NNuclear Physics A 00 (2020) 1–8
NuclearPhysics A / locate / procedia XXVIIIth International Conference on Ultrarelativistic Nucleus-Nucleus Collisions(Quark Matter 2019)
Probing QGP with flow: An experimental overview
Katar´ına Kˇr´ıˇzkov´a Gajdoˇsov´a
Czech Technical University in Prague, Faculty of Nuclear Sciences and Physical Engineering, Bˇrehov´a 7, 115 19, Prague, CzechRepublic
Abstract
An experimental overview of anisotropic flow measurements and their ability to probe the properties and the natureof the system created in ultra-relativistic hadron collisions is given in these proceedings. The aim is to discuss thestate-of-the-art measurements ranging from small to large systems at di ff erent collision energies. Keywords: quark-gluon plasma, heavy-ion collisions, small collision systems, anisotropic flow
1. Introduction
Relativistic Heavy Ion Collider (RHIC) at BNL and the Large Hadron Collider (LHC) at CERN aremachines standing at the forefront of the research of a hot and dense, strongly interacting QCD medium,called the quark-gluon plasma (QGP). One of their main purposes is to recreate the QGP via ultra-relativisticheavy-ion collisions in order to study its properties at extreme conditions, using collisions of smaller systems(such as p-A or pp) as a reference with absence of such medium. It is known for a couple of years, thatmeasurements in small collision systems with high particle multiplicities reveal features similar to thoseobserved in the collectively expanding medium present in heavy-ion collisions.In collisions of heavy ions, spatial anisotropy of their overlap region in the transverse plane is translatedto an anisotropic distribution of final-state particles via parton interactions during the deconfined phase. Theazimuthal distribution of emitted particles can be decomposed into Fourier series as P ( ϕ ) = π ∞ (cid:88) n = −∞ V n e − in ϕ , (1)where v n is the magnitude (also called the flow coe ffi cient) and Ψ n is the phase (also called the symmetryplane) of the flow vector V n = v n e in Ψ n . The values of flow vectors reflect the hydrodynamic response ofthe medium to the initial state eccentricity. Thus, measurements of flow vectors, their fluctuations and / orcorrelations provide an important ingredient for validation of the existing theoretical models, in particularfor determination of initial conditions and transport coe ffi cients of the medium [1, 2, 3]. a r X i v : . [ nu c l - e x ] J u l / Nuclear Physics A 00 (2020) 1–8
In small collision systems, the same observables are used to resolve, whether a medium of a similarorigin as in heavy-ion collisions is formed. As opposed to heavy-ion collisions, it is not yet clear, to whatextent are the flow vectors driven by the initial spatial anisotropy, and whether the gluon field momentumcorrelations from the initial state persist and contribute to the observed final state anisotropy [4, 5]. In ad-dition, an overwhelming contamination from non-flow e ff ects, arising mainly from correlations of particleswithin jets, leads to inevitable complications in measurements performed in small systems. A template fitmethod used in measurements of two-particle correlation functions [6] and the subevent method used in two-and multi-particle cumulants [7] are the state-of-the-art approaches to suppress contributions from non-flow.Any interpretation of measurements in small systems should not be advanced without appropriate treatmentof this contamination.Since the debates about large and small collision systems di ff er in their essence, the measurementspresented here will be separated into two sections. Yet, the discussion about small systems is often accom-panied by references and comparisons with the collective AA collisions. I would like to note that this is notan exhaustive summary of the results presented at the conference, but rather a general overview of the latestexperimental developments in the field. For a theory overview please refer to Ref. [8].
2. Large collision systems
Our understanding of the QGP has improved significantly with the measurements of magnitudes offlow vectors, v n [1, 2, 3]. However, details of the initial conditions and / or the dynamics of the subsequentdeconfined phase cannot be resolved with measurements of v n alone. Fortunately, a wealth of experimentaldata collected over the past years allow us to dive deeper into the investigations of the QGP and improve ourknowledge about this phase of QCD matter.Event-by-event fluctuations of the initial state geometry cause the flow vectors (their magnitudes andsymmetry plane angles), constructed in di ff erent p T or η ranges, to fluctuate around the event-averagedvalues. Investigations of these fluctuations pose important constraints on initial conditions, which in turncontribute to more precise modeling of the final state dynamics. This can be addressed with several mea-surements, such as the event-by-event fluctuations of the v n , decorrelations of flow vectors in p T and η , andin performing the event shape engineering (ESE). A selection of the most recent developments in each ofthese measurements is discussed below.The flow probability density function (p.d.f.), the P ( v n ), can be accessed either via the unfolding proce-dure as used in [9, 10], or by investigating the degeneracy of higher order cumulants by measuring deviationsof their ratios from unity [11, 12]. Measurements of inclusive charged hadrons revealed that flow fluctua-tions are neither Gaussian, nor Bessel-Gaussian [9, 10, 13, 14]. Impressive advancement in the collecteddata at the LHC and in the analysis techniques allowed to study flow fluctuations di ff erentially in p T . Theobserved p T -dependence of multi-particle cumulant ratios, skewness and kurtosis, shown in Fig. 1 (left),indicates influence of final state fluctuations [15]. While deviations from a Bessel-Gaussian flow p.d.f. wereconfirmed at p T < / c , degeneration of multi-particle cumulants found at intermediate p T indicates arecovery of this parametrisation. First measurements of the four-particle cumulant v { } ( p T ) of identifiedparticles allowed to study their relative flow fluctuations via the F ( v n ) = σ v n / (cid:104) v n (cid:105) (see Fig. 1 (right)) [15].While the iEBE-VISHNU hydrodynamic model [16] able to reproduce the v n measurements, predicts a par-ticle species dependence of flow fluctuations, data seem to disfavor this result, showing their potential forfurther constraints on theoretical calculations. Finally, it should be noted that flow fluctuations were foundto be a ff ected by the so-called volume fluctuations within a fixed centrality bin, which arise from variationsof the sources used to determine the event centrality [14]. Since values of v n change with collision centrality,such e ff ects will naturally lead to additional flow fluctuations, manifested e.g. by a “wrong” (positive) signof the four-particle cumulant. The e ff ect was assumed to be most pronounced in ultra-central collisions.However, it was found that volume fluctuations may a ff ect the results up to mid-central collisions [14].In addition to the standard centrality selection, the method of event shape engineering (ESE) imposesfurther selection based on the event-by-event variation of v n within a fixed centrality bin. Measurementsperformed with this method can help to further understand the initial conditions of a heavy-ion collision.Results of inclusive charged hadrons showed that indeed a modification of anisotropic flow is achieved Nuclear Physics A 00 (2020) 1–8 − − γ γ ALICE PreliminaryV0M 30-40% = 5.02 TeV NN s Pb-Pb, γ γ N γ ) c (GeV/ T p ALI−PREL−327438
ALI-PREL-331191
Fig. 1. Left: Skewness γ and kurtosis γ of the flow p.d.f. measured as a function of p T in Pb–Pb collisions at √ s NN = .
02 TeV [15].Right: Relative flow fluctuations of identified hadrons measured as a function of p T in Pb–Pb collisions at √ s NN = .
02 TeV [15],compared to iEBE-VISHNU model with AMPT initial conditions [16]. by employing the ESE technique [17]. New measurements of identified hadrons confirm these observa-tions [18], and extend them with a selection on triangularity. The results show no dependence on p T andparticle type.Flow vector decorrelations were extensively studied with the ratio v n { } v n [2] ( p T ) [19], or the factorisationratio r n ( p T , η ) [19, 20, 21]. Absence of decorrelation would yield ratios equal to 1, while decorrelationof v n and / or Ψ n would cause the ratios to deviate from unity. These measurements can provide importantconstraints on the fluctuation driven dynamics of the medium, especially its longitudinal structure needed forfurther improvements of three-dimensional hydrodynamic models. Collisions of smaller nuclei, or collisionsat lower energies, are more influenced by event-by-event fluctuations. Comparison of r n in Pb–Pb and Xe–Xe collisions recently presented in [22] provides additional sensitivity to the fluctuating initial geometryand viscous corrections. Hydrodynamic model [23, 24], tuned to describe the v n in both Xe–Xe and Pb–Pbcollisions, fails to reproduce the r n , as can be seen in Fig. 2 (left) for the 3 rd hadrmonic. Measurements of r n in Au–Au collisions at the new energy √ s NN =
27 GeV [25] provide further important input to theoreticalmodelling of heavy-ion collisions. In particular, it was found that longitudinal decorrelation of r is strongerat lower energy (see Fig. 2 (right)). This e ff ect is much more pronounced for r in comparison to r , probablydue to larger sensitivity of higher order flow harmonics to the fluctuating initial state. These, and additionalresults from the RHIC BES program, may significantly contribute to the advancements in the field. Fig. 2. Left: Ratios of flow decorrelations ( F -ratio) and v coe ffi cients from Xe–Xe to Pb–Pb collisions at √ s NN = .
44 and 5.02 TeV,respectively [22], compared to hydrodynamic model [23, 24]. Right: Factorisation ratio r ( η ) from Au–Au collisions at √ s NN = r ( η ) from Pb–Pb collisions at √ s NN = .
76 TeV and 5.02 TeV [20, 21]. / Nuclear Physics A 00 (2020) 1–8
Studies of higher order V n , in particular their linear and non-linear modes, and correlations betweendi ff erent orders of flow coe ffi cients or symmetry planes, provide a detailed insight into the hydrodynamicresponse of the system to the initial density profiles [26, 27, 28, 29, 30]. The main assumption in suchstudies lies in a linear response of V and V to the initial eccentricities [31], while higher order V n can beexpressed in term of linear and non-linear modes, each being proportional to the same order eccentricityor lower order eccentricities and / or their products, respectively [32, 33]. Particularly strong constraints tomodels of the hydrodynamic phase can be imposed by measurements of non-linear response coe ffi cients χ n , m , k . They are sensitive to the shear viscosity over entropy density ratio η/ s at freeze-out, which cannotbe addressed by any other measurement performed so far [33, 34]. A significant progress has been madein this direction. Results of non-linear modes of V n extending to harmonics of high orders showed thatnone of the model calculations used for comparison was able to simultaneously describe all the presentedobservables [35, 36]. In particular, the χ , revealed a remarkable ability to further constrain both initialconditions, and transport coe ffi cients of the medium [35]. A new testing ground for modeling of heavy-ioncollisions was recently provided by measurements of non-linear flow modes of identified hadrons [37] andby the first results of energy and system size dependence of χ , and symmetry planes correlation ρ , [38].As can be seen in Fig. 3 (left), no variation with collision energy is found for χ , .Since the first measurements of a finite flow of heavy flavor particles [39, 40] it has become evident thatsuch measurements opened a new window for theory validation due to the early formation of heavy flavorparticles and their subsequent participation in the collective expansion of the system. Investigating flowof open heavy flavor particles or quarkonia at low p T can give further insight into the way of how heavyquarks interact with the medium created in heavy-ion collisions. The increasing amount of collected databoth at RHIC and the LHC now allows to measure the flow of heavy flavor particles with unprecedentedprecision [41, 42, 43, 44]. Measurements of open heavy flavor hadrons confirm a significant non-zero v ,with the possibility to disentangle contributions from bottom quark, which is found to flow with less strengththan the charm quark [41, 44]. Flow of quarkonia is of particular interest to study the genuine interaction ofcharm and beauty with the medium. While J / Ψ exhibits a significant flow signal [45, 46], the v of Υ (1S)measured for the first time by ALICE [47] and CMS [48] does not reveal any signs of a non-zero v (fora comparison of the two measurements, see Fig. 3 (right)). Further investigation on this topic is of greatinterest. Fig. 3. Left: Non-linear response coe ffi cient χ , = v NL4 / √(cid:104) v (cid:105) measured in Au–Au collisions at various collision energies [38].Right: Transverse momentum dependence of v of Υ (1S) measured in Pb–Pb collisions at √ s NN = .
02 TeV [47, 48].
3. Small collision systems
Small collision systems with large number of produced particles have attracted a lot of attention sincethe first appearance of the near-side ridge in the year 2010 [49]. Whether this is a manifestation of a collec-
Nuclear Physics A 00 (2020) 1–8 tive behaviour of the system created in such collisions was the main question, which required tremendousimprovements in experimental techniques (especially in suppressing non-flow contamination as mentionedin section 1) to be answered. Multi-particle correlations spanning (very) longe range in pseudorapidity, indi-cating presence of collectivity, were observed by several experiments down to the smallest collision systemsand to very low energies [50, 51, 52, 53, 54, 55]. The origin of this apparent collectivity is however stillnot resolved. Do we create a strongly interacting fluid-like medium, or rather a dilute system where partonsundergo only few scatterings? Is flow generated as a response to the initial geometry via interactions in thefinal state, similarly as in heavy-ion collisions? To which extent do initial momentum correlations influ-ence the observed flow signal? These and many more questions are still not clearly answered and requireattention from both experimental and theoretical side.Many features of the results obtained from small systems suggest that final state e ff ects may be respon-sible for our observations. One of the key findings in pA collisions is the ordering of v depending on themass of the studied particle [50, 56, 57]. This is consistent with observations in heavy-ion collisions, ex-plained within a hydrodynamic picture as a result of a collective radial expansion of the medium. Final statescenario is further supported by the measurements of ratios of multi-particle cumulants which appear to bedriven by the initial-state geometry [58], in particular its fluctuations instead of an overall shape. Indeed,subnucleon fluctuations were found to be crucial for a correct description of the results in small collisionsystems by hydrodynamic models [59, 60]. This was recently confirmed by v n measurements in p-Au, d-Auand He-Au collisions, in particular by observing no dependence of v on collision systems geometry [61],as opposed to the system-dependent results reported in [62]. A comparison of the v and v measurementsfrom STAR and PHENIX experiments is shown in Fig. 4. While pA collisions tend to support a scenarioof strong final state interactions dependent on the fluctuating initial geometry, it is not so clear in pp colli-sions. Until now, only few hints toward a final state description were provided by an indication of a massordering [50] and hydrodynamic description of charged particle v n [60]. Nevertheless, these are challengedby several other observations, such as the inability of a full hydrodynamic simulation to translate a nega-tive c (cid:15) { } to a negative c { } [63], or by alternative explanations with just few parton scatterings in a dilutesystem within the transport model [64], or the string shoving mechanism in the PYTHIA model [65]. Fig. 4. Measurements of v (left) and v (right) in p–Au, d–Au and He–Au collisions at √ s NN =
200 GeV [61, 62].
Even though a general consensus on the origin of collectivity has not been reached yet, the resultspresented here tend to support the idea of a final state scenario with a small sized fluid being created insmall systems, at least in pA collisions. However, it should be kept in mind that considering the smallsize and short living time of the (possibly) created medium, influence of correlations from the initial stateshould not be neglected. It was discussed already few years ago [68] and studied again recently [69]. Itseems reasonable to rather focus on finding the relative balance of the two di ff erent approaches (initial vs.final state scenario), or finding a place at which one overwhelms the other. One way to study this may bethe anisotropic flow of heavy flavor quarks in small systems. Since these quarks are created at very earlytimes of a collision, they can o ff er a unique opportunity to disentangle the contribution from initial statecorrelations to the measurements of anisotropic flow. Increasing quality of the collected data at the LHC / Nuclear Physics A 00 (2020) 1–8 recch N - v ATLAS -1 =13 TeV, 150 pbs pp <6 GeV T hD mfi c mfi b Fig. 5. Left: Elliptic flow v of muons originating from charm or bottom quark as a function of N recch measured in pp collisions at √ s =
13 TeV [66]. Right: Elliptic flow of non-prompt D originating from decays of hadrons containing a b quark measured in p–Pbcollisions at √ s NN = .
16 TeV [67]. allowed to measure flow of heavy flavor particles in both p–Pb and pp collisions [42, 66, 67, 70, 71, 72].Recent measurements showed that while charm quark is observed to have a significant v , comparable to thatof light quarks, the bottom quark did not exhibit flow [66, 67], as can be seen in Fig. 5. This is in contrastto large collision systems, where similar measurements revealed a non-zero v (it is only the bottomoniumthat doesn’t exhibit flow, as discussed in section 2). AT p - - n v Zhang, Liao size a v size b v size a v size b v v v ATLAS -1 , 165 nb = 8.16 TeV NN s+Pb p Fig. 6. Measurement of v and v at very high p T in p–Pb collisions at √ s NN = .
16 TeV [73], compared to theoretical calculations [74].
One of the main drawbacks of the theory of having a medium similar to that created in heavy-ioncollisions, is the absence of jet quenching. Until now, no apparent jet modification by the medium wasfound [75], as opposed to AA collisions, where the nuclear modification factor R AA <
1. A possibility of abias in the way the normalisation of the modification ratio is obtained from simulations is discussed in [76].Measurements of anisotropic flow at high p T o ff er a di ff erent way of studying the parton energy loss withthe advantage of absence of such biases. As can be seen in Fig. 6, finite values of v and v were measuredat high p T of high-multiplicity p–Pb collisions [73], which would suggest a path length dependence of theparton energy loss in a medium. Indeed, hydrodynamic model [74] invokes a strong parton coupling to themedium in the attempt to reproduce the measurements. However, it also unavoidably leads to rather strongsuppression R pPb <
1, which is in contrast to the findings from experimental measurements [75, 77]. Thesefindings therefore suggest, that the observed finite azimuthal anisotropy at high p T must originate from ayet unknown mechanism. Further investigations in the direction of simultaneous description of finite v n andlack of suppression in R AA at high p T are therefore desirable. Nuclear Physics A 00 (2020) 1–8
4. Summary
The wealth of experimental results summarised in section 2 demonstrates the level of precision thatthe research of large collision systems has reached. The large amount of data collected in the past yearsallow to measure complex observables sensitive enough to provide unprecedented constraints to parametersof models aiming to describe the deconfined matter created in heavy-ion collisions. The most prominentexamples presented here are the system and energy dependence of longitudinal flow vector decorrelations,di ff erential studies of flow fluctuations, and measurements of non-linear response coe ffi cients of harmonicsup to very high orders and at variety of collision energies. Nevertheless, further studies, ideally usingobservables with exclusive sensitivity to either initial conditions, or the transport coe ffi cients (especiallythe η/ s or ζ/ s ), are still desired to reach a complete understanding of the deconfined medium created inheavy-ion collisions.The origin of the collective e ff ects observed in small collision systems is still elusive. Measurementsdriven by the fluctuating initial geometry suggest a large influence from final state e ff ects, although it doesnot provide a clear answer on whether these are manifestation of a small fluid-like medium, or rather resem-ble a dilute system with just few parton scatterings. Nevertheless, contributions from initial state correlationsshould not be neglected in our considerations. In spite of the impressive advancements in both experimentaland theoretical fields, we are still not able to conclude on what is the relative contribution of initial and finalstate e ff ects to our experimental observations. This is an area of active development and hopefully someanswers will be presented at the next edition of the Quark Matter conference.
5. Acknowledgements