betaFeedback

WorksPhysics and Astronomy

Uncertainty quantification of holographic transport and energy loss for the hot and baryon-dense QGP

Abstract

We investigate several transport coefficients across the phase diagram of a holographic Einstein-Maxwell-Dilaton (EMD) model of hot and dense QCD with Nf=2+1N_f=2+1 flavors. Our results are obtained from an open-source implementation of this model in C++, publicly available as a module within the MUSES Framework. This code includes a new numerical method to extract thermodynamic quantities from near-boundary asymptotics in holographic models, introduced here for the first time, which greatly improves numerical stability and performance in comparison to earlier implementations. Thanks to this improved technique, we are able to compute results for many realizations of our holographic model, sampled from a Bayesian posterior distribution constrained by lattice QCD results at zero chemical potential. This allows us to propagate lattice QCD error bars to predictions of transport coefficients in a wide window of temperature and baryon chemical potential, covering the crossover region, the neighborhood of the predicted critical point, and the line of first-order phase transition. The physical observables include baryon and thermal conductivities, baryon diffusion, shear and bulk viscosities, the jet-quenching parameter, the heavy-quark drag force, and Langevin diffusion coefficients. At vanishing baryon density, we compare our results to estimates extracted by the JETSCAPE Collaboration from heavy-ion data, with which we find good agreement.

Introduction

Understanding the properties of strongly interacting matter across the QCD phase diagram is a central goal of contemporary nuclear physics [1, 2, 3]. Temperatures of T≳150T \gtrsim150 MeV can be explored in relativistic heavy-ion collisions, where a tiny and short-lived quark-gluon plasma (QGP) is formed [4, 5, 6, 7, 8, 9, 10, 11]. More than two decades of experimental and theoretical efforts have been dedicated to exploring and characterizing this system both at RHIC and at CERN, leading to substantial progress in understanding its properties and evolution, including precise constraints on collective flow and transport properties [3, 12, 13, 14, 15]. Among these advances, it has been established that the QGP found in these collisions exhibits an exceptionally low shear viscosity to entropy ratio, characteristic of strongly coupled systems [16, 17, 18].

While collisions at top energies probe low values of the baryon chemical potential, there is increasing interest in investigating the QGP in the baryon-rich regime by exploring lower beam energies, of the order of a few to a few tens of GeV per nucleon. A major goal of experimental programs such as the RHIC Beam Energy Scan and the CBM experiment planned at FAIR (GSI) is that of mapping the QCD phase diagram to determine whether and under which conditions the smooth crossover to the QGP becomes a discontinuous phase transition—namely, a first-order transition line starting from a second-order critical end point (CEP) [2, 19, 20, 21, 22, 23, 24, 25, 26, 27]. Nonetheless, a realistic theoretical description of the baryon-dense QGP formed in these collisions still faces major challenges. First-principles lattice QCD (LQCD) simulations cannot yet reliably determine the baryon-dense equation of state [28, 29], and transport properties also remain poorly constrained [3]. In this context, one must rely on surrogate descriptions such as effective models of QCD, which should be constructed so as to reproduce current knowledge—for instance, the low-density equation of state and the nearly inviscid behavior of the QGP.

An approach capable of capturing strongly coupled results and reproducing several features of QCD can be found in some holographic models, based on the gauge-gravity duality [30, 31, 32, 33]—see, for instance [34, 35, 36, 37, 38, 39]. The Einstein–Maxwell–Dilaton (EMD) class of models represents a particularly flexible bottom-up holographic approach, and can be tuned to reproduce LQCD results at vanishing baryon density [40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76]. Recently, in Ref. [63], this procedure was made more systematic by applying Bayesian inference tools to constrain the EMD model with 2 + 1-flavors LQCD results for the entropy density and baryon susceptibility, both evaluated at zero baryon chemical potential. This Bayesian analysis showed that LQCD results at vanishing baryon density can place stringent constraints on features of the holographic model phase diagram and equation of state at higher baryon densities, revealing a strong preference for the existence of a critical point within two different functional realizations of the EMD model [63]. Moreover, results were shown to be in good quantitative agreement with state-of-the-art LQCD results at finite baryon density from the T′T'-expansion [77], even though they were not used to construct the model, lending further support to its phenomenological applicability [39, 63].

Unlike other later analyses [72, 74], the Bayesian analysis of Ref. [63] employed fully numerical solutions to the EMD equations of motion, instead of relying on analytical solutions applicable to simpler realizations of the model. For this reason, improvements in numerical performance and stability were essential for carrying out the Bayesian sampling. In particular, that work benefited from a novel numerical strategy which replaces least-squares fits to near-boundary data by a relaxation procedure to isolate the ultraviolet asymptotic coefficients used to extract thermodynamic observables. This approach leads to substantial gains in numerical stability and computational efficiency and was crucial for enabling the exploration of Bayesian posterior distributions. It has been implemented in a C++ code developed within the MUSES Framework [78], which was employed in [63].

Here, we introduce this new method for the first time in the context of holographic models. We also take advantage of the Bayesian posterior samples from [63] to propagate LQCD uncertainties to predictions for a broad set of transport coefficients throughout the phase diagram, and find them to be tightly constrained within the EMD model. This is achieved with a new version of the MUSES holographic module that incorporates the calculation of baryon and thermal conductivities, shear and bulk viscosities, the jet-quenching parameter, the heavy-quark drag force, and Langevin diffusion coefficients, which is described below and released publicly [78]. Our results for the conductivities and viscosities are directly relevant for heavy-ion collisions at Beam Energy Scan and FAIR energies, while results for heavy-quark transport and jet quenching—although not expected to play a central phenomenological role at such low beam energies—are presented for completeness.

Finally, we note that, during the development of the present work, two interesting papers [73, 74] appeared with which our manuscript has some overlap. In [73], the jet-quenching parameter and the energy loss of heavy and light quarks have been investigated in the framework of holographic EMD models at finite temperature and baryon density, adjusted to LQCD data with Nf=2N_f = 2, 2+12 + 1, and 2+1+12 + 1 + 1 flavors. They analyzed these observables in the crossover region, and also in the phase transition region comprising the CEP and the first order phase transition line of their EMD models, but without quantifying the statistical uncertainties in their predictions through Bayesian inference. In [74], a Bayesian analysis of an EMD model with Nf=2+1N_f = 2 + 1 flavors was employed to investigate the heavy quark drag force, jet-quenching parameter, heavy quark diffusion coefficient, and the bulk and shear viscosities strictly at zero chemical potential. On the other hand, in the present work, we investigate a broader set of physical observables, including the baryon susceptibility, baryon and thermal conductivities, baryon diffusion, shear and bulk viscosities, jet-quenching parameter, heavy quark drag force and Langevin diffusion coefficients across the phase diagram of the EMD model using Bayesian analysis with two different sets of functional forms for the free functions of the holographic setup, further asserting the robustness of the predictions provided here. Indeed, our approach provides the first in-depth analysis in the literature regarding the behavior of all these physical observables across the crossover and phase transition regions at finite baryon density with Bayesian uncertainty bands fully included.

The present manuscript is organized as follows. In section II, we review the basics of the holographic EMD setup and the Bayesian inference procedure, while also presenting our new numerical scheme for a more efficient, fast, and numerically stable extraction of the ultraviolet expansion coefficients required for the holographic computation of several physical observables. In section III A, we employ the aforementioned methods to compute the baryon susceptibility, the baryon and thermal conductivities, and the baryon diffusion coefficient at finite temperature and baryon density. In section III B, we compute the shear and bulk viscosities, while in section III C we evaluate the heavy quark drag force, the Langevin diffusion coefficients, and the jet-quenching parameter. In section IV, as a way of assessing theoretical uncertainties, we compare results obtained within two different parametric forms, or ansatze, for the holographic potentials in our EMD model. We present our main conclusions in section V. In appendix A, we further check the internal robustness of our holographic predictions for the QGP transport coefficients by performing a Bayesian analysis with different functional forms for the free functions of the EMD setup, as was done for thermodynamic observables in [63]. The numerical implementation developed in this work is released through the MUSES Collaboration and is publicly available at Ref. [78]. The Bayesian posterior samples employed in the analysis are likewise publicly available in a separate repository [79].

Throughout this work we use natural units with c=ℏ=kB=1c = \hbar= k_B = 1 and the mostly plus metric signature.

Holographic EMD setup at finite temperature and baryon chemical potential

One of the earliest realizations of the holographic gauge-gravity duality (then known as AdS-CFT correspondence) corresponds to the dual relationship between pure gravity on (asymptotically) AdS spacetimes and a supersymmetric and conformal field theory (CFT) called N=4\mathcal{N} = 4 Supersymmetric Yang-Mills (SYM) theory, living on the four-dimensional conformally flat boundary of the asymptotically AdS bulk geometry [30]. Since the QGP has no supersymmetry and is highly nonconformal around the hadronization/deconfinement crossover region, one needs to break supersymmetry and conformal symmetry in a rather specific way in order to quantitatively reproduce the behavior of QCD thermodynamics at finite temperature and baryon density using holographic gauge-gravity models. One of the most successful approaches in this regard was originally proposed in [40, 41, 42, 43, 44] by making use of bottom-up holographic models comprising a dilaton field with an adequately engineered potential. We refer the reader to the comprehensive review in Ref. [39] for refinements and extensions of such approach, applications for computing several observables relevant for characterizing the QGP, as well as a discussion of different formulations of other bottom-up holographic models that are used in the literature.

In the holographic approach used here, supersymmetry and conformal invariance are broken by introducing a real tachyonic scalar dilaton field ϕ\phi in the gravity action for the bulk theory, whose boundary value sources a QFT operator triggering a relevant deformation of the renormalization group (RG) flow of the dual gauge theory at the boundary. Moreover, a baryon chemical potential at the boundary QFT is associated with the boundary value of a bulk Maxwell field AμA^\mu, which promotes the global U(1)BU(1)_B symmetry associated to baryon number conservation at the boundary QFT, to a local gauge symmetry in the bulk gravity theory. This allows one to consider charged black hole duals carrying a conserved charge, which we identity with baryon number.

The EMD action for the bulk gravity theory may be written as follows [43],

S=∫M5d5x L=12κ52∫M5d5x −g[R−(∂μϕ)22−V(ϕ)−f(ϕ)Fμν24],(1)S = \int_{\mathcal{M}_5} d^5x\,\mathcal{L} = \frac{1}{2\kappa_5^2}\int_{\mathcal{M}_5} d^5x\,\sqrt{-g}\left[R - \frac{(\partial_\mu\phi)^2}{2} - V(\phi) - \frac{f(\phi)F_{\mu\nu}^2}{4}\right], \tag*{(1)}

where M5\mathcal{M}_5 is the five-dimensional bulk manifold, gg is the determinant of the metric tensor gμνg_{\mu\nu}, RR is the Ricci curvature scalar, Fμν=∂μAν−∂νAμF_{\mu\nu} = \partial_\mu A_\nu- \partial_\nu A_\mu is the Maxwell strength tensor, κ52≡8πG5\kappa_5^2 \equiv8\pi G_5, where G5G_5 is the five-dimensional Newton’s constant, and V(ϕ)V(\phi) is the dilaton potential. Since the dilaton is a real scalar field, there is no minimal coupling with the Maxwell field. Instead, the Fμν2F_{\mu\nu}^{2} term in the EMD action is scaled by the Maxwell-dilaton coupling function, f(ϕ)f(\phi), so that the RG flow is coupled to the baryon current without breaking U(1)BU(1)_{B}. We remark that along with κ2\kappa^{2}, V(ϕ)V(\phi), and f(ϕ)f(\phi), there is another free parameter in the EMD setup which is the asymptotic AdS radius, LL. As in previous works, we set here L=1L=1, and exchange it with an energy scale Λ\Lambda as a free parameter of the holographic model. In this way, the number of free parameters of the bottom-up setup is preserved, and Λ\Lambda is then interpreted as a characteristic energy scale of the non-conformal holographic model (playing a role analogous to the QCD dimensional transmutation scale, ΛQCD\Lambda_{\mathrm{QCD}}). In practice, we use powers of Λ\Lambda to express dimensionful observables of the dual QFT at the boundary in powers of MeV. These observables are originally computed on the gravity side of the holographic gauge-gravity correspondence in powers of the inverse of the asymptotic AdS radius LL of the bulk geometry.

The general EMD field equations are obtained by extremizing the EMD bulk action (1) with respect to the fields gμνg_{\mu\nu}, AμA^{\mu}, and ϕ\phi, respectively,

Rμν−gμν3[V(ϕ)−f(ϕ)4Fαβ2]−12∂μϕ∂νϕ−f(ϕ)2gαβFμαFνβ=0,(2)R_{\mu\nu}-\frac{g_{\mu\nu}}{3}\left[V(\phi)-\frac{f(\phi)}{4}F_{\alpha\beta}^{2}\right]-\frac{1}{2}\partial_{\mu}\phi\partial_{\nu}\phi-\frac{f(\phi)}{2}g^{\alpha\beta}F_{\mu\alpha}F_{\nu\beta}=0, \tag*{(2)}
∂μ(−gf(ϕ)gμαgνβFαβ)=0,(3)\partial_{\mu}\left(\sqrt{-g}f(\phi)g^{\mu\alpha}g^{\nu\beta}F_{\alpha\beta}\right)=0, \tag*{(3)}
1−g∂μ(−ggμν∂νϕ)−∂V(ϕ)∂ϕ−Fμν24∂f(ϕ)∂ϕ=0.(4)\frac{1}{\sqrt{-g}}\partial_{\mu}\left(\sqrt{-g}g^{\mu\nu}\partial_{\nu}\phi\right)-\frac{\partial V(\phi)}{\partial\phi}-\frac{F_{\mu\nu}^{2}}{4}\frac{\partial f(\phi)}{\partial\phi}=0. \tag*{(4)}

In order to describe isotropic and homogeneous states at finite temperature and baryon charge density in thermodynamic equilibrium, we consider a charged black hole Ansatz for the EMD bulk fields depending solely on the holographic radial coordinate, rr, which is the fifth dimension perpendicular to the spacetime directions parallel to the boundary. This fifth dimension corresponds to a geometrization of the RG flow of the dual QFT at the boundary, with ultraviolet information being encoded close to the boundary, while infrared information is encoded close to the black hole event horizon. Then, with a particular choice of the gauge fixing associated to diffeomorphism invariance, the Ansatz for the EMD fields can be written as follows [39, 43, 47, 54, 57],

ds2=gμνdxμdxν=e2A(r)[−h(r)dt2+dx⃗2]+dr2h(r),ϕ=ϕ(r),Aμ=Φ(r)δμ0,(5)ds^{2}=g_{\mu\nu}dx^{\mu}dx^{\nu}=e^{2A(r)}\left[-h(r)dt^{2}+d\vec{x}^{2}\right]+\frac{dr^{2}}{h(r)},\qquad\phi=\phi(r),\qquad A_{\mu}=\Phi(r)\delta_{\mu}^{0}, \tag*{(5)}

where the boundary lies at r→∞r\to\infty and the radial location of the black hole horizon rHr_{H} is given by the largest root of the blackening function, h(rH)=0h(r_{H})=0. By substituting the above Ansatz into the EMD field equations, one obtains,

ϕ′′(r)+[h′(r)h(r)+4A′(r)]ϕ′(r)−1h(r)[∂V(ϕ)∂ϕ−e−2A(r)Φ′(r)22∂f(ϕ)∂ϕ]=0,(6)\phi^{\prime\prime}(r)+\left[\frac{h^{\prime}(r)}{h(r)}+4A^{\prime}(r)\right]\phi^{\prime}(r)-\frac{1}{h(r)}\left[\frac{\partial V(\phi)}{\partial\phi}-\frac{e^{-2A(r)}\Phi^{\prime}(r)^{2}}{2}\frac{\partial f(\phi)}{\partial\phi}\right]=0, \tag*{(6)}
Φ′′(r)+[2A′(r)+dln⁡f(ϕ)dϕϕ′(r)]Φ′(r)=0,(7)\Phi^{\prime\prime}(r)+\left[2A^{\prime}(r)+\frac{d\ln f(\phi)}{d\phi}\phi^{\prime}(r)\right]\Phi^{\prime}(r)=0, \tag*{(7)}
A′′(r)+ϕ′(r)26=0,(8)A^{\prime\prime}(r)+\frac{\phi^{\prime}(r)^{2}}{6}=0, \tag*{(8)}
h′′(r)+4A′(r)h′(r)−e−2A(r)f(ϕ)Φ′(r)2=0,(9)h^{\prime\prime}(r)+4A^{\prime}(r)h^{\prime}(r)-e^{-2A(r)}f(\phi)\Phi^{\prime}(r)^{2}=0, \tag*{(9)}
h(r)[24A′(r)2−ϕ′(r)2]+6A′(r)h′(r)+2V(ϕ)+e−2A(r)f(ϕ)Φ′(r)2=0.(10)h(r)\left[24A^{\prime}(r)^{2}-\phi^{\prime}(r)^{2}\right]+6A^{\prime}(r)h^{\prime}(r)+2V(\phi)+e^{-2A(r)}f(\phi)\Phi^{\prime}(r)^{2}=0. \tag*{(10)}

which is a set of coupled ordinary differential equations discussed in detail in [43, 47, 54, 57]. Below we briefly review some of the main points involved and, in subsection II A, we present a new and very efficient way of numerically solving these differential equations.

The set of dynamical equations of motion for the unknown radial functions ϕ(r)\phi(r), Φ(r)\Phi(r), A(r)A(r), and h(r)h(r), given by Eqs. (6)–(9), together with the constraint Eq. (10), are solved numerically. Two different sets of coordinates are usually employed to analyze the structure of these differential equations and the asymptotic behavior of the bulk fields near the black hole horizon, in the deep infrared, and close to the boundary, corresponding to a nontrivial ultraviolet (UV) fixed point of the holographic RG flow. One of these sets of coordinates corresponds to the so-called standard coordinates, denoted here with a tilde, in terms of which the blackening function goes to unity at the boundary, h~(r→∞)=1\tilde{h}(r\to\infty)=1, and also A~(r→∞)→r~\tilde{A}(r\to\infty)\to\tilde{r}, so that holographic formulas for physical observables are written in standard form in these coordinates. The other set of coordinates are the so-called numerical coordinates, denoted here without a tilde, which are defined in terms of convenient rescalings of the standard coordinates. These rescalings are done with the purpose of fixing definite numerical values for the radial position of the event horizon, rH=0r_{H}=0, and also for some of the initially undetermined infrared Taylor-expansion coefficients of the bulk fields close to the event horizon, since for starting the numerical integration of the bulk equations of motion at the black hole horizon (which then proceeds up to the boundary), all the infrared Taylor-expansion coefficients must have definite numerical values. In fact, after those rescalings, all the near-horizon coefficients are expressed as functions of just two initially undetermined infrared coefficients, ϕ0≡ϕ(rH=0)\phi_{0} \equiv\phi(r_{H}=0) and Φ1≡Φ′(rH=0)\Phi_{1} \equiv\Phi'(r_{H}=0). These are the two near-horizon conditions which one needs to choose in order to construct different numerical solutions for the above specified field equations. In that way, each pair of event horizon conditions (ϕ0,Φ1)(\phi_{0},\Phi_{1}) numerically generates a charged black hole solution which is dual to a given thermodynamic equilibrium state at the boundary QFT, with definite values of temperature (TT), baryon chemical potential (μB\mu_{B}), entropy density (ss), and baryon charge density (ρB\rho_{B}) [43]. By generating a large ensemble of charged black hole solutions one can therefore construct the holographic phase diagram of the model in the (T,μB)(T,\mu_{B})-plane.

In the numerical coordinates, one can show from Eqs. (6)–(9) that the EMD fields scale in the UV limit r→∞r \to\infty, near the boundary and far from the event horizon, according to [43, 47, 54, 57],

A(r)=α(r)+O(e−2να(r)),(11)A(r)=\alpha(r)+O\left(e^{-2\nu\alpha(r)}\right), \tag*{(11)}
h(r)=h0far+O(e−4α(r)),(12)h(r)=h_{0}^{\mathrm{far}}+O\left(e^{-4\alpha(r)}\right), \tag*{(12)}
ϕ(r)=ϕAe−να(r)+O(e−2να(r)),(13)\phi(r)=\phi_{A}e^{-\nu\alpha(r)}+O\left(e^{-2\nu\alpha(r)}\right), \tag*{(13)}
Φ(r)=Φ0far+Φ2fare−2α(r)+O(e−(2+ν)α(r)),(14)\Phi(r)=\Phi_{0}^{\mathrm{far}}+\Phi_{2}^{\mathrm{far}}e^{-2\alpha(r)}+O\left(e^{-(2+\nu)\alpha(r)}\right), \tag*{(14)}

where α(r)=A−1farr+A0far\alpha(r)=A_{-1}^{\mathrm{far}}r+A_{0}^{\mathrm{far}}. The UV coefficient A−1far=1/h0farA_{-1}^{\mathrm{far}}=1/h_{0}^{\mathrm{far}} in α(r)\alpha(r) can be found by using the constraint Eq. (10), while the UV coefficient Φ2far\Phi_{2}^{\mathrm{far}} can be found by using the radially conserved Gauss charge [43, 54],

Φ2far=−h0far2f(0)f(ϕ0)Φ1.(15)\Phi_{2}^{\mathrm{far}}=-\frac{\sqrt{h_{0}^{\mathrm{far}}}}{2f(0)}f(\phi_{0})\Phi_{1}. \tag*{(15)}

The remaining UV coefficients, A0farA_{0}^{\mathrm{far}}, ϕA\phi_{A}, h0farh_{0}^{\mathrm{far}}, and Φ0far\Phi_{0}^{\mathrm{far}}, are obtained by matching the UV expansions (11)–(14) to the near-boundary behavior of the numerical solutions of the EMD equations of motion. In particular, the UV coefficients ϕA\phi_{A}, h0farh_{0}^{\mathrm{far}}, Φ0far\Phi_{0}^{\mathrm{far}}, and Φ2far\Phi_{2}^{\mathrm{far}} are relevant for the calculation of thermodynamic observables. Finally, the exponent in the leading UV contribution for ϕ(r→∞)\phi(r \to\infty) in Eq. (13) involves ν≡d−Δ\nu\equiv d-\Delta, where,

Δ≡d+d2+4m22,(16)\Delta\equiv\frac{d+\sqrt{d^{2}+4m^{2}}}{2}, \tag*{(16)}

is the scaling dimension of the boundary QFT operator dual to the bulk dilaton field ϕ\phi, with −d2/4=−4≤m2≡d2V(0)/dϕ2<0-d^{2}/4=-4 \leq m^{2}\equiv d^{2}V(0)/d\phi^{2}<0 for a relevant operator satisfying the Breitenlohner-Freedman bound for massive scalar fields on stable asymptotically AdS5 spacetimes [80, 81] (see also [82]).

The holographic formulas for the thermodynamic observables referred above are given by [39, 43, 47, 54, 57],

T=−g~tt′g~rr′4π∣r~=r~HΛ=eA~(r~H)4π∣h~′(r~H)∣Λ=14πϕA1/νh0farΛ,(17)T=\left.\frac{\sqrt{-\tilde{g}_{tt}^{\prime}\tilde{g}_{rr}^{\prime}}}{4\pi}\right|_{\tilde{r}=\tilde{r}_{H}}\Lambda=\frac{e^{\tilde{A}(\tilde{r}_{H})}}{4\pi}\left|\tilde{h}^{\prime}(\tilde{r}_{H})\right|\Lambda=\frac{1}{4\pi\phi_{A}^{1/\nu}\sqrt{h_{0}^{\mathrm{far}}}}\Lambda, \tag*{(17)}
μB=lim⁡r~→∞Φ~(r~)Λ=Φ0farϕA1/νh0farΛ,(18)\mu_{B}=\lim_{\tilde{r}\to\infty}\tilde{\Phi}(\tilde{r})\Lambda=\frac{\Phi_{0}^{\mathrm{far}}}{\phi_{A}^{1/\nu}\sqrt{h_{0}^{\mathrm{far}}}}\Lambda, \tag*{(18)}
s=SVΛ3=AH4G5VΛ3=2πκ52e3A~(r~H)Λ3=2πκ52ϕA3/νΛ3,(19)s=\frac{S}{V}\Lambda^{3}=\frac{A_{H}}{4G_{5}V}\Lambda^{3}=\frac{2\pi}{\kappa_{5}^{2}}e^{3\tilde{A}(\tilde{r}_{H})}\Lambda^{3}=\frac{2\pi}{\kappa_{5}^{2}\phi_{A}^{3/\nu}}\Lambda^{3}, \tag*{(19)}
ρB=lim⁡r~→∞∂L∂(∂r~Φ~)Λ3=−Φ2farκ52ϕA3/νh0farΛ3,(20)\rho_{B}=\lim_{\tilde{r}\to\infty}\frac{\partial\mathcal{L}}{\partial(\partial_{\tilde{r}}\tilde{\Phi})}\Lambda^{3}=-\frac{\Phi_{2}^{\mathrm{far}}}{\kappa_{5}^{2}\phi_{A}^{3/\nu}\sqrt{h_{0}^{\mathrm{far}}}}\Lambda^{3}, \tag*{(20)}

where AHA_{H} is the area of the black hole event horizon. Above, we first wrote the holographic formulas in the standard coordinates and next converted them to the corresponding expressions in the numerical coordinates, where the numerical solutions are made available. As mentioned before, we note the use of adequate powers of the energy scale Λ\Lambda employed to express the corresponding observables in powers of MeV.

The dimensionless ratio,

χ^2B≡χ2BT2≡∂2(P/T4)∂(μB/T)2=∂(ρB/T3)∂(μB/T),(21)\hat{\chi}_{2}^{B}\equiv\frac{\chi_{2}^{B}}{T^{2}}\equiv\frac{\partial^{2}(P/T^{4})}{\partial(\mu_{B}/T)^{2}}=\frac{\partial(\rho_{B}/T^{3})}{\partial(\mu_{B}/T)}, \tag*{(21)}

is called the reduced second order baryon susceptibility, where PP is the plasma pressure. By considering this expression in the limit of zero chemical potential, one may derive the following holographic formula [43, 47],

χ^2B(T,μB=0)=116π2sT31f(0)∫rH∞dr e−2A(r)f(ϕ(r))−1,(22)\hat{\chi}_2^B(T,\mu_B=0)=\frac{1}{16\pi^2}\frac{s}{T^3}\frac{1}{f(0)}\int_{r_H}^{\infty}dr\,e^{-2A(r)}f(\phi(r))^{-1}, \tag*{(22)}

which is to be evaluated on top of the numerical EMD solutions generated with the horizon condition Φ1=0\Phi_1=0, corresponding to setting μB=0\mu_B=0. As discussed in [47, 54, 57], in order to numerically evaluate the integral (22) one must make the substitutions rH→rstartr_H\to r_{\mathrm{start}} and ∞→rmax\infty\to r_{\mathrm{max}} in the limits of integration, with rstartr_{\mathrm{start}} being a small infrared cutoff, typically of O(10−8)O(10^{-8}), which is employed to avoid the singular point of the EMD field equations at the rescaled numerical horizon rH=0r_H=0, and rmaxr_{\mathrm{max}} corresponding to a numerical parametrization of the radial location of the boundary of the bulk geometry, which ideally should lie at r→∞r\to\infty. However, because it is not possible to use infinity as a number in numerical computations, one needs to stop the radial integration of the equations of motion at rmaxr_{\mathrm{max}}, which is some number typically around 2−102-10. In practice, this number is chosen so that R(rmax)≈RAdS5=−20R(r_{\mathrm{max}})\approx R_{\mathrm{AdS}_5}=-20 (in units of L=1L=1) within some small numerical tolerance, where RAdS5R_{\mathrm{AdS}_5} is the Ricci scalar curvature of the AdS5\mathrm{AdS}_5 spacetime corresponding to the strongly coupled UV fixed point of the holographic RG flow.

It is important to remark that classical gauge-gravity models have a nontrivial strongly coupled UV fixed point, contrary to the trivial UV fixed point of QCD, which implies that differently from QCD, classical gauge-gravity models are not asymptotically free, but just asymptotically safe. This phenomenological limitation of classical holographic plasmas becomes clearly manifest by the constant value of shear viscosity over entropy density, η/s=1/4π\eta/s=1/4\pi, which is valid for all values of temperature and baryon chemical potential [16, 83, 84]. This is in striking contrast to the temperature-dependent profile for η/s\eta/s in QCD, which increases with temperature in the asymptotically free limit of very high temperatures [85] (see [86, 87] for the phenomenological implications of this in heavy-ion collisions). In fact, as it is well-known, classical gauge-gravity models can only possibly provide a good description of some target system when such a system is operating in a strongly coupled regime. However, we also remark that, based on the theoretical foundations of the holographic duality, one expects that higher curvature corrections to the bulk gravity action should drive the running coupling towards lower values, and it has been indeed shown in [88] that η/s\eta/s may acquire a temperature-dependent profile in higher curvature holographic models with a dilaton field employed to break conformal symmetry. In fact, this has been recently explored in [74], where a holographic Bayesian profile was provided for the shear viscosity of the QGP at zero baryon density using higher derivative corrections to the EMD setup. This approach provided a qualitative improvement in the agreement with the JETSCAPE result for the shear viscosity to entropy density ratio [13, 15], relatively to the constant value η/s=1/4π\eta/s=1/4\pi valid for holographic gauge-gravity models with just two derivatives of the metric field in the bulk action. Such a qualitative improvement refers to the production of a minimum for η/s\eta/s in the crossover region. However, the obtained numerical variation on the temperature profile for η/s\eta/s is so small that, quantitatively, the approach followed in [74] is almost indistinguishable from the usual constant holographic result η/s=1/4π\eta/s=1/4\pi. This means that one still needs to devise another approach to try quantitatively reproducing the temperature profile for η/s\eta/s in QCD by using holographic models.

New numerical scheme for UV near-boundary asymptotics

Regarding the set of UV coefficients {ϕA,h0far,Φ0far,Φ2far}\{\phi_A,h_0^{\mathrm{far}},\Phi_0^{\mathrm{far}},\Phi_2^{\mathrm{far}}\} relevant for the computation of thermodynamic observables in Eqs. (17)–(20), we saw that Φ2far\Phi_2^{\mathrm{far}} can be evaluated according to Eq. (15), once h0farh_0^{\mathrm{far}} is known. Two of the remaining three UV coefficients can be easily obtained from the numerical solutions and from Eqs. (12) and (14) by setting h0far=h(rmax)h_0^{\mathrm{far}}=h(r_{\mathrm{max}}) and Φ0far=Φ(rmax)\Phi_0^{\mathrm{far}}=\Phi(r_{\mathrm{max}}). However, the extraction of ϕA\phi_A is more challenging, as it appears in Eq. (13) scaled by an exponentially suppressed factor near the boundary. Naively dividing ϕ(rmax)\phi(r_{\mathrm{max}}) by this exponentially small factor near the boundary to obtain ϕA(naive)=ϕ(rmax)/e−νA(rmax)\phi_A^{(\mathrm{naive})}=\phi(r_{\mathrm{max}})/e^{-\nu A(r_{\mathrm{max}})} would involve a ratio between two very small values, raising numerical issues with precision and noise.

This issue has been previously addressed by fitting the UV scaling form for ϕ(r)\phi(r) in Eq. (13) to the numerical solutions within some adaptive radial region [54, 57]. However, performing this fit: i) demands the judicious choice of the fitting region; ii) is computationally costly; and iii) often leads to numerical noise, which may be particularly large in the phase transition regions of the phase diagram of the model [57]. Here we discuss a new method, already successfully employed in Ref. [63] (although not discussed there), which we found to be much more stable and computationally efficient.

The UV coefficient ϕA\phi_A can be extracted via the inclusion of an extra field C(r)C(r), to which corresponds an extra differential equation of relaxational form,

dC(r)dr=−ΓC[C(r)−ϕ(r)eνA(r)].(23)\frac{dC(r)}{dr}=-\Gamma_C\left[C(r)-\phi(r)e^{\nu A(r)}\right]. \tag*{(23)}

New numerical scheme for UV near-boundary asymptotics

The solution to Eq. (23) can be written as,

C(r)=C(rH)e−ΓC(r−rH)+ΓC∫rHrdr′e−ΓC(r−r′)ϕ(r′)eνA(r′).(24)C(r) = C(r_H)e^{-\Gamma_C(r-r_H)}+\Gamma_C\int_{r_H}^{r}dr'e^{-\Gamma_C(r-r')}\phi(r')e^{\nu A(r')}. \tag*{(24)}

For r−rH≫1/ΓCr-r_H\gg1/\Gamma_C, the dependence on the horizon conditions is damped away and C≈⟨ϕ(r)eνA(r)⟩exp⁡(−ΓCΔr)C\approx\langle\phi(r)e^{\nu A(r)}\rangle_{\exp(-\Gamma_C\Delta r)} becomes the average of ϕA(naive)(r)=ϕ(r)eνA(r)\phi_A^{(\mathrm{naive})}(r)=\phi(r)e^{\nu A(r)} with an exponential weight. While ϕA(naive)(r)\phi_A^{(\mathrm{naive})}(r) is subject to numerical noise, noisy features are eliminated by averaging over a range Δr∼1/ΓC\Delta r\sim1/\Gamma_C, with exponentially decaying weight.

The convergence of the method can be evaluated by checking that ∣C−ϕeνA∣/C≪1\lvert C-\phi e^{\nu A}\rvert/C\ll1. In practice, we take the method to have converged whenever ∣C−ϕeνA∣/C\lvert C-\phi e^{\nu A}\rvert/C is smaller than a precision parameter ϵC≪1\epsilon_C\ll1. The UV coefficient ϕA\phi_A can then be directly estimated by setting ϕA=C(rmax⁡)\phi_A=C(r_{\max}).

This new method has the following advantages:

  1. Efficiency: ϕA\phi_A can be extracted directly from the final value of C(r)C(r), with no need for a costly fitting procedure on a judicious choice of radial interval;

  1. Stability: the resulting values of ϕA\phi_A turn out to be less subject to numerical noise — we believe this is so because ϕA\phi_A in this new method depends linearly on the values of ϕ(r)\phi(r), which is not the case for a least squares fitting;

  1. Computational time: with the improved stability against numerical noise, the application of filters to the numerical results is rendered unnecessary, drastically reducing the computational cost of the final results.

Although there is no need for a judicious choice of fitting range, the rate ΓC\Gamma_C must be carefully chosen. If ΓC\Gamma_C is too large, the value of ϕA\phi_A will be averaged over just a few points and the result will be susceptible to numerical noise. On the other hand, if ΓC\Gamma_C is too small, relaxation to the correct value will take longer, and convergence will become slow. Moreover, requiring an improved precision on ϕA\phi_A and on the Ricci scalar RR will lead to convergence at larger values of the AdS radius, where the extraction of ϕA\phi_A becomes increasingly challenging. Therefore, the value of ΓC\Gamma_C must be adjusted according to precision. The new numerical method is not only computationally robust but it also solves many issues related to noise associated with the previously applied numerical method, as can be clearly seen in the left panel of Figure 1.

Figure 1. Left panel: A constant ϕ0\phi_0-line starting at (T,μB)=(60,0)(T,\mu_B)=(60,0) MeV, computed with a prefactor ϕA\phi_A extracted by different means. The dot-dashed blue curve shows results obtained using the methods presented in Ref. [57], while the solid red curve represents results extracted with the present relaxational approach. Right panel: Ultraviolet scaling coefficient ϕA\phi_A, obtained from Eq. (24), with ΓC/Λ=10\Gamma_C/\Lambda=10 (dashed blue curves) and ΓC/Λ=5\Gamma_C/\Lambda=5 (magenta dot-dashed curves), for ramdomly selected values of 115 MeV≤T≤250 MeV115\ \mathrm{MeV}\le T\le250\ \mathrm{MeV} and μB≤3.5T\mu_B\le3.5T.

In order to test how the convergence of this new method depends on the choice of ΓC\Gamma_C, the right panel of Figure 1 shows how our new approach to extract ϕA\phi_A evolves with increasing radius, for different values of the relaxation rate ΓC\Gamma_C and randomly selected values of 115 MeV≤T≤250 MeV115\ \mathrm{MeV}\le T\le250\ \mathrm{MeV} and μB≤3.5T\mu_B\le3.5T. Notice that, in the right panel of Figure 1, we do not show results up to the maximum value of r=rmax⁡r=r_{\max} where numerical convergence is found. The reason behind this is that the results being compared become harder to distinguish when values of the radius up to r=rmax⁡r=r_{\max} are shown.

New algorithm to locate the critical point

Our goal in the Bayesian analysis of the EMD model presented in Ref [63] was to draw predictions for the critical point from our posterior distribution. Here, we describe the CEP-locating algorithm that was devised to extract the location of the conjectured CEP in the prior and posterior samples of the Bayesian analysis. In this manuscript, we use this algorithm for a fast determination of the CEP location for each Bayesian posterior we utilize to construct our uncertainty bands.

The phase diagram in our model is of the form shown in the left panel of Figure 2, where lines of constant ϕ0\phi_0 bend and often intersect each other. This clearly indicates the existence of several dual black-hole solutions corresponding to the same μB\mu_B and TT, and therefore to different phases. In the left panel of Figure 2, the region within which constant ϕ0\phi_0 lines cross each other is delimited by the spinodal lines, shown as dashed black lines. In this region there are three black hole duals for each point, corresponding to thermodynamically stable, metastable and unstable solutions. We observe that the two spinodal lines shown in the left panel of Figure 2 meet at a cusp where multiple ϕ0\phi_0 lines all cross each other, so that ϕ0\phi_0 is undefined. As a consequence, the EoS becomes non-analytic, indicating the location of the second-order CEP, marked with a star. This crossing of multiple constant ϕ0\phi_0 lines can be used to quickly locate the CEP.

Phase diagram and critical-point search

Figure 2. Left panel: Lines of constant ϕ0\phi_0 are displayed as thin solid lines across the phase diagram, colored according to the value of ϕ0\phi_0. Stable, metastable and unstable states are realized as different dual black holes corresponding to the same value of (μB,T)(\mu_B,T). Thick dashed lines correspond to the spinodal lines, where metastable solutions become unstable. The star marks the position of the critical endpoint, where these lines end. Right panel: The process of finding the critical point at which constant ϕ0\phi_0-lines cross each other. Parabolas interpolating previously found values of (μB,T)(\mu_B,T) on a constant ϕ0\phi_0 line are used to find the critical point.

To find crossings between two lines of constant ϕ0\phi_0, we interpolate these lines with quadratic functions T(ϕ0)(μB)=a(ϕ0)μB2+b(ϕ0)μB+c(ϕ0)T^{(\phi_0)}(\mu_B)=a^{(\phi_0)}\mu_B^2+b^{(\phi_0)}\mu_B+c^{(\phi_0)}. From the interpolated curves, checking for crossings at real (μB,T)(\mu_B,T) pairs becomes trivial. By also interpolating Φ1(ϕ0)(μB)\Phi_1^{(\phi_0)}(\mu_B) for each of these lines, we can estimate the values of (ϕ0,Φ1)(\phi_0,\Phi_1) where the two lines intersect. Using these values as initial conditions to solve Eqs. (6) to (9), we can find the actual values of (μB,T)(\mu_B,T) and check if they are indeed the same within specified precision and accuracy goals. If they are not the same within the requested errors, we can use the new point to update our quadratic interpolation of T(ϕ0)(μB)T^{(\phi_0)}(\mu_B) and Φ1(ϕ0)(μB)\Phi_1^{(\phi_0)}(\mu_B) and repeat the process up to a maximum number of trials, until either a crossing between the two lines is found, the two lines move away from each other, or the crossing enters unphysical values of ϕ0\phi_0, Φ1\Phi_1 and TT and μB\mu_B (e.g., T<0T<0 or Φ1>Φ1max⁡\Phi_1>\Phi_1^{\max}).

We start by interpolating T(i)(μB)T^{(i)}(\mu_B) and Φ1i(μB)\Phi_1^i(\mu_B) for different values of ϕ0=ϕ0(i)\phi_0=\phi_0^{(i)}, i=0,1,…,Nϕ0i=0,1,\ldots,N_{\phi_0}, within a given range, and looking for crossings between neighboring lines T(i)(μB)T^{(i)}(\mu_B), T(i+1)(μB)T^{(i+1)}(\mu_B). To increase the number of crossings we are able to find, we select values of ϕ0(i)\phi_0^{(i)} for which an intersection between neighboring lines was found, and search for additional intersections between every pair of non-neighboring lines. Again, this is done by iteratively interpolating these lines until the process converges or fails.

The process of finding the points at which constant ϕ0\phi_0 lines cross each other is illustrated in the right panel of Figure 2, where different line styles corresponds to different iterations, ntry=0,1,…,4n_{\mathrm{try}}=0,1,\ldots,4, in the search for the intersection points, marked by a ‘×\times’ symbol. After just a few trials, the remaining intersection points converge to the region between the spinodal lines, as should be expected. Finally, we take the crossing point at the lowest value of μB\mu_B as the location of the critical point. This point is marked as an unfilled star in the left panel, and as a red star in the right panel of Figure 2. The entire process can be performed in a few seconds, which is sufficiently fast for use in our Bayesian analysis.

Bayesian inference and lattice QCD results

In this section, we present the functional forms chosen for the free functions V(ϕ)V(\phi) and f(ϕ)f(\phi) and review how to use Bayesian inference to fix all the free parameters of the EMD model by taking continuum extrapolated LQCD data with Nf=2+1N_f = 2 + 1 flavors and physical quark masses as first principles inputs for the equation of state and baryon susceptibility of the QGP at zero chemical potential.

With the holographic formulas (17), (18), (19), and (20) describing the basic thermodynamic observables defining the equation of state for the strongly coupled plasma, one may fix {V(ϕ),κ52,Λ}\{V(\phi),\kappa_5^2,\Lambda\} by imposing that the equation of state of the holographic model at zero baryon chemical potential matches the corresponding LQCD data [89]. Moreover, by numerically solving the integral (22) over the background EMD solutions at μB=0\mu_B = 0 and by requiring that the baryon susceptibility at zero chemical potential in the holographic model matches the corresponding LQCD result [77], one can fix f(ϕ)f(\phi).

Here we make use of two ansatze for V(ϕ)V(\phi) and f(ϕ)f(\phi). The first functional form is a polynomial-hyperbolic Ansatz (PHA), given by

V(ϕ)=−12cosh⁡(γϕ)+b2ϕ2+b4ϕ4+b6ϕ6,(25)V(\phi) = -12\cosh(\gamma\phi) + b_2\phi^2 + b_4\phi^4 + b_6\phi^6, \tag*{(25)}
f(ϕ)=sech⁡(c1ϕ+c2ϕ2+c3ϕ3)1+d1+d11+d1sech⁡(d2ϕ).(26)f(\phi) = \frac{\operatorname{sech}(c_1\phi+ c_2\phi^2 + c_3\phi^3)}{1+d_1} + \frac{d_1}{1+d_1}\operatorname{sech}(d_2\phi). \tag*{(26)}

The second one is a parametric Ansatz (PA), with the following functional form

V(ϕ)=−12cosh⁡[(γ1ΔϕV2+γ2ϕ2ΔϕV2+ϕ2)ϕ],(27)V(\phi) = -12\cosh\left[\left(\frac{\gamma_1\Delta\phi_V^2+\gamma_2\phi^2}{\Delta\phi_V^2+\phi^2}\right)\phi\right], \tag*{(27)}
f(ϕ)=1−(1−A1)[12+12tanh⁡(ϕ−ϕ1δϕ1)]−A2[12+12tanh⁡(ϕ−ϕ2δϕ2)].(28)f(\phi) = 1-(1-A_1)\left[\frac{1}{2}+\frac{1}{2}\tanh\left(\frac{\phi-\phi_1}{\delta\phi_1}\right)\right]-A_2\left[\frac{1}{2}+\frac{1}{2}\tanh\left(\frac{\phi-\phi_2}{\delta\phi_2}\right)\right]. \tag*{(28)}

A few considerations are taken into account in order to propose the above ansatze for V(ϕ)V(\phi) and f(ϕ)f(\phi) (see the recent review [39] for detailed discussions). Regarding the dilaton potential V(ϕ)V(\phi), it is expected by the standards of the holographic dictionary that in the UV regime (ϕ→0)(\phi\to0) one recovers in the bulk action, from V(ϕ)V(\phi), twice the value of the cosmological constant of the AdS5_5 spacetime. Therefore, one requires that, V(0)=−12V(0)=-12 (in units of L=1L=1). In this way, the functional forms (25) and (27) for V(ϕ)V(\phi) are such that, close to the boundary at ϕ(r→∞)→0\phi(r\to\infty)\to0, the Ricci scalar approaches R=−20R=-20 (in units of L=1L=1), which allows us to have asymptotically AdS5_5 background solutions, as prescribed by the holographic dictionary. Besides that, the functional forms proposed for V(ϕ)V(\phi) are engineered so as to have the holographic entropy density at zero chemical potential matching the corresponding LQCD result from Ref. [89]. As for the Maxwell-dilaton coupling function f(ϕ)f(\phi), it is constrained by matching to the LQCD result from Ref. [77] for the baryon susceptibility at zero chemical potential, the corresponding holographic formula given by Eq. (22). An appealing feature of χ^2B(T,μB=0)\hat{\chi}_2^B(T,\mu_B=0) in Ref. [77] is its fast inflection around T=140T=140 MeV. The freedom in AμA_\mu allows us to set a normalization such that f(0)=1f(0)=1, as originally proposed in Ref. [43]. Then, to reproduce this feature of the baryon susceptibility from LQCD, we must consider a function f(ϕ)f(\phi) that presents a fast variation close to the boundary ϕ(r→∞)→0\phi(r\to\infty)\to0 [39]. These features can be achieved with Eqs. (25) and (26) corresponding to the PHA ansatz. The PA ansatz, represented by Eqs. (27) and (28), reproduce the main features of Eqs. (25) and (26) in a more transparent way, in the sense that it is easier to tune specific features of V(ϕ)V(\phi) and f(ϕ)f(\phi) by adjusting the free parameters in the PA ansatz [63]. We note that both ansatze lead to mostly similar results, at least in the regime of applicability of the present model. Results for the transport coefficients obtained within the two ansatze are compared in section IV below.

These specific ansatze were introduced in Ref. [63], where Bayesian inference was employed to determine the free parameters in the functions that yield optimal agreement with lattice QCD results for each chosen functional form of V(ϕ)V(\phi) and f(ϕ)f(\phi). The prior distributions for V(ϕ)V(\phi) and f(ϕ)f(\phi) led to critical points widely scattered across the QCD phase diagram, and in some cases resulted in no critical point at all. However, posterior distributions constrained by lattice QCD data at zero chemical potential revealed holographic critical points tightly clustered within the ranges T=101T=101–108108 MeV and μB=560\mu_B=560–625625 MeV, at 95% confidence level. In Ref. [63], the Bayesian analysis focused primarily on thermodynamic quantities. In this work, our primary objective is to use the posterior distributions obtained from the previous analysis to examine transport coefficients and energy-loss observables in the QGP. Specifically, we aim to investigate the behavior of these physical observables near the critical end point and across the first order phase transition line and to assess its robustness against the Bayesian analysis uncertainty band. All figures presented in the main text correspond to the PHA Ansatz. Results for the PA Ansatz are shown in the Appendix A for completeness. As a reminder of the thermodynamic predictions of our model at finite baryon chemical potential, Figure 3 shows the results based on our best-fit parameter for the PHA Ansatz, for s/T3s/T^{3} and P/T4P/T^{4} at finite (T,μB)(T,\mu_{B}) compared to the corresponding LQCD results. We note that since finite μB\mu_{B} lattice results were not used to fix the free parameters of the holographic model, such results follow as true predictions of the holographic setup. Comparing such predictions with LQCD results at finite baryon density, as done in Figure 3, is a clear way of accessing the phenomenological reliability of the present holographic EMD framework. Further comparisons involving holographic predictions for transport coefficients, like the bulk viscosity and the jet quenching parameter, and the profiles favored in multistage Bayesian models simultaneously matching several heavy ion data are presented in Figure 10, to be discussed in the upcoming sections.

Figure 3. Comparison between the Equation of State at finite density obtained from the best fit to lattice results at zero chemical potential for the PHA Ansatz, and the state-of-the-art lattice QCD results from [77]. For detailed comparison between the lattice results and PA, PHA models, see Ref. [63].

Figure 10. Left panel: Holographic bulk viscosity ζ/s\zeta/s Bayesian posterior band at 95% confidence intervals, and best fit (solid line) for PA and PHA ansatze both at μB=0\mu_B = 0 compared with the 90% credible intervals of the JETSCAPE Bayesian model from [13, 15], and the Duke Group results [12]. Right panel: Jet-quenching parameter Bayesian posterior band at 95% confidence level, and best fit (solid line) from EMD model for PA and PHA ansatze at μB=0\mu_B = 0 for three representative values of the ‘t Hooft parameter λt={1/25,1/16,1/9}\lambda_t = \{1/25, 1/16, 1/9\}, compared to the JETSCAPE result from Ref. [13].

As a final remark regarding the physical relevance of the bulk free parameters κ52≡8πG5\kappa_{5}^{2}\equiv8\pi G_{5} and Λ\Lambda, their values can be fixed, together with the form of the dilaton potential V(ϕ)V(\phi), by matching the holographic formulas (17) and (19) to the LQCD result for the entropy density as a function of temperature at zero chemical potential [63]. In top-down holographic models (which is not the case of our bottom-up EMD setup), one can explicitly show that κ52=8πG5∼1/Nc2\kappa_{5}^{2}=8\pi G_{5}\sim1/N_{c}^{2}, where NcN_{c} is the number of colors of the gauge group SU(Nc)SU(N_{c}) of the dual QFT at the boundary. For instance, in the most well-known case corresponding to the strongly interacting N=4\mathcal{N}=4 Supersymmetric Yang-Mills (SYM) plasma with gauge group SU(Nc)SU(N_{c}), the relation (in units of the AdS radius L=1L=1) is κ52=4π2/Nc2\kappa_{5}^{2}=4\pi^{2}/N_{c}^{2}. In the standard holographic gauge-gravity dictionary, NcN_{c} is taken to be very large, so that κ52\kappa_{5}^{2} should be very small, implying weakly-coupled gravity in the bulk. On the other hand, in our approach we do not match our holographic entropy density at zero chemical potential to a large NcN_{c} QCD result, but to the physical SU(Nc=3)SU(N_{c}=3) QCD result extrapolated to the continuum limit with physical values of the quark masses. Although the precise relation between the bulk gravitational constant κ52≡8πG5\kappa_{5}^{2}\equiv8\pi G_{5} and the rank NcN_{c} of the dual QFT gauge group SU(Nc)SU(N_{c}) is unknown in bottom-up models, if we heuristically take into consideration the N=4\mathcal{N}=4 SYM relation κ52=4π2/Nc2\kappa_{5}^{2}=4\pi^{2}/N_{c}^{2} and use Nc=3N_{c}=3, we obtain κ52≈4.4\kappa_{5}^{2}\approx4.4, which is not a small number. In fact, this heuristic estimate has the same order of magnitude of κ52≈11.4\kappa_{5}^{2}\approx11.4, which is the value favored by the Bayesian analyses of LQCD data within the holographic EMD setup performed by considering both, the PHA and PA ansatze — see the Maximum A Posteriori (MAP) values in Table II of Ref. [63]. The bulk gravitational constant itself is not directly observable at the dual boundary QFT, although it appears in holographic formulas for several physical observables of the boundary QFT.

Regarding the energy scale factor Λ\Lambda (which replaces 1/L1/L as a free parameter in our gauge-gravity setup), it plays a role analogous to the ΛQCD\Lambda_{\mathrm{QCD}} scale within the holographic EMD model. Although the value of ΛQCD\Lambda_{\mathrm{QCD}} is not directly observable and depends on the number of active quarks and the particular renormalization scheme considered, a commonly quoted value (for 3 active quark flavors in the MS scheme) is ΛQCD∼330\Lambda_{\mathrm{QCD}}\sim330 MeV. In Table II of Ref. [63], the MAP values of the energy scale factor Λ\Lambda in the holographic EMD setup are ΛEMD(PHA)=1129\Lambda_{\mathrm{EMD}(\mathrm{PHA})}=1129 MeV ≈3.4ΛQCD\approx3.4\Lambda_{\mathrm{QCD}} and ΛEMD(PA)=955\Lambda_{\mathrm{EMD}(\mathrm{PA})}=955 MeV ≈2.9ΛQCD\approx2.9\Lambda_{\mathrm{QCD}}. Like the dimensional transmutation scale ΛQCD\Lambda_{\mathrm{QCD}} in QCD, the energy scale factor Λ\Lambda in the holographic EMD model is not directly observable, although its value is employed to express gravity results in physical units at the dual boundary QFT.

Holographic transport coefficients and energy loss

We revisit here the holographic calculation of several transport coefficients [47, 48, 53, 60] in view of the Bayesian inference procedure discussed in the previous section. The merits of such approach are twofold: a) the Bayesian analysis allows for a much more systematic and precise determination of the best-fit model given some choice for the functional forms of the free functions V(ϕ)V(\phi) and f(ϕ)f(\phi) of the EMD setup; and b) the Bayesian procedure allows for the construction of uncertainty bands for the theoretical predictions of the EMD setup based on statistical posterior distributions, which in turn provides a way of checking the internal robustness of such holographic predictions against variations of the free parameters of the model.

In the EMD setup, the holographic formulas for computing the baryon and thermal conductivities as well as the baryon diffusion coefficient were derived in [44, 48], while the holographic formulas for the shear and bulk viscosities were obtained in [44] — see also [53, 60]. In terms of parton energy loss, the original holographic formulas for the heavy quark drag force, the Langevin diffusion coefficients, and the jet-quenching parameter were derived for the conformal SYM plasma in [90, 91, 92, 93, 94, 95, 96, 97, 98], later extended for non-conformal dilatonic models in [99, 100], and then adapted for EMD setups in [47, 50, 56], see also [60]. Since those formulas are well-known, we will not review their derivations here and refer the interested reader to consult the above references for detailed discussions. However, before presenting the relevant holographic formulas and discussing the corresponding numerical results, we make some brief comments regarding the general ideas involved.

For the isotropic EMD model considered here, there is a spatial SO(3)SO(3) rotation symmetry which separates the diffeomorphism and gauge invariant combinations of linearized perturbations of the bulk fields into different irreducible representations. The SO(3)SO(3) triplet channel is related to the baryon conductivity to be discussed in subsection III A. The SO(3)SO(3) quintuplet and singlet channels are related, respectively, to the shear and bulk viscosities, to be analyzed in subsection III B. The different irreducible representations do not mix at the linear level, so that one needs to solve decoupled differential equations for the corresponding linear perturbations on top of the EMD black hole backgrounds [44]. One can show that the perturbation in the quintuplet channel leads to η/s=1/4π\eta/s = 1/4\pi for all values of (T>0,μB≥0)(T > 0, \mu_B \ge0), as expected for any holographic model which is homogeneous, isotropic, and with at most two derivatives of the bulk metric field in the gravity action [16, 83, 84]. We remark, however, that the dimensionless combination entering the hydrodynamic expression for the viscous part of the boundary gauge theory energy-momentum tensor at finite baryon chemical potential is ηT/(ϵ+P)\eta T/(\epsilon+ P) instead of η/s\eta/s [101, 102], where PP and ϵ\epsilon are, respectively, the pressure and the energy density of the plasma. The dimensionless combination ηT/(ϵ+P)\eta T/(\epsilon+ P) reduces to η/s\eta/s when μB=0\mu_B = 0. However, ηT/(ϵ+P)\eta T/(\epsilon+ P) displays a nontrivial dependence on TT and μB\mu_B at finite baryon density, as we are going to see in subsection III B.

Regarding physical observables describing energy loss for partons traversing the strongly coupled medium, they are typically computed in holography by using the Nambu-Goto (NG) action for probe strings defined over the background field solutions. This probe string action involves the square root of the ‘t Hooft coupling, λt=L2/α′=L2/ℓs2=1/ℓs2\sqrt{\lambda_t} = L^2/\alpha' = L^2/\ell_s^2 = 1/\ell_s^2 (in units of L=1L = 1), where ℓs\ell_s is the fundamental string length, which is typically taken as a free parameter in bottom-up constructions. However, by the standards of the holographic dictionary, one expects that ℓs\ell_s should be very small and, consequently, λt\lambda_t should be very large in the classical limit corresponding to the gauge-gravity duality. Therefore, as done e.g. in [47, 51, 56, 60], by considering infinitely heavy probe quarks traversing the plasma we neglect both the coupling between the background dilaton field and the Ricci scalar induced on the probe string worldsheet, and also the minimal coupling between the background Maxwell field and the string endpoint attached to a probe quark at the boundary, since both couplings are of order zero in the ‘t Hooft coupling, while the NG action is of order 1/21/2 [103, 104]. Thus, both terms do not contribute to leading order in the ‘t Hooft coupling for the calculations of the heavy quark drag force and the Langevin diffusion coefficients. In the case of the jet-quenching parameter related to the energy loss of light partons, we remark that the minimal coupling term is automatically zero because the contributions from each string endpoint at the boundary in this computation cancel out for such a coupling.

Baryon transport: baryon susceptibility, baryon and thermal conductivities and baryon diffusion

In the relativistic Navier-Stokes approximation [105], the baryon current for a hot and baryon dense fluid at the flat four-dimensional boundary gauge theory with Minkowski metric ημν\eta_{\mu\nu} is given by [106]

JBμ=ρBuμ+ΔJBμ,ΔJBμ=TσB(ημν−uμuν)∂ν(μBT),(29)J_B^\mu= \rho_B u^\mu+ \Delta J_B^\mu,\qquad\Delta J_B^\mu= T\sigma_B\left(\eta^{\mu\nu} - u^\mu u^\nu\right)\partial_\nu\left(\frac{\mu_B}{T}\right), \tag*{(29)}

where uμu^\mu is the local velocity field of the fluid and ΔJBμ\Delta J_B^\mu is the dissipative part of the baryon current, which is characterized by the baryon conductivity σB(T,μB)\sigma_B(T,\mu_B). This transport coefficient therefore quantifies how the baryon current responds to gradients of μB/T\mu_B/T in the medium. By the Nernst-Einstein relation [106, 107], it is also proportional to the diffusion of baryon charge in the medium, DB(T,μB)D_B(T,\mu_B), with the proportionality factor being the baryon susceptibility χ2B(T,μB)≡(∂2P/∂μB2)T=(∂ρB/∂μB)T\chi_2^B(T,\mu_B) \equiv(\partial^2 P/\partial\mu_B^2)_T = (\partial\rho_B/\partial\mu_B)_T,

TDB(T,μB)=TσB(T,μB)χ2B(T,μB)=σ^B(T,μB)χ^2B(T,μB),(30)T D_B(T,\mu_B) = T \frac{\sigma_B(T,\mu_B)}{\chi_2^B(T,\mu_B)} = \frac{\hat{\sigma}_B(T,\mu_B)}{\hat{\chi}_2^B(T,\mu_B)}, \tag*{(30)}

where σ^B≡σB/T\hat{\sigma}_B \equiv\sigma_B/T and χ^2B≡χ2B/T2\hat{\chi}_2^B \equiv\chi_2^B/T^2 are dimensionless ratios. On the other hand, the thermal conductivity σT(T,μB)\sigma_T(T,\mu_B), which measures the ability of the medium to conduct heat, is also related to the baryon conductivity as follows [108],

σT=χ2BDBT(ϵ+PρB)2=TσB(sρB+μBT)2,(31)\sigma_T = \frac{\chi_2^B D_B}{T}\left(\frac{\epsilon+P}{\rho_B}\right)^2 = T\sigma_B\left(\frac{s}{\rho_B}+\frac{\mu_B}{T}\right)^2, \tag*{(31)}

where we made use of the fundamental thermodynamic relation, ϵ=Ts+μBρB−P\epsilon=Ts+\mu_B\rho_B-P, and the Nernst-Einstein relation. Since ρB(T,μB→0)→μBχ2B(T,μB→0)→0\rho_B(T,\mu_B\to0)\to\mu_B\chi_2^B(T,\mu_B\to0)\to0, in order to construct a well-defined dimensionless combination in the limit of zero baryon chemical potential, one may consider as in [109],

[μB2σTηT](T,μB)=4πσBTμB2Ts(TsρB+μB)2,(32)\left[\frac{\mu_B^2\sigma_T}{\eta T}\right](T,\mu_B)=4\pi\frac{\sigma_B}{T}\frac{\mu_B^2}{Ts}\left(\frac{Ts}{\rho_B}+\mu_B\right)^2, \tag*{(32)}
→4π(χ2B/T2)2σBTsT3=4πσ^B(χ^2B)2sT3,as μB→0,(33)\to\frac{4\pi}{(\chi_2^B/T^2)^2}\frac{\sigma_B}{T}\frac{s}{T^3}=\frac{4\pi\hat{\sigma}_B}{(\hat{\chi}_2^B)^2}\frac{s}{T^3},\quad\text{as }\mu_B\to0, \tag*{(33)}

where we used the holographic relation η/s=1/4π\eta/s=1/4\pi in writing the RHS of (32). While the limit (33) is finite, the LHS of (32) also depends on the shear viscosity η\eta besides the thermal conductivity σT\sigma_T. Since it is the combination μB2σT\mu_B^2\sigma_T which remains finite in the μB→0\mu_B\to0 limit, a simpler dimensionless combination which does not depend on any other transport coefficient besides the thermal conductivity may be constructed as done in [48, 60],

[μB2σTT4](T,μB)=σBT(μBρBsT+μB2T2)2,(34)\left[\frac{\mu_B^2\sigma_T}{T^4}\right](T,\mu_B)=\frac{\sigma_B}{T}\left(\frac{\mu_B}{\rho_B}\frac{s}{T}+\frac{\mu_B^2}{T^2}\right)^2, \tag*{(34)}
→σBT(sT3T2χ2B)2=σ^B(χ^2B)2(sT3)2,as μB→0.(35)\to\frac{\sigma_B}{T}\left(\frac{s}{T^3}\frac{T^2}{\chi_2^B}\right)^2=\frac{\hat{\sigma}_B}{(\hat{\chi}_2^B)^2}\left(\frac{s}{T^3}\right)^2,\quad\text{as }\mu_B\to0. \tag*{(35)}

In Ref. [110], a hybrid phenomenological approach based on 3+13+1 hydrodynamics and hadronic transport was used to simulate relativistic heavy-ion collisions producing a hot and baryon dense QGP with subsequent hadronization. The baryon conductivity influences the difference between proton and antiproton mean transverse momentum, and also the difference between proton and antiproton elliptic flow. Therefore, the investigation of the dissipative dynamics of baryon charge is relevant for the theoretical modeling of heavy-ion collisions at lower beam energies, when finite baryon chemical potential effects become important.

Notice from the above discussion that in order to obtain the baryon diffusion coefficient (30) and the thermal conductivity (34), besides the basic thermodynamic quantities given in Eqs. (17)–(20), one needs to compute the numerical derivative corresponding to the baryon susceptibility, χ2B(T,μB)=(∂ρB/∂μB)T\chi_2^B(T,\mu_B)=(\partial\rho_B/\partial\mu_B)_T, and the baryon conductivity σB(T,μB)\sigma_B(T,\mu_B). The latter can be obtained from the analysis of the diffeomorphism and gauge invariant perturbation a(r,ω)a(r,\omega) in the SO(3) triplet channel of the homogeneous and isotropic holographic EMD model, corresponding to any of the three spatial components aia^i of the Maxwell field perturbation aμa^\mu. The equation of motion for this perturbation is given by [44],

a′′(r,ω)+[2A′(r)+h′(r)h(r)+∂ϕf(ϕ)f(ϕ)ϕ′(r)]a′(r,ω)+e−2A(r)h(r)[ω2h(r)−f(ϕ)Φ′(r)2]a(r,ω)=0,(36)a''(r,\omega)+\left[2A'(r)+\frac{h'(r)}{h(r)}+\frac{\partial_\phi f(\phi)}{f(\phi)}\phi'(r)\right]a'(r,\omega)+\frac{e^{-2A(r)}}{h(r)}\left[\frac{\omega^2}{h(r)}-f(\phi)\Phi'(r)^2\right]a(r,\omega)=0, \tag*{(36)}

where the prime denotes derivative with respect to rr, and ω\omega is the (angular) frequency of the homogeneous plane wave Ansatz used for the bulk perturbation. One needs to solve (36) on top of the numerical EMD backgrounds by using the infalling wave condition at the event horizon, rstartr_{\mathrm{start}}, which corresponds to solving for the retarded thermal correlator of the boundary QFT baryon current operator, with the additional requirement that the perturbation is normalized to unity at the boundary, rmax⁡r_{\max} [44]. These two boundary conditions are systematically implemented via [60],

a(r,ω)=r−iωG(r,ω)rmax⁡−iωG(rmax⁡,ω).(37)a(r,\omega)=\frac{r^{-i\omega}G(r,\omega)}{r_{\max}^{-i\omega}G(r_{\max},\omega)}. \tag*{(37)}

where G(r,ω)G(r,\omega) must be a regular function at the event horizon, with its equation of motion being obtained by substituting (37) into (36). The holographic formula for the baryon conductivity expressed in physical units of MeV is [44, 48, 53],

σB(T,μB)=−Λ2κ52ϕA1/νlim⁡ω→01ω(e2A(r)h(r)f(ϕ)Im⁡[a∗(r,ω)a′(r,ω)])∣on−shell(38)\sigma_B(T,\mu_B)=-\frac{\Lambda}{2\kappa_5^2\phi_A^{1/\nu}}\lim_{\omega\to0}\frac{1}{\omega}\left(e^{2A(r)}h(r)f(\phi)\operatorname{Im}\left[a^*(r,\omega)a'(r,\omega)\right]\right)\bigg|_{\mathrm{on-shell}} \tag*{(38)}

The term between brackets in Eq. (38) is a radially conserved flux which can then be evaluated at any value of rr [44]. In practice, as discussed in [60], the limit of zero frequency is approximated in numerical computations by evaluating (38) at some small but nonzero value ω=ωeval\omega=\omega_{\mathrm{eval}}, which is typically chosen to be O(10−5)\mathcal{O}(10^{-5}). In order to check whether this is in fact a good approximation to the zero frequency limit, we tested for several different backgrounds at finite (T,μB)(T,\mu_B) if σB(T,μB)\sigma_B(T,\mu_B) remains unchanged, within some small numerical tolerance, when computed using different small values of ωeval\omega_{\mathrm{eval}}. This is satisfied for ωeval∼10−7−10−2\omega_{\mathrm{eval}}\sim10^{-7}-10^{-2}. On the other hand, for very small frequencies there are spurious divergences related to issues with numerical precision, while for ωeval≳10−1\omega_{\mathrm{eval}}\gtrsim10^{-1} the evaluation frequency is not small enough to serve as a good approximation to the limit of zero frequency in (38).

In Figure 4, we show our Bayesian results for the baryon conductivity. The Bayesian posterior bands at 95% confidence level are relatively narrow near the crossover. This occurs because the EMD parameters are closely constrained by the lattice data in this temperature range. As a result, there is strong consistency among the posterior samples. The dimensionless ratio σB/T\sigma_B/T increases with the temperature, as shown in the left panel of Figure 4. Within the temperature range T∼150−180T\sim150-180 MeV, the bands and corresponding best-fit lines of σB/T\sigma_B/T at fixed μB\mu_B cross each other. The baryon conductivity increases with increasing μB\mu_B below the temperature range T∼150T\sim150 MeV. However, the opposite behavior is observed above the temperature range T∼180T\sim180 MeV. At the critical point, the baryon conductivity remains finite. Across the first-order phase transition line at higher values of μB\mu_B and lower values of TT, it develops a small discontinuity gap.

Figure 4. The left panel shows posterior bands at 95% confidence level, and corresponding best-fit curves (solid lines) of σB/T\sigma_B/T as a function of TT for different values of μB\mu_B while the right panel displays the σB/T\sigma_B/T as a function of TT and μB\mu_B computed from the best-fit parameterization.

In Figure 5, we show our Bayesian results for the dimensionless baryon susceptibility χ^2B≡χ2BT2\hat{\chi}_2^B\equiv\chi_2^B T^2 across the phase diagram of the holographic EMD setup at finite temperature and baryon chemical potential. The uncertainty band comprises several different EMD models in the posterior distribution of the Bayesian inference procedure. One can notice χ^2B\hat{\chi}_2^B diverges at the critical point, as expected for a second-order phase transition point.

Figure 5. The left panel shows posterior bands at 95% confidence level, and corresponding best-fit curves (solid lines) for χB2\chi^2_B as a function of TT for different values of μB\mu_B while the right panel displays the χB2\chi^2_B as a function of TT and μB\mu_B computed from the best-fit parameterization.

Figure 6 presents our Bayesian results for the baryon diffusion coefficient. One can see that by increasing μB\mu_B, the baryon diffusion is generally suppressed at high temperatures, and that it also increases with temperature in the high temperature regime, asymptotically saturating. At the critical point DB→0D_B\to0, because χ2B\chi_2^B diverges while σB\sigma_B remains finite, and TDB=σ^B/χ^2T D_B=\hat{\sigma}_B/\hat{\chi}_2. Across the first-order phase transition, the baryon diffusion shows a small discontinuity gap, which is visible in the best fit line; this gap increases as we increase the baryon chemical potential μB\mu_B.

Figure 6. The left panel shows posterior bands at 95% confidence level, and corresponding best-fit curves (solid lines) of scaled baryon diffusion coefficient TDBT D_B as a function of TT for different values of μB\mu_B, while the right panel displays TDBT D_B as a function of TT and μB\mu_B computed from the best-fit parameterization.

Finally, in Figure 7, we show our Bayesian results for the thermal conductivity. The dimensionless holographic thermal conductivity reveals a minimum within the temperature range T=100−150T=100-150 MeV for μB=0\mu_B=0. With increasing μB\mu_B the minima move toward the critical point and turn into a cusp. Beyond the critical point, this observable develops a small discontinuity gap along the first-order phase transition line.

Figure 7. The left panel shows posterior bands at 95% confidence level, and corresponding best-fit curves (solid lines) of holographic thermal conductivity as a dimensionless combination given in Eq. (34) as functions of TT for different values of μB\mu_B. The right panel shows the same observable as a function of TT and μB\mu_B computed from the best-fit parameterization.

Shear and bulk viscosities

The shear viscosity represents a measure of the resistance of the fluid to sheared flow. It can be obtained from the analysis of the diffeomorphism and gauge invariant perturbation Z(r,ω)Z(r,\omega) in the SO(3)SO(3) quintuplet channel of the EMD model, corresponding to any of the five independent spatial components of the graviton perturbation [44].1 This

Shear and bulk viscosities

perturbation satisfies the equation of motion for a massless scalar field on top of the numerical EMD backgrounds. It has been shown in [44] that the shear viscosity for the EMD model satisfies η/s=1/4π\eta/s = 1/4\pi for all values of temperature T>0T > 0 and chemical potential μB≥0\mu_B \ge0, as previously stated. However, as mentioned before, the dimensionless combination involving the shear viscosity that naturally appears in hydrodynamics at finite baryon density is ηT/(ϵ+P)\eta T/(\epsilon+ P), which reduces to η/s\eta/s in the limit of zero chemical potential [101, 102]. This dimensionless combination presents a non-trivial behavior as a function of (T,μB)(T, \mu_B) in holographic models at finite baryon density. In fact, taking into account that in the EMD model η/s=1/4π\eta/s = 1/4\pi, it follows that

[ηTϵ+P](T,μB)=14π(1+μBρBTs),(39)\left[\frac{\eta T}{\epsilon+ P}\right](T,\mu_B) = \frac{1}{4\pi\left(1+\frac{\mu_B\rho_B}{Ts}\right)}, \tag*{(39)}
→14π=ηs, as μB→0.(40)\to\frac{1}{4\pi} = \frac{\eta}{s},\ \text{as }\mu_B \to0. \tag*{(40)}

Note that (39) can be computed solely in terms of the basic thermodynamic quantities given in Eqs. (17) — (20).

The small values of the ratio η/s∼1/4π≈0.08\eta/s \sim1/4\pi\approx0.08 — 0.20.2 within the temperature window associated to the QCD deconfinement crossover at μB∼0\mu_B \sim0 is the main fingerprint of the nearly-perfect fluidity associated to the strongly coupled behavior of the QGP produced at current collider energies. This was recently confirmed by the JETSCAPE Collaboration, through a Bayesian analysis that uses a hybrid phenomenological model matching several high energy heavy-ion data [13, 15].

One can see in Figure 8 that ηT/(ϵ+P)\eta T/(\epsilon+ P) in the EMD model decreases with increasing baryon chemical potential, indicating that the strongly coupled medium further approaches the nearly-perfect fluidity in the baryon dense regime. This behavior is characterized by the presence of a minimum and an inflection point which moves towards the CEP. Beyond the critical end point, there is a discontinuity gap signaling the phase transition taking place in the medium, as already seen in the other observables.

Figure 8. The left panel shows posterior bands at 95% confidence level, and corresponding best-fit curves (solid lines) of holographic shear viscosity times temperature over enthalpy density obtained from Eq. (39) for fixed μB\mu_B as functions of temperature TT, whereas the right panel presents a surface plot of the same observable obtained by using the best-fit parameterization.

Bulk viscosity measures the resistance of the fluid against compressions or expansions. It can be obtained from the analysis of the diffeomorphism and gauge invariant perturbation H(r,ω)\mathcal{H}(r,\omega) in the SO(3)SO(3) singlet channel of the EMD model, which corresponds to a combination of the spatial trace of the graviton and the dilaton perturbation. This perturbation satisfies the following equation of motion [44],

H′′+[4A′+h′h+2ϕ′′ϕ′−2A′′A′]H′+[e−2Aω2h2+h′h(A′′A′−ϕ′′ϕ′)+e−2Ahϕ′(3A′∂ϕf(ϕ)−f(ϕ)ϕ′)ϕ′2]H=0,(41)\mathcal{H}''+\left[4A'+\frac{h'}{h}+\frac{2\phi''}{\phi'}-\frac{2A''}{A'}\right]\mathcal{H}'+\left[\frac{e^{-2A}\omega^2}{h^2}+\frac{h'}{h}\left(\frac{A''}{A'}-\frac{\phi''}{\phi'}\right)+\frac{e^{-2A}}{h\phi'}\left(3A'\partial_\phi f(\phi)-f(\phi)\phi'\right)\phi'^2\right]\mathcal{H}=0, \tag*{(41)}

which is to be solved with infalling boundary condition at the event horizon, with the solution being normalized to unity at the boundary. These two boundary conditions are implemented via [60],

H(r,ω)=r−iωF(r,ω)rmax⁡−iωF(rmax⁡,ω),(42)\mathcal{H}(r,\omega) = \frac{r^{-i\omega}F(r,\omega)}{r_{\max}^{-i\omega}F(r_{\max},\omega)}, \tag*{(42)}

where F(r,ω)F(r,\omega) must be a regular function at the event horizon, with its equation of motion being obtained by substituting (42) into (41). The holographic formula for the dimensionless ratio between bulk viscosity and entropy density in the EMD model reads as follows [44, 53, 60]

[ζs](T,μB)=−136πlim⁡ω→01ω(e4A(r)h(r)ϕ′(r)2Im⁡[H∗(r,ω)H′(r,ω)]A′(r)2)on−shell,(43)\left[\frac{\zeta}{s}\right](T,\mu_B)=-\frac{1}{36\pi}\lim_{\omega\to0}\frac{1}{\omega}\left(\frac{e^{4A(r)}h(r)\phi'(r)^2\operatorname{Im}\left[\mathcal{H}^{*}(r,\omega)\mathcal{H}'(r,\omega)\right]}{A'(r)^2}\right)_{\mathrm{on-shell}}, \tag*{(43)}

where the term between brackets is a radially conserved flux which can then be computed at any value of rr [44]. Similar observations to those below Eq. (38) also apply here.

From (43), by using the fundamental thermodynamic relation ϵ=Ts+μBρB−P\epsilon= Ts + \mu_B\rho_B - P, and the relation η/s=1/4π\eta/s = 1/4\pi, which is valid for the holographic EMD model, one obtains the following dimensionless combination

[ζTϵ+P](T,μB)=[ζs](T,μB)11+μBρBTs,(44)\left[\frac{\zeta T}{\epsilon+P}\right](T,\mu_B)=\left[\frac{\zeta}{s}\right](T,\mu_B)\frac{1}{1+\frac{\mu_B\rho_B}{Ts}}, \tag*{(44)}
→ζs,as μB→0,(45)\to\frac{\zeta}{s}, \quad\text{as } \mu_B \to0, \tag*{(45)}

which is naturally featured in the hydrodynamic expression for the bulk viscous pressure of charged fluids.

Together with the shear viscosity, bulk viscosity plays an important role in the description of the QGP, affecting e.g. the azimuthal momentum anisotropy, the transverse momentum spectra, and the multiplicity of charged hadrons produced in heavy-ion collisions [12, 13, 15, 111, 112, 113, 114]. In Figure 9, we present the results for the best-fit parametrization of the EMD model for the dimensionless combination (44) along with its Bayesian uncertainty bands from the posterior distribution. Similar to shear viscosity, the overall behavior is characterized by a decrease of bulk viscosity with increasing baryon density in the medium. This prediction is also present in previous holographic EMD setups [44, 53, 60], making it a recurrent aspect of such holographic models. Another feature that appears in this EMD model is the shifting of the peak towards increasing temperature as μB\mu_B increases, both in the polynomial-hyperbolic Ansatz, as shown in Figure 9, and in the parametric Ansatz, as shown in Figure 24 in Appendix A. The peak height becomes smaller with increasing chemical potential, exhibiting therefore a suppression of viscosity effects in the baryon dense regime. This aspect is also present in [53, 60], but not in [44], where the peak height in the bulk viscosity remains nearly the same as one increases the baryon density. This highlights the model-dependent aspect of the bulk viscosity in the EMD setup.

Figure 9. Holographic bulk viscosity, normalized as ζT/(ϵ+P)\zeta T/(\epsilon+ P). Posterior bands at 95% confidence level along with best-fit curves (solid lines) for several values of μB\mu_B (left) and the same observable as a function of TT and μB\mu_B generated from the best-fit parameterization (right).

Figure 24. Posterior bands for the holographic bulk viscosity at 95% confidence level and corresponding best-fit curves (solid lines) as a function of TT for different values of μB\mu_B.

The normalized bulk viscosity develops a dip at finite baryon density which moves towards the CEP, located at μB=598\mu_B = 598 MeV. At the CEP, it acquires an infinite slope, similar to what happens with the shear viscosity. Beyond the critical point, across the first order line, there is a discontinuity gap as well, signaling the phase transition in this observable.

One can also notice that the peak present at μB=0\mu_B = 0 seems to be a robust feature of this observable compared to other Bayesian analyses of the quantity ζ/s\zeta/s, such as those from the JETSCAPE Collaboration [13, 15] and the DUKE Group [12], illustrated in Figure 10, where we compare the best-fit curve of the present EMD model and its Bayesian uncertainty band with the Bayesian analyses of those groups. It is important to remark that, while our result for ζ/s\zeta/s displayed in this comparison is a genuine prediction coming from a microscopic holographic calculation, the Bayesian profiles shown from the JETSCAPE Collaboration and the Duke Group are not microscopic calculations or predictions for ζ/s\zeta/s, being instead adjusted by matching the results from the underlying hybrid phenomenological models to several different heavy-ion experimental data. Therefore, the overall good agreement shown between the EMD prediction for ζ/s\zeta/s and the hybrid model Bayesian results further strengthens the phenomenological reliability of the holographic EMD setup.

Energy loss: jet-quenching parameter, heavy quark drag force and Langevin diffusion coefficients

The energy loss experienced by a heavy quark traversing a strongly coupled medium can be computed in holography by using the trailing string method, originally proposed for conformal models in [90, 91] and later generalized for nonconformal dilatonic models in [99]. In the trailing string method, a probe heavy quark traveling within the medium at a constant velocity vv along, say, the xx direction, is holographically modeled by an open string where one of the string endpoints is attached to the heavy quark at the boundary, while the rest of the string trails behind it within the higher dimensional bulk, with the other string endpoint being attached to a two-dimensional event horizon formed on top of the string worldsheet [94, 97, 100]. As the heavy quark traverses the strongly coupled medium at the boundary, it loses energy and momentum due to the drag force Fdrag=dpx/dtF_{\mathrm{drag}} = dp_x/dt, which is calculated via the energy flow dE/dxdE/dx from the string endpoint at the boundary towards the string endpoint at the worldsheet event horizon in the interior of the higher dimensional bulk.

It was shown in [47] that the heavy quark drag force for the isotropic EMD model can be written as

[FdragλtT2](T,μB;v)=−8πvh0fare2/3 ϕ(r∗)+2A(r∗),(46)\left[\frac{F_{\mathrm{drag}}}{\sqrt{\lambda_t T^2}}\right](T,\mu_B;v)=-8\pi v h_0^{\mathrm{far}}e^{\sqrt{2/3}\,\phi(r_*)+2A(r_*)}, \tag*{(46)}

where r∗r_* is the radial location of the string worldsheet event horizon given by the solution of the equation [47]

h(r∗)=h0far,2v2.(47)h(r_*) = h_0^{\mathrm{far},2}v^2 . \tag*{(47)}

In the conformal limit of high temperatures, the EMD background solutions tend to the AdS5_5-Schwarzschild black hole metric. Consequently, the conformal limit of (46) tends to the SYM result [90, 91],

lim⁡T→∞[FdragλtT2](T,μB;v)→−πγ(v)v2=−πv21−v2.(48)\lim_{T\to\infty}\left[\frac{F_{\mathrm{drag}}}{\sqrt{\lambda_t}T^2}\right](T,\mu_B;v) \to-\frac{\pi\gamma(v)v}{2} = -\frac{\pi v}{2\sqrt{1-v^2}} . \tag*{(48)}

In Figure 11 and Figure 12 we show our Bayesian results for the heavy quark drag force at finite temperature and baryon chemical potential. The magnitude of the drag force, which quantifies the energy loss of a heavy quark moving through the plasma, increases with the medium’s baryon density. This enhancement becomes more pronounced at larger quark velocities. When comparing, for instance, v=0.5v = 0.5 and v=0.99v = 0.99, one clearly observes that the separation between curves at fixed μB\mu_B becomes more visible for higher velocities. Although our calculation employs the large-mass (probe) limit, where the quark mass does not explicitly enter, the qualitative trend suggests that quarks capable of reaching higher velocities, thus less massive ones, would experience a stronger sensitivity to the baryon density of the surrounding medium. In both scenarios, the magnitude of the drag force also increases as the temperature of the medium lowers down. In particular, the predictions of both the Bayesian bands and their best fit curves show that the fixed μB\mu_B curves develop a dip in the crossover region that increases towards the CEP. Across the first order line, as already seen in the other transport coefficients, there is also a discontinuity gap indicating that the system undergoes a phase transition.

Figure 11. Heavy quark drag force at v=0.5v = 0.5. The left panel displays posterior bands at 95% confidence level, and corresponding best-fit curves (solid lines) as functions of TT keeping μB\mu_B constant and the right panel shows the same observable as a function of TT and μB\mu_B using the best-fit parameterization.

Figure 12. Heavy quark drag force at v=0.99v = 0.99. The left panel displays posterior bands at 95% confidence level, and corresponding best-fit curves (solid lines) as functions of TT keeping μB\mu_B constant and the right panel shows the same observable as a function of TT and μB\mu_B using the best-fit parameterization.

The holographic method for treating stochastic diffusion processes was originally proposed for conformal models in [94, 97], and later extended for non-conformal dilatonic setups in [100]. In the trailing string picture, the effects of thermal fluctuations on the propagation of a heavy probe quark are neglected. On the other hand, Refs. [94, 97, 100] took into account the Brownian motion of the heavy quark as it traverses the medium, which can be approximated by a linearized local Langevin equation describing the thermal fluctuations in the heavy quark trajectory. By the standards of the holographic dictionary, from the minimal coupling interaction for a probe heavy quark at the boundary, ∫dt δXμ(r→∞,t)Fμ(t)\int dt\,\delta X^\mu(r \to\infty,t)F_\mu(t), one can identify the boundary value of the fluctuating trailing string worldsheet embedding, δXμ(r→∞,t)\delta X^\mu(r \to\infty,t), as the source for a gauge invariant QFT operator at the boundary, Fμ(t)F_\mu(t), which effectively describes a stochastic force acting upon the heavy quark. The momentum dependent Langevin diffusion coefficients are thus computed from the symmetrized real time two-point correlation function of this boundary QFT operator. Here, we consider the zero frequency limit of these coefficients, describing the long time behavior of the stochastic diffusion process within the hot and baryon dense medium.

The EMD holographic formulas for the heavy quark Langevin diffusion coefficients, perpendicular and parallel to the heavy quark velocity, are given by [47]

[κ⊥λtT3](T,μB;v)=16πv(h0far)3/2e23ϕ(r∗)+3A(r∗)h′(r∗)[4A′(r∗)+83ϕ′(r∗)+h′(r∗)h(r∗)](49)\left[\frac{\kappa_{\perp}}{\sqrt{\lambda_t T^3}}\right](T,\mu_B;v)=16\pi v\left(h_0^{\mathrm{far}}\right)^{3/2}e^{\sqrt{\frac{2}{3}}\phi(r_\ast)+3A(r_\ast)}\sqrt{h'(r_\ast)\left[4A'(r_\ast)+\sqrt{\frac{8}{3}}\phi'(r_\ast)+\frac{h'(r_\ast)}{h(r_\ast)}\right]} \tag*{(49)}
[κ∥λtT3](T,μB;v)=16πv3(h0far)5/2e23ϕ(r∗)+3A(r∗)h′(r∗)2(h′(r∗)[4A′(r∗)+83ϕ′(r∗)+h′(r∗)h(r∗)])3/2(50)\left[\frac{\kappa_{\parallel}}{\sqrt{\lambda_t T^3}}\right](T,\mu_B;v)=16\pi v^3\frac{\left(h_0^{\mathrm{far}}\right)^{5/2}e^{\sqrt{\frac{2}{3}}\phi(r_\ast)+3A(r_\ast)}}{h'(r_\ast)^2}\left(h'(r_\ast)\left[4A'(r_\ast)+\sqrt{\frac{8}{3}}\phi'(r_\ast)+\frac{h'(r_\ast)}{h(r_\ast)}\right]\right)^{3/2} \tag*{(50)}

and the corresponding conformal limits of high temperatures are given by [94, 97]

lim⁡T→∞[κ⊥λtT3](T,μB;v)→πγ(v)1/2=π(1−v2)1/4,(51)\lim_{T\to\infty}\left[\frac{\kappa_{\perp}}{\sqrt{\lambda_t T^3}}\right](T,\mu_B;v)\to\pi\gamma(v)^{1/2}=\frac{\pi}{(1-v^2)^{1/4}}, \tag*{(51)}
lim⁡T→∞[κ∥λtT3](T,μB;v)→πγ(v)5/2=π(1−v2)5/4.(52)\lim_{T\to\infty}\left[\frac{\kappa_{\parallel}}{\sqrt{\lambda_t T^3}}\right](T,\mu_B;v)\to\pi\gamma(v)^{5/2}=\frac{\pi}{(1-v^2)^{5/4}}. \tag*{(52)}

As discussed in detail in [100], one can define two velocity-dependent jet-quenching parameters as follows

q^⊥(T,μB;v)=⟨p⊥2⟩vt=2κ⊥(T,μB;v)v,q^∥(T,μB;v)=⟨Δp∥2⟩vt=κ∥(T,μB;v)v,(53)\hat{q}_{\perp}(T,\mu_B;v)=\frac{\langle p_{\perp}^{2}\rangle}{vt}=\frac{2\kappa_{\perp}(T,\mu_B;v)}{v},\qquad\hat{q}_{\parallel}(T,\mu_B;v)=\frac{\langle\Delta p_{\parallel}^{2}\rangle}{vt}=\frac{\kappa_{\parallel}(T,\mu_B;v)}{v}, \tag*{(53)}

where ⟨p⊥2⟩\langle p_{\perp}^{2}\rangle and ⟨Δp∥2⟩\langle\Delta p_{\parallel}^{2}\rangle correspond to linear approximations for, respectively, the noise-averaged transverse and longitudinal momentum fluctuations of a probe heavy quark after it traversed a distance vtvt within the medium. In particular, q^⊥(T,μB;v)\hat{q}_{\perp}(T,\mu_B;v) gives the transverse momentum broadening of a probe heavy quark traveling with velocity vv through the medium [94, 100].

Figure 13 and Figure 14 show the behavior of the Langevin diffusion coefficients κ∥\kappa_{\parallel} and κ⊥\kappa_{\perp}. Similar to the drag force results, the coefficients increase with the medium’s baryon density, and their sensitivity to μB\mu_B becomes more pronounced at higher quark velocities. This is reflected in the larger separation between fixed-μB\mu_B curves as vv increases, suggesting that quarks moving at higher velocities experience stronger medium-induced diffusion. One can also notice how the magnitude of both Langevin diffusion coefficients increases by lowering temperature and/or by increasing the medium’s baryon density. In contrast to the drag force, κ∥\kappa_{\parallel} and κ⊥\kappa_{\perp} develop a peak towards the CEP along the crossover region, with a discontinuity gap over the first order phase transition line. One important feature of the predictions of both coefficients is that they are consistent with the universal inequality κ∥≥κ⊥\kappa_{\parallel} \ge\kappa_{\perp} derived in [100].

Figure 13. Perpendicular Langevin diffusion coefficient κ⊥\kappa_{\perp} obtained by using Eq. (49). The panels on the left show posterior bands at 95% confidence level, and the best-fit curves (solid lines) as functions of TT for fixed μB\mu_B, with v=0.5v=0.5 (top left) and v=0.99v=0.99 (bottom left). The panels on the right display κ⊥\kappa_{\perp} surface plots as a function of TT and μB\mu_B using the best-fit parameterization, with v=0.5v=0.5 (top right) and v=0.99v=0.99 (bottom right).

Figure 14. Parallel Langevin diffusion coefficient κ∥\kappa_{\parallel} obtained by using Eq. (50). The panels on the left show the posterior bands at 95% confidence level, and the best-fit curves (solid lines) as functions of TT for fixed μB\mu_B, with v=0.5v=0.5 (top left) and v=0.99v=0.99 (bottom left). The panels on the right display the κ⊥\kappa_{\perp} surface plots as functions of TT and μB\mu_B using the best-fit parameterization, with v=0.5v=0.5 (top right) and v=0.99v=0.99 (bottom right).

Finally, we consider the phenomenon of jet quenching in the QGP formed in high-energy heavy-ion collisions. High-momentum partons produced in the initial hard scatterings radiate gluons and develop parton showers prior to hadronization. In a dense deconfined medium, these showers are modified through multiple scatterings with the medium constituents, resulting in parton energy loss and the suppression of high-transverse-momentum hadrons and jets [115, 116]. This suppression provides a powerful probe of the QGP, as the initial jet production can be computed perturbatively while its subsequent medium-induced modification encodes information on the properties of the plasma [117, 118, 119, 120]. The combined collisional and radiative energy loss is commonly characterized by the jet-quenching parameter q^\hat{q}, also defined as the rate of transverse momentum broadening [121, 122].

The holographic prescription for computing the jet-quenching parameter q^\hat{q} associated with the transverse momentum broadening of light partons moving at the speed of light within the QGP [121], originally proposed for conformal holographic models in [95, 96], was later adapted to non-conformal EMD models in [47], leading to the following integral formula,

[q^λtT3](T,μB)=64π2h0far∫rstartrmaxdr e−2/3ϕ(r)−3A(r)h(r)[h0far−h(r)](54)\left[\frac{\hat{q}}{\sqrt{\lambda_t}T^3}\right](T,\mu_B)=\frac{64\pi^2 h_0^{\mathrm{far}}}{\int_{r_{\mathrm{start}}}^{r_{\mathrm{max}}} \mathrm{d}r\,\frac{e^{-\sqrt{2/3}\phi(r)-3A(r)}}{\sqrt{h(r)[h_0^{\mathrm{far}}-h(r)]}}} \tag*{(54)}

with the corresponding conformal limit of high temperatures being given by [95],

lim⁡T→∞[q^λtT3](T,μB)=π3/2Γ(3/4)Γ(5/4)≈7.52814.(55)\lim_{T \to\infty}\left[\frac{\hat{q}}{\sqrt{\lambda_t T^3}}\right](T,\mu_B)=\frac{\pi^{3/2}\Gamma(3/4)}{\Gamma(5/4)}\approx7.52814. \tag*{(55)}

In Figure 15 we display our Bayesian results for the jet-quenching parameter. One notes that the energy loss also grows in magnitude as the medium temperature decreases and its baryon density increases, indicating more jet suppression in the baryon-dense regime. A peak is also developed towards the CEP around the crossover region, with a discontinuity gap over the first-order transition line.

Figure 15. Left: scaled jet-quenching parameter q^/λt1/2T3\hat{q}/\lambda_t^{1/2}T^3 posterior bands at 95% confidence level, and the best-fit curves (solid lines) obtained from Eq. (54) as functions of TT for several values of μB\mu_B. Right: scaled jet-quenching parameter as a function of TT and μB\mu_B, from the best-fit parameterization.

In the right panel of Figure 10, we show the holographic EMD prediction for the jet-quenching parameter at zero baryon chemical potential compared to the Bayesian profile obtained with the hybrid phenomenological model of the JETSCAPE Collaboration [13], which is adjusted to describe heavy ion data on partonic energy loss. As in any holographic calculation involving the NG action, there is an extra free parameter corresponding to the ‘t Hooft coupling, λt\lambda_t. In the right panel of Figure 10, we adjusted the height of the holographic jet-quenching parameter by choosing three representative values for λt={1/25,1/16,1/9}\lambda_t=\{1/25,1/16,1/9\}, all of them lying within the broad JETSCAPE uncertainty band. Since q^∝λt\hat{q}\propto\sqrt{\lambda_t}, the overall normalization of the holographic result is controlled by the choice of λt\lambda_t. Therefore, the holographic model can only make a prediction for the shape of the jet quenching parameter as a function of (T,μB)(T,\mu_B), but not for its magnitude. In this sense, the aforementioned values for λt\lambda_t, employed to plot the holographic result for the jet quenching parameter on the right panel of Figure 10, should be regarded as a phenomenological normalization used for comparison with the JETSCAPE result, rather than as a determination of the fundamental QCD ‘t Hooft coupling.

Indeed, we remark that such small values for λt=gQCD2Nc\lambda_t=g_{\mathrm{QCD}}^2N_c are in tension with one of the basic assumptions of the standard holographic gauge-gravity dictionary, which prescribes that the ‘t Hooft coupling should be very large. Such a tension is also similar to what was discussed at the end of Section II concerning the value of the gravitational constant in the present bottom-up EMD model, κ52≡8πG5=11.4\kappa_5^2\equiv8\pi G_5=11.4. By the standards of the holographic gauge-gravity dictionary, one should expect the gravitational constant κ52\kappa_5^2 to be very small, in association with the large NcN_c limit of the dual gauge theory at the boundary. However, in our bottom-up phenomenological approach, we have never matched any of our results to large NcN_c QCD, but used instead inputs from the physical SU(Nc=3)SU(N_c=3) QCD. Although the approach followed here provides in practice holographic predictions in good agreement with first principles LQCD results, as shown in Figure 3, and also in reasonable agreement with the JETSCAPE results for the bulk viscosity and the jet quenching parameter inferred from heavy-ion experimental data, as displayed in Figure 10, from a purely theoretical perspective, the resolution of the aforementioned tensions with holographic top-down expectations is still something to be understood.

Comparison Between PHA and PA ansatze

The results for transport coefficients and energy loss presented in Section III, were obtained using the polyhyperbolic ansatz (PHA), while the corresponding results obtained with the parametric ansatz (PA) are shown in Appendix A. Since both PHA and PA ansatze were constrained by the same lattice QCD data [77] through Bayesian analysis in [63], a direct comparison of PA and PHA 95% credible bands offers a quantitative measure of how sensitively the transport coefficient predictions depend on the choice of ansatz. To assist this comparison, we show in Figure 16 – Figure 20 the PA–PHA envelope as the dotted black lines, which were constructed by taking the smallest lower limit and the largest upper limit of the two credible bands. Its width provides a visual measure of the ansatz dependence. We discuss such a comparison below.

Figure 16. Panels display normalized baryon conductivity, σB/T\sigma_B/T, obtained from PA and PHA ansatze. Bands denote the 95 % credible intervals, while solid curves represent the corresponding best-fit predictions. The black dotted curves indicate the combined envelope of the PA and PHA results. Left panel: μB=0\mu_B = 0 MeV. Right panel: μB=600\mu_B = 600 MeV.

Figure 20. Comparison of the bulk viscosity, normalized as ζT/(ϵ+P)\zeta T/(\epsilon+ P). Obtained using the PA and PHA ansatze. Shaded bands represent 95% credible intervals while solid lines correspond to the best-fit curves. The dotted black lines indicate the combined PA/PHA envelope. Left panel: μB=0\mu_B = 0 MeV. Right panel: μB=600\mu_B = 600 MeV.

Figure 16 compares the bands for the normalized baryon conductivity, σB/T\sigma_B/T, as a function of temperature obtained from PA and PHA ansatze at μB=0\mu_B = 0 MeV and μB=600\mu_B = 600 MeV. The two bands overlap almost completely across the temperature range, resulting in a very narrow PA-PHA envelope. This near-perfect agreement demonstrates that predictions for σB/T\sigma_B/T are remarkably robust to the choice of ansatz.

Figure 17 compares the posterior bands for the scaled jet-quenching parameter, q^/(λt1/2T3)\hat{q}/(\lambda_t^{1/2}T^3), obtained from PA and PHA ansatze at fixed chemical potential. The 95% credible bands show significant overlap throughout the temperature range, indicating that the predicted behavior of q^/(λt1/2T3)\hat{q}/(\lambda_t^{1/2}T^3) is largely insensitive to the ansatz choice. A modest difference does emerge around the peak region at μB=600\mu_B = 600 MeV, where the PA-PHA envelope somewhat broadens, signaling a slightly stronger ansatz dependence around the peak.

Figure 17. Comparison of the scaled jet-quenching parameter q^/(λt1/2T3)\hat{q}/(\lambda_t^{1/2}T^3) obtained using the PA and PHA ansatze. Shaded bands represent 95% credible intervals while solid lines correspond to the best-fit curves. The dotted black lines indicate the combined PA/PHA envelope. Left panel: μB=0\mu_B = 0 MeV. Right panel: μB=600\mu_B = 600 MeV.

Figure 18 compares the posterior bands for the scaled heavy quark drag force at v=0.99v = 0.99 obtained using the PA and PHA ansatze for μB=0\mu_B = 0 and μB=600\mu_B = 600 MeV as function of TT. The 95% credible bands overlap almost completely across the entire temperature range, indicating negligible dependence on the choice of ansatz.

Figure 18. Comparison of the heavy quark drag force with v=0.99v = 0.99 obtained using the PA and PHA ansatze. Bands represent 95% credible intervals while solid lines correspond to the best-fit curves. The dotted black lines indicate the combined PA/PHA envelope. Left panel: μB=0\mu_B = 0 MeV. Right panel: μB=600\mu_B = 600 MeV.

Figure 19 compares the scaled parallel component of the Langevin diffusion coefficient at v=0.99v = 0.99 obtained from PA and PHA ansatze for μB=0\mu_B = 0 and 600600 MeV. The 95% credible bands overlap almost completely across the entire temperature range, indicating that the heavy-quark drag force is largely insensitive to the choice of holographic ansatz.

Figure 19. Comparison of the Parallel Langevin diffusion coefficient κ∥\kappa_{\parallel} with v=0.99v = 0.99 obtained using the PA and PHA ansatze. Bands represent 95% credible intervals while solid lines correspond to the best-fit curves. The dotted black lines indicate the combined PA/PHA envelope. Left panel: μB=0\mu_B = 0 MeV. Right panel: μB=600\mu_B = 600 MeV.

Figure 20 compares the posterior bands for the normalized bulk viscosity ζT/(ϵ+P)\zeta T/(\epsilon+ P) at μB=0\mu_B = 0 and 600600 MeV. In contrast to other observables considered in this work, the bulk viscosity exhibits stronger dependence on the choice of holographic ansatz. At μB=0\mu_B = 0, both PA and PHA bands show qualitatively different behavior for T≲150T \lesssim150 MeV. The PHA band decreases toward lower temperature, whereas the PA band remains nearly flat. Moreover, the PHA credible band becomes noticeably wider in this region, implying that low-temperature bulk viscosity is less constrained. This low-temperature sensitivity is similar to the larger differences between the PA and PHA trace anomaly, (ϵ−3P)/T4(\epsilon- 3P)/T^4, compared to the other thermodynamic observables shown in Figure 8 of Ref. [63]. At μB=600\mu_B = 600 MeV, both ansatze show the same qualitative features, such as a clear dip at low temperature and a broad peak at intermediate temperatures. The PA–PHA envelope is wider around this temperature region than for the other transport observables, but the two posterior bands overlap, especially at high temperatures.

For each observable, we overlay the PA and PHA 95% credible bands and quantify their separation through the PA–PHA envelope. The envelope is defined by the lower bound and upper bound of the two bands at each temperature. The width of the PA–PHA envelope quantifies ansatz dependence of the observables. A narrow envelope shows that the observable is largely insensitive to the choice of ansatz. Whereas a broader envelope shows greater model dependence. This comparison also provides an estimate of the uncertainty associated with the choice of holographic parametrization. Note also from the comparison done in Figure 16 – Figure 20 that the PA ansatz exhibits smaller posterior uncertainty than the PHA ansatz (which is also the case for thermodynamic observables, as originally disclosed in Ref. [63]).

Conclusions

In the present work, we employed the holographic Einstein-Maxwell-Dilaton model supplemented with Bayesian inference tools [63] to evaluate several transport coefficients for the QGP with Nf=2+1N_f = 2+1 flavors throughout the QCD phase diagram. The model employed in this work shows good overall agreement with state-of-the-art (2 + 1)-flavor lattice QCD thermodynamics and with the finite-density equation of state obtained via the T′T'-expansion [77, 123].

We introduced a new relaxational numerical strategy for extracting ultraviolet asymptotics in holographic models, leading to substantial gains in stability and performance, which in turn enabled the Bayesian sampling of lattice QCD–constrained posterior distributions in the analysis of Ref. [63].

A publicly available C++ implementation of the EMD model incorporating this method was extended to include the calculation of several transport coefficients: baryon and thermal conductivities, the baryon diffusion coefficient, shear and bulk viscosities, the heavy-quark drag force and Langevin diffusion coefficients, and the jet-quenching parameter

[78]

Combining this improved C++ code with Bayesian samples from [63, 79], we obtained predictions for these transport coefficients across the phase diagram, including Bayesian uncertainty bands constrained by lattice QCD. Our results go beyond previous work in the literature (see e.g. [74]) by presenting lattice QCD-constrained uncertainty bands at finite baryon chemical potential, and by considering baryon and thermal conductivities. Moreover, by exploring two different sets of functional forms for the holographic potentials, we were able to confirm the robustness of our predictions in the crossover and phase-transition regions. Within the regime of validity of the model, the lattice-constrained posterior leads to tightly bounded transport coefficients. The Bayesian uncertainty bands obtained in the present work are typically broader at low temperatures, as expected, since this region of the phase diagram lies beyond the regime of applicability of the model, and was not constrained by lattice QCD data at zero baryon density.

We found an overall suppression of the diffusion of baryon charge, shear viscosity, and bulk viscosity with increasing baryon density, indicating that the strongly interacting QGP becomes even closer to the perfect-fluid limit at large baryon chemical potentials. On the other hand, the jet-quenching parameter and the heavy-quark energy loss and Langevin momentum-diffusion coefficients were found to be enhanced with increasing baryon density. Furthermore, we found that our prediction for the bulk viscosity is in good agreement with the uncertainty band obtained from the Bayesian analysis of heavy-ion data by the JETSCAPE Collaboration [13, 15]. This provides further support for the holographic EMD description of dissipative transport in the quark-gluon plasma. For the jet-quenching parameter, we found that the predicted temperature dependence is in agreement with the JETSCAPE band after fixing the overall holographic normalization for this observable by phenomenologically choosing λt∈[1/25,1/9]\lambda_t \in[1/25, 1/9]. Since q^/T3∝λt\hat{q}/T^3 \propto\sqrt{\lambda_t}, the choice of λt\lambda_t only fixes the overall magnitude of the holographic jet quenching parameter, while the (T,μB)(T,\mu_B)-dependence remains unchanged.

Acknowledgments

We thank J. Jahan for his assistance in running all posterior samples on the cluster. This material is based upon work supported by the National Science Foundation under grants No. PHY-2208724, PHY-2116686 and PHY-2514763, and within the framework of the MUSES collaboration, under Grant No. OAC-2103680. This material is also based upon work supported by the U.S. Department of Energy, Office of Science, Office of Nuclear Physics, under Award Numbers DE-SC0022023, DE-SC0026065, and DE-SC0023861, as well as by the National Aeronautics and Space Agency (NASA) under Award Number 80NSSC24K0767. A.N. acknowledges the financial support by the Brazilian National Council for Scientific and Technological Development (CNPq) under process number 176343/2025-3. M.H. was supported by the Brazilian National Council for Scientific and Technological Development (CNPq) under process No. 313638/2025-0. R.R. acknowledges financial support by the Brazilian National Council for Scientific and Technological Development (CNPq) under grants number 407162/2023-2 and 305466/2024-0.

Holographic Bayesian transport with the Parametric Ansatz

In the main section, we discussed the results obtained using the polyhyperbolic Ansatz (PHA) given in Eqs. (26) and (25). Here, we present the corresponding transport coefficient and energy loss results computed using the parametric Ansatz (PA), Eqs. (28) and (27). The baryon conductivity, baryon diffusion, and thermal conductivity, along with the shear viscosity, are shown in Figure 21. Drag force and perpendicular Langevin diffusion coefficient are given in Figure 22 and Figure 23, respectively. Finally, the bulk viscosity is shown in Figure 24, while the jet-quenching parameter is presented in Figure 25.

Figure 21. Panel (a) display posterior bands at 95% confidence level and corresponding best-fit curves (solid lines) for σB/T\sigma_B/T as a function of TT for different values of μB\mu_B. Panel (b) presents posterior bands at 95% confidence level and corresponding best-fit curves (solid lines) for the scaled baryon diffusion coefficient TDBT D_B as a function of TT for different values of μB\mu_B. Panel (c) shows posterior bands for the holographic thermal conductivity at 95% confidence level and corresponding best-fit curves (solid lines) as a function of TT for different values of μB\mu_B. Panel (d) reveals holographic shear viscosity times temperature over enthalpy density. Posterior bands at 95% confidence level and the corresponding best-fit curves (solid lines) as a function of TT for several values of μB\mu_B.

Figure 22. Heavy quark drag force at v=0.5v = 0.5 (left panel) and v=0.99v=0.99 (right panel). Both figures show posterior band at 95% confidence level and corresponding best-fit curves (solid lines).

Figure 23. Perpendicular Langevin diffusion coefficient κ⊥\kappa_{\perp} at v=0.5v = 0.5 (left panel) and v=0.99v=0.99 (right panel). Both figures show posterior band at 95% confidence level and the corresponding best-fit curves (solid lines).

Figure 25. Scaled jet-quenching parameter. Posterior bands at 95% confidence level, and the corresponding best-fit curves (solid lines) of σB/T\sigma_B/T as a function of TT for different values of μB\mu_B.

References

References

  1. [1]Kenji Fukushima and Tetsuo Hatsuda. The phase diagram of dense QCD. Rept. Prog. Phys., 74:014001, 2011. arXiv:1005.4814, doi:10.1088/0034-4885/74/1/014001.DOI
  2. [2]Alessandro Lovato et al. Long Range Plan: Dense matter theory for heavy-ion collisions and neutron stars. 11 2022. arXiv:2211.02224.arxiv.org/abs/2211.02224
  3. [3]Rajesh Kumar et al. Theoretical and experimental constraints for the equation of state of dense and hot matter. Living Rev. Rel., 27(1):3, 2024. arXiv:2303.17021, doi:10.1007/s41114-024-00049-6.DOI
  4. [4]Ulrich W. Heinz and Maurice Jacob. The state of nuclear matter: An Assessment of the results from the CERN lead beam program. 1 2000. arXiv:nucl-th/0002042.
  5. [5]Miklos Gyulassy and Larry McLerran. New forms of QCD matter discovered at RHIC. Nucl. Phys. A, 750:30–63, 2005. arXiv:nucl-th/0405013, doi:10.1016/j.nuclphysa.2004.10.034.DOI
  6. [6]K. Adcox et al. Formation of dense partonic matter in relativistic nucleus-nucleus collisions at RHIC: Experimental evaluation by the PHENIX collaboration. Nucl. Phys. A, 757:184–283, 2005. arXiv:nucl-ex/0410003, doi:10.1016/j.nuclphysa.2005.03.086.DOI
  7. [7]I. Arsene et al. Quark gluon plasma and color glass condensate at RHIC? The Perspective from the BRAHMS experiment. Nucl. Phys. A, 757:1–27, 2005. arXiv:nucl-ex/0410020, doi:10.1016/j.nuclphysa.2005.02.130.DOI
  8. [8]B. B. Back et al. The PHOBOS perspective on discoveries at RHIC. Nucl. Phys. A, 757:28–101, 2005. arXiv:nucl-ex/0410022, doi:10.1016/j.nuclphysa.2005.03.084.DOI
  9. [9]John Adams et al. Experimental and theoretical challenges in the search for the quark gluon plasma: The STAR Collaboration’s critical assessment of the evidence from RHIC collisions. Nucl. Phys. A, 757:102–183, 2005. arXiv:nucl-ex/0501009, doi:10.1016/j.nuclphysa.2005.03.085.DOI
  10. [10]Berndt Muller, Jurgen Schukraft, and Boleslaw Wyslouch. First Results from Pb+Pb collisions at the LHC. Ann. Rev. Nucl. Part. Sci., 62:361–386, 2012. arXiv:1202.3233, doi:10.1146/annurev-nucl-102711-094910.DOI
  11. [11]Wit Busza, Krishna Rajagopal, and Wilke van der Schee. Heavy Ion Collisions: The Big Picture, and the Big Questions. Ann. Rev. Nucl. Part. Sci., 68:339–376, 2018. arXiv:1802.04801, doi:10.1146/annurev-nucl-101917-020852.DOI
  12. [12]Jonah E. Bernhard, J. Scott Moreland, and Steffen A. Bass. Bayesian estimation of the specific shear and bulk viscosity of quark–gluon plasma. Nature Phys., 15(11):1113–1117, 2019. doi:10.1038/s41567-019-0611-8.DOI
  13. [13]D. Everett et al. Phenomenological constraints on the transport properties of QCD matter with data-driven model averaging. Phys. Rev. Lett., 126(24):242301, 2021. arXiv:2010.03928, doi:10.1103/PhysRevLett.126.242301.DOI
  14. [14]Govert Nijs, Wilke van der Schee, Umut Gürsoy, and Raimond Snellings. Transverse Momentum Differential Global Analysis of Heavy-Ion Collisions. Phys. Rev. Lett., 126(20):202301, 2021. arXiv:2010.15130, doi:10.1103/PhysRevLett.126.202301.DOI
  15. [15]D. Everett et al. Multisystem Bayesian constraints on the transport coefficients of QCD matter. Phys. Rev. C, 103(5):054904, 2021. arXiv:2011.01430, doi:10.1103/PhysRevC.103.054904.DOI
  16. [16]P. Kovtun, Dan T. Son, and Andrei O. Starinets. Viscosity in strongly interacting quantum field theories from black hole physics. Phys. Rev. Lett., 94:111601, 2005. arXiv:hep-th/0405231, doi:10.1103/PhysRevLett.94.111601.DOI
  17. [17]Ulrich Heinz and Raimond Snellings. Fluid dynamics and viscosity in relativistic heavy-ion collisions. Ann. Rev. Nucl. Part. Sci., 63:123–151, 2013. arXiv:1301.2826, doi:10.1146/annurev-nucl-102212-170540.DOI
  18. [18]Matthew Luzum and Hannah Petersen. Initial State Fluctuations and Final State Correlations in Relativistic Heavy-Ion Collisions. J. Phys. G, 41:063102, 2014. arXiv:1312.5503, doi:10.1088/0954-3899/41/6/063102.DOI
  19. [19]D. Almaalol et al. QCD Phase Structure and Interactions at High Baryon Density: Continuation of BES Physics Program with CBM at FAIR. 9 2022. arXiv:2209.05009.arxiv.org/abs/2209.05009
  20. [20]D. Cebra, S. G. Brovko, C. E. Flores, B. A. Haag, and J. L. Klay. Coulomb effect in Au+Au and Pb+Pb collisions as a function of collision energy. 8 2014. arXiv:1408.1369.arxiv.org/abs/1408.1369
  21. [21]J. Adamczewski-Musch et al. Probing dense baryon-rich matter with virtual photons. Nature Phys., 15(10):1040–1045, 2019. doi:10.1038/s41567-019-0583-8.
  22. [22]V. Friese. The CBM experiment at GSI/FAIR. Nucl. Phys. A, 774:377–386, 2006. doi:10.1016/j.nuclphysa.2006.06.018.DOI
  23. [23]N. A. Tahir et al. Proposal for the Study of Thermophysical Properties of High-Energy-Density Matter Using Current and Future Heavy-Ion Accelerator Facilities at GSI Darmstadt. Phys. Rev. Lett., 95:035001, 2005. doi:10.1103/PhysRevLett.95.035001.DOI
  24. [24]M. F. M. Lutz et al. Physics Performance Report for PANDA: Strong Interaction Studies with Antiprotons. 3 2009. arXiv:0903.3905.arxiv.org/abs/0903.3905
  25. [25]M. Durante et al. All the fun of the FAIR: fundamental physics at the facility for antiproton and ion research. Phys. Scripta, 94(3):033001, 2019. arXiv:1903.05693, doi:10.1088/1402-4896/aaf93f.DOI
  26. [26]V. Kekelidze, A. Kovalenko, R. Lednicky, V. Matveev, I. Meshkov, A. Sorin, and G. Trubnikov. Feasibility study of heavy-ion collision physics at NICA JINR. Nucl. Phys. A, 967:884–887, 2017. doi:10.1016/j.nuclphysa.2017.06.031.DOI
  27. [27]V. Kekelidze, A. Kovalenko, R. Lednicky, V. Matveev, I. Meshkov, A. Sorin, and G. Trubnikov. Prospects for the dense baryonic matter research at NICA. Nucl. Phys. A, 956:846–849, 2016. doi:10.1016/j.nuclphysa.2016.03.019.DOI
  28. [28]Owe Philipsen. The QCD equation of state from the lattice. Prog. Part. Nucl. Phys., 70:55–107, 2013. arXiv:1207.5999, doi:10.1016/j.ppnp.2012.09.003.
  29. [29]Claudia Ratti. Lattice QCD and heavy ion collisions: a review of recent progress. Rept. Prog. Phys., 81(8):084301, 2018. arXiv:1804.07810, doi:10.1088/1361-6633/aabb97.DOI
  30. [30]Juan Martin Maldacena. The Large N limit of superconformal field theories and supergravity. Adv. Theor. Math. Phys., 2:231–252, 1998. arXiv:hep-th/9711200, doi:10.4310/ATMP.1998.v2.n2.a1.DOI
  31. [31]S. S. Gubser, Igor R. Klebanov, and Alexander M. Polyakov. Gauge theory correlators from noncritical string theory. Phys. Lett. B, 428:105–114, 1998. arXiv:hep-th/9802109, doi:10.1016/S0370-2693(98)00377–3.DOI
  32. [32]Edward Witten. Anti-de Sitter space and holography. Adv. Theor. Math. Phys., 2:253–291, 1998. arXiv:hep-th/9802150, doi:10.4310/ATMP.1998.v2.n2.a2.DOI
  33. [33]Edward Witten. Anti-de Sitter space, thermal phase transition, and confinement in gauge theories. Adv. Theor. Math. Phys., 2:505–532, 1998. arXiv:hep-th/9803131, doi:10.4310/ATMP.1998.v2.n3.a3.DOI
  34. [34]Johanna Erdmenger, Nick Evans, Ingo Kirsch, and Ed Threlfall. Mesons in Gauge/Gravity Duals - A Review. Eur. Phys. J. A, 35:81–133, 2008. arXiv:0711.4467, doi:10.1140/epja/i2007-10540-1.DOI
  35. [35]Stanley J. Brodsky, Guy F. de Téramond, Hans Günter Dosch, and Joshua Erlich. Light-Front Holographic QCD and Emerging Confinement. Phys. Rept., 584:1–105, 2015. arXiv:1407.8131, doi:10.1016/j.physrep.2015.05.001.DOI
  36. [36]Umut Gürsoy, Elias Kiritsis, Liuba Mazzanti, Georgios Michalogiorgakis, and Francesco Nitti. Improved Holographic QCD. Lect. Notes Phys., 828:79–146, 2011. arXiv:1006.5461, doi:10.1007/978-3-642-04864-7_4.DOI
  37. [37]Oliver DeWolfe, Steven S. Gubser, Christopher Rosen, and Derek Teaney. Heavy ions and string theory. Prog. Part. Nucl. Phys., 75:86–132, 2014. arXiv:1304.7794, doi:10.1016/j.ppnp.2013.11.001.DOI
  38. [38]Carlos Hoyos, Niko Jokela, and Aleksi Vuorinen. Holographic approach to compact stars and their binary mergers. Prog. Part. Nucl. Phys., 126:103972, 2022. arXiv:2112.08422, doi:10.1016/j.ppnp.2022.103972.DOI
  39. [39]Romulo Rougemont, Joaquin Grefa, Mauricio Hippert, Jorge Noronha, Jacquelyn Noronha-Hostler, Israel Portillo, and Claudia Ratti. Hot QCD phase diagram from holographic Einstein–Maxwell–Dilaton models. Prog. Part. Nucl. Phys., 135:104093, 2024. arXiv:2307.03885, doi:10.1016/j.ppnp.2023.104093.DOI
  40. [40]Steven S. Gubser and Abhinav Nellore. Mimicking the QCD equation of state with a dual black hole. Phys. Rev. D, 78:086007, 2008. arXiv:0804.0434, doi:10.1103/PhysRevD.78.086007.DOI
  41. [41]Steven S. Gubser, Abhinav Nellore, Silviu S. Pufu, and Fabio D. Rocha. Thermodynamics and bulk viscosity of approximate black hole duals to finite temperature quantum chromodynamics. Phys. Rev. Lett., 101:131601, 2008. arXiv:0804.1950, doi:10.1103/PhysRevLett.101.131601.DOI
  42. [42]Steven S. Gubser, Silviu S. Pufu, and Fabio D. Rocha. Bulk viscosity of strongly coupled plasmas with holographic duals. JHEP, 08:085, 2008. arXiv:0806.0407, doi:10.1088/1126-6708/2008/08/085.DOI
  43. [43]Oliver DeWolfe, Steven S. Gubser, and Christopher Rosen. A holographic critical point. Phys. Rev. D, 83:086005, 2011. arXiv:1012.1864, doi:10.1103/PhysRevD.83.086005.DOI
  44. [44]Oliver DeWolfe, Steven S. Gubser, and Christopher Rosen. Dynamic critical phenomena at a holographic critical point. Phys. Rev. D, 84:126014, 2011. arXiv:1108.2029, doi:10.1103/PhysRevD.84.126014.DOI
  45. [45]Stefano I. Finazzo, Romulo Rougemont, Hugo Marrochio, and Jorge Noronha. Hydrodynamic transport coefficients for the non-conformal quark-gluon plasma from holography. JHEP, 02:051, 2015. arXiv:1412.2968, doi:10.1007/JHEP02(2015)051.DOI
  46. [46]Romulo Rougemont, Renato Critelli, and Jorge Noronha. Holographic calculation of the QCD crossover temperature in a magnetic field. Phys. Rev. D, 93(4):045013, 2016. arXiv:1505.07894, doi:10.1103/PhysRevD.93.045013.DOI
  47. [47]Romulo Rougemont, Andrej Ficnar, Stefano Finazzo, and Jorge Noronha. Energy loss, equilibration, and thermodynamics of a baryon rich strongly coupled quark-gluon plasma. JHEP, 04:102, 2016. arXiv:1507.06556, doi:10.1007/JHEP04(2016)102.DOI
  48. [48]Romulo Rougemont, Jorge Noronha, and Jacquelyn Noronha-Hostler. Suppression of baryon diffusion and transport in a baryon rich strongly coupled quark-gluon plasma. Phys. Rev. Lett., 115(20):202301, 2015. arXiv:1507.06972, doi:10.1103/PhysRevLett.115.202301.DOI
  49. [49]Stefano Ivo Finazzo and Romulo Rougemont. Thermal photon, dilepton production, and electric charge transport in a baryon rich strongly coupled QGP from holography. Phys. Rev. D, 93(3):034017, 2016. arXiv:1510.03321, doi:10.1103/PhysRevD.93.034017.DOI
  50. [50]Stefano Ivo Finazzo, Renato Critelli, Romulo Rougemont, and Jorge Noronha. Momentum transport in strongly coupled anisotropic plasmas in the presence of strong magnetic fields. Phys. Rev. D, 94(5):054020, 2016. [Erratum: Phys.Rev.D 96, 019903 (2017)]. arXiv:1605.06061, doi:10.1103/PhysRevD.94.054020.DOI
  51. [51]Renato Critelli, Romulo Rougemont, Stefano I. Finazzo, and Jorge Noronha. Polyakov loop and heavy quark entropy in strong magnetic fields from holographic black hole engineering. Phys. Rev. D, 94(12):125019, 2016. arXiv:1606.09848, doi:10.1103/PhysRevD.94.125019.
  52. [52]J. Knaute, R. Yaresko, and B. Kämpfer. Holographic QCD phase diagram with critical point from Einstein–Maxwell–dilaton dynamics. Phys. Lett. B, 778:419–425, 2018. arXiv:1702.06731, doi:10.1016/j.physletb.2018.01.053.DOI
  53. [53]Romulo Rougemont, Renato Critelli, Jacquelyn Noronha-Hostler, Jorge Noronha, and Claudia Ratti. Dynamical versus equilibrium properties of the QCD phase transition: A holographic perspective. Phys. Rev. D, 96(1):014032, 2017. arXiv:1704.05558, doi:10.1103/PhysRevD.96.014032.DOI
  54. [54]Renato Critelli, Jorge Noronha, Jacquelyn Noronha-Hostler, Israel Portillo, Claudia Ratti, and Romulo Rougemont. Critical point in the phase diagram of primordial quark-gluon matter from black hole physics. Phys. Rev. D, 96(9):096026, 2017. arXiv:1706.00455, doi:10.1103/PhysRevD.96.096026.DOI
  55. [55]Romulo Rougemont, Renato Critelli, and Jorge Noronha. Nonhydrodynamic quasinormal modes and equilibration of a baryon dense holographic QGP with a critical point. Phys. Rev. D, 98(3):034028, 2018. arXiv:1804.00189, doi: 10.1103/PhysRevD.98.034028.DOI
  56. [56]Romulo Rougemont. Jet quenching parameters in strongly coupled anisotropic plasmas in the presence of magnetic fields. Phys. Rev. D, 102(3):034009, 2020. arXiv:2002.06725, doi:10.1103/PhysRevD.102.034009.DOI
  57. [57]Joaquim Grefa, Jorge Noronha, Jacquelyn Noronha-Hostler, Israel Portillo, Claudia Ratti, and Romulo Rougemont. Hot and dense quark-gluon plasma thermodynamics from holographic black holes. Phys. Rev. D, 104(3):034002, 2021. arXiv:2102.12042, doi:10.1103/PhysRevD.104.034002.DOI
  58. [58]Rico Zöllner and Burkhard Kämpfer. Holographic bottomonium formation in a cooling strong-interaction medium at finite baryon density. Phys. Rev. D, 104(10):106005, 2021. arXiv:2109.05824, doi:10.1103/PhysRevD.104.106005.DOI
  59. [59]Rong-Gen Cai, Song He, Li Li, and Yuan-Xu Wang. Probing QCD critical point and induced gravitational wave by black hole physics. Phys. Rev. D, 106(12):L121902, 2022. arXiv:2201.02004, doi:10.1103/PhysRevD.106.L121902.DOI
  60. [60]Joaquim Grefa, Mauricio Hippert, Jorge Noronha, Jacquelyn Noronha-Hostler, Israel Portillo, Claudia Ratti, and Romulo Rougemont. Transport coefficients of the quark-gluon plasma at the critical point and across the first-order line. Phys. Rev. D, 106(3):034024, 2022. arXiv:2203.00139, doi:10.1103/PhysRevD.106.034024.DOI
  61. [61]Yidian Chen, Danning Li, and Mei Huang. The dynamical holographic QCD method for hadron physics and QCD matter. Commun. Theor. Phys., 74(9):097201, 2022. arXiv:2206.00917, doi:10.1088/1572-9494/ac82ad.DOI
  62. [62]Zhibin Li, Jingmin Liang, Song He, and Li Li. Holographic study of higher-order baryon number susceptibilities at finite temperature and density. Phys. Rev. D, 108(4):046008, 2023. arXiv:2305.13874, doi:10.1103/PhysRevD.108.046008.DOI
  63. [63]Mauricio Hippert, Joaquim Grefa, T. Andrew Manning, Jorge Noronha, Jacquelyn Noronha-Hostler, Israel Portillo Vazquez, Claudia Ratti, Romulo Rougemont, and Michael Trujillo. Bayesian location of the QCD critical point from a holographic perspective. Phys. Rev. D, 110(9):094006, 2024. arXiv:2309.00579, doi:10.1103/PhysRevD.110.094006.DOI
  64. [64]Yan-Qing Zhao, Song He, Defu Hou, Li Li, and Zhibin Li. Phase structure and critical phenomena in two-flavor QCD by holography. Phys. Rev. D, 109(8):086015, 2024. arXiv:2310.13432, doi:10.1103/PhysRevD.109.086015.DOI
  65. [65]Xun Chen and Mei Huang. Machine learning holographic black hole from lattice QCD equation of state. Phys. Rev. D, 109(5):L051902, 2024. arXiv:2401.06417, doi:10.1103/PhysRevD.109.L051902.DOI
  66. [66]Qingxuan Fu, Song He, Li Li, and Zhibin Li. Revisiting holographic model for thermal and dense QCD with a critical point. JHEP, 06:221, 2025. arXiv:2404.12109, doi:10.1007/JHEP06(2025)221.DOI
  67. [67]Bing Chen, Xun Chen, Xuanhua Li, Zhou-Hun Zhu, and Kai Zhou. Exploring transport properties of quark-gluon plasma in flavor-dependent systems with a holographic model. Phys. Rev. D, 111(8):086033, 2025. arXiv:2404.18217, doi:10.1103/PhysRevD.111.086033.DOI
  68. [68]Niko Jokela, Matti Järvinen, and Aleksi Piispa. Refining holographic models of the quark-gluon plasma. Phys. Rev. D, 110(12):126013, 2024. arXiv:2405.02394, doi:10.1103/PhysRevD.110.126013.DOI
  69. [69]Xun Chen and Mei Huang. Flavor dependent critical endpoint from holographic QCD through machine learning. JHEP, 02:123, 2025. arXiv:2405.06179, doi:10.1007/JHEP02(2025)123.DOI
  70. [70]Rong-Gen Cai, Song He, Li Li, and Hong-An Zeng. Neural ordinary differential equations for mapping the magnetic QCD phase diagram via holography. Sci. China Phys. Mech. Astron., 69(4):240414, 2026. arXiv:2406.12772, doi:10.1007/s11433-025-2914-x.DOI
  71. [71]Xuanmin Cao and Hui Lin. Impact of the phase transition on quark-gluon plasma with an extremely strong magnetic field in holographic QCD. Phys. Rev. D, 111(2):026008, 2025. arXiv:2408.00467, doi:10.1103/PhysRevD.111.026008.DOI
  72. [72]Liqiang Zhu, Xun Chen, Kai Zhou, Hanzhong Zhang, and Mei Huang. Bayesian inference of the critical end point in a (2+1)-flavor system from holographic QCD. Phys. Rev. D, 112(2):026019, 2025. arXiv:2501.17763, doi:10.1103/wpts-lbtr.DOI
  73. [73]Zhibin Li, Danning Li, and Mei Huang. Jet quenching and quark energy loss from holographic QCD. Phys. Rev. D, 111(12):126019, 2025. arXiv:2504.04147, doi:10.1103/52sx-qyvg.DOI
  74. [74]Bing Chen, Liqiang Zhu, Xun Chen, Defu Hou, and Xurong Chen. Transport Properties of QGP within a Bayesian Holographic QCD Model. 8 2025. arXiv:2508.16167.arxiv.org/abs/2508.16167
  75. [75]Jin-Yang Shen, Xin-Yi Liu, Jin-Rui Wu, Yue-Jiang Liu, and Zhen Fang. Toward a holographic realization of the 2+1-flavor QCD phase structure. Phys. Rev. D, 112(11):L111504, 2025. arXiv:2511.11273, doi:10.1103/3plt-wlt7.DOI
  76. [76]Hong-An Zeng, Lingxiao Wang, and Mei Huang. HoloNet: Toward a Unified Einstein-Maxwell-Dilaton Framework of QCD. 12 2025. arXiv:2512.06044.arxiv.org/abs/2512.06044
  77. [77]S. Borsányi, Z. Fodor, J. N. Günther, R. Kara, S. D. Katz, P. Parotto, A. Pásztor, C. Ratti, and K. K. Szabó. Lattice QCD equation of state at finite chemical potential from an alternative expansion scheme. Phys. Rev. Lett., 126(23):232001, 2021. arXiv:2102.06660, doi:10.1103/PhysRevLett.126.232001.DOI
  78. [78]Yumu Yang and Mauricio Hippert. Holographic equation of state module, February 2025. doi:10.5281/zenodo.14695242.
  79. [79]Mauricio Hippert, Joaquim Grefa, Thomas A. Manning, Jorge Noronha, Jacquelyn Noronha-Hostler, Israel Portillo Vazquez, Claudia Ratti, Romulo Rougemont, and Michael Trujillo. Bayesian analysis of the equation of state of qcd from a holographic model — open data set. Zenodo, 2024. doi:10.5281/zenodo.13830379.
  80. [80]Peter Breitenlohner and Daniel Z. Freedman. Stability in Gauged Extended Supergravity. Annals Phys., 144:249, 1982. doi:10.1016/0003-4916(82)90116-6.DOI
  81. [81]Peter Breitenlohner and Daniel Z. Freedman. Positive Energy in anti-De Sitter Backgrounds and Gauged Extended Supergravity. Phys. Lett. B, 115:197–201, 1982. doi:10.1016/0370-2693(82)90643-8.DOI
  82. [82]Alfonso V. Ramallo. Introduction to the AdS/CFT correspondence. Springer Proc. Phys., 161:411–474, 2015. arXiv:1310.4319, doi:10.1007/978-3-319-12238-0_10.DOI
  83. [83]G. Policastro, Dan T. Son, and Andrei O. Starinets. The Shear viscosity of strongly coupled N=4 supersymmetric Yang-Mills plasma. Phys. Rev. Lett., 87:081601, 2001. arXiv:hep-th/0104066, doi:10.1103/PhysRevLett.87.081601.DOI
  84. [84]Alex Buchel and James T. Liu. University of the shear viscosity in supergravity. Phys. Rev. Lett., 93:090602, 2004. arXiv:hep-th/0311175, doi:10.1103/PhysRevLett.93.090602.DOI
  85. [85]Peter Brockway Arnold, Guy D. Moore, and Laurence G. Yaffe. Transport coefficients in high temperature gauge theories. 1. Leading log results. JHEP, 11:001, 2000. arXiv:hep-ph/0010177, doi:10.1088/1126-6708/2000/11/001.
  86. [86]Emma McLaughlin, Jacob Rose, Travis Dore, Paolo Parotto, Claudia Ratti, and Jacquelyn Noronha-Hostler. Building a testable shear viscosity across the QCD phase diagram. Phys. Rev. C, 105(2):024903, 2022. arXiv:2103.02090, doi:10.1103/PhysRevC.105.024903.DOI
  87. [87]Isabella Danhoni, Jordi Salinas San Martin, and Jacquelyn Noronha-Hostler. Shear viscosity from perturbative quantum chromodynamics to the hadron resonance gas at finite baryon, strangeness, and electric charge densities. Phys. Rev. D, 111(11):116006, 2025. arXiv:2406.04968, doi:10.1103/gckr-c2yg.DOI
  88. [88]Sera Cremonini, Umut Gursoy, and Phillip Szepietowski. On the Temperature Dependence of the Shear Viscosity and Holography. JHEP, 08:167, 2012. arXiv:1206.3581, doi:10.1007/JHEP08(2012)167.DOI
  89. [89]Szabolcs Borsanyi, Zoltan Fodor, Christian Hoelbling, Sandor D. Katz, Stefan Krieg, and Kalman K. Szabo. Full result for the QCD equation of state with 2+1 flavors. Phys. Lett. B, 730:99–104, 2014. arXiv:1309.5258, doi:10.1016/j.physletb.2014.01.007.DOI
  90. [90]Steven S. Gubser. Drag force in AdS/CFT. Phys. Rev. D, 74:126005, 2006. arXiv:hep-th/0605182, doi:10.1103/PhysRevD.74.126005.DOI
  91. [91]C. P. Herzog, A. Karch, P. Kovtun, C. Kozcaz, and L. G. Yaffe. Energy loss of a heavy quark moving through N=4 supersymmetric Yang-Mills plasma. JHEP, 07:013, 2006. arXiv:hep-th/0605158, doi:10.1088/1126-6708/2006/07/013.DOI
  92. [92]Jorge Casalderrey-Solana and Derek Teaney. Heavy quark diffusion in strongly coupled N=4 Yang-Mills. Phys. Rev. D, 74:085012, 2006. arXiv:hep-ph/0605199, doi:10.1103/PhysRevD.74.085012.
  93. [93]Steven S. Gubser. Comparing the drag force on heavy quarks in N=4 super-Yang-Mills theory and QCD. Phys. Rev. D, 76:126003, 2007. arXiv:hep-th/0611272, doi:10.1103/PhysRevD.76.126003.DOI
  94. [94]Steven S. Gubser. Momentum fluctuations of heavy quarks in the gauge-string duality. Nucl. Phys. B, 790:175–199, 2008. arXiv:hep-th/0612143, doi:10.1016/j.nuclphysb.2007.09.017.DOI
  95. [95]Hong Liu, Krishna Rajagopal, and Urs Achim Wiedemann. Calculating the jet quenching parameter from AdS/CFT. Phys. Rev. Lett., 97:182301, 2006. arXiv:hep-ph/0605178, doi:10.1103/PhysRevLett.97.182301.DOI
  96. [96]Hong Liu, Krishna Rajagopal, and Urs Achim Wiedemann. Wilson loops in heavy ion collisions and their calculation in AdS/CFT. JHEP, 03:066, 2007. arXiv:hep-ph/0612168, doi:10.1088/1126-6708/2007/03/066.DOI
  97. [97]Jorge Casalderrey-Solana and Derek Teaney. Transverse Momentum Broadening of a Fast Quark in a N=4 Yang Mills Plasma. JHEP, 04:039, 2007. arXiv:hep-th/0701123, doi:10.1088/1126-6708/2007/04/039.
  98. [98]Francesco D’Eramo, Hong Liu, and Krishna Rajagopal. Transverse Momentum Broadening and the Jet Quenching Parameter, Redux. Phys. Rev. D, 84:065015, 2011. arXiv:1006.1367, doi:10.1103/PhysRevD.84.065015.DOI
  99. [99]Umut Gursoy, Elias Kiritsis, Georgios Michalogiorgakis, and Francesco Nitti. Thermal Transport and Drag Force in Improved Holographic QCD. JHEP, 12:056, 2009. arXiv:0906.1890, doi:10.1088/1126-6708/2009/12/056.DOI
  100. [100]Umut Gursoy, Elias Kiritsis, Liuba Mazzanti, and Francesco Nitti. Langevin diffusion of heavy quarks in non-conformal holographic backgrounds. JHEP, 12:088, 2010. arXiv:1006.3261, doi:10.1007/JHEP12(2010)088.DOI
  101. [101]Jinfeng Liao and Volker Koch. On the Fluidity and Super-Criticality of the QCD matter at RHIC. Phys. Rev. C, 81:014902, 2010. arXiv:0909.3105, doi:10.1103/PhysRevC.81.014902.
  102. [102]Gabriel S. Denicol, Charles Gale, Sangyong Jeon, and Jorge Noronha. Fluid behavior of a baryon-rich hadron resonance gas. Phys. Rev. C, 88(6):064901, 2013. arXiv:1308.1923, doi:10.1103/PhysRevC.88.064901.DOI
  103. [103]U. Gursoy, E. Kiritsis, and F. Nitti. Exploring improved holographic theories for QCD: Part II. JHEP, 02:019, 2008. arXiv:0707.1349, doi:10.1088/1126-6708/2008/02/019.DOI
  104. [104]Elias Kiritsis and George Pavlopoulos. Heavy quarks in a magnetic field. JHEP, 04:096, 2012. arXiv:1111.0314, doi:10.1007/JHEP04(2012)096.DOI
  105. [105]Gabriel S. Rocha, David Wagner, Gabriel S. Denicol, Jorge Noronha, and Dirk H. Rischke. Theories of Relativistic Dissipative Fluid Dynamics. Entropy, 26(3):189, 2024. arXiv:2311.15063, doi:10.3390/e26030189.DOI
  106. [106]J. I. Kapusta and C. Young. Causal Baryon Diffusion and Colored Noise. Phys. Rev. C, 90(4):044902, 2014. arXiv:1404.4894, doi:10.1103/PhysRevC.90.044902.DOI
  107. [107]Nabil Iqbal and Hong Liu. Universality of the hydrodynamic limit in AdS/CFT and the membrane paradigm. Phys. Rev. D, 79:025023, 2009. arXiv:0809.3808, doi:10.1103/PhysRevD.79.025023.DOI
  108. [108]Joseph I. Kapusta and Juan M. Torres-Rincon. Thermal Conductivity and Chiral Critical Point in Heavy Ion Collisions. Phys. Rev. C, 86:054911, 2012. arXiv:1209.0675, doi:10.1103/PhysRevC.86.054911.DOI
  109. [109]Sachin Jain. Holographic electrical and thermal conductivity in strongly coupled gauge theory with multiple chemical potentials. JHEP, 03:101, 2010. arXiv:0912.2228, doi:10.1007/JHEP03(2010)101.DOI
  110. [110]Gabriel S. Denicol, Charles Gale, Sangyong Jeon, Akihiko Monnai, Björn Schenke, and Chun Shen. Net baryon diffusion in fluid dynamic simulations of relativistic heavy-ion collisions. Phys. Rev. C, 98(3):034916, 2018. arXiv:1804.10557, doi:10.1103/PhysRevC.98.034916.DOI
  111. [111]Jacquelyn Noronha-Hostler, Gabriel S. Denicol, Jorge Noronha, Rone P. G. Andrade, and Frederique Grassi. Bulk Viscosity Effects in Event-by-Event Relativistic Hydrodynamics. Phys. Rev. C, 88(4):044916, 2013. arXiv:1305.1981, doi:10.1103/PhysRevC.88.044916.
  112. [112]J. Noronha-Hostler, J. Noronha, and F. Grassi. Bulk viscosity-driven suppression of shear viscosity effects on the flow harmonics at energies available at the BNL Relativistic Heavy Ion Collider. Phys. Rev. C, 90(3):034907, 2014. arXiv:1406.3333, doi:10.1103/PhysRevC.90.034907.
  113. [113]S. Ryu, J. F. Paquet, C. Shen, G. S. Denicol, B. Schenke, S. Jeon, and C. Gale. Importance of the Bulk Viscosity of QCD in Ultrarelativistic Heavy-Ion Collisions. Phys. Rev. Lett., 115(13):132301, 2015. arXiv:1502.01675, doi: 10.1103/PhysRevLett.115.132301.DOI
  114. [114]Jonah E. Bernhard, J. Scott Moreland, Steffen A. Bass, Jia Liu, and Ulrich Heinz. Applying Bayesian parameter estimation to relativistic heavy-ion collisions: simultaneous characterization of the initial state and quark-gluon plasma medium. Phys. Rev. C, 94(2):024907, 2016. arXiv:1605.03954, doi:10.1103/PhysRevC.94.024907.DOI
  115. [115]A. Majmudar and M. Van Leeuwen. The Theory and Phenomenology of Perturbative QCD Based Jet Quenching. Prog. Part. Nucl. Phys., 66:41–92, 2011. arXiv:1102.2206, doi:10.1016/j.ppnp.2010.09.001.
  116. [116]Karen M. Burke et al. Extracting the jet transport coefficient from jet quenching in high-energy heavy-ion collisions. Phys. Rev. C, 90(1):014909, 2014. arXiv:1312.5003, doi:10.1103/PhysRevC.90.014909.DOI
  117. [117]S. S. Adler et al. Absence of suppression in particle production at large transverse momentum in S(NN)**(1/2) = 200-GeV d + Au collisions. Phys. Rev. Lett., 91:072303, 2003. arXiv:nucl-ex/0306021, doi:10.1103/PhysRevLett.91.072303.
  118. [118]J. Adams et al. Evidence from d + Au measurements for final state suppression of high p(T) hadrons in Au+Au collisions at RHIC. Phys. Rev. Lett., 91:072304, 2003. arXiv:nucl-ex/0306024, doi:10.1103/PhysRevLett.91.072304.DOI
  119. [119]B. B. Back et al. Centrality dependence of charged hadron transverse momentum spectra in d + Au collisions at S(NN)**1/2 = 200 GeV. Phys. Rev. Lett., 91:072302, 2003. arXiv:nucl-ex/0306025, doi:10.1103/PhysRevLett.91.072302.
  120. [120]I. Arsene et al. Transverse momentum spectra in Au+Au and d+Au collisions at s**(1/2) = 200-GeV and the pseudorapidity dependence of high p(T) suppression. Phys. Rev. Lett., 91:072305, 2003. arXiv:nucl-ex/0307003, doi:10.1103/PhysRevLett.91.072305.
  121. [121]R. Baier, Yuri L. Dokshitzer, Alfred H. Mueller, S. Peigne, and D. Schiff. Radiative energy loss and p(T) broadening of high-energy partons in nuclei. Nucl. Phys. B, 484:265–282, 1997. arXiv:hep-ph/9608322, doi:10.1016/S0550-3213(96)00581-0.DOI
  122. [122]B. G. Zakharov. Radiative energy loss of high-energy quarks in finite size nuclear matter and quark - gluon plasma. JETP Lett., 65:615–620, 1997. arXiv:hep-ph/9704255, doi:10.1134/1.567389.DOI
  123. [123]R. Bellwied, S. Borsanyi, Z. Fodor, S. D. Katz, A. Pasztor, C. Ratti, and K. K. Szabo. Fluctuations and correlations in high temperature QCD. Phys. Rev. D, 92(11):114505, 2015. arXiv:1507.04627, doi:10.1103/PhysRevD.92.114505.DOI

Paper details

vlnc:W_cndhjmwj
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents