-tight” selection), or by requiring that they are not isolated such that Eîso_T >5 GeV in pp collisions or Eîso_T >10 GeV in Pb + ^Pb collisions. Regions A and B are defined as those containing tight photons which are isolated and non-isolated,respectively,withregion A correspondingtothesig- nalphoton selection.Regions C and D containnon-tight photons which are isolated andnon-isolated, respectively. The numberof photon candidatesineach regionis generallyamixture ofsignal andbackgroundphotons,i.e.thosearisingfromneutralmesonsin- sidejets.The Eîso_T distributionforbackgroundphotonsisexpected tobethesameforthetightandnon-tightselectionssuchthatthe distributionofbackgroundphotons“factorises”alongisolationand identificationaxes.Separately,theprobability thata promptpho- tonisfound inregions B, C or D isdetermined fromsimulation.

This information andthe background factorisation assumption is

thenappliedtothedatatodeterminethepurityofphotonsinre- gion A, deﬁnedas the ratioof the number of signal photons to all selected photons. The purity increasessystematically with pγ T over themeasured pT range.In pp collisions,itrises from≈^85%

at pγ

T =^{80 GeV to}^more ^than^95% ^at^{100 GeV,}^while ⁱⁿ^Pb+^Pb collisionsitistypically≈^75–90%^over^the^same^kinematic^range.

The background-corrected promptphoton yields are then cor- rectedfortheresolutionofthepγ

T measurement.Thisisperformed by comparingtheyields,evaluated separatelyasafunctionofre- constructed andgenerator-level pT,insimulation.Giventhegood pT resolution,thesedifferby2%atmost,andthissmallresulting correctionisappliedtotheyieldsindata.

5.2. Jetbackgroundsubtraction

Therawjetyields,measuredasafunctionofxJγ,arecorrected for two background components using data-driven methods. The correctionsareperformedseparatelyforeach pγ

T intervalandsep- aratelyinppcollisionsandPb+^Pb^collisions^of^different^central- ityranges.

The ﬁrstbackgroundarisesfromthecombinationofahigh-p_T photon withjetsunrelatedto thephoton-producing hardscatter- ing. Theseinclude jetsfromseparate hard parton–partonscatter- ings and UE ﬂuctuations reconstructed as jets. This background is negligible in pp collisions. Because of the inclusive jet selection inthe analysis, the combinatoricbackgroundis purely addi- tive and can be statistically subtracted after scaling to the total photon yield. The combinatoric jet yields are determined in the data-overlay simulation, by examining the yield of reconstructed jets separatedfrom a generator-level photon by φ >7

π

/8. Re- constructedjetsthat arenotconsistent withagenerator-level jet, i.e.nogenerator-level jetwith p_T>20 GeV within R<0.4,are deemedtoarisefromtheoriginalPb+^Pb^dataêventândâre^thus labelledas“combinatoric”jets.Thecombinatoricjetyieldsaresub- tractedfromthemeasuredx_Jγ distributionsindata.

The second background is related to the estimated purity of the selectedphotons. The xJγ yields forphoton candidates inre- gion A contain an admixture ofdijets, speciﬁcallyjets correlated with misidentiﬁed neutral mesons. Since thesehadrons pass experimental isolation requirements,they maybe, forexample,the leadingfragment insideajet.Theshapeofthisbackgroundinthe xJγ distributionisdeterminedbyrepeatingtheanalysisforphoton candidates in region C, since this region contains mostly neutral mesons that remain isolated at the detector level. The resulting per-photon xJγ distributions are scaled to match the number of backgroundphotons,asdeterminedaboveinSection5.1,andtheir yieldsarestatisticallysubtractedfromthejetyieldsforphotonsin region A.

Fig.1showsthesizeofthesebackgroundsinthelowest-pγ T interval,wheretheyarethelargest.Thecombinatoricjetbackground for Pb+ ^Pb ^collisions contributesprimarily to kinematicregions populated by p^jet_T <50 GeV. It also dependsstrongly on centrality,beinglargestin0–10% collisionsbutnearly negligiblealready in 30–50%collisions.The dijetbackgroundcontributestoa broad rangeofp^jet_T valuesincludingtheregionxJγ>1,sincethepTratio ofajettooneofthehadronsinthebalancingjetcangenerallybe aboveunity.Thisbackgroundhasasimilarshapeinalleventtypes.

However,sincethephotonpurityislowerinPb+^Pb^events^than in ppevents,thiscorrectionislargerintheformer.

5.3. Unfolding

The background-subtracted xJγ yields are corrected for bin- migration effects due to detector resolution via a Bayesian unfolding procedure [55,56]. To accomplish this, the reconstructed

(6)

Fig. 1.Distributionsofthephoton–jetpT-balancexJγ forthephotontransversemomentumintervalp^γ_T=63.1–79.6 GeV for(left)pp,(centre)50–80%centralityand(right) 0–10%centralityPb+Pbevents.Solidgrey,dottedred,anddashedbluehistogramsshowtherawjetyields,theestimateofthecombinatoricbackground(non-existent forppevents),andthedijetbackground,respectively.Blackpointsshowthebackground-subtracteddatabeforeunfolding,withtheverticalbarsrepresentingthecombined statisticaluncertaintyfromthedataandbackgroundsubtractionprocedure.

yields are arranged in a two-dimensional (pγ

T,xJγ) matrix with binedgesthat are evenlyspacedon logarithmicscales (andwith valuesmatchingthose usedinpreviousjet measurements),anda two-dimensionalunfoldingisperformedsimilartothatfordijetp_T correlationsinRef. [12].TheunfoldingisperformedinxJγ directly topreservetheﬁnecorrelationbetweenp^jet_T andpγ

T whichwould bewashed out ifthe unfolding wereperformed in(pγ

T,p^jet_T ). Al- thoughthemigrationalongthepγ

T axisissmall,itisnecessaryto includeitsincethedegreeofbinmigrationinxJγ dependsonthe pTofthejets.

Tofullyaccountfortheeffectsofbinmigrationacrosstheanal- ysisselection, the axes of the matrix are extended over a larger rangeofpγ

T andx_Jγ than theﬁducialregioninwhichtheresults are reported.A responsematrix isdetermined by matching each pairof(pγ

T,xJγ)valuesatthegeneratorleveltotheircounterparts atthereconstruction level,separately for pp eventsandforeach Pb+^Pbcentrality.

TheBayesianunfoldingmethodrequiresachoiceforthenum- ber of iterations, niter, and an assumption for the prior for the initial particle-level distribution. The Pythia simulation does not include the effects of jet energy loss, and thus the underlying particle-leveldistribution indataisexpectedtohaveashape dif- ferentfromthedefaultprior inthe simulation.An initialunfold- ing usingthe defaultPythia prior is performedforeach centralityselection, and the ratios of the unfolded distributions to the generator-level priors in Pythia are ﬁtted with a smooth func- tioninx_Jγ ineach pγ

T interval. Thisfunctionis evaluatedto give aweight w=^w(x_Jγ,pγ

T) that isused to reweightthe generator- leveldistributioninsimulationandthusconstructanominalprior.

Alternativereweightings, usedinevaluating thesensitivity tothe choice of prior, are determined by applying √

w (the geometric meanof thenominal reweighting andno reweighting)and w³^/² tothesample.Thereconstruction-levelx_Jγ distributionsinsimula- tionaftereachofthesereweightingswereexaminedtoensurethat theyspanareasonablerangeofvaluescomparedtothatobserved atthereconstructionlevelindata.

Beforeapplyingtheunfolding proceduretodata,it wastested on simulation. After the nominal reweighting, the Monte Carlo samplesweresplit intotwo statisticallyindependent subsamples.

Onesubsamplewas usedto populatethe responsematrix,which was then used to unfold the reconstruction-level distribution in theothersubsample.Theunfolded resultwascomparedwiththe original generator-level distribution in the latter sample, which

werefoundtoberecoveredwithinthelimitsofthestatisticalpre- cisionofthesamples.

The values of n_iter used for the nominal results are chosen following the sameprocedure as inRef. [12]. For each centrality selection,theunfoldeddistributionsareexaminedasafunctionof niter.Foreachvalueofniter,a totaluncertaintyisformedbyadding two components in quadrature: (1) the statistical uncertainty of theunfoldeddata,whichgrowsslowly withn_iter,and(2)thesum ofsquare differencesbetweentheresultsandthoseobtainedwith an alternative prior, which decreases quickly withniter.The ﬁnal values ofniter are chosen to minimise the total uncertainty, and arebetweentwoandfour.

TheunfoldedxJγ resultsarecorrectedforthejetreconstruction eﬃciency,evaluatedinsimulationasthepγ

T-dependentprobability that ageneratedjet atthegivenxJγ is successfullyreconstructed within the total (pγ

T,xJγ) range used in the unfolding. This eﬃ- ciency is typically >99% for all events in the kinematic regions populatedbyjetswithpT>50 GeV.Inppcollisions,thiseﬃciency falls to ≈^96% ⁱⁿ ^the ^lowest-xJγ region for each pγ

T interval. In Pb+^Pbcollisions,theeﬃciencyatﬁxed xJγ decreasesmonotoni- callyinincreasinglycentralevents,reaching aminimumof≈75%

inthelowest-x_Jγ regionin0–10%centralityevents.

6. Systematicuncertainties

Theprimary sourcesofsystematicuncertaintycanbe grouped intothreemajorcategories:themeasurementofp^jet_T ;theselection ofthephotonandmeasurementofpγ

T;themodellingandsubtrac- tionofthecombinatoricbackground;andtheunfoldingprocedure.

Foreachvariationdescribedbelow,theentireanalysisisrepeated includingthebackgroundcorrectionsteps andunfolding.Thedif- ferences betweenthe resulting xJγ values andthe nominal ones aretakenasanestimateoftheuncertaintyfromeachsource.

Astandard setofuncertainties in thejet p_T scale andresolu- tion, followingthe strategy described in Ref. [57] andcommonly used for measurements in 2015 Pb + ^Pb ^and ^pp data [54,58], are usedinthis analysis.The impact oftheuncertainties is evaluated by modifying the response matrix according to the given variations in the reconstructed jet p_T. These include uncertaintiesinthe p_T scalederived frominsitustudiesofthecalorimeter response [47,59], an uncertainty in the resolution derived using data-driventechniques [60],anduncertaintiesinbothwhichresult froma small relativeenergy-scale difference betweentheheavy- ion jet reconstruction procedure and that used in √

s=^{13 TeV}

(7)

Fig. 2.Unfoldeddistributionsandsummaryofsystematicuncertaintiesintheper-photonjet-yieldmeasurementforp^γ_T=63.1–79.6 GeV in(left)ppeventsand(right)0–10%

centralityPb+Pbevents.Toppanelsshowthephoton–jetpT-balancexJγ distributionsandtotaluncertainties,whilethebottompanelsshowtheabsoluteuncertainties fromjet-related,photon-related,andmodellingorunfoldingsources,aswellasthetotaluncertainty.

ppcollisions [53].All oftheabove uncertainties apply equallyto jetsin pp and Pb + ^Pb ^events. A separate, centrality-dependent uncertainty isincluded in 0–60% Pb + ^Pb collisions. This uncer- taintyaccountsforapossiblemodiﬁcationofthejetresponseafter energy loss and is evaluated through insitu comparisons of the charged-particletrack-jetandcalorimeter-jetpTvaluesindataand simulation. More details are provided in Refs. [54,57]. No addi- tionaluncertaintyisincludedfor60–80%centralityevents.

Uncertaintiesinthe photonpurityestimateare determinedby varying the non-tight identiﬁcationand isolation criteriaused to selecthadronbackgroundcandidatesandbyconsideringapossible non-factorisationofthehadronbackgroundalongtheaxesusedin thedouble-sidebandprocedure.Thesensitivitytothemodellingof photonshowershapesinsimulationisevaluatedbyremovingthe data-driven correctionsto these quantities [50]. Finally,the photon pTscaleandresolutionuncertaintiesaredescribedindetailin Ref. [51],andtheirimpactisevaluatedbyapplyingthemasvaria- tionstotheresponsematricesusedinunfolding.

Modelling- or unfolding-related systematic uncertainties arise fromseveralsources.Theestimateofthecombinatoricphoton–jet rateinthedata-overlaysimulationissensitivetotherequirement on the minimum p_T of a generator-level jet in the classiﬁcation ofa given reconstructedjet asa combinatoric jet, asopposedto aphoton-correlatedjet.Toprovideoneestimate ofthesensitivity tothisthreshold,itisvaried intherange20±^{10 GeV.}^To^assess thesensitivitytothechoiceofprior,theunfoldingisrepeatedus- ing the alternative priors which are systematically closer to and farther from the original Pythia prior. The sensitivity to statistical limitations of the simulation samples is determined through pseudo-experiments, resampling entriesin the response matrices accordingtotheir uncertainty.Finally,theanalysisisrepeatedus- ingtheSherpasimulationtoperformthecorrectionsandunfold- ing,sincethisgeneratorprovidesadifferentdescriptionofphoton–

jetproductiontopologies.

Fig. 2 summarises the systematic uncertainties in each cate- gory, aswell asthe total uncertainty, forthe lowest-pγ

T interval inppand0–10%Pb+^Pb^events.^Thejet-relateduncertaintiesare generally the dominant ones, except in more central events and

lower-pγ

T intervals, wheretheunfoldingandmodellinguncertain- tiesbecomeco-dominant.

Asanadditionalcheckonthefeaturesintheunfoldedx_Jγ distributions observed in data, the analysis was repeated with two modiﬁcationswhichchangethesignalphoton–jetdeﬁnition.First, the photon–jet φ requirement was changed from >7

π

/8 to

>3

π

/4.Withthisalteration,thecorrelatedjetyieldchangesonly by a smallamount,while thecombinatoricbackground,which is constant in φ, doubles. Second, the analysis was repeated, but selecting only the leading (highest-p_T) jet in the event if it fell withintheφ window.Inthiscase,thecombinatoricbackground contributionisnolongerpurelyadditiveandtheinefficiencywhen a higher-pT uncorrelated jet is selected instead of the photon- correlated jetmust be accountedfor,similarto Ref. [12].In both cases, thedistributions in Pb+ ^Pb êxhibit â qualitatively similar modification pattern compared to themain results asa function ofxJγ.

7. Results

The unfolded (1/Nγ)(dN/dxJγ) distributions in pp collisions are shownforeach pγ

T intervalinFig.3.Thedistributions arere- portedforall xJγ binswherethejet minimum pT requirementis fully eﬃcient. Also shown are the corresponding generator-level distributions from thePythia, SherpaandHerwig samples.Each generator describes the data fairly well, with Herwig generally overpredictingtheyieldatlarge-xJγ andSherpashowingthebest agreementoverthefullx_Jγ range.

The unfolded (1/Nγ)(dN/dxJγ) distributions in Pb + ^Pb ^collisions arepresentedinFigs.4through 7,witheach ﬁgurerepre- sentingadifferentpγ

T interval.Sincetheresultsarefullycorrected, they maybe directly compared withthe analogous x_Jγ distributions in pp collisions, which are reproduced in each panel for convenience.

Forall pγ

T intervals,thex_Jγ distributions inPb+ ^Pb^collisions evolvesmoothlywithcentrality.Forperipheralcollisionswithcen- trality50–80%,theyaresimilartothosemeasuredinppcollisions.

However, inincreasingly more centralcollisions, thedistributions become progressively more modiﬁed. For the pγ

T <100 GeV in-

(8)

Fig. 3.Photon–jetpT-balancedistributions(1/Nγ)(dN/dxJγ)inppcollisions,eachpanelshowingadifferentphoton-pTinterval.Theunfoldedresultsarecomparedwith theparticle-leveldistributionsfromthreeMonteCarloeventgenerators.Bottompanelsshowtheratiosofthegeneratorstotheppdata.Totalsystematicuncertaintiesare shownasboxes,whilestatisticaluncertaintiesareshownasverticalbars.

tervalsshown inFigs. 4 and5,the xJγ distributions in the most central0–10% eventsare sostrongly modiﬁedthat they decrease monotonically over the measured xJγ range and no peak is observed. For the pγ

T >100 GeV region shown in Fig. 6, the xJγ distributionsretain a peak atornearx_Jγ ≈⁰.9 eveninthe most centralcollisions.However,themagnitudeofthepeakislowerand signiﬁcantlywiderthanthesharppeakinppevents.Inbothcases, thejetyield atsmallxJγ issystematically higherthanthat in pp collisions,byup toafactoroftwo.Inlesscentral events,a peak- likestructure develops atthe same positionas the maximumin pp events,near xJγ ≈⁰.9.Forthe lowest-pγ

T interval,this occurs onlyfor50–80%centralityevents,whileinthehighesttwopγ

T in- tervalsthedistributionin0–10%eventsis consistentwithalocal peak.

Asanother wayof characterising howthe modiﬁed x_Jγ distributions depend oncentrality and pγ

T,Fig. 8 presentstheir mean value,

x_Jγ

, and integral, Rγ, with both values calculated in the

region xJγ>0.5.Thesequantities areshownasa functionof the meannumberofparticipatingnucleonsNpart inthecorresponding centralityselection,andareplottedfortheﬁrstthree pγ

T intervals wheretheyhavesmallstatisticaluncertainties.Whenmeasuredin theregionxJγ>0.5,thevalueof

xJγ

inppcollisionsisobserved to be ≈⁰.89 for all pγ

T intervals. Simulation studies show that, at generator level,the jet yield atx_Jγ >0.5 corresponds to only theleading(highest-pT)photon-correlatedjetineachevent.Thus, xJγ

can be interpretedas aconditional per-jet fractional energy loss,and Rγ canbeinterpreted asthefractionofphotonswitha leading jetabove x_Jγ =⁰.5.In pp collisions, Rγ rangesfrom0.65 to0.75 inthethree pγ

T intervalsshown,whichisbelowunitydue tothejetselectioncriteria(φ >7

π

/8,|

η

|<2.8).

InPb+^Pb^events, xJγ

decreasesmonotonicallyfromthevalue inppcollisionsasthecollisionsbecomemorecentral.Inthemost central collisions,it isbelowthe pp value by0.04–0.06,depend- ing onthe pγ

T interval, whilein peripheralcollisions itreachesa

(9)

Fig. 4.Photon–jetpT-balancedistributions(1/Nγ)(dN/dxJγ)inPb+Pbevents (redcircles)witheachpanelshowingadifferentcentralityselectioncomparedtothatin ppevents(bluesquares).Thesepanelsshowresultsforp^γ_T=⁶³.1–79.6 GeV.Totalsystematicuncertaintiesareshownasboxes,whilestatisticaluncertaintiesareshownas verticalbars.

Fig. 5.Photon–jetpT-balancedistributions(1/Nγ)(dN/dxJγ)inPb+Pbevents (redcircles)witheachpanelshowingadifferentcentralityselectioncomparedtothatin ppevents(bluesquares).Thesepanelsshowresultsforp^γ_T=⁷⁹.6–100 GeV.Totalsystematicuncertaintiesareshownasboxes,whilestatisticaluncertaintiesareshownas verticalbars.

valuewhichisstatisticallycompatiblewiththatin ppevents.The Rγ value also decreases monotonically as the collisions become morecentral, reﬂectingthe overall shiftofthe xJγ value oflead- ingjetsbelowxJγ=⁰.5.Atlow pγ

T incentral Pb+^Pbcollisions, Rγ reachesthevalueof0.5,whichisonly≈^75%^of^its^valueⁱⁿ^pp collisions.

Theresultsarecomparedwiththefollowingtheoreticalpredic- tions whichinclude MonteCarlogenerators andanalytical calculations ofjet energy loss: (1)a pQCD calculationwhich includes Sudakov resummation to describe the vacuum distributions and energy loss in Pb + ^Pb ^collisions ^as ^described ⁱⁿ ^the ^BDMPS-Z formalism [26], (2) a perturbative calculation within the frame- work of soft-collinear effective ﬁeld theory with Glauber gluons (SCETG)inthesoftgluonemission(energy-loss)limit [27],(3)the JEWELMonteCarloeventgeneratorwhichsimulatesQCDjetevo- lutioninheavy-ioncollisionsandincludesenergy-losseffectsfrom

radiative andelasticscatteringprocesses [28], and(4)theHybrid Strong/WeakCoupling model [29] which combinesinitialproduc- tion usingPythia witha parameterisation ofenergy lossderived fromholographicmethods,andincludesback-reactioneffects.

Figs. 9 and 10 compare a selection of the measured xJγ distributions withtheresultsofthesetheoretical predictions,where possible. Before testing the description of energy-loss effects in Pb+^Pb^events,^the^predicted^xJγ distributionsarecomparedwith ppdatainFig.9.TheHybridmodelandJEWEL,whichusePythia forthephoton–jet productioninvacuum, givea gooddescription of pp events over the measured x_Jγ range in both pγ

T intervals shown. TheBDMPS-ZandSCETG perturbative calculationscapture the generalfeatures but predict distributions that are more and lesspeaked,respectively,thanthoseindata.

In Pb + ^Pb ^events ^with ^low ^p^γ_T^, ^shown ⁱⁿ ^the ^left ^panel ^of Fig.10,theJEWEL,Hybrid,andSCET_G modelssuccessfullycapture

(10)

Fig. 6.Photon–jetpT-balancedistributions(1/Nγ)(dN/dxJγ)inPb+Pbevents (redcircles)witheachpanelshowingadifferentcentralityselectioncomparedtothatin ppevents(bluesquares).Thesepanelsshowresultsfor p^γ_T=100–158 GeV.Totalsystematicuncertaintiesareshownasboxes,whilestatisticaluncertaintiesareshownas verticalbars.

Fig. 7.Photon–jetpT-balancedistributions(1/Nγ)(dN/dxJγ)inPb+Pbevents (redcircles)witheachpanelshowingadifferentcentralityselectioncomparedtothatin ppevents(bluesquares).Thesepanelsshowresultsfor p^γ_T=158–200 GeV.Totalsystematicuncertaintiesareshownasboxes,whilestatisticaluncertaintiesareshownas verticalbars.

severalkey featuresofthe x_Jγ distribution,includingtheabsence ofavisiblepeak,andthemonotonicallyincreasingbehaviourwith decreasingxJγ.TheBDMPS-Zmodelpredictsasuppressionofthe yieldnearxJγ≈⁰.9 relativetowhatispredictedinppevents,con- sistentwiththetrendindata.However,itunderestimatestheyield atlow xJγ in both pp andPb + ^Pbcollisions. In the higher-pγ T interval, the Hybrid model and JEWEL successfully describe the reappearanceofalocalisedpeaknearxJγ ≈⁰.9.However,noneof themodelsconsidered heredescribe theincrease ofthejet yield at x_Jγ <0.5 above that observed in pp events. Additional comparisons between these data and theoretical calculations which are differential in both pγ

T and centrality will further constrain the description of the strongly coupled medium in these mod- els.

8. Conclusion

This Letter presents a study of photon–jet transverse momentum correlations for photons with 63.1< pγ

T <200 GeV in Pb+ ^Pb^collisions ^at√

sNN=⁵.02 TeV and ppcollisions at√ s= 5.02 TeV.Thedatawere recordedwiththe ATLASdetectoratthe LHC andcorrespond to integratedluminosities of 0.49 nb⁻¹ and 25 pb⁻¹,respectively. The data are correctedfor thepresence of combinatoricphoton–jetpairsandofdijetpairswhereoneofthe jetsismisidentiﬁedasaphoton.Themeasured quantitiesindata are fullycorrected for detectoreffects andreportedat theparticle level. Per-photon distributions of the jet-to-photon pT ratio, xJγ=^p^jetT /pγ

T,aremeasuredforpairswithanazimuthallybalanced conﬁguration,φ >7

π

/8. In pp events,thedata arewell repro- ducedbyeventgeneratorsormodelsthatdependonthem,butare

(11)

Fig. 8.Summaryof(left)themeanjet-to-photonpTratio xJγ

and(right)thetotalper-photonjetyieldR^γ,calculatedintheregionxJγ>0.5.Thevaluesarepresentedasa functionofthemeannumberofparticipatingnucleonsNpartintoppanels.Eachcolourandsymbolrepresentsadifferentp^γ_T interval,wherethelowestandhighestintervals aredisplacedhorizontallyforclarity.Thepointsplottedat Npart=2 correspondtoppcollisions.ThebottompanelsshowthedifferencebetweenthePb+Pbcentrality selectionandppcollisions.Boxesshowthetotalsystematicuncertaintywhiletheverticalbarsrepresentstatisticaluncertainties.

Fig. 9.Photon–jetpT-balancedistributions(1/Nγ)(dN/dxJγ)inppcollisionsfor(left)p^γ_T=63.1–79.6 GeV and(right)p^γ_T=100–158 GeV.Theunfoldedresultsarecompared withthetheoreticalcalculationsshownasdashedcolouredlines(seetext).Totalsystematicuncertaintiesareshownasboxes,whilestatisticaluncertaintiesareshownas verticalbars.

Fig. 10.Photon–jetpT-balancedistributions(1/Nγ)(dN/dxJγ)in0–10%Pb+Pbcollisionsfor(left)p^γ_T =63.1–79.6 GeV and(right) p^γ_T =100–158 GeV.Theunfolded resultsarecomparedwiththetheoreticalcalculationsshownasdashedcolouredlinesdenotingcentralvaluesorcolouredbandswhichcorrespondtoarangeoftheoretical parameters(seetext).Totalsystematicuncertaintiesareshownasboxes,whilestatisticaluncertaintiesareshownasverticalbars.