Search for collectivity with azimuthal J/ψ -hadron correlations in high multiplicity p–Pb collisions at √sNN = 5.02 and 8.16 TeV

This is an electronic reprint of the original article. This reprint may differ from the original in pagination and typographic detail. Author(s): Title: Year: Version: Please cite the original version: All material supplied via JYX is protected by copyright and other intellectual property rights, and duplication or sale of all or part of any of the repository collections is not permitted, except that material may be duplicated by you for your research use or educational purposes in electronic or print form. You must obtain permission for any other use. Electronic or print copies may not be offered, whether for sale or otherwise to anyone who is not an authorised user. Search for collectivity with azimuthal J/ψ -hadron correlations in high multiplicity p–Pb collisions at √sNN = 5.02 and 8.16 TeV ALICE Collaboration ALICE Collaboration. (2018). Search for collectivity with azimuthal J/ψ -hadron correlations in high multiplicity p–Pb collisions at √sNN = 5.02 and 8.16 TeV. Physics Letters B, 780, 7-20. https://doi.org/10.1016/j.physletb.2018.02.039 2018 Physics Letters B 780 (2018) 7–20 Contents lists available at ScienceDirect Physics Letters B www.elsevier.com/locate/physletb Search for collectivity with azimuthal J/ψ-hadron correlations in high multiplicity p–Pb collisions at √sNN =5.02 and 8.16 TeV .ALICE Collaboration a r t i c l e i n f o a b s t r a c t Article history: Received 27 September 2017 Received in revised form 14 February 2018 Accepted 14 February 2018 Available online xxxx Editor: M. Doser We present a measurement of azimuthal correlations between inclusive J/ψand charged hadrons in p–Pb collisions recorded with the ALICE detector at the CERN LHC. The J/ψare reconstructed at forward (pgoing, 2.03 <y <3.53) and backward (Pb-going, −4.46 <y <−2.96) rapidity via their μ+μ−decay channel, while the charged hadrons are reconstructed at mid-rapidity (|η| <1.8). The correlations are expressed in terms of associated charged-hadron yields per J/ψtrigger. A rapidity gap of at least 1.5 units is required between the trigger J/ψand the associated charged hadrons. Possible correlations due to collective effects are assessed by subtracting the associated per-trigger yields in the low-multiplicity collisions from those in the high-multiplicity collisions. After the subtraction, we observe a strong indication of remaining symmetric structures at ϕ≈0and ϕ≈π, similar to those previously found in two-particle correlations at middle and forward rapidity. The corresponding second-order Fourier coefficient (v2) in the transverse momentum interval between 3 and 6 GeV/cis found to be positive with a significance of about 5σ. The obtained results are similar to the J/ψv2coefficients measured in Pb–Pb collisions at √sNN =5.02 TeV, suggesting a common mechanism at the origin of the J/ψv2. ©2018 The Author. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Funded by SCOAP3. 1. Introduction The measurement of angular correlations between particles produced in hadron and nucleus collisions is a powerful tool to study the particle production mechanisms. Usually the two-particle correlation function is expressed in terms of differences in the azimuthal angle (ϕ) and pseudorapidity (η) of the emitted particles. In minimum-bias proton–proton (pp) collisions, the dominant structures in the correlation function are a near-side peak at (ϕ, η) ≈(0, 0)and an away-side ridge located at ϕ≈π and elongated in η[1]. The near-side peak originates from jet fragmentation, resonance decays and femtoscopic correlations. The away-side ridge results from fragmentation of recoil jets. In collisions of heavy ions, the two-particle correlation function exhibits additional long-range structures elongated in η[2]. These structures are usually interpreted as signatures of collective particle flow produced during the hydrodynamic evolution of the fireball. They are analyzed in terms of the Fourier coefficients of the relative angle distributions. Assuming factorization, these coefficients are then related to the Fourier coefficients (vn) of the particle azimuthal distribution relative to the common symmetry plane of the colliding nuclei’s overlap area. E-mail address: alice -publications @cern .ch. The discovery of a near-side ridge in high-multiplicity pp [3] and p–Pb [4] collisions has increased the interest in two-particle angular correlations in small collision systems. These discoveries were followed by the observation that the near-side ridge in p–Pb collisions is accompanied by an away-side one [5,6]. Long-range structures have also been reported in two-particle correlations in d–Au collisions at RHIC [7,8]. Further studies using multi-particle correlations have proven that the observed long-range correlations are of a collective origin [9–11]. Moreover, the transversemomentum and particle-mass dependencies of the vncoefficients in p–Pb collisions have been found to be similar to those measured in A–A collisions, suggesting a common hydrodynamic origin of the observed correlations [12,13]. Alternative interpretations, including Color-Glass Condensate based models [14] and final-state parton– parton scattering [15], have also been proposed. Long-range correlations of forward and backward muons with mid-rapidity hadrons have also been found in p–Pb collisions at a center-of-mass energy per nucleon pair √sNN =5.02 TeV [16]. The results show that these correlations persist across wide rapidity ranges and extend into the high muon transverse-momentum interval, which is dominated by decays of heavy flavors. In pp collisions, the J/ψresonance is formed mainly from pairs of c and ¯ cquarks produced in hard scattering reactions during the initial stage of the collision. The theoretical models describing the https://doi.org/10.1016/j.physletb.2018.02.039 0370-2693/©2018 The Author. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Funded by SCOAP3. 8ALICE Collaboration / Physics Letters B 780 (2018) 7–20 J/ψproduction combine calculations of the production of c¯ cpairs within a perturbative Quantum Chromodynamics approach with the subsequent non-perturbative formation of the c¯ cbound state [17]. In p–Pb collisions, the production is affected by the modification of parton distribution functions inside the nucleus [18]as well as possible energy loss and inelastic scattering inside nuclear matter [19,20]. In A–A collisions, there are two additional competing phenomena that influence the J/ψproduction. First is the suppressed production due to the dissociation of the c¯ cpairs in the quark–gluon plasma [21]. Second is the J/ψenhancement via recombination of charm quarks thermalized in the medium [22, 23]. The recombination is expected to become prevalent in central collisions at the LHC energies. Recently, the ALICE Collaboration has published a precise measurement of the second-order Fourier coefficient, v2, of the azimuthal distribution of the J/ψproduction in Pb–Pb collisions at √sNN =5.02 TeV [24]. The results show significant v2in central and semi-central collisions. The measured J/ψv2at low and intermediate transverse momentum can be qualitatively described by a transport model in which the J/ψazimuthal anisotropy is inherited from that of recombined charm quarks [25,26]. However, at higher transverse momentum the data still indicates significant v2while the transport model predicts significantly smaller values coming mostly from path-length dependent suppression in the almond-shaped interaction region of the colliding nuclei and from non-prompt J/ψproduced from b-hadron decays assuming thermalized b quarks. Given these results in Pb–Pb collisions, it is of interest to study the J/ψ-hadron azimuthal correlations also in the smaller p–Pb system. The recombination of charm quarks, if any, should have much smaller impact, due to the smaller number of initially produced charm quarks with respect to Pb–Pb collisions. The small system size should not lead to a sizeable path-length dependent suppression. Nevertheless, the study of the J/ψ-hadron azimuthal correlations could allow to determine whenever J/ψ production is affected by the medium possibly created in these collisions [27–29]. In this Letter, we present results for long-range correlations between forward (p-going, 2.03 <y <3.53) and backward (Pbgoing, −4.46 <y <−2.96) inclusive J/ψand mid-rapidity charged hadrons in p–Pb collisions at √sNN =5.02 and 8.16 TeV. Inclusive J/ψrefers to both prompt J/ψ(direct and decays from higher mass charmonium states) and non-prompt J/ψ(feed down from b-hadron decays). 2. Experimental setup and data samples A detailed description of the ALICE apparatus can be found in Ref. [30]. Below, we briefly describe the detector systems essential for the present analysis. In the following, ηand ylab will denote the pseudorapidity and rapidity in the ALICE laboratory system. The muons are reconstructed in the muon spectrometer covering the range of −4 <η< −2.5. The spectrometer contains a front absorber located between 0.9 and 5m from the nominal interaction point. The absorber is followed by five tracking stations, each made of two planes of Cathode Pad Chambers. The third station is placed inside a dipole magnet with 3 Tm field integral. The tracking stations are followed by an iron wall with a thickness of 7.2 interaction lengths and two trigger stations, each one consisting of two planes of Resistive Plate Chambers. The position of the interaction point is obtained using the clusters reconstructed in the Silicon Pixel Detector (SPD) [31,32]. The SPD is located in the central barrel of the ALICE apparatus and operated inside a large solenoidal magnet providing a uniform 0.5 T magnetic field parallel to the beam line. The SPD consists of two cylindrical layers which cover |η| <2.0 and |η| <1.4with respect to the nominal interaction-point, for the inner and outer layer, respectively. The associated charged hadrons at mid-rapidity are reconstructed via the so-called SPD tracklets, short track segments formed from the clusters in the two layers of the SPD and the primary vertex [32]. The V0 detector [33] consists of two rings of 32 scintillator counters each, covering 2.8 <η<5.1(V0-A) and −3.7 <η<−1.7 (V0-C), respectively. It is used for triggering and event-multiplicity estimation. The data samples presented here were collected during the 2013 and 2016 p–Pb LHC runs. The collision energy was √sNN = 5.02 and 8.16 TeV for the 2013 and 2016 data samples, respectively. Part of the 5.02 TeV data were collected during the 2016 p–Pb run. Data with both beam configurations, namely Pb–nucleus momentum (denoted as Pb–p collisions) or proton momentum (denoted as p–Pb collisions) oriented towards the muon spectrometer, have been analyzed. The asymmetric beam energies, imposed by the two-in-one LHC magnet design, resulted in collisions whose nucleon–nucleon center-of-mass reference system is shifted in rapidity by 0.465 in the direction of the proton beam with respect to the ALICE laboratory system. The data were taken with a trigger that required coincidence of minimum-bias (MB) and dimuon triggers. The MB trigger was provided by the V0 detector requesting a signal in both V0-A and V0-C rings. Its efficiency is found to be about 98% [34]. The dimuon trigger required at least a pair of opposite-sign track segments in the muon trigger system, each with a transverse momentum (pT) above the threshold of the online trigger algorithm. This threshold was set to provide 50% efficiency for muon tracks with pT=0.5GeV/c. The collected data samples of p–Pb and Pb–p collisions at 5.02 TeV (8.16 TeV) correspond to integrated luminosities of 8.1 and 5.8 (8.7 and 12.9) nb−1, respectively. The maximum interaction pile-up probability ranged up to 3% and 8% during 2013 and 2016 data taking, respectively. 3. Event, track and dimuon selection The beam-induced background is rejected by requiring that the timing signals from both rings of the V0 detector are compatible with particles coming from collision events. Events containing multiple collisions (pile-up) are rejected by requiring one single interaction vertex reconstructed in the SPD and by exploiting the correlation between the number of clusters in the two layers of the SPD and the number of the reconstructed SPD tracklets. The longitudinal position of the reconstructed primary vertex (zvtx) is required to be within ±10 cm from the nominal interaction point. The reconstructed SPD tracklets are selected by applying a zvtx-dependent pseudorapidity cut. The cut is adjusted to exclude the contribution from the edges of the SPD where the detector acceptance is low. For example, we select tracklets within −1.8 <η<0.5, −1.3 <η<1.3 and −0.5 <η<1.8for events with zvtx =10, 0 and −10 cm, respectively. The contribution from fake and secondary tracklets is reduced by applying a || <5mrad cut on the difference between the azimuthal angles of the clusters in the two layers of the SPD with respect to the primary vertex. With this cut, the mean pTof the selected charged hadrons is found to be approximately 0.75 GeV/c[16]. The tracks reconstructed in the muon spectrometer are required to emerge at a radial transverse position between 17.6 and 89.5 cm from the end of the front absorber in order to avoid regions with higher material budget. The tracks reconstructed in the tracking chambers are identified as muons by requiring their matching with corresponding track segments in the trigger chambers. Background tracks are removed with a selection on the product of ALICE Collaboration / Physics Letters B 780 (2018) 7–20 9 Fig. 1. The Mμμ distribution in the 3 <pμμ T<6GeV/cinterval fitted with a combination of a CB2 function for the signal and a VWG function for the background, for high-multiplicity (left panel) and low-multiplicity (right panel) p–Pb collisions at √sNN =8.16 TeV. the total track momentum and the distance of closest approach to the primary vertex in the transverse plane [35]. The selected dimuons are defined as pairs of opposite-sign muon tracks having −4 <yμμ lab <−2.5, transverse momentum pμμ Tbetween 0 and 12 GeV/cand invariant mass Mμμ between 1 and 5GeV/c2. Only events with at least one dimuon satisfying these selection criteria are considered. The data samples are split into multiplicity classes based on the total charge deposited in the two rings (V0-A and V0-C) of the V0 detector (V0M) [34]. The high-multiplicity (low-multiplicity) event class is defined as 0–20% (40–100%) of the MB trigger event sample. 4. Analysis The Mμμ distribution in each event-multiplicity class and pμμ T bin is fit with the combination of an extended Crystal Ball (CB2) function for the J/ψsignal and a Variable-Width Gaussian (VWG) function for the background [36]. The tail parameters of the CB2 function were fixed to the values used in [37,38]. The J/ψpeak position and width were obtained from the fit in the 0–100% event class and fixed to these values in the other event-multiplicity classes. Examples of the Mμμ fit in the 0–20% and the 40–100% event classes in the 3 <pμμ T<6GeV/cinterval are shown in Fig. 1. The angular correlations between J/ψand charged hadrons are obtained from the associated-particle (SPD tracklets) yields per dimuon trigger. The yields are defined as Yi(zvtx,Mμμ,pμμ T,ϕ,η) =1 Ni trig(zvtx,Mμμ,pμμ T) d2Ni assoc(zvtx,Mμμ,pμμ T) dϕdη =1 Ni trig(zvtx,Mμμ,pμμ T) SEi(zvtx,Mμμ,pμμ T,ϕ,η) MEi(zvtx,Mμμ,pμμ T,ϕ,η),(1) where Ni trig(zvtx, Mμμ, pμμ T)is the number of dimuons, Ni assoc(zvtx, Mμμ, pμμ T)is the number of associated SPD tracklets corrected for acceptance and combinatorial effects (as shown in the second line of the equation and described below), ϕand η=yμμ lab −ηtracklet are the azimuthal angle and (pseudo)rapidity difference between the trigger dimuon and the associated SPD tracklet. The yields are calculated separately in each event-multiplicity class (index i) and 1 cm-wide zvtx interval. The distribution SEi(zvtx,Mμμ,pμμ T,ϕ,η)=d2Ni same(zvtx,Mμμ,pμμ T) dϕdη is the yield of associated SPD tracklets from the same event. The distribution MEi(zvtx,Mμμ,pμμ T,ϕ,η) =αi(zvtx,Mμμ,pμμ T)d2Ni mixed(zvtx,Mμμ,pμμ T) dϕdη is constructed using the event-mixing technique, i.e. combining dimuons from one event with SPD tracklets from other events selected in the same event-multiplicity class and zvtx interval. It serves both to correct for detector acceptance and efficiency and to take into account the combinatorial background. The normalization factor αi(zvtx, Mμμ, pμμ T)is defined as 1/(d2Ni mixed(zvtx, Mμμ, pμμ T)/dϕdη)in the ηregion corresponding to the maximal acceptance [16]. Within each event-multiplicity class and bin of Mμμ, pμμ T, ϕ and η, the yields Yiaveraged over zvtx are obtained by fitting the distribution YiNtrig(zvtx)iMEi(zvtx)to the distribution SEi(zvtx). A Poisson likelihood fit is used in order to properly deal with the cases of low number of tracklets. Then, the average yields are projected on the ϕaxis in the range of 1.5 <|η| <5using the method described in [16]. In order to extract the yields per J/ψtrigger, the yields per dimuon trigger in each event-multiplicity class, pμμ Tand ϕbins are fit as a function of Mμμ using the following superposition Yi(Mμμ)=S S+BYi J/ψ +B S+BYi B(Mμμ), (2) where Sand Bare the number of J/ψand the background dimuons in each bin of Mμμ obtained from the invariant mass fit (using a CB2 function for the J/ψsignal and a VWG function for the background) described above, YJ/ψ is the associated yield corresponding to the J/ψtrigger and YB(Mμμ)is a second-order polynomial function aimed to describe the associated yields corresponding to the background. The fit range is chosen between 1.5 and 4.5 GeV/c2. Examples of fits in high-multiplicity and lowmultiplicity event classes are shown in Fig. 2. Fig. 3shows the obtained associated tracklet yields per J/ψtrigger for p–Pb and Pb–p collisions at √sNN =5.02 and 8.16 TeV. 10 ALICE Collaboration / Physics Letters B 780 (2018) 7–20 Fig. 2. Example of associated tracklet yields per dimuon trigger in the 3 <pμμ T<6GeV/cinterval for high-multiplicity (left panel) and low-multiplicity (right panel) p–Pb collisions at √sNN =8.16 TeV. The result of the fit with the function from Eq. (2)is represented with the blue solid line. The dashed red line corresponds to the associated tracklet yields per background dimuon. (For interpretation of the colors in the figure(s), the reader is referred to the web version of this article.) As expected, in low-multiplicity collisions we observe a significant correlation structure on the away side (Fig. 3, top panels), presumably originating from the fragmentation of recoil jets. In high-multiplicity collisions (Fig. 3, middle panels), a possible enhancement on both near (ϕ≈0) and away (ϕ≈π) side can be spotted on top of the away-side structure. In order to isolate possible correlations due to collective effects between the J/ψand the associated tracklets, we apply the same subtraction method as in previous measurements [5,6,12,16], namely subtracting the YJ/ψ yields in low-multiplicity collisions from those in high-multiplicity collisions (Fig. 3, bottom panels). The subtraction method relies on the assumptions that the jet correlations on the away side remain unmodified as a function of the event multiplicity and that there are no significant correlations due to collective effects in low-multiplicity collisions (see discussion in Section 6). In order to quantify the remaining correlation structures, the subtracted yields Ysub J/ψ (ϕ)are fit with a0+2a1cosϕ+2a2cos2ϕ.(3) The second-order Fourier coefficient V2{J/ψ −tracklet,sub}of the azimuthal correlation between the J/ψand the associated charged hadrons is finally calculated as a2/bhigh 0. The denominator bhigh 0= a0+blow 0corresponds to the combinatorial baseline of the highmultiplicity collisions, where the parameter blow 0is the combinatorial baseline of the low-multiplicity collisions obtained at the minimum of the per-trigger yields, namely in ϕ<π/6. The parameter blow 0is the normalization factor used in Fig. 3. The parameter a1, which describes the strength of the remaining away-side correlation structure, is found to be compatible with zero in practically all pJ/ψ Tintervals, in both p–Pb and Pb–p collisions at both 5.02 and 8.16 TeV. As an alternative extraction method, the calculation of blow 0, the subtraction of low-multiplicity from high-multiplicity collision yields and the fit to Eq. (3)is done in each bin of Mμμ separately. Then the V2{J/ψ −tracklet,sub}coefficient is extracted by fitting V2{μμ −tracklet,sub}(Mμμ)with a superposition similar to the one defined in Eq. (2) V2{μμ −tracklet,sub}(Mμμ) =S S+BV2{J/ψ −tracklet,sub} +B S+BVB 2{μμ −tracklet,sub}(Mμμ), (4) where the VB 2{μμ −tracklet,sub}(Mμμ)is the second-order Fourier coefficient of the azimuthal correlation between the background dimuons and associated tracklets. The background coefficient VB 2{μμ −tracklet,sub}(Mμμ)is parameterized with a second-order polynomial function. This parameterization is chosen since it reproduces the dimuon v2(Mμμ)constructed from the measured muon v2coefficient [16] assuming that the dominant part of the background is combinatorial. An example of the V2{μμ −tracklet,sub}(Mμμ)fit is shown in Fig. 4. Following the procedure used in Refs. [5,12,16], the V2{J/ψ − tracklet,sub}coefficient is factorized into a product of J/ψand charged-hadron v2coefficients. Thus, the J/ψsecond-order Fourier azimuthal coefficient vJ/ψ 2{2,sub}is obtained as vJ/ψ 2{2,sub}=V2{J/ψ −tracklet,sub}/vtracklet 2{2,sub},(5) where the vtracklet 2{2,sub}is the tracklet second-order Fourier azimuthal coefficient obtained by performing the analysis considering SPD tracklets as both trigger and associated particles. The obtained values of vtracklet 2{2,sub}are between 0.067 and 0.069 depending on the beam configuration and collision energy, with 1–2% relative statistical uncertainty and 5–6.5% relative systematic uncertainty. 5. Systematic uncertainties The combined statistical and systematic uncertainties of the measured vtracklet 2{2,sub}coefficient for each beam configuration and collision energy are taken as global systematic uncertainties of the corresponding vJ/ψ 2{2,sub}coefficients. All the other systematic uncertainties of the vJ/ψ 2{2,sub}coefficients are obtained for each data sample and pTinterval separately. The following sources are considered. A possible inaccurate correction for the SPD acceptance is assessed by varying the zvtx range between ±8 and ±12 cm. Systematic uncertainties are assigned only in the cases of a significant change of the results. The significance is defined according to the procedure described in Ref. [39]. The systematic effect related to the uncertainty of the shape of the dimuon background yields YB(Mμμ)is estimated by performing the fit with Eq. (2)using a linear function for the background term and varying the fit range. The systematic effect coming from the uncertainty of the signal-to-background ratio S/Bis checked ALICE Collaboration / Physics Letters B 780 (2018) 7–20 11 Fig. 3. Associated tracklet yields per J/ψtrigger in 3 <pJ/ψ T<6GeV/cin p–Pb and Pb–p collisions at √sNN =5.02 TeV (left panels) and 8.16 TeV (right panels). The top and the middle panels correspond to the low-multiplicity and the high-multiplicity event classes, respectively. The bottom panels show the yields after the subtraction of the low-multiplicity collision yields from the high-multiplicity collision ones. The solid line represent the fit to the data as described in the text. The dashed, dot-dashed and dotted lines correspond to the individual terms of the fit function defined in Eq. (3). All the yields are normalized to the value in ϕ<π/6in the low-multiplicity (40–100%) event class. Only the statistical uncertainties are shown. (For interpretation of the colors in the figure(s), the reader is referred to the web version of this article.) by employing various invariant mass fit functions, both for the background and for the J/ψsignal. The maximal difference of the results obtained with the above checks with respect to the default approach is taken as the corresponding systematic uncertainty. The uncertainty arising from the employed analysis approach is obtained as the difference between the two extraction methods described in Section 4. As described in Section 4, by default the mixed-event distribution ME(ϕ, η)is normalized to unity in the ηregion corresponding to the maximal acceptance. As an alternative approach, normalizing the integral of ME(ϕ, η)to unity is used. No significant effect on the obtained results is observed and thus no systematic uncertainty is assigned. The used event-mixing technique can introduce systematic biases. The event multiplicity distribution of the selected dimuons (1 <Mμμ <5GeV/c2) differs from that of the J/ψsignal. Since the charged-hadron spectra and the charged-hadron density as a function of ηchange with event multiplicity [34], the non-uniform (both in the azimuthal and longitudinal directions) SPD acceptance can introduce a bias. The corresponding systematic uncertainty is evaluated by doing the event mixing in finer event-multiplicity bins. 12 ALICE Collaboration / Physics Letters B 780 (2018) 7–20 Table 1 Summary of absolute systematic uncertainties of the vJ/ψ 2{2,sub}coefficients. The uncertainties vary within the indicated ranges depending on pJ/ψ T. The values not preceded by a sign represent double-sided uncertainties. Source of systematics √sNN =5.02 TeV √sNN =8.16 TeV p–Pb Pb–p p–Pb Pb–p Acceptancecorrection 0to0.019 0to0.057 0to0.011 0to0.007 Background shape 0.007 to 0.013 0.015 to 0.056 0.011 to 0.013 0.003 to 0.012 Extraction method 0.003 to 0.015 0.010 to 0.040 0.002 to 0.011 0.008 to 0.018 Event mixing 0.003 to 0.015 0.004 to 0.025 0.002 to 0.008 0.004 to 0.012 Residual away-side jet correlation –−0.030 to 0 −0.018 to 0 – Total +0.009 to +0.024 +0.024 to +0.084 +0.013 to +0.019 +0.015 to +0.021 −0.009 to −0.024 −0.024 to −0.090 −0.015 to −0.026 −0.015 to −0.021 Fig. 4. Example of the fit from Eq. (4)in the 3 <pμμ T<6GeV/cinterval for p–Pb collisions at √sNN =8.16 TeV. The dashed line corresponds to the VB 2{μμ − tracklet,sub}(Mμμ). The non-uniform acceptance of the muon spectrometer coupled to sizeable correlations between the dimuons and SPD tracklets can bias azimuthally the sample of SPD tracklets used for event mixing. In order to check for possible effects on our measurement, the event mixing is performed in intervals of azimuthal angle of the selected dimuons. We observe no significant systematic effect as the obtained results show negligible deviations with respect to the results using the default event-mixing technique. The effect of a possible residual near-side peak is checked by varying the rapidity gap between the trigger dimuons and associated charged-hadrons from 1.0 to 2.0 units. We observe no indication of increasing v2with reduced gap and thus consider the default gap of 1.5 units sufficient to eliminate any significant residual near-side peak contribution. As shown in Section 4, the recoil-jet away-side correlation structure in the high-multiplicity event class is greatly diminished after the subtraction of the low-multiplicity event class. By default, any remaining away-side structure is supposed to be taken into account by the cosϕterm in Eq. (3). In order to check for residual effects we proceed in the following way. First, the correlation function in the low-multiplicity event class is fit with a Gaussian function centered at ϕ=π. Then, the correlation function in the high-multiplicity event class is fit with the function from Eq. (3), where the cosϕterm is replaced by a Gaussian function with a width fixed to the value obtained from the fit in the low-multiplicity collisions. No clear signature of systematic change of the results is seen, except some hints of a possible effect in the highest pJ/ψ Tinterval. Conservatively, we assign systematic uncertainty as the difference with respect to the default analysis approach. Since the typical values of the Gaussian width are around 1rad, one-sided (negative) systematic uncertainty is assigned. In Table 1we present a summary of the assigned systematic uncertainties of the vJ/ψ 2{2,sub}coefficients. No sizeable correlations between the pJ/ψ Tintervals are observed and therefore in the following the uncertainties are considered uncorrelated. Our measurement is for inclusive J/ψ. The fraction of J/ψfrom decays of b-hadrons reaches up to about 15% at pJ/ψ T≈6GeV/cin p–Pb collisions at √sNN =5.02 [40] and 8.16 TeV [41]. Therefore the feed-down contribution is unlikely to influence significantly our results. In principle, a possible strong multiplicity dependence of the feed-down fraction can potentially affect the subtraction approach. However, no evidence for such a strong dependence is observed in pp collisions [42]. As additional cross-checks the analysis is done using alternative event-multiplicity estimators, varying the tracklet ||cut, applying a cut on the asymmetry of transverse momentum of the two muon tracks, removing the pile-up cuts and excluding the SPD regions with non-uniform acceptance in pseudorapidity. The corresponding results are found to be compatible with those obtained with the default analysis approach and therefore no further systematic uncertainties are assigned. 6. Results In Fig. 5we report the measured vJ/ψ 2{2,sub}coefficients as a function of pJ/ψ Tfor p–Pb and Pb–p collisions at √sNN =5.02 and 8.16 TeV. Up to pJ/ψ Tof 3 GeV/c, no significant deviation from zero is observed for either p–Pb or Pb–p collisions at the two collision energies. On the contrary, in the pJ/ψ Tinterval between 3 and 6 GeV/c, the vJ/ψ 2{2,sub}is found to be positive although with large uncertainties. As also shown in Fig. 5, the vJ/ψ 2coefficients in 2.5 <y <4in central Pb–Pb collisions at √sNN =5.02 TeV reach maximal values in the same pJ/ψ Tinterval [24]. Two methods are employed in order to obtain the probability that the vJ/ψ 2{2,sub}is zero in the 3 <pJ/ψ T<6 GeV/cinterval. In the first method, the vJ/ψ 2{2,sub}values in the two pJ/ψ Tintervals (3 <pJ/ψ T<4GeV/cand 4 <pJ/ψ T<6GeV/c) are combined into a weighted average for each rapidity and collision energy. The obtained probabilities are 0.13% and 0.13% (7.8% and 0.23%) for p–Pb and Pb–p collisions, respectively, at √sNN =8.16 TeV (5.02 TeV). Combining all eight vJ/ψ 2{2,sub}values yields a total probability of 1.7 ×10−7. This corresponds to a 5.1σsignificance of the measured positive vJ/ψ 2{2,sub}coefficient. The second method is Fisher’s combined probability test [43]. With this method one obtains probabilities of 0.14% and 0.23% (10.3% and 0.41%) for p–Pb and Pb–p collisions at √sNN =8.16 TeV (5.02 TeV), respectively. ALICE Collaboration / Physics Letters B 780 (2018) 7–20 13 Fig. 5. vJ/ψ 2{2,sub}in bins of pJ/ψ Tfor p–Pb, 2.03 <y <3.53 (left panels), and Pb–p, −4.46 <y <−2.96 (right panels), collisions at √sNN =5.02 TeV (top panels) and 8.16 TeV (bottom panels). The results are compared to the vJ/ψ 2{EP}coefficients measured in central Pb–Pb collisions at √sNN =5.02 TeV in forward rapidity (2.5 <y <4) using event plane (EP) based methods [24]. The statistical and uncorrelated systematic uncertainties are represented by lines and boxes, respectively. The quoted global systematic uncertainties correspond to the combined statistical and systematic uncertainties of the measured vtracklet 2{2,sub}coefficient. The total probability is 1.4 ×10−6which corresponds to a 4.7σ significance. In the calculation of the above probabilities, both statistical and systematic uncertainties of the measured values are taken into account. The global systematic uncertainty is not taken into account as it is irrelevant in the case of the zero hypothesis. The analysis method presented in this Letter relies on the assumption that there are no significant correlations due to collective effects in the low-multiplicity event class. In case of a presence of such correlations, the measured V2{J/ψ −tracklet,sub}is equal to V2{J/ψ −tracklet,high}−blow 0 bhigh 0 V2{J/ψ −tracklet,low},(6) where V2{J/ψ −tracklet,high}and V2{J/ψ −tracklet,low}are the second-order Fourier coefficients of the azimuthal correlation between the J/ψand the associated charged hadrons in the highmultiplicity and the low-multiplicity collisions, respectively, and blow 0/bhigh 0≈1/3 is the ratio of the combinatorial baseline in the low-multiplicity and high-multiplicity collisions (see Fig. 3). As is demonstrated in Ref. [44], the assumption of no significant collective correlations in the low-multiplicity collisions is certainly questionable for light-flavor hadrons. Our data indicates the same, as we observe a statistically significant increase of the measured values of vtracklet 2{2,sub}when subtracting a lower event-multiplicity, e.g. 60–100%, class. Ultimately, the value of the vtracklet 2coefficient is found to be about 17% higher in case no subtraction is applied. Therefore, replacing the subtracted vtracklet 2{2,sub}coefficient in Eq. (5)by the non-subtracted coefficient would mean that the vJ/ψ 2 coefficients are up to 17% lower with respect to the measured vJ/ψ 2{2,sub}coefficients. However, assuming that the vJ/ψ 2coefficients follow the same trend as a function of event multiplicity as the vtracklet 2coefficient, they would be up to 17% higher with respect to the measured vJ/ψ 2{2,sub}coefficients. Subtracting lower event-multiplicity classes in the measurement of the vJ/ψ 2{2,sub} coefficient does not improve the precision of our measurement, because of the limited amount of J/ψsignal in the low-multiplicity collisions. The nuclear modification factor of J/ψin p–Pb and Pb–p collisions [37,38]as well as the charged-particle v2coefficient [45–47] in pp collisions show no significant √sNN dependence. As seen in Fig. 5, the measured vJ/ψ 2{2,sub}coefficients at √sNN =5.02 and 8.16 TeV also appear to be consistent with each other. The largest absolute difference between the results at the two collision energies is observed in Pb–p collisions in the 3 <pJ/ψ T<6GeV/c interval. The significance of this difference is rather low (below 1.5σ), because of the large uncertainties of the measurement at √sNN =5.02 TeV. Hence, the data for the two collision energies are combined as a weighted average taking into account both statistical and systematic uncertainties. In Fig. 6, we present these combined results for p–Pb and Pb–p collisions together with measurements and model calculations for Pb–Pb collisions at √sNN = 5.02 TeV [25]. In Pb–Pb collisions, the positive vJ/ψ 2coefficients at pJ/ψ Tbelow 3–4 GeV/care believed to originate from the recombination of charm quarks thermalized in the medium and are described fairly well by the transport model [25](see Fig. 6). In p–Pb collisions, the amount of produced charm quarks is small and therefore the contribution from recombination should be negligible. Our measured values at pJ/ψ T<3GeV/care compatible with zero, in line with this expectation. There is one publication [28] which suggests that even in p–Pb collisions a sizeable contribution from recombination could occur due to canonical enhancement effects. The uncertainties of our results do not allow to confirm or to rule out this scenario. In Pb–Pb collisions, the measured vJ/ψ 2coefficients exceed substantially the theoretical predictions at pJ/ψ T>4GeV/c, where the 14 ALICE Collaboration / Physics Letters B 780 (2018) 7–20 Fig. 6. Combined vJ/ψ 2{2,sub}coefficients in p–Pb and Pb–p collisions compared to the results in central and semi-central Pb–Pb collisions at √sNN =5.02 TeV [24]and the transport model calculations for semi-central Pb–Pb collisions at √sNN =5.02 TeV [25]. The solid line corresponds to the contribution from pathlength dependent suppression inside the medium. The band shows the resulting vJ/ψ 2including also the recombination of thermalized charm quarks and the feeddown from b-hadron decays assuming thermalization of b quarks. main contribution to vJ/ψ 2is expected to come from path-length dependent suppression inside the medium [25](see Fig. 6). In p–Pb collisions, the medium, if any, has a much smaller size [48] and hence very little, if any, path-length dependent effects are expected. In principle, the feed-down from decays of b-hadrons can give a positive vJ/ψ 2at high transverse momentum in case of a positive b quark v2. However, the latter would have to reach unreasonably high values given the magnitude of the measured vJ/ψ 2{2,sub}and the small feed-down fraction. Despite these considerations, the measured positive vJ/ψ 2coefficients would imply that the J/ψparticipates in the collective behavior of the p–Pb collision system. 7. Summary We presented a measurement of the angular correlations between forward and backward J/ψand mid-rapidity charged hadrons in p–Pb and Pb–p collisions at √sNN =5.02 and 8.16 TeV. The data indicate persisting long-range correlation structures at ϕ≈0 and ϕ≈π, reminiscent of the double ridge previously found in charged-particle correlations at midand forward rapidity. The corresponding vJ/ψ 2{2,sub}coefficients in 3 <pJ/ψ T< 6GeV/care found to be positive with a total significance of 4.7σto 5.1σ. The obtained values, albeit with large uncertainties, are comparable with those measured in Pb–Pb collisions at √sNN =5.02 TeV in forward rapidity. Although the underlying mechanism is not understood, the comparable magnitude of the vJ/ψ 2coefficients at high transverse momentum in p–Pb and Pb–Pb collisions indicates that this mechanism could be similar in both collision systems. Acknowledgements The ALICE Collaboration would like to thank all its engineers and technicians for their invaluable contributions to the construction of the experiment and the CERN accelerator teams for the outstanding performance of the LHC complex. The ALICE Collaboration gratefully acknowledges the resources and support provided by all Grid centres and the Worldwide LHC Computing Grid (WLCG) collaboration. The ALICE Collaboration acknowledges the following funding agencies for their support in building and running the ALICE detector: A.I. Alikhanyan National Science Laboratory (Yerevan Physics Institute) Foundation (ANSL), State Committee of Science and World Federation of Scientists (WFS), Armenia; Austrian Academy of Sciences and Nationalstiftung für Forschung, Technologie und Entwicklung, Austria; Ministry of Communications and High Technologies, National Nuclear Research Center, Azerbaijan; Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Universidade Federal do Rio Grande do Sul (UFRGS), Financiadora de Estudos e Projetos (Finep) and Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP), Brazil; Ministry of Science & Technology of China (MSTC), National Natural Science Foundation of China (NSFC) and Ministry of Education of China (MOEC), China; Ministry of Science, Education and Sports and Croatian Science Foundation, Croatia; Ministry of Education, Youth and Sports of the Czech Republic, Czech Republic; The Danish Council for Independent Research – Natural Sciences, the Carlsberg Foundation and Danish National Research Foundation (DNRF), Denmark; Helsinki Institute of Physics (HIP), Finland; Commissariat à l’Energie Atomique (CEA) and Institut National de Physique Nucléaire et de Physique des Particules (IN2P3) and Centre National de la Recherche Scientifique (CNRS), France; Bundesministerium für Bildung, Wissenschaft, Forschung und Technologie (BMBF) and GSI Helmholtzzentrum für Schwerionenforschung GmbH, Germany; General Secretariat for Research and Technology, Ministry of Education, Research and Religions, Greece; National Research, Development and Innovation Office, Hungary; Department of Atomic Energy, Government of India (DAE), Department of Science and Technology, Government of India (DST), University Grants Commission, Government of India (UGC) and Council of Scientific and Industrial Research (CSIR), India; Indonesian Institute of Science, Indonesia; Centro Fermi – Museo Storico della Fisica e Centro Studi e Ricerche Enrico Fermi and Istituto Nazionale di Fisica Nucleare (INFN), Italy; Institute for Innovative Science and Technology, Nagasaki Institute of Applied Science (IIST), Japan Society for the Promotion of Science (JSPS) KAKENHI and Japanese Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan; Consejo Nacional de Ciencia (CONACYT) y Tecnología, through Fondo de Cooperación Internacional en Ciencia y Tecnología (FONCICYT) and Dirección General de Asuntos del Personal Academico (DGAPA), Mexico; Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO), Netherlands; The Research Council of Norway, Norway; Commission on Science and Technology for Sustainable Development in the South (COMSATS), Pakistan; Pontificia Universidad Católica del Perú, Peru; Ministry of Science and Higher Education and National Science Centre, Poland; Korea Institute of Science and Technology Information and National Research Foundation of Korea (NRF), Republic of Korea; Ministry of Education and Scientific Research, Institute of Atomic Physics and Romanian National Agency for Science, Technology and Innovation, Romania; Joint Institute for Nuclear Research (JINR), Ministry of Education and Science of the Russian Federation and National Research Centre Kurchatov Institute, Russia; Ministry of Education, Science, Research and Sport of the Slovak Republic, Slovakia; National Research Foundation of South Africa, South Africa; Centro de Aplicaciones Tecnológicas y Desarrollo Nuclear (CEADEN), Cubaenergía, Cuba, Ministerio de Ciencia e Innovacion and Centro de Investigaciones Energéticas, Medioambientales y Tecnológicas (CIEMAT), Spain; Swedish Research Council (VR) and Knut & Alice Wallenberg Foundation (KAW), Sweden; European Organization for Nuclear Research, Switzerland; National Science and Technology Development Agency (NSDTA), Suranaree University of Technol-