Resummed Hydrodynamic Scheme

• Resummed hydrodynamic schemes are reorganization methods that absorb infinite gradient and moment contributions into effective objects, enhancing modeling of non-equilibrium fluids.
• They integrate techniques from kinetic theory, attractor analysis, and nonlinear constitutive laws to ensure stability, causality, and convergence even under strong gradients.
• These schemes enable systematic extensions of viscous hydrodynamics in applications such as heavy-ion collisions, supporting frameworks like Israel-Stewart, DNMR, and anisotropic hydrodynamics.

A resummed hydrodynamic scheme is, in current usage, a reorganization of hydrodynamics or its kinetic-theory derivation in which infinite subsets of gradient, moment, or transient contributions are absorbed into finite dynamical objects, effective transport coefficients, or modified evolution equations. In relativistic theory, this label has been applied to Gaussian-weighted moment hierarchies for Boltzmann-Vlasov-Maxwell dynamics, Borel- and attractor-based reinterpretations of divergent gradient series, stress- or gradient-dependent transport coefficients in Israel-Stewart and DNMR-type theories, and covariant order-by-order resummations of the hydrodynamic initial-value problem (Tinti et al., 2018, Romatschke, 2017, Chiu et al., 7 Aug 2025, Heller et al., 11 Nov 2025). At lowest nontrivial truncation such schemes typically reduce to familiar second-order viscous hydrodynamics or Israel-Stewart-like equations, while away from equilibrium they are designed to remain finite, stable, or causality-compatible (Tinti et al., 2018, Chiu et al., 7 Aug 2025).

1. Conceptual scope

The expression does not denote a single formalism. In the cited literature it refers to several related constructions whose common aim is to replace a naive finite truncation by a better organized description of far-from-equilibrium evolution. In one class of works, the problem is the infrared pathology of ordinary moment hierarchies once effective masses or gauge fields drive the hierarchy toward negative energy-index moments; the remedy is a resummed moment basis that packages infinitely many ordinary moments into Gaussian-weighted generating functions (Tinti et al., 2018). In another class, the problem is the divergence of the hydrodynamic gradient expansion itself; the remedy is to identify hydrodynamics with a Borel-resummed attractor branch rather than with any finite-order gradient truncation (Romatschke, 2017). Other works use the term for constitutive laws in which transport coefficients become nonlinear functions of gradient strength or inverse Reynolds numbers, thereby suppressing large-stress configurations and enforcing nonlinear causality bounds (Denicol et al., 2020, Chiu et al., 7 Aug 2025).

The motivations are correspondingly diverse but structurally similar. Standard gradient expansions are asymptotic and can have zero radius of convergence; ordinary moment hierarchies can become ill-defined in the presence of mean fields; and finite-order viscous theories may become unreliable when the Knudsen number or inverse Reynolds numbers are large. Resummation, in these settings, denotes an attempt to preserve hydrodynamic content while reorganizing the expansion so that the relevant far-from-equilibrium sectors are either summed, regularized, or dynamically suppressed (Tinti et al., 2018, Strickland, 2024).

A recurrent misconception is that “resummed hydrodynamics” always means a formally exact summation theorem. The literature is more heterogeneous. In some cases the resummation is an exact change of variables at the level of kinetic theory; in others it is a physically motivated effective completion of second-order hydrodynamics; in still others it is an attractor-based interpretation of the late-time transseries sector (Tinti et al., 2018, Chiu et al., 7 Aug 2025, Romatschke, 2017).

2. Resummed moment schemes from kinetic theory

The most explicit kinetic-theory realization begins from the relativistic method of moments. For the Boltzmann equation one introduces moments

$${\cal F}^{\mu_1\cdots \mu_l}_r=\int_{\mathbf p}(p\cdot u)^r\,p^{\mu_1}\cdots p^{\mu_l}\,f,$$

with hydrodynamic tensors such as $T^{\mu\nu}={\cal F}_0^{\mu\nu}$ contained among the lowest moments. When long-range effects are added through a space-time-dependent effective mass $m(x)$ and/or gauge fields $F^{\mu\nu}$, the exact Boltzmann-Vlasov hierarchy couples moments of a given rank to lower and lower energy indices $r$. In the ultrarelativistic limit $m/T\to 0$, moments with $r+l<-2$ diverge, so the straightforward higher-order extension becomes infrared divergent or otherwise ill-defined (Tinti et al., 2018).

The proposed remedy is to replace the ordinary hierarchy by resummed moments. One central definition is

$$\Phi^{\mu_1\cdots\mu_s}_r(x,\xi^2)=\int_p (p\cdot u)^r\,p^{\mu_1}\cdots p^{\mu_s}\,e^{-\xi^2(p\cdot u)^2}\,f(x,p),$$

or, in the fixed-index form emphasized in the 2018 formulation,

$$\Phi^{\mu_1\cdots\mu_l}(x,\xi^2)=\int_{\mathbf p}(p\cdot u)\,p^{\mu_1}\cdots p^{\mu_l}\,e^{-\xi^2(p\cdot u)^2}\,f(x,p).$$

The Gaussian factor makes each $\Phi$ a generating function for infinitely many ordinary moments, since expanding $T^{\mu\nu}={\cal F}_0^{\mu\nu}$0 produces an infinite tower of even energy-index contributions. Ordinary moments are recovered by $T^{\mu\nu}={\cal F}_0^{\mu\nu}$1-derivatives or $T^{\mu\nu}={\cal F}_0^{\mu\nu}$2-integrals, and in the general formulation one has identities such as $T^{\mu\nu}={\cal F}_0^{\mu\nu}$3 and $T^{\mu\nu}={\cal F}_0^{\mu\nu}$4 (Tinti et al., 2018).

This reorganization removes the negative-index singularity by trading problematic moments for regularized $T^{\mu\nu}={\cal F}_0^{\mu\nu}$5-dependent objects. It also changes the structure of truncation. In the Bjorken-reduced equations, higher-order couplings enter multiplied by factors such as $T^{\mu\nu}={\cal F}_0^{\mu\nu}$6, while the Gaussian kernel concentrates the support near $T^{\mu\nu}={\cal F}_0^{\mu\nu}$7. The papers identify this as a new truncation criterion unrelated to small Knudsen number or small pressure corrections: omitted higher moments are weak precisely in the $T^{\mu\nu}={\cal F}_0^{\mu\nu}$8-region that dominates the resummed integrals (Tinti et al., 2018).

The resulting hierarchy is formally exact before truncation. At leading nontrivial truncation it reproduces second-order viscous hydrodynamics or an Israel-Stewart-like theory, and increasing the maximal rank yields a systematic higher-order sequence. In $T^{\mu\nu}={\cal F}_0^{\mu\nu}$9-dimensional tests against exact solutions of the Boltzmann-Maxwell-Vlasov or coupled Boltzmann-Vlasov-Maxwell equations, the approximation converges rapidly and stably even in cases with large electric fields and large deviations from local equilibrium (Tinti et al., 2018, Tinti et al., 2018).

3. Attractors, Borel resummation, and hydrodynamic generators

A different use of resummation arises from the divergence of the relativistic gradient expansion. For conformal Bjorken flow, the series for $m(x)$0 has factorially growing coefficients, so the ordinary hydrodynamic series is asymptotic rather than convergent. In this setting, Borel transformation and transseries analysis lead to a non-analytic resummed object identified with a hydrodynamic attractor. The attractor is the universal solution toward which generic evolutions converge after non-hydrodynamic transients decay; the ordinary gradient expansion is only its late-time asymptotics (Romatschke, 2017).

Within this framework, hydrodynamics far from local equilibrium is not defined by small gradients or by local isotropy. It is defined by the decay of non-hydrodynamic modes and the persistence of an attractor branch. Romatschke’s formulation makes this explicit for rBRSSS, RTA kinetic theory, and strongly coupled $m(x)$1 SYM in Bjorken flow, and proposes an effective constitutive rewriting in terms of a gradient-dependent Borel-resummed viscosity $m(x)$2 whose small-gradient limit matches the equilibrium viscosity while its large-gradient behavior is strongly suppressed (Romatschke, 2017).

The same logic was recast directly in kinetic theory through the notion of a hydrodynamic generator. For the RTA Boltzmann equation, the exact solution can be written as a sum of an exponentially damped initial-state contribution and an integral over past equilibrium sources. In Bjorken flow this source term,

$m(x)$3

becomes, at late times and for constant relaxation time, exactly the Borel-resummed Chapman-Enskog series. In full $m(x)$4 dimensions the construction generalizes to an integral along kinetic characteristics. At finite times the same representation exhibits incomplete-gamma suppression factors multiplying the gradient corrections, thereby identifying how non-hydrodynamic modes regulate the onset of hydrodynamics (McNelis et al., 2020).

Anisotropic hydrodynamics provides yet another resummed realization. Its leading-order distribution is momentum-space anisotropic,

$m(x)$5

so the dominant longitudinal-transverse pressure splitting is treated nonperturbatively rather than through a perturbative expansion in the inverse Reynolds number. In the conformal Bjorken case this implies a bounded shear variable $m(x)$6, and when the aHydro evolution equation is rewritten in terms of $m(x)$7 it contains nonlinear functions of $m(x)$8, hence an infinite power series in $m(x)$9. The review literature emphasizes that this amounts to an all-orders resummation in the inverse Reynolds number and, through the attractor equation, an effective all-orders resummation of the gradient series. The aHydro attractor lies much closer to the exact RTA attractor than MIS or DNMR and maintains positive longitudinal pressure for all $F^{\mu\nu}$0 (Strickland, 2024).

4. Nonlinear constitutive resummations in viscous hydrodynamics

A more phenomenological use of the term appears in Israel-Stewart and DNMR-type theories when transport coefficients are promoted from fixed microscopic inputs to gradient- or stress-dependent effective quantities. In conformal Israel-Stewart theory at zero chemical potential, one may define the dimensionless shear tensor $F^{\mu\nu}$1 and perform a slow-roll expansion of its exact equation for general flow. At zeroth slow-roll order one obtains an algebraic relation implying $F^{\mu\nu}$2, which yields

$F^{\mu\nu}$3

with

$F^{\mu\nu}$4

This “resummed shear viscosity” equals $F^{\mu\nu}$5 for $F^{\mu\nu}$6, decreases as $F^{\mu\nu}$7 for $F^{\mu\nu}$8, and keeps $F^{\mu\nu}$9 finite even when gradients are large. The authors interpret it as an effective constitutive coefficient valid in the attractor regime, not as a redefinition of the underlying microscopic viscosity (Denicol et al., 2020).

A 2025 heavy-ion formulation applies the same general idea to full DNMR hydrodynamics with shear and bulk sectors. There the right-hand-side transport coefficients are multiplied by a regulator

$r$0

where $r$1 and $r$2 are shear and bulk inverse Reynolds numbers. The relaxation times $r$3 and $r$4 are left unchanged, while viscosities and second-order coefficients are scaled by powers of $r$5. Since the argument of $r$6 must satisfy $r$7, the evolution is automatically bounded by

$r$8

Near equilibrium one has $r$9, so the first correction is quadratic in inverse Reynolds numbers and enters only at third order in gradients; thus the scheme reduces to standard second-order DNMR in the small-stress regime (Chiu et al., 7 Aug 2025).

The stated purpose of this nonlinear completion is enforcement of nonlinear causality conditions. When $m/T\to 0$0, the source terms are turned off and the viscous equations become pure relaxation,

$m/T\to 0$1

so large stresses are exponentially quenched rather than amplified. For the lattice-QCD equation of state, the choice $m/T\to 0$2 is reported to satisfy the simplified necessary causality conditions for $m/T\to 0$3, implying $m/T\to 0$4 (Chiu et al., 7 Aug 2025).

5. Order-by-order and linear-response reformulations

Resummation also appears as a solution to the initial-value problem of gradient-truncated relativistic hydrodynamics. In the 2025 “stable evolution” framework, the hydrodynamic fields are expanded as

$m/T\to 0$5

and the central difficulty is secular growth in the coefficient fields rather than a failure of the effective theory itself. The proposed cure is a covariant redefinition of the coefficient fields that introduces inter-order transfer currents $m/T\to 0$6 while preserving the perturbative sum. At first order the modified hierarchy takes the form

$m/T\to 0$7

and the intermediate evolution equation for the zeroth-order fields is precisely BDNK-like, even though the full construction starts and ends in the Landau frame. Non-hydrodynamic modes appear in the intermediate coefficient-level dynamics but cancel in the reconstructed physical fields to the target order. The method introduces no new physical fields and requires no additional physical initial data beyond those of ideal hydrodynamics (Heller et al., 11 Nov 2025).

A different linear-response resummation is furnished by holography. For strongly coupled $m/T\to 0$8 SYM in a weakly curved background, the dissipative stress tensor can be written as a constitutive relation that is linear in amplitudes but exact to all orders in derivatives,

$m/T\to 0$9

The transport coefficients $r+l<-2$0 are operator-valued functions of $r+l<-2$1 and $r+l<-2$2, or ordinary functions of $r+l<-2$3 in Fourier space. The curvature-dependent coefficients were identified as gravitational susceptibilities of the fluid. The same work reformulates the constitutive law in terms of memory functions; the viscosity memory kernel vanishes for negative times, and the authors interpret this as the causal form of the all-order linear resummation (Bu et al., 2015).

These two approaches differ substantially in purpose, but both replace a naive derivative truncation by auxiliary structures that reorganize the derivative expansion while preserving the intended hydrodynamic content. In one case the auxiliary objects are inter-order currents in an EFT expansion; in the other they are momentum-dependent transport functions derived from AdS/CFT (Heller et al., 11 Nov 2025, Bu et al., 2015).

6. Extensions, applications, and open issues

The resummation logic has been extended beyond spinless relativistic viscous hydrodynamics. In dissipative relativistic spin hydrodynamics derived from quantum kinetic theory, the inverse-Reynolds dominance (IReD) scheme organizes the spin moment hierarchy by joint counting in Knudsen and inverse Reynolds numbers. The resulting second-order accurate closure reduces roughly thirty spin-related quantities to eleven equations: six for the components of the spin potential and five for a dissipative irreducible rank-two tensor. The final relaxation-type equations evolve $r+l<-2$4, $r+l<-2$5, and $r+l<-2$6, and explicit first- and second-order transport coefficients are computed for a simple truncation (Wagner, 2024).

Heavy-ion phenomenology supplies the main application domain. The resummed DNMR causality scheme was implemented in MUSIC inside the IP-Glasma + MUSIC + UrQMD hybrid framework and tested event by event for Pb+Pb and p+Pb collisions at $r+l<-2$7 TeV. With equilibrium initialization the effect is minimal in Pb+Pb and reaches up to about 10% in charged-hadron anisotropic flow coefficients in p+Pb; with far-from-equilibrium initialization the resummed scheme produces noticeable changes, especially in peripheral Pb+Pb and in p+Pb, and the authors interpret this as a theoretical uncertainty of small-system hydrodynamics (Chiu et al., 7 Aug 2025). The anisotropic-hydrodynamics program similarly argues that resummation of inverse Reynolds number contributions improves the description of early-time QGP evolution, freeze-out, and the extraction of transport coefficients at high temperature (Strickland, 2024).

Several limitations recur across the literature. Some resummed schemes are exact reorganizations of kinetic theory, but others are effective nonlinear completions not derived uniquely from microscopic dynamics. The attractor-based constructions do not replace the full microscopic evolution for arbitrary initial data; they isolate the universal sector that remains after transient modes decay. The kinetic resummed-moment program remains classical and explicitly identifies extension to the full relativistic quantum case, with Wigner quasiprobability functions replacing the classical distribution $r+l<-2$8, as a future direction (Chiu et al., 7 Aug 2025, Romatschke, 2017, Tinti et al., 2018).

This suggests that “resummed hydrodynamic scheme” is best understood not as a single theory but as a methodological program. Its defining feature is the replacement of a naive truncation by a reorganization that incorporates, regularizes, or dynamically controls sectors that would otherwise be encoded in an infinite gradient series, an ill-defined moment hierarchy, or unstable transient dynamics. Across kinetic theory, attractor theory, viscous constitutive modeling, holography, and spin hydrodynamics, resummation functions as a device for extending hydrodynamic reasoning deeper into regimes where local equilibrium and small-gradient assumptions are no longer parametrically justified (Tinti et al., 2018, Romatschke, 2017, Heller et al., 11 Nov 2025, Wagner, 2024).