Elliptic flow of muons from heavy-flavour hadron decays at forward rapidity in Pb-Pb collisions at √sNN = 2.76 TeV

Contents lists available atScienceDirect

Physics

Letters

B

www.elsevier.com/locate/physletb

Elliptic

flow

of

muons

from

heavy-flavour

hadron

decays

at

forward

rapidity

in

Pb–Pb

collisions

at

√

s

_NN

=

2

.

76 TeV

.

ALICE

Collaboration



a

r

t

i

c

l

e

i

n

f

o

a

b

s

t

r

a

c

t

Articlehistory: Received15July2015

Receivedinrevisedform19November2015 Accepted20November2015

Availableonline2December2015 Editor: L.Rolandi

Keywords: LHC

ALICEexperiment Pb–Pbcollisions

Heavy-flavourdecaymuons Ellipticflow

The elliptic flow, v2, of muons from heavy-flavour hadron decays atforward rapidity (2.5< y<4)

is measured inPb–Pb collisions at√sNN=2.76 TeV with the ALICE detector atthe LHC. The scalar

product,two- andfour-particleQ cumulantsandLee–Yangzerosmethodsareused.Thedependenceof the v2ofmuonsfromheavy-flavourhadrondecaysonthecollisioncentrality,intherange0–40%,and

ontransversemomentum, pT,isstudiedintheinterval 3<pT<10 GeV/c.Apositive v2is observed

withthe scalarproductandtwo-particle Q cumulantsinsemi-central collisions(10–20%and 20–40% centrality classes) for the pT interval from 3 to about 5 GeV/c with a significance larger than 3

σ

,

basedonthecombinationofstatisticalandsystematicuncertainties.Thev2magnitudetendstodecrease

towardsmorecentralcollisionsandwithincreasingpT.Itbecomescompatiblewithzerointheinterval

6<pT<10 GeV/c.Theresultsarecomparedtomodelsdescribingtheinteractionofheavyquarksand

openheavy-flavourhadronswiththehigh-densitymediumformedinhigh-energyheavy-ioncollisions.

©2015CERNforthebenefitoftheALICECollaboration.PublishedbyElsevierB.V.Thisisanopen accessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).FundedbySCOAP3_.

1. Introduction

Experimentswith ultra-relativistic heavy-ion collisions aim at investigatingtheproperties ofstrongly-interacting matter atvery hightemperaturesandenergydensities.Quantum Chromodynam-ics (QCD) calculations on the lattice predict, under these con-ditions, the formation of a Quark–Gluon Plasma (QGP), where color confinement vanishes and chiral symmetry is partially re-stored[1–5].Heavyquarks(charmandbeauty)arecreatedin ini-tialhard-scatteringprocessesonatimescaleshorterthantheQGP formationtime.Subsequently,theyinteractwiththemedium con-stituentsviainelastic[6,7]andelastic[8–10]processes.Therefore, heavyquarksare regardedaseffectiveprobes oftheQGP proper-ties.

Heavy-quark energy loss due to in-medium interactions can be studied by means ofthe nuclear modification factor RAA,

de-finedastheratiooftheyieldofheavy-flavourparticles measured in nucleus–nucleus (AA) collisions to that observed in proton– proton (pp) collisions scaled by the number of binary nucleon– nucleoncollisions.ThePHENIXandSTARCollaborationsmeasured, incentralAu–Aucollisions at

√

sNN

=

200 GeV, astrong

suppres-sion corresponding to a RAA of about 0.2–0.3 for heavy-flavour

decay electrons at mid-rapidity ( y) and transverse momentum

pT

>

5 GeV

/

c[11–17].Asimilarsuppressionwasalsomeasuredby

 _{E-mailaddress:alice-publications@cern.ch.}

theSTARCollaborationformid-rapidityD0 _mesons[18].A

signif-icantsuppression wasalso observedbythePHENIX Collaboration at forwardrapidity formuons fromheavy-flavour hadron decays in central Cu–Cucollisions at

√

sNN

=

200 GeV [19]. At the LHC,

theALICECollaborationreportedasimilar effectincentralPb–Pb collisions at

√

sNN

=

2

.

76 TeV for Dmesons atmid-rapidity [20]

and muons from heavy-flavour hadron decays at forward rapid-ity[21]intheinterval 2

<

pT

<

16 GeV

/

c and4

<

pT

<

10 GeV

/

c,

respectively. The CMS Collaboration measured a significant sup-pression of non-prompt J

/ψ

from beauty-hadron decays in the interval 6

.

5

<

pT

<

30 GeV

/

c (3

<

pT

<

30 GeV

/

c) and

|

y

|

<

2

.

4

(1

.

6

<

|

y

|

<

2

.

4)[22,23]. Afirstmeasurementofnon-prompt J

/ψ

by the ALICECollaborationat mid-rapidity(

|

y

|

<

0

.

8) andin the interval4

.

5

<

pT

<

10 GeV

/

c hasbeenrecentlypublished[24].

Further insights into the QGP evolution and the in-medium interactions can be gained from the study of the azimuthal anisotropy of particles carrying heavy quarks which, in contrast to light quarks, have experienced the full system evolution. The study of azimuthal anisotropy is a field of intense experimental andtheoreticalinvestigations(see [25]andreferencestherein).In non-central collisions,the initialspatial anisotropy ofthe overlap region, elongated in the direction perpendicular to the reaction plane, defined by the beam axis and the impact parameter of thecollision,isconvertedintoananisotropy inmomentumspace throughrescatterings[26].Experimentally,thestudyoftheparticle azimuthalanisotropyisbasedonaFourierexpansionofazimuthal distributionsgivenby:

http://dx.doi.org/10.1016/j.physletb.2015.11.059

0370-2693/©2015CERNforthebenefitoftheALICECollaboration.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense (http://creativecommons.org/licenses/by/4.0/).FundedbySCOAP3_.

d2_N dpTd

ϕ

=

1 2

π

dN dpT



1

+

2 ∞



n=1 vn

(

pT

)

cos

[

n

(

ϕ

− 

n

)

]



,

(1)

where

ϕ

and pT are the particleazimuthal angle andtransverse

momentum,respectively.The Fouriercoefficients, vn,characterize theanisotropyofproducedparticlesand



nistheazimuthalangle of the initial-state symmetry plane for the nth harmonic, intro-ducedtoaccount fortheevent-by-eventfluctuationsoftheinitial nucleondensityprofile. Thesecond Fourier coefficient, v2, which

canalsobe expressedas v2

= 

cos

[

2

(

ϕ

− 

2

)

]

,isnamed elliptic

flow.

The v2 of heavy-flavourhadronsis expectedto provide

infor-mation on the collective expansion anddegree of thermalization ofheavy quarksinthemedium atlow pT (pT

<

2–3 GeV

/

c).The

participation of heavy quarks in the collective expansion is ex-pected to give a positive v2 [26]. Moving towards intermediate pT (3

<

pT

<

6 GeV

/

c),the v2 Fouriercoefficient isalsoexpected

tobe sensitiveto thepresenceofrecombinationprocessesinthe hadronizationofheavyquarks[27,28].AthighpT (pT

>

6 GeV

/

c),

thev2 measurementcanconstrainthepath-lengthdependenceof

thein-medium partonenergyloss, whichbecomes the dominant contributionto theazimuthal anisotropy andis alsopredictedto givea positive v2 [29,30],thuscomplementingthe RAA

measure-ment.

The PHENIX Collaboration reported a positive v2 of

heavy-flavour decay electrons at mid-rapidity in Au–Au collisions at

√

sNN

=

200 GeV, reaching a maximum value of about 0.15 at pT

=

1

.

5 GeV

/

c in semi-central collisions [14,15,31]. A similar

behavior was also observed by the STAR Collaboration [32]. Re-cently,av2valuesignificantlylargerthanzerowasmeasuredforD

mesonsatmid-rapidityinPb–Pbcollisionsat

√

sNN

=

2

.

76 TeV[33, 34].Acomplementarymeasurementatthesameenergy,provided bytheheavy-flavourdecaymuonellipticflowatforwardrapidity (2

.

5

<

y

<

4), is of great interest in order to provide new con-straints for models that implement the heavy-quark interactions withthemedium. Finally,themeasurement isalsoimportantfor the interpretation of the J

/ψ

elliptic flow results at forward ra-pidity[35]intermsofaregenerationproductionfromdeconfined charmquarksinthemedium.

InthisLetter, wepresentthemeasurementoftheellipticflow of muons from heavy-flavour hadron decays at forward rapidity (2

.

5

<

y

<

4) in Pb–Pb collisions at

√

sNN

=

2

.

76 TeV recorded

withthe ALICE detector.The elliptic flow is measured using dif-ferentmethods:scalarproduct[36],two- andfour-particle Q

cu-mulants [37,38] andLee–Yang zeros [39–41]. These methods ex-hibitdifferentsensitivitiestoflowfluctuationsandcorrelationsnot relatedtotheazimuthal asymmetryintheinitial geometry (non-floweffects).The v2 coefficientismeasuredasafunctionofpT in

theinterval 3

<

pT

<

10 GeV

/

c and inthree centralityclasses in

therange0–40%.Thecentralitydependenceofv2 ispresentedin

theinterval3

<

pT

<

5 GeV

/

c.

TheLetterisorganized asfollows.TheALICEdetector,withan emphasisonthemuonspectrometer,andthedatasampleare pre-sentedin Section 2.The analysisdetails, the methods forthe v2

measurement, the inclusive muon v2 determination, the

proce-dureforthesubtractionofthebackgroundofmuonsfromdecays oflight-flavourhadronsandthestudyofsystematicuncertainties, aredescribed inSection3.The v2 resultsformuonsfrom

heavy-flavourdecaysarepresentedinSection4.The v2 measurementin

semi-centralcollisionsaswellasthepublished RAAincentral

col-lisions are compared to model calculations in Section 5. Finally, conclusionsaregiveninSection6.

2. ALICEexperimentanddatasample

The ALICE detector is described in detail in [42,43]. The ap-paratus is composedof aset ofcentral barreldetectors (pseudo-rapidity coverage

|

η

|

<

0

.

9)located insideasolenoidmagnetthat generates a field of0.5 T parallelto the beamdirection, amuon spectrometer (

−

4

<

η

<

−

2

.

51) and a set of detectors forevent characterization andtriggering located in the forward and back-ward

η

regions.Themuonspectrometerconsistsofapassivefront absorber made of carbon, concrete and steel, a beam shield, a 3 T mdipole magnet,trackingchambers,a muonfilter(ironwall) andtrigger chambers. The muon trackingsystemis composed of five stations, each includingtwo planes ofcathod pad chambers, withthe thirdstationinside thedipole magnet. Themuon track-ing systemis completed by four trigger planes of resistiveplate chambers downstream of the iron wall, which absorbs hadrons thatpunchthroughthefrontabsorber,aswell assecondary parti-clesproducedinsideitandlowmomentummuons(p

<

4 GeV

/

c).

Twoscintillatorarrays(V0)coveringthepseudo-rapidity inter-vals

−

3

.

7

<

η

<

−

1

.

7 and 2

.

8

<

η

<

5

.

1 are used for triggering, forcollision centralitydeterminationandforbeam-induced back-ground rejection.The Zero DegreeCalorimeters (ZDC), located at 114 m fromthecentre ofthe detectoron both sides, can detect spectator protons and neutrons andare also used for the offline rejection ofbeam-inducedbackgroundandelectromagnetic inter-actions. The Silicon PixelDetector (SPD), that composes the two innermost layers of the Inner Tracking System (ITS), is used for theinteractionvertexreconstruction.TheTimeProjectionChamber (TPC), whichmeasurescharged-particletrackswithfullazimuthal coveragein

|

η

|

<

0

.

9,isusedinthisanalysisforthemeasurement ofthereferenceparticles(Section3.1).

TheresultspresentedinthisLetterareobtainedfromthedata sample recordedwithALICEduringthe2011Pb–Pbrun.Thedata were collected withaminimum-biastrigger requiringthe coinci-denceofsignalsinthetwoV0arraysinsynchronizationwiththe passage oftwo crossing bunches.In addition,the recordedevent sample was enriched with central and semi-central Pb–Pb colli-sionsbyapplyingthresholds,atthetriggerlevel,ontheV0signal amplitude.Thebeam-inducedbackground(beam–gasinteractions) was reduced by using the timing information from the V0 and ZDCdetectors.Furthermore,a minimalenergydepositintheZDC was requiredto rejectthe contributionfromelectromagneticPb– Pb interactions. Onlyevents witha reconstructed primary vertex within

±

10 cmfromthenominalpositionofthe interaction ver-texalongthebeamdirectionareanalyzed.ThePb–Pbcollisionsare classified accordingtotheir degree of centralityby means ofthe sum ofthe amplitudesof thesignals inthe V0detectorand the centrality classesare definedaspercentiles ofthe total hadronic Pb–Pb crosssection[44].The analysisiscarriedout inthree cen-trality classes: 0–10% (using the sample with trigger on central collisions),10–20%and20–40%(using thesamplewithtriggeron semi-central collisions).The analyzeddatasample corresponds to anintegratedluminosityof11.3 μb−1 inthe0–10%centralityclass andof3.5 μb−1 _{intheothertwocentralityclasses.}

3. Dataanalysis

The elliptic flow of muonsfrom heavy-flavourhadron decays,

vμ₂←HF,isobtainedfromthemeasurement oftheinclusivemuon ellipticflow,vμ₂,bysubtractingtheellipticflowofmuonsfrom

pri-1 _{IntheALICEreferenceframe,themuonspectrometercoversanegativeηrange} andconsequentlyanegativey range.Inthefollowing,giventhatthecolliding sys-temissymmetric,theresultsarepresentedwithapositivey notation.

marychargedpionandkaondecaysv₂μ←π,K(Sections3.1and3.4), as: vμ₂←HF

=

v μ 2

−

fμ←π,K

·

v μ←π,K 2 1

−

fμ←π,K

,

(2)

where f μ←π,K _{is the muon background fraction, definedas the}

ratiooftheyield ofmuonsfromprimary chargedpionandkaon decaystothatofinclusivemuons.Themeasurementofthevμ₂←HF

coefficientiscarriedoutintheinterval3

<

pT

<

10 GeV

/

c inorder

tolimitthesystematicuncertaintyonthesubtractionofthemuon backgroundcontribution.

3.1.Trackselection

Theselectioncriteriaforparticles ofinterest,muontracks,are similarto those usedin theprevious analyses ofpp collisions at

√

s

=

2

.

76 TeV and7TeVandPb–Pbcollisionsat

√

sNN

=

2

.

76 TeV [21,45].The tracks are requiredto be within thegeometrical ac-ceptance of the muon spectrometer, with

−

4

<

η

<

−

2

.

5 and 170◦

< θ

abs

<

178◦,where

θ

abs isthepolaranglemeasuredatthe

endoftheabsorber.Inordertoimprovethemuonidentification,a reconstructedtrackinthetrackingchambersisrequiredtomatch atracksegment inthe triggerchambers. Thisleads to avery ef-ficientrejectionofthebackgroundproduced by chargedhadrons, which are absorbed in the iron wall. Furthermore, a cut on the product p

·

DCA of the track momentum p anddistance of clos-estapproach(DCA)totheprimaryvertexisappliedtoremovethe beam-inducedbackgroundtracksandfaketrackscomingfromthe superpositionofseveralparticlescrossingthemuonspectrometer. Due to multiple scatteringin the front absorber, the DCA

distri-bution of tracks coming form the interaction vertex is expected to be described by a Gaussian function, its width being depen-dentontheabsorbermaterialandproportionalto1/p.Background trackshaveaverybroaddistributioninp

·

DCA andareeffectively rejected by a cut at 6

σ

, where

σ

is extracted from a Gaussian fit to the p

·

DCA distribution measured intwo intervalsof

θ

abs,

correspondingtodifferentmaterialsinthefrontabsorber.The rel-ative momentum resolution of reconstructed tracks varies from about1%to4%fortrackswithmomentumbetween20 GeV

/

c and

100 GeV

/

c.Afterthecutsareapplied,intheregion pT

>

3 GeV

/

c

theresidualbackgroundtoheavy-flavourdecaymuonsconsistsof muonsfrom decaysof primary charged pions andkaons2 andit amounts to 5–15%, depending on pT and on collision centrality

(Section3.4).

Themid-rapiditycharged-particletracksusedtodeterminethe flowvector Q



n orthegeneratingfunction(Section3.2)are called in the following reference particles. They are defined as tracks measured in the TPCin

|

η

|

<

0

.

8. These are required to have at least 70 associated space points out of the maximum of 159, a

χ

2 _{per degree of freedom (ndf) for the momentum fit in the}

range

χ

2

_/

_ndf

_<

_{2 and a transverse momentum value in the}

in-terval0

.

2

<

pT

<

5 GeV

/

c.Additionally,tracksarerejectediftheir

distanceof closest approachto the primary vertex islarger than 3 cmintheplanetransversetothebeamdirectionorinthe lon-gitudinaldirection.

3.2.Flowanalysismethods

The elliptic flow measurement is carried out using various methodsthat havedifferentsensitivities to flow fluctuationsand

2 _{Notethatthecontributionofmuonsfromsecondarylighthadrondecays} pro-ducedinsidethefrontabsorberisnegligibleforpT>3 GeV/c[45].

non-flow effects [46]. Flow fluctuationsare mainlydue to event-by-eventfluctuationsoftheinitial densityprofile,whilenon-flow effects correspond to correlations not related to the azimuthal anisotropy inthe initial state,such asresonance decays,jetsand Bose–Einsteincorrelationsbetweenidentical particles.It isworth mentioning that, inthe present analysis, mostof thesenon-flow effects arestrongly suppressedby introducing an

η

gapbetween reference particles andparticles of interest [47]. In thisanalysis, the scalarproduct [36],two- and four-particle Q cumulants [37, 38] andLee–Yang zeros [39–41] methods are employed. The de-scriptionofthesemethodswill belimitedtothefeaturesspecific to the present analysis. The following notations are introduced:

vμ₂(μ←HF)

{

SP

}

,referstothemeasurementusingthescalarproduct,

v₂μ(μ←HF)

{

2

}

and v₂μ(μ←HF)

{

4

}

correspond to the ones using the two-particle Q cumulants andfour-particle Q cumulants, while

v₂μ(μ←HF)

{

LYZ-Prod

}

and v₂μ(μ←HF)

{

LYZ-Sum

}

are obtained using Lee–Yang zeros with product and sum generating functions. The superscripts

μ

and

μ

←

HF refer to inclusivemuonsandmuons from heavy-flavourhadron decays, respectively. It is worth men-tioning that thesemethods are more accurate than the standard eventplane method,which yields a measurement lying between the event-averaged mean value and the root-mean-square value inthe presenceofflow fluctuations[48,49].Moreover, the multi-particlecorrelation methods(four-particle Q cumulantsandLee– Yang zeros)are less affected by non-flow correlationsthan two-particle correlation methods, but they cannot be used reliably when the muon flow magnitude is small and when the num-ber of muonsis small in the selected phase-spaceregion e.g. in centralandperipheralcollisions,respectively[37,39].Underthese conditions,thescalarproductandtwo-particlecumulantmethods providea v2 measurementinawidercentralityrange.

The scalarproduct method[36,48], derived fromthestandard eventplane technique [48], is basedon the measurement of the flow vector Q



n [36] computed fromreference particles. In order to determine the ellipticflow, the Q



2 vector in a givenevent is

expressedas:



Q2

=



_

N j=1 cos2

ϕ

j

,

N



j=1 sin2

ϕ

j



,

(3)

where

ϕ

jistheparticleazimuthalangleandN isthemultiplicity ofreferenceparticles.

Withthismethodthe2ndharmoniccoefficientisgivenby: v2

{

SP

} =

 

Q2

· 

u2,i

(

η

,

pT

)



2



 

Q₂A

· 

Q₂B



,

(4)

where the brackets in the numerator indicate the average over muonsat forwardrapidity,in all events.The vector Q2 is

calcu-latedfromEq.(3)andthevector



u2,i

= (

cos 2

ϕ

i

,

sin 2

ϕ

i

)

istheunit vectoroftheithmuon.Inthedenominator,eachsampleof refer-enceparticlesusedtocompute Q



2isdividedintotwosub-samples

ofsamemultiplicity insymmetrical

η

intervals,

−

0

.

8

<

η

<

−

0

.

5 and 0

.

5

<

η

<

0

.

8, separated by a

η

gap of one unit of pseudo-rapidity, labeled withthe superscripts A and B and the brackets correspondtotheaverageoverevents.

Thecumulanttechnique[37,38]isbasedonacumulant expan-sionofmulti-particleazimuthalcorrelations.Differentorder cumu-lantshavedifferentsensitivitiestoflowfluctuations.Inthepresent analysis, two- andfour-particlecumulantsareusedtoextractthe muonellipticflow. Theresultspresentedinthe followingare ob-tained from a direct calculation of multi-particle cumulants per-formed by usingthe Q -cumulanttechnique [38], whichis based on the moments of the magnitude of the flow vector Q



2. It is

worth mentioning that in this approach the cumulants are not biasedby the interferences between various harmonics. The ref-erenceellipticflowvaluesV2evaluatedfromthe2ndorder

cumu-lant c2

{

2

}

and4thorder cumulantc2

{

4

}

withreferenceparticles

are given by V2

{

2

}

=

√

c2

{

2

}

and V2

{

4

}

=

√

4

−

c2

{

4

}

,respectively.

Oncethereferenceellipticflowisestimated,themuonellipticflow withrespecttothereferenceellipticflowisobtainedfromthe2nd and4thordercumulantsaccordingto:

v2

{

2

} =

d2

{

2

}

V2

{

2

}

and v2

{

4

} =

d2

{

4

}

V2

{

4

}

3

,

(5)

where d2

{

2

}

andd2

{

4

}

are the 2nd and 4thorder cumulants of

selectedmuons[38].

The Lee–Yangzeros method [39–41] relieson correlations in-volving all particles in the event. This is the limit of cumulants whentheorderofcumulantsgoestoinfinity.Themethodisbased on the location ofthe zeros in thecomplex plane, of a generat-ing functionof azimuthal correlations, whichrelates theposition ofthefirstminimumofthegeneratingfunctiontothemagnitude ofthereferenceellipticflowV2 definedas:

V2

≡



_

M j=1 cos

[

2(

ϕ

j

− 

2

)

]

events

,

(6)

whereM isthemultiplicityofreferenceparticlesandtheaverage istakenover allevents.Forthispurpose,thefollowing complex-valuedgeneratingfunctionisevaluatedasafunctionofapositive realvariabler andfew,typicallyfive,equallyspacedreference an-gles

ϑ

(LYZ-Prodmethod):

Gϑ

(

ir

)

≡



M j=1

(1

+

ir cos

[

2(

ϕ

j

− ϑ)])

events

.

(7)

Thefirstpositive minimumof

|

Gϑ

₍

_ir

₎

_|

_,denotedasrϑ

0,allowsone

to estimate Vϑ

2, which can be written as V2ϑ

=

j01

/

r0ϑ, where j01

2

.

405 isthe firstrootofthe Besselfunction. Oncethefirst

minimum rϑ

0 is determined, the differential muon elliptic flow

is estimated with respect to the reference flow Vϑ

2 as detailed

in[41].Finally,theresultis averagedover all

ϑ

angles.An alter-nativeformofthegeneratingfunctionprovidedwiththeLYZ-Sum methodis: Gϑ

(

ir

)

≡



exp



ir M



j=1 cos

[

2(

ϕ

j

− ϑ)]



events

.

(8)

Theversionofthemethodinvolvingaproductfortheconstruction of the generation function (Eq. (7)) was designedto disentangle interferences betweendifferentharmonics, whichis not the case withthegeneratingfunction usinga sumoftheindividual refer-enceparticlecontributions.Both generatingfunctionsareusedin thisanalysis.

Notethat, for all methods, autocorrelationeffects are avoided because the particles (muons) used in the determination of the flowarenotincludedintheestimationofthereferenceflow.

3.3. Inclusivemuonellipticflow

The ellipticflow ofinclusivemuons, vμ₂,isstudied with two-particle correlation methods (scalar product and two-particle Q

cumulants)inthecentralityintervals0–10%,10–20%and20–40%. In the 20–40% centrality interval, the multi-particle correlation methods(four-particle Q cumulantsandLee–Yangzeros)arealso used.

Severalsourcesofsystematicuncertaintyaffectingthemuon el-liptic flow measurement are considered. Thesetake into account the changes dueto thevariations ofthe referenceparticle selec-tion criteriaasin[33,34,50],to allowusto check therobustness ofthevμ₂ measurement.Sincethecollisionimpactparameter dis-tribution could slightly depend on the observable used for the centralitydetermination,a systematicuncertaintyis estimatedby repeating the analysisusingthe numberofclustersin the outer-most layer of the SPD and the number of tracks in the TPC as centralityestimators,insteadoftheV0signalamplitude. The sys-tematicuncertaintyduetotheeffectofTPCtracksfromdifferent Pb–Pbcollisions piled-upinthesamerecordedeventisestimated byapplyingatightercuttoremoveoutliersinthemultiplicity dis-tribution ofreferenceparticles.Thisisdone byrequiringthat the centralityvaluesdeterminedusingtheV0signalamplitudeandthe numberofTPCtracksdonotdifferbymorethan5%.Anadditional systematic uncertaintyspecific to the scalar product isevaluated by varying the

η

gapbetweenthe two sub-events from1to 0.8

η

-units (see Eq. (4) and[36]). The various systematic uncertain-tiesareaddedinquadrature.Theytendtoincreasewithincreasing

pT (see Fig. 1).Asummaryofthesystematicuncertainties,inthe

interval3

<

pT

<

4

.

5 GeV

/

c,ispresentedinTable 1.

Fig. 1showsthe pT-differentialmuonellipticflow(vμ2) inthe

0–10%, 10–20% and 20–40% centrality classes as obtained using thevariousmethods.Thevaluesofvμ₂ slightlyincreasefrom cen-tral tosemi-central collisions andthiseffect ismorepronounced in the pT interval 3

<

pT

<

4

.

5 GeV

/

c. The two-particle

correla-tion methods(scalarproduct andtwo-particle Q cumulants)give consistent results over the whole pT range,indicating that these

methodshaveasimilarsensitivitytonon-floweffects3andin par-ticular to flow fluctuations. A similar agreement is found when comparingthemulti-particlecorrelation methods(four-particle Q

cumulantsandLee–Yangzeros)toeachother.Nosignificant differ-encebetweenthe vμ₂ resultsextractedwithLee–Yangzerosusing either thesum orproduct generatingfunction is seen, hence in-dicating thatinterferencesbetweenharmonicsarenegligible [51]. Moreover, four-particle Q cumulants give comparable results as Lee–Yang zeros.The four-particle Q cumulants andLee–Yang ze-rosareexpectedtobelessaffectedbynon-floweffectsthanscalar productortwo-particle Q cumulants[52].However,asmentioned non-flow effects are expected to be negligible, even with two-particle correlation techniques, due to the large

η

between ref-erenceparticlesandinclusivemuons.Finally,thecentralvaluesof

vμ₂ obtainedwithfour-particleQ cumulantsorLee–Yangzerosare systematicallysmallerthanwithtwo-particlecorrelationmethods, althoughcompatiblewithinuncertainties.Suchdifferencesmay in-dicatethatinitialfluctuationsplayaroleinthedevelopmentofthe finalmomentum-spaceanisotropy.

3.4. Muonbackgroundsubtraction

The subtraction of the muon background contribution to the measured vμ₂ requires an estimate of the elliptic flow of muons from charged pion and kaon decays, v₂μ←π,K, and of the back-ground fraction, f μ←π,K (see Eq. (2)). The determination of the vμ₂←π,K coefficient requires two steps. First, the pT- and

η

-differential v2 of charged particles measured in

|

η

|

<

2

.

5 by

the ATLAS Collaboration inPb–Pb collisions [53] andthe pT

dis-tributions of charged pions and kaons measured in

|

y

|

<

0

.

8 by

3 _{Notethat,inthisanalysis,mostnon-flowcorrelationsaresuppressed,evenwith} two-particlecorrelationmethodssincereferenceparticlesandinclusivemuonsare separatedbyatleast1.7 η-units.However,itisworthmentioningthatthemain differencebetweenthetwomethodsistheηgapbetweenthetwosub-samples usedtocomputeQ2(Eq.(4))whichalsoallowstopartlyremovenon-floweffects.

Table 1

Systematicuncertaintysourcesaffectingtheinclusivemuonellipticflowmeasurementinthe0–10%,10–20%and 20–40%centralityclassesfortheinterval3<pT<4.5 GeV/c.Theyaregivenasapercentageofthev2value.

vμ₂ analysis Source Systematic uncertainty (%)

0–10% 10–20% 20–40% vμ2{SP} Reference particles 3 1 3 Centrality selection 6 1 4 TPC pile-up 2 4 2 ηgap 13 1 1 vμ2{2} Reference particles 13 3 2 Centrality selection 14 3 6 TPC pile-up 8 1 4 vμ2{4} Reference particles 10 Centrality selection 1 TPC pile-up 1

vμ2{LYZ-Sum} Reference particles 4

Centrality selection 7

TPC pile-up 2

vμ2{LYZ-Prod} Reference particles 2

Centrality selection 8

TPC pile-up 2

Fig. 1. pT-differentialinclusivemuonv2in2.5<y<4 andvariouscentralityintervals,inPb–Pbcollisionsat√sNN=2.76 TeV.Thesymbolsareplacedatthecentreofthe

pTintervaland,forvisibility,thepointsfromtwo-particleQ cumulantsandLee–Yangzeroswithproductgeneratingfunctionareshiftedhorizontally.Theverticalerrorbars representthestatisticaluncertainty,thehorizontalerrorbarscorrespondtothewidthofthebin(notshownfortheshifteddatapoints)andtheopenboxesarethesystematic uncertainties.ThepTintervalsusedwiththeLee–Yangzerosmethodaredifferentwithrespecttotheothermethods.Upperpanels:resultsfromtwo-particlecorrelationflow methods(scalarproductandtwo-particle Q cumulants)inthe0–10%(left)and10–20%(right)centralityintervals.Lowerpanels:resultsinthe20–40%centralityinterval fromtwo-particlecorrelationflowmethods(scalarproductandtwo-particleQ cumulants)andfromfour-particleQ cumulants(left),andfromfour-particleQ cumulants andLee–Yangzeros(right).

the ALICE Collaboration in pp and Pb–Pb collisions [54,55] are extrapolated to forward rapidity. Then, the pT distributions of

muons from charged pion and kaon decays, needed to estimate

f μ←π,K_andvμ←π,K

2 ,aregeneratedaccordingtoasimulation

tak-inginto accountthe decaykinematicsandtheeffectof thefront absorber.

The pT- and

η

-differentialelliptic flowof chargedparticles in

|

η

|

<

2

.

5,vch

2 ,isextrapolatedtoforwardrapidityusing:

vch₂

(

pT

,

η

)

=

F

(

η

)

·

vch2

(

pT

,2

<

|

η

| <

2.5), (9)

where vch₂

(

pT

,

2

<

|

η

|

<

2

.

5

)

is themeasured charged-particle

the vch₂

(

pT

)

measured by the ATLAS Collaboration is affected by

statistical fluctuations, it is assumed that in the interval 10

<

pT

<

20 GeV

/

c,neededto simulatethedecaymuonsup to pT

=

10 GeV

/

c, vch

2 remains constant with a value given by the one

measured in the interval 10

<

pT

<

12 GeV

/

c. The extrapolation

factor F

(

η

)

iscalculated by parameterizingthe

η

-differential vch₂

measuredbytheATLASCollaborationinvarious pT intervalswith

asecondorderpolynomial.Intheinterval7

<

pT

<

20 GeV

/

c,the

ATLAS vch₂ doesnotshowa dependenceon

η

in

|

η

|

<

2

.

5. There-fore,for pT

>

7 GeV

/

c, F

(

η

)

iscomputedastheaveragebetween

aflatextrapolationfunctionandtheextrapolationfactorobtained withtheparabolicparameterizationin4

<

pT

<

7 GeV

/

c.

Themid-rapiditychargedpionandkaon pT distributions

mea-suredinPb–Pbcollisionsareextrapolatedtoforwardrapidityusing thesamestrategyasin[21]andsummarizedinthefollowing. As-sumingthatthenuclearmodificationfactorRπ_AA,K ofchargedpions andkaonsin Pb–Pbcollisions doesnot dependon rapidityup to

y

=

4[21,56],the pT distributionsofchargedpionsandkaons at

forwardrapiditycanbeexpressedas:

dN_PbPbπ,K dpTd y

= 

TAA

 ·

d

σ

ppπ,K dpTd y

· [

Rπ_AA,K

(

pT

)

]

y=0

,

(10)

where



TAA



is the average nuclear overlap function in

central-ity classesunder study, estimated asdescribed in [57]. The sys-tematic uncertainty introduced by the assumption on Rπ_AA,K will be discussed later. The rapidity extrapolation ofthe mid-rapidity pionandkaon pT-differentialcrosssectionsmeasuredinpp

colli-sions[21,58]isdoneaccordingto:

d2

σ

ppπ,K dpTd y

=



d2

σ

ppπ,K dpTd y



y=0

·

exp

−

y 2 2

σ

2 y



,

(11)

σ

y beingestimated fromMonte-Carlo eventgenerators (see [21] fordetails).

The elliptic flow of muons from charged pion and kaon de-cays, vμ₂←π,K, in 2

.

5

<

y

<

4 and in various centrality classes,4

is obtained by means of fast simulations using vch

2

(

η

,

pT

)

given

by Eq. (9) and charged pion and kaon pT distributions as

ob-tained from Eqs. (10)–(11). The absorber effect is accounted for by rejectingthe pionsandkaonsthat donot decaywithin a dis-tancecorrespondingtoone interactionlengthfromthebeginning of the absorber. The simulation was repeated twice, considering thatchargedparticlesareeitherallpionsorallkaons.

Thebackground fraction, f μ←π,K_{,is calculatedastheratio of}

thepT-differentialyieldofmuonsfromchargedpionandkaon

de-cays in 2

.

5

<

y

<

4 obtained in the simulation to the measured

pT-differentialyieldofinclusivemuons.

The systematic uncertainties affecting the estimated vμ₂←π,K

aresummarizedinTable 2.Theyoriginatefromi)themethodused tomeasurethecharged-particle vch₂ inATLAS,ii)the

η

andpT

ex-trapolation of vch₂ and iii) the treatment of the charged-particle

vch₂ inthe fastsimulation procedure. As theeventplane method was used for the vch₂ measurement in ATLAS, the results range betweenthemean (



vch₂



)and R.M.S.(



(

vch₂

)

2

_

_{) ofthe true v}ch 2

valuesdue tofluctuations, depending onthe eventplane resolu-tionwhichvarieswiththecollisioncentrality[49].Accordingtoa Monte-CarloGlaubermodel[49],theratio





v2₂

/

v2



isexpected 4 _Thevμ←π,K

2 ofmuonsfromchargedpionandkaondecaysinthe20–40% cen-tralityclassisthenobtainedfromthemeanofthecharged-particlev2in20–30% and30–40%centralityclasses,withanadditionalsystematicuncertaintyprovided bythedifferencewithrespecttotheresultsinthesetwocentralityclasses.

Table 2

Systematicuncertaintysourcesaffectingtheestimatedvμ2←π,Kfortheinterval3<

pT<10 GeV/c.Theyarestatedasapercentageofthev2 value.Thegivenrange reflectsthedependenceonthecollisioncentrality.

Source Systematic uncertainty (%)

Input vch 2 bias 9 vch 2 ηextrapolation 9–12 vch 2 high pTextrapolation 13–15 πand K in fast simulations <1

to varyfromabout1.06 to1.15.Consequently,a conservative sys-tematicuncertaintyof15%isappliedtoaccount forthisbiasand ispropagatedto v₂μ←π,K.Thesystematicuncertaintyduetothe

η

extrapolation of vch

2 is evaluated using severalfit functions(first

andthirdorderpolynomials,andGaussian function)intheregion

pT

<

7 GeV

/

c, andfor larger pT values an additional systematic

uncertaintyduetotheextrapolationprocedureisconsidered. The latter is determined by comparing the results obtainedwith the twoextrapolationfunctionsusedintheinterval pT

>

7 GeV

/

c.The

systematic uncertainty due to the assumption on vch

2 in the

re-gion pT

>

10 GeV

/

c is estimatedby varying vch2 between0 and

thevaluein10

<

pT

<

12 GeV

/

c inthefastsimulations.Such

un-certaintyaffectsmainlythehighpT region(pT

>

7 GeV

/

c).Finally,

the systematicuncertainty obtainedby treating charged particles separately aspionsandkaonsisfound tobenegligible.The vari-oussystematicuncertaintysourcesarepropagatedtotheestimated

vμ₂←π,Kandaddedinquadrature.

The systematic uncertainty on f μ←π,K_{, detailed in [21],}

in-cludestheuncertaintyonthegenerated pTdistributionsofmuons

from charged pionand kaon decays,and the uncertainty on the measured inclusivemuon pT distributions. The former originates

from theinput chargedpion andkaon distributions, therapidity extrapolation andthe absorber effect.The systematicuncertainty on the measured inclusive muon yields contains the systematic uncertaintyondetectorresponse,residualmis-alignmentand cen-trality dependenceof the efficiency.This givesa total systematic uncertainty on f μ←π,K of about 21% in the interval 3

<

pT

<

4

.

5 GeV

/

c withalmost nodependenceonthecollision centrality. Finally, asdone forthe measurement of theheavy-flavour decay muon RAA [21], the systematic uncertainty dueto the unknown

suppression of chargedparticles at forwardrapidity is calculated by varying f μ←π,K _{from 0 to two times the estimated value.}

This corresponds to a variation of R_AAπ,K

(

pT

)

at forward rapidity

from0uptotwotimes

[

Rπ_AA,K

(

pT

)

]

y=0.Thissystematicuncertainty

amountsto10–30%intheinterval3

<

pT

<

4

.

5 GeV

/

c,depending

onthecollisioncentralityandtheflowanalysismethod.

Fig. 2presentstheestimatedbackgroundellipticflow(vμ₂←π,K, left)andbackgroundfraction( f μ←π,K_{,right)asafunctionofp}

Tin

the0–10%,10–20%and20–40%centralityclasses.Theopenboxes representthesystematicuncertaintiespreviouslydiscussed,except forthesystematicuncertaintyduetotheunknownsuppressionof charged particles atforward rapidity which is treatedseparately. The estimated vμ₂←π,K and f μ←π,K _{decrease withincreasing p}

T.

Adecreasingtrendofthemagnitudeofvμ₂←π,Kfromsemi-central collisionstowardscentralcollisionsisalsoobserved.

Finally,thesystematicuncertaintyontheellipticflowofmuons fromheavy-flavourdecays,vμ₂←HF,containstwocontributions:the systematic uncertainties on vμ₂, v₂μ←π,K and f μ←π,K _propagated

accordingtothedefinitionofv₂μ←HF giveninEq.(2),andthe sys-tematic uncertainty due to the unknown suppression of charged particles atforward rapidity. The final systematic uncertainty on

Fig. 2. Estimatedbackgroundv2(vμ2←π,K,left)andbackgroundfraction( fμ←π,K,right)asafunctionofpTin2.5<y<4 andvariouscentralityintervals,inPb–Pbcollisions at√sNN=2.76 TeV.ThesymbolsareplacedatthecentreofthepTintervaland,forvisibility,thepointsforthecentralityclasses10–20%and20–40%areshiftedhorizontally. Thehorizontalerrorbarscorrespondtothewidthofthebin(notshownfortheshiftedvalues)andtheopenboxesarethesystematicuncertainties.Seethetextfordetails.

Fig. 3. pT-differentialellipticflowofmuonsfromheavy-flavourdecays,vμ2←HF,in2.5<y<4 andvariouscentralityintervals,inPb–Pbcollisionsat √_s

NN=2.76 TeV.The symbolsareplacedatthecentreofthepTintervaland,forvisibility,thepointsfromtwo-particleQ cumulantsandLee–Yangzeroswithproductgeneratingfunctionare shiftedhorizontally.ThemeaningofthesymbolsisthesameasinFig. 1.Thehorizontalerrorbarsarenotplottedforshifteddatapoints.ThepTintervalsusedwiththe Lee–Yangzerosmethodaredifferentwithrespecttotheothermethods.Upperpanels:resultsfromtwo-particlecorrelationflowmethods(scalarproductandtwo-particle Q cumulants)inthe0–10%(left)and10–20%(right)centralityintervals.Lowerpanels:resultsinthe20–40%centralityintervalfromtwo-particlecorrelationflowmethods (scalarproductandtwo-particle Q cumulants)andfromfour-particleQ cumulants(left),andfromfour-particleQ cumulantsandLee–Yangzeros(right).Seethetextfor details.

Itamountsto about12%–36%inthe interval3

<

pT

<

4

.

5 GeV

/

c,

dependingonthecollisioncentralityandtheflowanalysismethod.

4. Results

Fig. 3 presentsthe pT-differential ellipticflow ofmuons from

heavy-flavourhadrondecays, vμ₂←HF, calculatedwithEq.(2).The resultsareshownforthe0–10%(upper,left),10–20%(upper,right) and20–40%(bottom)centralityclassesusingthesameflow meth-ods as for the measurement of the inclusive muon elliptic flow

(Fig. 1). When comparing the results to those obtained for in-clusive muons (Fig. 1), one can notice that vμ₂←HF and vμ₂ are similarduetothesmallbackgroundfraction(5%to15%)inthe pT

interval 3–10 GeV

/

c. The differences betweenthe various meth-ods are similar to those discussed for the measurement of the inclusive muon vμ₂ i.e. i) scalar product and two-particle Q

cu-mulants give compatible results, ii) consistent results are also found with four-particle Q cumulants and Lee–Yang zeros, and iii) the vμ₂←HF values extracted from these multi-particle corre-lation methods are smaller, although still compatible within

un-Fig. 4. Ellipticflow ofmuonsfromheavy-flavourhadron decaysas afunctionofthe collisioncentralityin2.5<y<4 and3<pT<5 GeV/c,inPb–Pbcollisionsat √

sNN=2.76 TeV.Theresultsareobtainedwithscalarproductandtwo-particle Q cumulants.Verticalbars(openboxes)representthestatistical(systematic)uncertainty, thehorizontalerrorbarscorrespondtothewidthofthecentralitybin.Forvisibility,thepointsfromscalarproductareshiftedhorizontallyandthehorizontalerrorbarsare notplotted.

certainties, than the ones obtained with two-particle correlation methods. As mentioned in Section 3.3, such differences are ex-pectedif initial-statefluctuationsplay a role inthe development ofthefinalmomentum-spaceanisotropy.

Apositivevμ₂←HFisobservedatintermediatepTforthe20–40%

and10–20% centrality classeswith a significancelarger than 3

σ

when combining statistical and systematic uncertainties. In the 20–40% centrality class, the values of the significance in the in-terval 3

<

pT

<

4 GeV

/

c (4

<

pT

<

5

.

5 GeV

/

c) are4

σ

(3

.

2

σ

) and

4

.

3

σ

(3

.

8

σ

) with scalar product and two-particle Q cumulants,

respectively. In the 10–20% centrality class and in the interval 3

<

pT

<

4

.

5 GeV

/

c, the valuesof the significancecorrespond to

4

.

4

σ

bothwithscalarproductandtwo-particle Q cumulants.This behavior resultsfromtheinterplay betweenthesignificant inter-actionofheavyquarkswiththeexpandingmedium andthe path-length dependenceofin-medium partonenergyloss [29,30].The

vμ₂←HF ofmuonsfromheavy-flavourhadrondecaysdecreaseswith increasingpTandbecomescompatiblewithzerointhehighpT

re-gion.

Fig. 4shows the centrality dependenceof the ellipticflow of muonsfromheavy-flavourhadrondecaysintheinterval3

<

pT

<

5 GeV

/

c.Itisinvestigatedwithscalarproductandtwo-particle Q

cumulants,whichcanbeappliedinawiderevent-multiplicity(i.e. centrality)intervalcomparedtomulti-particlecorrelationmethods. Asignificant decrease ofthe v2 magnitude towardscentral

colli-sionsisobserved.Thisisexpectedfromthedecreaseoftheinitial spatialanisotropyfromsemi-centraltocentralcollisions.

ALICEhas measured the ellipticflow of prompt Dmesons in

|

y

|

<

0

.

8 in three centrality classes in the interval 0–50% with varioustwo-particlecorrelationmethods[33,34].Similartrendsas thosereportedhereformuonsfromheavy-flavourdecaysare ob-served,althoughindifferentpTandrapidityintervals.Inparticular,

apositive v2wasobservedforDmesonsinsemi-centralcollisions

in2

<

pT

<

6 GeV

/

c withasignificanceof5

.

7

σ

.

Thepositive ellipticflow ofmuonsfromheavy-flavourhadron decayshasbeenobservedina pTintervalfrom3toabout5 GeV/c

where the charm contribution is expected to be dominant with respect to the beautycomponent according to perturbative QCD calculations[21].Thismeasurementsupportstheinterpretationof theJ

/ψ

positive v2 atforwardrapidity [35]interms ofa

signifi-cantcontributiontoJ

/ψ

productionfromrecombinationofflowing charmquarksinthedeconfinedmedium.

5. Comparisonwithmodels

The resultspresented in this publication may constrain mod-els describing theinteractions ofheavy quarks withthemedium via elastic (collisional) and inelastic (radiative) processes, and in particular the partonenergyloss dependenceon the path-length withinthemedium.

The ellipticflow coefficient and the nuclear modification fac-tor of muons from heavy-flavour hadron decays [21] are com-paredtothefollowingthreemodels.TheMC@sHQ

+

EPOS trans-port model [59] treats the propagation of heavy quarks in the medium including collisional and radiative energy loss, within a 3

+

1 dimensional fluid dynamical expansion based on the EPOS model [60,61]. The hadronization of heavy quarks takes place at the transition temperature via recombination atlow pT

and fragmentation at intermediate and high pT. The final-state

hadronic interactions are not included in the model. TAMU [62]

is a transport model including only collisional processes via the Langevin equation. The hydrodynamical expansion is constrained by pT spectraandelliptic flow dataoflight-flavour hadrons. The

hadronization is modeled including a component of recombina-tion of heavy quarks with light-flavourhadrons in the QGP. The diffusionofheavy-flavourmesonsinthehadronicphaseisalso in-cluded.BAMPS[63–65]isapartonictransportmodelbasedonthe Boltzmann approachto multi-partonscatterings. Itincludes colli-sionalprocesseswitharunningstrongcouplingconstant.Thelack of radiative contributions is accounted for by scaling the binary cross section witha correction factor, tuned to describe the nu-clearmodificationfactorandellipticflowresultsatRHICenergies. Vacuumfragmentationfunctionsareusedforthehadronization.

Fig. 5 shows a comparison of the three models with the measurement of the pT-differential elliptic flow of muons from

heavy-flavour hadron decays in the 20–40% centrality class (up-per panel) and of the pT-differential nuclear modification factor

of muons from heavy-flavour hadron decays in the 0–10% cen-tralityclass [21] (lower panel).Inthe interval 3

<

pT

<

5 GeV

/

c,

theBAMPS modeldescribesthe vμ₂←HF datawithin uncertainties, while the TAMU andMC@sHQ

+

EPOS models give vμ₂←HF values lower than thedata. The three models describe the vμ₂←HF data athigher pT,althoughthesizeable experimentaluncertainties

af-fectthesignificanceofthecomparison.TheBAMPSmodeltendsto slightly underestimate the RAA of muons fromheavy-flavour

Fig. 5. Upperpanel:pT-differentialellipticflowofmuonsfromheavy-flavourhadron decaysin2.5<y<4,inPb–Pb collisions at √sNN=2.76 TeV for the central-ityclass20–40% compared tovarious transportmodel predictions:MC@sHQ + EPOS[59–61],TAMU[62]andBAMPS [63–65].TheTAMU modelis shownwith atheoretical uncertaintyband.Lowerpanel: pT-differential RAA ofmuonsfrom heavy-flavourhadrondecaysforthecentralityclass0–10%from[21]comparedto thesamemodelsasforvμ2←HF.

model tends to overestimate it. The TAMU model describes the

RAA measurement over the entire pT interval within

uncertain-ties. These comparisons indicate that it is challenging to simul-taneouslydescribethestrongsuppressionofhigh-pT muonsfrom

heavy-flavourhadrondecaysincentralcollisionsandtheazimuthal anisotropy in semi-central collisions. Similar trends are also ob-servedinthemid-rapidityregionfromthecomparisonoftheRAA

andv2ofDmesonswithmodelcalculations[34]. 6. Conclusions

Insummary,wehavereportedonameasurementoftheelliptic flowofmuonsfromheavy-flavourhadrondecaysatforward rapid-ityincentralandsemi-centralPb–Pbcollisionsat

√

sNN

=

2

.

76 TeV

withtheALICEdetectorattheLHC.

Measurements have been carried out using several methods whichexhibit differentsensitivity toinitial-statefluctuations and non-flow correlations. The systematiccomparison of scalar prod-uct,two- andfour-particleQ cumulantsandLee–Yangzeroshelps inunderstandingtheprocesses thatbuild upthe observed differ-encesbetweentwo-particlecorrelationmethodsandmulti-particle correlationmethodsandsuggeststhatflowfluctuationsare signif-icant.

Themagnitudeoftheellipticflowofmuonsfromheavy-flavour hadrondecaysincreasesfromcentraltosemi-centralcollisionsand

decreases with increasing pT, becoming compatiblewith zero at

highpT. Theresultsindicateapositiveellipticflowwiththescalar

product and two-particle Q cumulants in semi-central collisions (10–20%and20–40%centralityclasses)forthe pT intervalfrom3

to about5 GeV/c with asignificance larger than 3

σ

. The elliptic flow in semi-central collisions and the previously published nu-clear modification factorin the 10% most central collisions were compared with transport model calculations. These comparisons show that a simultaneous description of RAA and v2 over the

whole pTintervalremainsachallenge.Theresultsreportedinthis

Letter invariouscentralityclassesmayprovidefurther important constraintstothemodels.

Acknowledgements

The ALICE Collaboration would like to thank all its engineers andtechniciansfortheir invaluablecontributions tothe construc-tionoftheexperimentandtheCERNacceleratorteamsforthe out-standingperformanceoftheLHCcomplex.TheALICECollaboration gratefully acknowledges the resources and support provided by all Gridcentres andtheWorldwide LHC ComputingGrid (WLCG) collaboration. The ALICE Collaboration acknowledges the follow-ing funding agencies for their support in building and running theALICEdetector:StateCommittee ofScience,WorldFederation of Scientists (WFS)and SwissFonds Kidagan, Armenia; Conselho Nacional de DesenvolvimentoCientífico e Tecnológico (CNPq), Fi-nanciadorade Estudos eProjetos(FINEP),Fundação de Amparoà Pesquisa do Estado de São Paulo (FAPESP); National Natural Sci-enceFoundation ofChina (NSFC), theChinese Ministryof Educa-tion(CMOE)andtheMinistryofScienceandTechnologyofChina (MSTC); Ministry of Education andYouth ofthe Czech Republic; Danish Natural Science Research Council, the Carlsberg Founda-tion andthe DanishNationalResearch Foundation;TheEuropean ResearchCouncilundertheEuropeanCommunity’sSeventh Frame-work Programme; Helsinki Institute of Physics andthe Academy of Finland; French CNRS-IN2P3, the ‘Region Pays de Loire’, ‘Re-gion Alsace’, ‘Region Auvergne’ and CEA, France; German Bun-desministerium für Bildung, Wissenschaft, Forschung und Tech-nologie(BMBF)andtheHelmholtzAssociation;GeneralSecretariat forResearchandTechnology,MinistryofDevelopment,Greece; Na-tionalResearch,DevelopmentandInnovationOffice(NKFIH), Hun-gary;DepartmentofAtomicEnergyandDepartmentofScienceand TechnologyoftheGovernmentofIndia;IstitutoNazionalediFisica Nucleare(INFN) and CentroFermi –Museo Storico dellaFisica e CentroStudieRicerche“EnricoFermi”,Italy;JapanSocietyforthe PromotionofScience(JSPS)KAKENHIandMEXT,Japan;Joint Insti-tuteforNuclearResearch,Dubna;NationalResearchFoundationof Korea(NRF); ConsejoNacionaldeCiencayTecnologia(CONACYT), Direccion General de Asuntos del Personal Academico (DGAPA), México,AmeriqueLatineFormationacademique–European Com-mission (ALFA-EC) and the EPLANET Program (European Particle PhysicsLatinAmericanNetwork);StichtingvoorFundamenteel On-derzoekder Materie(FOM)andtheNederlandseOrganisatievoor WetenschappelijkOnderzoek(NWO),Netherlands; Research Coun-cilof Norway(NFR); NationalScienceCentre, Poland;Ministryof NationalEducation/InstituteforAtomicPhysicsandNational Coun-cil of Scientific Research in Higher Education (CNCSI-UEFISCDI), Romania;MinistryofEducationandScienceofRussianFederation, Russian Academy of Sciences, Russian Federal Agency of Atomic Energy, Russian Federal Agency for Science and Innovations and The RussianFoundation forBasicResearch;Ministry ofEducation ofSlovakia; Departmentof ScienceandTechnology, SouthAfrica; Centro de Investigaciones Energeticas, Medioambientales y Tec-nologicas (CIEMAT), E-Infrastructure shared between Europe and Latin America (EELA), Ministerio de Economía y Competitividad

(MINECO) of Spain, Xunta de Galicia (Consellería de Educación), Centro de Aplicaciones Tecnológicas y Desarrollo Nuclear (CEA-DEN), Cubaenergía, Cuba, and IAEA (International Atomic Energy Agency);SwedishResearchCouncil(VR)andKnut& Alice Wallen-berg Foundation (KAW); Ukraine Ministry of Education and Sci-ence; United Kingdom Science and Technology Facilities Council (STFC);TheUnitedStatesDepartmentofEnergy,theUnitedStates NationalScience Foundation, the State of Texas,and theState of Ohio; Ministry of Science, Education and Sports of Croatia and Unity throughKnowledge Fund, Croatia; Council ofScientific and IndustrialResearch(CSIR),NewDelhi,India;PontificiaUniversidad CatólicadelPerú.

References

[1]F. Karsch, Lattice simulations of the thermodynamics of strongly interacting elementary particles and the exploration of new phases of matter in rela-tivistic heavy ion collisions, J. Phys. Conf. Ser. 46 (2006) 122–131, arXiv:hep-lat/0608003.

[2]Wuppertal-Budapest Collaboration, S. Borsanyi, et al., Is there still any Tc mys-tery in lattice QCD? Results with physical masses in the continuum limit III, J. High Energy Phys. 1009 (2010) 073, arXiv:1005.3508 [hep-lat].

[3]S. Borsanyi, G. Endrodi, Z. Fodor, A. Jakovac, S.D. Katz, et al., The QCD equa-tion of state with dynamical quarks, J. High Energy Phys. 1011 (2010) 077, arXiv:1007.2580 [hep-lat].

[4]A. Bazavov, T. Bhattacharya, M. Cheng, C. DeTar, H. Ding, et al., The chiral and deconfinement aspects of the QCD transition, Phys. Rev. D 85 (2012) 054503, arXiv:1111.1710 [hep-lat].

[5]P. Petreczky, Review of recent highlights in lattice calculations at finite tem-perature and finite density, PoS ConfinementX (2012) 028, arXiv:1301.6188 [hep-lat].

[6]M. Gyulassy, M. Plumer, Jet quenching in dense matter, Phys. Lett. B 243 (1990) 432–438.

[7]R. Baier, Y.L. Dokshitzer, A.H. Mueller, S. Peigne, D. Schiff, Radiative energy loss and pTbroadening of high-energy partons in nuclei, Nucl. Phys. B 484 (1997)

265–282, arXiv:hep-ph/9608322.

[8]M.H. Thoma, M. Gyulassy, Quark damping and energy loss in the high temper-ature QCD, Nucl. Phys. B 351 (1991) 491–506.

[9]E. Braaten, M.H. Thoma, Energy loss of a heavy fermion in a hot plasma, Phys. Rev. D 44 (1991) 1298–1310.

[10]E. Braaten, M.H. Thoma, Energy loss of a heavy quark in the quark–gluon plasma, Phys. Rev. D 44 (1991) 2625–2630.

[11]PHENIX Collaboration, K. Adcox, et al., Measurement of single electrons and implications for charm production in Au +Au collisions at √sNN=130 GeV,

Phys. Rev. Lett. 88 (2002) 192303, arXiv:nucl-ex/0202002.

[12]PHENIX Collaboration, S. Adler, et al., Centrality dependence of charm pro-duction from single electrons measurement in Au +Au collisions at √sNN=

200 GeV, Phys. Rev. Lett. 94 (2005) 082301, arXiv:nucl-ex/0409028.

[13]PHENIX Collaboration, S. Adler, et al., Nuclear modification of electron spectra and implications for heavy quark energy loss in Au +Au collisions at √sNN=

200 GeV, Phys. Rev. Lett. 96 (2006) 032301, arXiv:nucl-ex/0510047.

[14]PHENIX Collaboration, A. Adare, et al., Energy loss and flow of heavy quarks in Au +Au collisions at √sNN=200 GeV, Phys. Rev. Lett. 98 (2007) 172301, arXiv:nucl-ex/0611018.

[15]PHENIX Collaboration, A. Adare, et al., Heavy quark production in p +p and energy loss and flow of heavy quarks in Au +Au collisions at √sNN=200 GeV, Phys. Rev. C 84 (2011) 044905, arXiv:1005.1627 [nucl-ex].

[16]PHENIX Collaboration, A. Adare, et al., System-size dependence of open-heavy-flavor production in nucleus–nucleus collisions at √sNN=200 GeV, Phys. Rev. C 90 (3) (2014) 034903, arXiv:1310.8286 [nucl-ex].

[17]STAR Collaboration, B. Abelev, et al., Transverse momentum and centrality de-pendence of high-pTnon-photonic electron suppression in Au +Au collisions

at √sN N=200 GeV, Phys. Rev. Lett. 98 (2007) 192301, arXiv:nucl-ex/0607012; STAR Collaboration, B. Abelev, et al., Phys. Rev. Lett. 106 (2011) 159902 (Erra-tum).

[18]STAR Collaboration, L. Adamczyk, et al., Observation of D0_{meson nuclear}

mod-ifications in Au +Au collisions at √sNN=200 GeV, Phys. Rev. Lett. 113 (14) (2014) 142301, arXiv:1404.6185 [nucl-ex].

[19]PHENIX Collaboration, A. Adare, et al., Nuclear-modification factor for open-heavy-flavor production at forward rapidity in Cu +Cu collisions at √sNN=

200 GeV, Phys. Rev. C 86 (2012) 024909, arXiv:1204.0754 [nucl-ex]. [20]ALICE Collaboration, B. Abelev, et al., Suppression of high transverse

momen-tum D mesons in central Pb–Pb collisions at √sNN=2.76 TeV, J. High Energy

Phys. 1209 (2012) 112, arXiv:1203.2160 [nucl-ex].

[21]ALICE Collaboration, B. Abelev, et al., Production of muons from heavy flavour decays at forward rapidity in pp and Pb–Pb collisions at √sNN=2.76 TeV,

Phys. Rev. Lett. 109 (2012) 112301, arXiv:1205.6443 [hep-ex].

[22] CMSCollaboration,J/ψ resultsfromCMSinPb–Pbcollisions,with150 μb−1 data,Tech.Rep.CMS-PAS-HIN-12-014,CERN,Geneva,2012,https://cds.cern.ch/ record/1472735.

[23]CMS Collaboration, S. Chatrchyan, et al., Suppression of non-prompt J/ψ,

prompt J/ψ, and ϒ(1S)in Pb–Pb collisions at √sNN=2.76 TeV, J. High

En-ergy Phys. 1205 (2012) 063, arXiv:1201.5069 [nucl-ex].

[24]ALICE Collaboration, J. Adam, et al., Inclusive, prompt and non-prompt J/ψ

production at mid-rapidity in Pb–Pb collisions at √sNN=2.76 TeV, arXiv: 1504.07151 [nucl-ex].

[25]S.A. Voloshin, A.M. Poskanzer, R. Snellings, in: Relativistic Heavy Ion Physics, in: Landolt–Boernstein New Series, vol. 1/23, Springer-Verlag, 2010, p. 5. [26]J.-Y. Ollitrault, Anisotropy as a signature of transverse collective flow, Phys. Rev.

D 46 (1992) 229–245.

[27]V. Greco, C. Ko, R. Rapp, Quark coalescence for charmed mesons in ultrarel-ativistic heavy ion collisions, Phys. Lett. B 595 (2004) 202–208, arXiv:nucl-th/0312100.

[28]A. Andronic, P. Braun-Munzinger, K. Redlich, J. Stachel, Statistical hadronization of charm in heavy ion collisions at SPS, RHIC and LHC, Phys. Lett. B 571 (2003) 36–44, arXiv:nucl-th/0303036.

[29]M. Gyulassy, I. Vitev, X. Wang, High pT azimuthal asymmetry in noncentral

A+A at RHIC, Phys. Rev. Lett. 86 (2001) 2537–2540, arXiv:nucl-th/0012092. [30]E. Shuryak, The azimuthal asymmetry at large pTseem to be too large for a

‘jet quenching’, Phys. Rev. C 66 (2002) 027902, arXiv:nucl-th/0112042. [31]PHENIX Collaboration, S. Adler, et al., Measurement of single electron event

anisotropy in Au +Au collisions at √sNN=200 GeV, Phys. Rev. C 72 (2005) 024901, arXiv:nucl-ex/0502009.

[32]STAR Collaboration, L. Adamczyk, et al., Elliptic flow of non-photonic electrons in Au +Au collisions at √sNN=200, 62.4 and 39 GeV, arXiv:1405.6348

[hep-ex].

[33]ALICE Collaboration, B. Abelev, et al., D meson elliptic flow in non-central Pb– Pb collisions at √sNN=2.76 TeV, Phys. Rev. Lett. 111 (2013) 102301, arXiv:

1305.2707 [nucl-ex].

[34]ALICE Collaboration, B.B. Abelev, et al., Azimuthal anisotropy of D meson pro-duction in Pb–Pb collisions at √sNN=2.76 TeV, Phys. Rev. C 90 (3) (2014)

034904, arXiv:1405.2001 [nucl-ex].

[35]ALICE Collaboration, E. Abbas, et al., J_√ /ψ elliptic flow in Pb–Pb collisions at

sNN=2.76 TeV, Phys. Rev. Lett. 111 (2013) 162301, arXiv:1303.5880

[nucl-ex].

[36]STAR Collaboration, C. Adler, et al., Elliptic flow from two and four particle correlations in Au +Au collisions at √sNN=130 GeV, Phys. Rev. C 66 (2002) 034904, arXiv:nucl-ex/0206001.

[37]N. Borghini, P.M. Dinh, J.-Y. Ollitrault, Flow analysis from multiparticle az-imuthal correlations, Phys. Rev. C 64 (2001) 054901, arXiv:nucl-th/0105040. [38]A. Bilandzic, R. Snellings, S. Voloshin, Flow analysis with cumulants: direct

cal-culations, Phys. Rev. C 83 (2011) 044913, arXiv:1010.0233 [nucl-ex].

[39]R. Bhalerao, N. Borghini, J. Ollitrault, Analysis of anisotropic flow with Lee–Yang zeroes, Nucl. Phys. A 727 (2003) 373–426, arXiv:nucl-th/0310016.

[40]R. Bhalerao, N. Borghini, J. Ollitrault, Genuine collective flow from Lee–Yang zeroes, Phys. Lett. B 580 (2004) 157–162, arXiv:nucl-th/0307018.

[41]N. Borghini, R. Bhalerao, J. Ollitrault, Anisotropic flow from Lee–Yang zeroes: a practical guide, J. Phys. G 30 (2004) S1213–S1216, arXiv:nucl-th/0402053. [42]ALICE Collaboration, K. Aamodt, et al., The ALICE experiment at the CERN LHC,

J. Instrum. 3 (2008), S08002.

[43]ALICE Collaboration, B.B. Abelev, et al., Performance of the ALICE experiment at the CERN LHC, Int. J. Mod. Phys. A 29 (2014) 1430044, arXiv:1402.4476 [nucl-ex].

[44]ALICE Collaboration, B. Abelev, et al., Centrality dependence of π, K, p produc-tion in Pb–Pb collisions at √sNN=2.76 TeV, Phys. Rev. C 88 (2013) 044910,

arXiv:1303.0737 [hep-ex].

[45]ALICE Collaboration, B. Abelev, et al., Heavy flavour decay muon production at forward rapidity in proton–proton collisions at √s=7 TeV, Phys. Lett. B 708 (2012) 265–275, arXiv:1201.3791 [hep-ex].

[46]ALICE Collaboration, K. Aamodt, et al., Higher harmonic anisotropic flow mea-surements of charged particles in Pb–Pb collisions at √sNN=2.76 TeV, Phys.

Rev. Lett. 107 (2011) 032301, arXiv:1105.3865 [nucl-ex].

[47]M. Luzum, Collective flow and long-range correlations in relativistic heavy ion collisions, Phys. Lett. B 696 (2011) 499–504, arXiv:1011.5773 [nucl-th]. [48]A.M. Poskanzer, S. Voloshin, Methods for analyzing anisotropic flow in

rel-ativistic nuclear collisions, Phys. Rev. C 58 (1998) 1671–1678, arXiv:nucl-ex/9805001.

[49]M. Luzum, J.-Y. Ollitrault, Eliminating experimental bias in anisotropic-flow measurements of high-energy nuclear collisions, Phys. Rev. C 87 (4) (2013) 044907, arXiv:1209.2323 [nucl-ex].

[50]ALICE Collaboration, B. Abelev, et al., Anisotropic flow of charged hadrons, pi-ons and (anti-)protpi-ons measured at high transverse momentum in Pb–Pb col-lisions at √sNN=2.76 TeV, Phys. Lett. B 719 (2013) 18–28, arXiv:1205.5761

[nucl-ex].

[51]STAR Collaboration, B. Abelev, et al., Centrality dependence of charged hadron and strange hadron elliptic flow from √sNN=200 GeV Au +Au collisions, Phys. Rev. C 77 (2008) 054901, arXiv:0801.3466 [nucl-ex].