Viscous Anisotropic Hydrodynamics (VAH)

• VAH is a framework that reorganizes conventional hydrodynamics to nonperturbatively treat large momentum-space anisotropies, crucial for early-stage heavy-ion collisions.
• It combines an anisotropic background distribution with perturbative corrections to handle extreme pressure differences and smoothly reduce to standard second-order viscous hydrodynamics.
• VAH has been benchmarked against exact solutions and applied in Bayesian calibration studies, enhancing the modeling accuracy of heavy-ion collision dynamics and parameter extraction.

Searching arXiv for core VAH references and recent developments to ground the article in the literature. Viscous anisotropic hydrodynamics (VAH), often also denoted vaHydro in the earlier literature, is a reorganization of relativistic dissipative hydrodynamics in which the dominant momentum-space anisotropies of the quark-gluon plasma are incorporated already at leading order in the one-particle distribution function, while the remaining dissipative flows are evolved perturbatively. It was developed to address the regime in which the longitudinal and transverse pressures differ strongly, especially during the first fractions of a fm/$c$ after a heavy-ion collision and, in nonconformal settings, near the QCD crossover where bulk viscous effects become significant. In that sense VAH interpolates between far-from-equilibrium anisotropic expansion and late-time second-order viscous hydrodynamics, and it has been benchmarked against exact solutions of the relaxation-time-approximation (RTA) Boltzmann equation, extended to nonconformal $3+1$-dimensional evolution, and used in Bayesian calibration studies of heavy-ion data (Heinz et al., 2014, Bazow et al., 2015, McNelis et al., 2018, Heinz et al., 2023).

1. Historical development and conceptual role

The central motivation for anisotropic hydrodynamics was the observation that standard second-order viscous hydrodynamics expands around a locally isotropic equilibrium distribution and therefore becomes strained when the local rest frame pressure anisotropy is $O(1)$, as occurs in rapidly expanding quark-gluon plasma. Early reviews emphasized that such anisotropies are expected not only at very early times but also in dilute edge regions, and that standard viscous frameworks may then generate unphysical negative pressures (Strickland, 2014).

Leading-order anisotropic hydrodynamics addressed this by promoting the dominant longitudinal-transverse pressure splitting into the background distribution itself. VAH was the next step: it generalized this idea to include the residual dissipative currents through relaxation-type equations derived from kinetic theory, usually within RTA and a 14-moment or moments-based construction. In the formulation introduced in "Viscous hydrodynamics for strongly anisotropic expansion" (Heinz et al., 2014), the framework already contained a complete set of dissipative currents on top of a spheroidally deformed background. Subsequent work extended the formalism to nonconformal systems with finite masses and explicit bulk degrees of freedom (Bazow et al., 2015), to a new leading-order formulation consistent with Israel-Stewart theory in the near-equilibrium limit (Tinti, 2014), and to optimized $3+1$-dimensional formulations incorporating a lattice-QCD equation of state through quasiparticle dynamics (McNelis et al., 2018).

Within this lineage, VAH occupies an intermediate position between pure leading-order anisotropic hydrodynamics and conventional viscous hydrodynamics. Its defining feature is not merely the use of an anisotropic background, but the simultaneous nonperturbative treatment of the largest pressure anisotropies and perturbative treatment of the smaller residual stresses. This structure is what allows VAH to remain well behaved when $P_L \ll P_\perp$, yet reduce smoothly to second-order viscous hydrodynamics when the anisotropy relaxes.

2. Kinetic ansatz and macroscopic tensor structure

The kinetic starting point is a decomposition of the distribution function into an anisotropic leading piece and a residual correction,

$f(x,p)=f_a(x,p)+\delta\tilde f(x,p),$

with $f_a$ taken in generalized Romatschke-Strickland form. In a common covariant notation,

$$f_a(x,p)=f_{\rm iso}\!\Bigl(\sqrt{p_\mu \,\Xi^{\mu\nu}(x)\, p_\nu}/\Lambda(x)\Bigr),$$

where $\Lambda(x)$ is an effective momentum scale and $\Xi^{\mu\nu}$ encodes the momentum-space deformation (Heinz et al., 2014, McNelis et al., 2018). In nonconformal formulations the local-rest-frame form is often written with two deformation parameters,

$3+1$0

so that $3+1$1 encode the leading anisotropic state and, when supplemented by a mean field $3+1$2, reproduce realistic QCD thermodynamics (McNelis et al., 2018).

A standard VAH decomposition of the energy-momentum tensor introduces a timelike flow vector $3+1$3, a spacelike unit vector $3+1$4 selecting the longitudinal or beam direction, and the projector

$3+1$5

In this basis,

$3+1$6

where $3+1$7 is the energy density, $3+1$8 and $3+1$9 are the longitudinal and transverse pressures, $O(1)$0 is a longitudinal momentum-diffusion current, and $O(1)$1 is the residual transverse shear stress (McNelis et al., 2018, Peng et al., 4 Sep 2025).

This form is directly related to the conventional viscous decomposition

$O(1)$2

through

$O(1)$3

Accordingly, VAH treats the largest components of shear and, in nonconformal versions, bulk pressure through $O(1)$4 and $O(1)$5 themselves, leaving only the residual tensor structures to be handled perturbatively (McNelis et al., 2018, McNelis et al., 2018).

3. Closure, matching, and dynamical equations

The hydrodynamic backbone of VAH remains energy-momentum conservation,

$O(1)$6

supplemented by relaxation-type equations for the anisotropic pressures and residual flows. Closure requires more than ordinary Landau matching because the anisotropic background contains additional parameters beyond the local temperature and flow velocity. In the nonconformal $O(1)$7-dimensional quasiparticle construction, generalized matching conditions impose that the leading-order distribution alone reproduces the full local energy density, longitudinal pressure, and transverse pressure: $O(1)$8 These conditions determine $O(1)$9 as functions of the macroscopic fields (McNelis et al., 2018, McNelis et al., 2018).

The dynamical equations are obtained from moments of the Boltzmann equation. In the full $3+1$0-dimensional nonconformal setup, one evolves ten macroscopic variables,

$3+1$1

using four conservation equations and six relaxation-type equations for the dissipative flows (McNelis et al., 2018, McNelis et al., 2018). In conformal boost-invariant $3+1$2-dimensional reductions, the structure becomes especially transparent: $3+1$3

$3+1$4

$3+1$5

with the transport coefficients computed from underlying kinetic theory in RTA (Peng et al., 4 Sep 2025).

An essential formal property is the near-equilibrium limit. Several formulations showed explicitly that, for small anisotropy parameters and small residual stresses, VAH reduces to Israel-Stewart or DNMR-type viscous hydrodynamics with the appropriate relaxation times and second-order transport coefficients (Tinti, 2014, Tinti, 2015). The distinction is therefore not a different late-time theory, but a different resummation of the large inverse-Reynolds sector at early times.

4. Symmetry-reduced systems, exact benchmarks, and attractors

Much of the foundational evidence for VAH comes from comparisons with exactly solvable or highly constrained kinetic-theory problems. In the $3+1$6-dimensional boost-invariant RTA Boltzmann equation, exact solutions permit direct comparison of the energy density and pressure anisotropy with hydrodynamic approximations. Already in early tests, anisotropic hydrodynamics provided a very good approximation to the exact solution, and the viscous anisotropic extension performed much better than other known second-order viscous-hydrodynamic approximations (Florkowski et al., 2013, Heinz et al., 2014).

The nonconformal extension sharpened this result by incorporating a scalar bulk degree of freedom and finite masses at leading order. In "Nonconformal viscous anisotropic hydrodynamics" (Bazow et al., 2015), the macroscopic theory was derived by taking moments of the Boltzmann equation around a spheroidally deformed distribution with an additional scalar field for bulk viscous effects, and in $3+1$7-dimensional boost-invariant expansion it was shown to approximate the exact RTA solution more accurately than any other known hydrodynamic approximation.

A particularly stringent benchmark is Gubser flow, which combines boost-invariant longitudinal expansion with azimuthally symmetric transverse expansion and is naturally formulated in de Sitter coordinates. In "Viscous anisotropic hydrodynamics for the Gubser flow" (Martinez et al., 2017), the closure was obtained by selecting the relaxation equation for the longitudinal pressure together with a suitable Landau matching condition, leading to two coupled differential equations for the energy density and longitudinal pressure that respect the $3+1$8 symmetry of Gubser flow. Their numerical solutions were compared with the exact RTA Boltzmann solution subject to the same flow, and the paper reports that the MNR description reproduces the space-time evolution of the system better than the other hydrodynamical approaches considered.

These benchmark studies connect VAH to the modern language of hydrodynamic attractors. The broader anisotropic-hydrodynamic framework has been interpreted as a resummation to all orders in the inverse Reynolds number, which explains why its attractors track the kinetic-theory attractor much more closely than MIS or DNMR attractors in far-from-equilibrium regimes (Strickland, 2024). This suggests that VAH is best viewed not as a small modification of second-order hydrodynamics, but as a different organization of the macroscopic expansion in which the dominant nonequilibrium structure is treated exactly.

5. $3+1$9-dimensional simulation and heavy-ion phenomenology

The transition from symmetry-reduced tests to realistic collision modeling required numerically stable $P_L \ll P_\perp$0-dimensional implementations. A key development was the Eulerian-grid solver presented in "Anisotropic fluid dynamical simulations of heavy-ion collisions" (McNelis et al., 2021), which evolves VAH on a grid using the Kurganov-Tadmor algorithm together with an adaptive Runge-Kutta method. That scheme was designed to start soon after the nuclear collision, thereby largely avoiding the need for a separate pre-equilibrium module, and it was validated against conformal and nonconformal Bjorken flow as well as conformal Gubser flow. In central Pb+Pb simulations, nonconformal anisotropic hydrodynamics and second-order viscous hydrodynamics produced similar transverse profiles once the system relaxed, but the former yielded a longitudinal flow profile that responded more consistently to gradients along spacetime rapidity (McNelis et al., 2021).

The phenomenological use of VAH was advanced further in the Bayesian calibration study of Pb-Pb collisions at $P_L \ll P_\perp$1 TeV (Heinz et al., 2023). In that hybrid model, the usual pre-hydrodynamic and viscous relativistic fluid dynamic stages were replaced by a VAH core that smoothly interpolates between the initial approximately boost-invariant longitudinally free-streaming regime and the later collision-dominated $P_L \ll P_\perp$2-dimensional viscous-fluid regime. The calibration used ALICE data for integrated charged multiplicities, identified-hadron yields, mean $P_L \ll P_\perp$3, and integrated $P_L \ll P_\perp$4 up to $P_L \ll P_\perp$5, and it yielded meaningful constraints on the temperature-dependent specific shear and bulk viscosities up to about $P_L \ll P_\perp$6 MeV.

The reported posterior structure includes a well-defined minimum

$P_L \ll P_\perp$7

at $P_L \ll P_\perp$8, rising gently at high temperature, and a bulk-viscosity peak

$P_L \ll P_\perp$9

centered at

$f(x,p)=f_a(x,p)+\delta\tilde f(x,p),$0

with width

$f(x,p)=f_a(x,p)+\delta\tilde f(x,p),$1

all quoted as $f(x,p)=f_a(x,p)+\delta\tilde f(x,p),$2 credible intervals in the summarized results (Heinz et al., 2023). With best-fit parameters, the calibrated model was reported to predict identified-hadron $f(x,p)=f_a(x,p)+\delta\tilde f(x,p),$3 spectra for $f(x,p)=f_a(x,p)+\delta\tilde f(x,p),$4, $f(x,p)=f_a(x,p)+\delta\tilde f(x,p),$5, and $f(x,p)=f_a(x,p)+\delta\tilde f(x,p),$6 within $f(x,p)=f_a(x,p)+\delta\tilde f(x,p),$7 over $f(x,p)=f_a(x,p)+\delta\tilde f(x,p),$8 GeV and differential flows $f(x,p)=f_a(x,p)+\delta\tilde f(x,p),$9, $f_a$0, at the few-percent level for observables not used in the calibration. The principal significance of these results is not merely improved fit quality, but the extension of sensitivity to much higher temperatures than in models that begin hydrodynamics only after an explicit pre-equilibrium stage.

6. Applicability limits, small systems, and current research directions

A recurring controversy in relativistic heavy-ion theory concerns the applicability of hydrodynamics in small or short-lived systems. VAH has become central to this discussion because its domain of validity is explicitly designed to extend into large-anisotropy, large-Knudsen regimes where traditional second-order viscous hydrodynamics is least reliable.

In the conformal boost-invariant $f_a$1-dimensional study with transverse expansion reported in "Extended applicability domain of viscous anisotropic hydrodynamics in (2+1)-D Bjorken flow with transverse expansion" (Peng et al., 4 Sep 2025), VAH and Müller-Israel-Stewart hydrodynamics were compared with RTA kinetic theory using common initial conditions and varying opacity. For small $f_a$2, all three descriptions coincided. As $f_a$3 increased and $f_a$4, MIS began to deviate in early-to-mid times, underpredicting transverse flow and elliptic response, whereas VAH remained within $f_a$5 of the RTA benchmark down to $f_a$6 and within $f_a$7–$f_a$8 down to $f_a$9 in final-state observables. The study attributes this to the anisotropic ansatz capturing the large early-time longitudinal-transverse pressure anisotropy and thereby mitigating the wrong-sign pressure corrections that appear in MIS near free streaming (Peng et al., 4 Sep 2025).

This issue is even sharper in proton-proton collisions. In "Exploring the fluid behavior in p+p collisions at $$f_a(x,p)=f_{\rm iso}\!\Bigl(\sqrt{p_\mu \,\Xi^{\mu\nu}(x)\, p_\nu}/\Lambda(x)\Bigr),$$0 with viscous anisotropic hydrodynamics" (Zhao et al., 4 Sep 2025), VAH was used to study collectivity in p+p, a setting whose fluid interpretation remains controversial. The reported results state that VAH provides a good description of $$f_a(x,p)=f_{\rm iso}\!\Bigl(\sqrt{p_\mu \,\Xi^{\mu\nu}(x)\, p_\nu}/\Lambda(x)\Bigr),$$1 and $$f_a(x,p)=f_{\rm iso}\!\Bigl(\sqrt{p_\mu \,\Xi^{\mu\nu}(x)\, p_\nu}/\Lambda(x)\Bigr),$$2 over a wide range of multiplicities and correctly reproduces the experimentally observed negative $$f_a(x,p)=f_{\rm iso}\!\Bigl(\sqrt{p_\mu \,\Xi^{\mu\nu}(x)\, p_\nu}/\Lambda(x)\Bigr),$$3. By contrast, traditional second-order viscous hydrodynamics could reproduce the measurements only with parameter choices for which much of the evolution was characterized by large shear Knudsen number and, according to the summary, could not capture the large longitudinal/transverse pressure anisotropy during the early evolution (Zhao et al., 4 Sep 2025).

These results do not eliminate the broader debate about hydrodynamics in small systems, but they change its technical content. A plausible implication is that the relevant question is no longer whether an isotropic-background gradient expansion is valid, but whether a resummed anisotropic-background theory with small residual stresses remains controlled. Current directions identified in the literature include extension to nonconformal media with bulk viscosity and finite quark masses, fully $$f_a(x,p)=f_{\rm iso}\!\Bigl(\sqrt{p_\mu \,\Xi^{\mu\nu}(x)\, p_\nu}/\Lambda(x)\Bigr),$$4-dimensional simulations with rapidity fluctuations, electromagnetic fields or spin, more realistic collision kernels beyond RTA, and the inclusion of conserved charges and diffusion currents (Peng et al., 4 Sep 2025, McNelis et al., 2021).