---
title: Non-Thermal Fixed Point Overview
url: https://www.emergentmind.com/topics/non-thermal-fixed-point
type: topic
---

# Non-Thermal Fixed Point Overview

Searching arXiv for recent and foundational papers on non-thermal fixed points.
arxiv_search.query({"search_query":"all:\"non-thermal fixed point\"","start":0,"max_results":10,"sort_by":"submittedDate","sort_order":"descending"})
arxiv_search.query({"search_query":"ti:\"non-thermal fixed point\" OR abs:\"non-thermal fixed point\"","start":0,"max_results":10,"sort_by":"submittedDate","sort_order":"descending"})
{"search_query":"all:\"non-thermal fixed point\"","start":0,"max_results":5,"sort_by":"submittedDate","sort_order":"descending"}
A non-thermal fixed point (NTFP) is a far-from-equilibrium dynamical attractor at which a many-body system exhibits universal self-similar evolution without being in thermal equilibrium. In this regime, correlation functions, structure factors, or occupation spectra become invariant in form under suitable rescalings of time, space, and momentum, so that the dynamics is “fixed” only in a scaling sense rather than by stationarity in the thermodynamic sense. The concept is the nonequilibrium analogue of an equilibrium renormalization-group fixed point, but it describes transient or long-lived scaling states reached after strong quenches, overoccupation, defect imprinting, or related extreme preparations in systems ranging from ultracold Bose gases and spinor condensates to non-Abelian plasmas, holographic superfluids, and dilute Fermi gases [1810.08143, 2304.12464].

## 1. Definition and scaling structure

The defining signature of an NTFP is self-similar evolution. In real space, this is expressed by scaling forms such as
\[
C(x,t)=t^{\alpha} f\!\left(t^{-\beta}x\right),
\]
while in momentum space one commonly writes
\[
n(k,t)=\left(\frac{t}{t_0}\right)^\alpha f\!\left[\left(\frac{t}{t_0}\right)^\beta k\right].
\]
After rescaling the amplitude by \(t^\alpha\) and the momentum or length axis by \(t^\beta\), data taken at different times collapse onto a single scaling function. This collapse is the operational criterion that the dynamics has entered the basin of an NTFP rather than merely exhibiting a transient power law [1810.08143, 2203.14752].

The time exponent \(\beta\) sets the drift of the characteristic scale, \(k_\Lambda(t)\sim t^{-\beta}\) or equivalently \(\ell(t)\sim t^\beta\), and the amplitude exponent \(\alpha\) determines how occupation or correlation weight redistributes in the scaling window. Conservation laws constrain these exponents. Particle-number conservation in the scaling regime implies
\[
\alpha=d\beta,
\]
whereas energy conservation implies
\[
\alpha=(d+z)\beta,
\]
with \(d\) the spatial dimension and \(z\) the dynamical exponent of the dispersion. Since both relations cannot generally hold simultaneously for a single scaling window, NTFP dynamics often involves bidirectional transport, with particle transport toward the infrared and energy transport toward the ultraviolet [1801.09490, 2304.12464].

An NTFP is not a thermal fixed point. It does not satisfy detailed balance, and it need not be stationary in laboratory time. Rather, the system approaches a scaling regime in which evolution is critically slowed: observables still change, but only through algebraic rescaling. This slowing is a recurrent diagnostic in Bose-gas turbulence, vortex coarsening, spinor dynamics, and expanding kinetic theories [1204.2487, 2203.14752].

## 2. Analytical formulations

A principal analytical route to NTFPs is non-perturbative kinetic theory derived from the Schwinger-Keldysh closed-time-path formalism. In this approach, one starts from nonequilibrium equations for statistical and spectral two-point functions and obtains a kinetic equation for an occupation distribution \(n(\mathbf{k},t)\). In the infrared, where occupations are large, the relevant scattering is not described by the bare coupling alone; a loop-resummed effective coupling \(g_{\mathrm{eff}}\) becomes universal in the collective-scattering regime. For dilute Bose gases this yields scaling relations such as
\[
\beta=\frac{1}{z},\qquad \alpha=\frac{d}{z},
\]
together with momentum exponents such as \(\kappa=d+(3z-4)/2+\eta\) and \(\zeta=d+z/2\) in the corresponding particle and occupation distributions [1801.09490, 1810.08143].

A complementary formulation is the low-energy effective field theory of multicomponent Bose gases. There, density fluctuations are integrated out and the remaining long-wavelength dynamics is expressed in terms of phase fields. For a \(\mathrm{U}(N)\)-symmetric gas, the low-energy theory contains one Bogoliubov-like total-phase mode and \(N-1\) quadratic relative-phase Goldstone modes. In the large-\(N\) limit, the fixed point dominated by relative phases is Gaussian with respect to these low-energy quasiparticles, and the theory predicts
\[
\beta=\frac12,\qquad \alpha=\frac d2,\qquad \kappa=d+1.
\]
This Gaussianity refers to the effective phase variables, not to the microscopic Bose field [1807.10228].

The functional renormalization group on a closed real-time path provides a more general fixed-point classification. For an \(O(N)\)-symmetric scalar theory, the scaling of statistical and spectral functions is written as
\[
F(p^0,\mathbf p)=s^{2+\kappa}F(s^z p^0,s\mathbf p),\qquad
\rho(p^0,\mathbf p)=s^{2-\eta}\rho(s^z p^0,s\mathbf p),
\]
and the nonequilibrium fixed-point analysis yields relations such as
\[
\kappa=-\eta+z+d
\]
as well as a second solution \(\kappa=-\eta+2z+d\). In this formulation, nonthermal fixed points appear as genuine infrared fixed points of quantum field theory, alongside vacuum and thermal equilibrium fixed points [0809.5208].

Recent work extends the theory from fixed-point identification to fixed-point stability. In longitudinally expanding non-Abelian kinetic theory, the self-similar attractor
\[
f(\tau,p_T,p_z)\sim \tau^\alpha f_S\!\left(\tau^\beta p_T,\tau^\gamma p_z\right)
\]
has fixed-point exponents
\[
\alpha_*=-\frac23,\qquad \beta_*=0,\qquad \gamma_*=\frac13,
\]
and the linearized stability matrix has eigenvalues \(-2\) and \(-4/3\), implying a stable attractor with power-law relaxation of the scaling exponents [2203.02299]. A related development introduces quasinormal modes of NTFPs: perturbations of the scaling state are written as \(B(t)^{i\Omega}\delta f_\Omega(\bar p)\), producing a tower of power-law decays rather than exponential thermal relaxation [2502.01622]. Another extension shows that weakly inhomogeneous perturbations around an isotropic NTFP support hydrodynamic shear and sound modes with time-dependent transport coefficients; in relativistic and nonrelativistic direct cascades, both \(\eta(t)\) and \(\tau_\pi(t)\) scale linearly in time [2504.18754].

## 3. Universality classes and microscopic mechanisms

The same scaling form can arise from distinct microscopic mechanisms. One widely discussed class is quasiparticle transport and wave turbulence, where particles and energy move through momentum space under kinetic constraints. Another is defect-driven coarsening, where the relevant degrees of freedom are vortices, solitons, domain walls, or phase-amplitude defects whose spacing grows algebraically in time [1810.08143, 1302.1448].

A particularly clear distinction appears in the quenched two-dimensional Bose gas. Semi-classical Gross-Pitaevskii simulations identify a strongly anomalous NTFP with
\[
\alpha_a = 0.402\pm 0.05,\qquad \beta_a=0.193\pm 0.05,\qquad \zeta_a=5.7\pm 0.3,
\]
and a near-Gaussian NTFP with
\[
\alpha_g = 1.10\pm 0.08,\qquad \beta_g=0.56\pm 0.08,\qquad \zeta_g=4.0\pm 0.1.
\]
The anomalous regime is associated with suppressed dissipative coupling, vortex clustering, and a three-vortex-collision-induced loss law \(N_d(t)\sim t^{-2/5}\), whereas the Gaussian regime is associated with pairwise annihilation and \(N_d(t)\sim t^{-1}\) [1611.01163].

The sine-Gordon model provides a different anomalous universality class. In its non-relativistic limit, classical-statistical simulations and kinetic analysis yield self-similar coarsening with
\[
\alpha=d\beta,\qquad \beta=\frac{1}{d+2},\qquad \kappa=2d+2,
\]
so that \(\beta=1/4\) in \(d=2\) and \(\beta=1/5\) in \(d=3\). The transport is not wave-turbulent in the local Kolmogorov-Zakharov sense; it is concentrated near the evolving infrared front and is interpreted as non-local in momentum space, with a corresponding real-space picture of growing coherent patches [2212.01162].

Spinor gases show that the microscopic symmetry alone does not determine the far-from-equilibrium universality class. In a quasi-one-dimensional \(^{87}\mathrm{Rb}\) spinor Bose-Einstein condensate evolving in the easy-plane phase under the same Hamiltonian symmetry, a polar initial condition yields approximately
\[
\alpha = 0.64 \pm 0.09_{\rm stat} \pm 0.50_{\rm sys},\qquad
\beta = 0.58 \pm 0.04_{\rm stat} \pm 0.26_{\rm sys},
\]
or \(\beta \approx 0.54\) under the constraint \(\alpha=\beta\). A rotation initial condition with six localized spin rotations instead yields
\[
\alpha = 0.24 \pm 0.04_{\rm stat} \pm 0.03_{\rm sys},\qquad
\beta = 0.28 \pm 0.04_{\rm stat} \pm 0.05_{\rm sys}.
\]
Spatially resolved measurement of the complex order parameter \(F_\perp(x)=|F_\perp(x)|e^{iL(x)}\) shows that the former regime is dominated by phase excitations, while the latter contains persistent phase-amplitude defects. This is direct evidence that one microscopic symmetry class can support more than one NTFP, selected by different basins of attraction [2306.16497].

In a one-dimensional antiferromagnetic spin-1 Bose gas, the relevant defects are magnetic solitons. After a strong quench across the quadratic-Zeeman threshold, thermalization proceeds through metastable Flemish-string bound states of opposite-sign solitons. The nematic correlation function becomes self-similar with exponents
\[
\beta \simeq 0.32,\qquad \gamma \simeq 0.11,
\]
while the soliton number decays as \(n_{\rm sol}\propto t^{-0.31}\), tying the NTFP directly to defect annihilation [1812.03581].

A quasi-two-dimensional dipolar Bose gas adds another variation. There, the anomalous coarsening exponent \(\beta\approx 0.2\) persists across contact-only and dipolar regimes, while roton excitations alter the vortex morphology: contact interactions favor clustering of like-sign vortices, whereas tilted dipolar interactions favor anticlustering or maximization of distances between defects. At late times the system crosses over to a diffusion-type NTFP with \(\beta\approx 1/2\) [2506.01653].

## 4. Experimental realizations in quantum gases

Experiments in Bose gases have established NTFPs through scaling collapses of momentum distributions, spin structure factors, and defect-separation statistics. The reported exponents depend on dimensionality, conserved quantities in the scaling regime, and the character of the excitations, but the central diagnostic is always the same: rescaled data from different times collapse onto a single scaling function [2203.14752].

| System | Observable | Reported scaling |
|---|---|---|
| Quasi-1D \(^{87}\mathrm{Rb}\) | Momentum distribution \(\bar n(k,t)\) | \(\alpha=0.09(5)\), \(\beta=0.10(4)\) |
| Spin-1 Bose gas | Spin structure factor \(f_\theta(k,t)\) | \(\alpha=0.33(8)\), \(\beta=0.54(6)\) |
| Homogeneous 3D \(^{39}\mathrm{K}\) | IR and UV momentum scaling | IR: \(\alpha=1.15(8)\), \(\beta=0.34(5)\); UV: \(\alpha=-0.70(7)\), \(\beta=-0.14(2)\) |
| Turbulent 3D \(^{87}\mathrm{Rb}\) | 2D-projected and reconstructed spectra | Projected: \(\alpha=-0.50(8)\), \(\beta=-0.2(4)\); reconstructed 3D: \(\alpha=-0.75\), \(\beta=-0.2\) |
| Quasi-1D \(^{6}\mathrm{Li}_2\) | Longitudinal momentum distribution | 1D window: \(\alpha \approx 0.09 \pm 0.01\), \(\beta \approx 0.10 \pm 0.01\); crossover: \(\alpha \approx 0.20 \pm 0.05\), \(\beta \approx 0.46 \pm 0.07\) |

The quasi-one-dimensional \(^{6}\mathrm{Li}_2\) experiment is especially significant for universality. A strongly interacting molecular Bose gas, driven far from equilibrium by a white-noise speckle phase imprint, exhibits the same 1D scaling exponents and the same universal scaling function as the previously observed weakly interacting \(^{87}\mathrm{Rb}\) case. In the reduced parameter space \((\xi_s/\xi_h,\mu/\hbar\omega_\perp)\), the observed 1D scaling window is associated with narrow solitonic defects, and the data are interpreted as evidence for a single universal NTFP with a large basin of attraction across quasi-1D bosonic systems [2505.20213].

The 2023 spinor-BEC experiment adds a further experimental refinement by combining scaling analysis of the transverse-spin structure factor with spatially resolved access to the complex order parameter. This makes it possible to identify not only the exponents but also the excitation content of the corresponding attractor, distinguishing phase-only coarsening from defect-rich phase-amplitude dynamics within the same microscopic easy-plane phase [2306.16497].

## 5. Beyond weakly interacting Bose gases

NTFPs are not confined to weakly interacting Gross-Pitaevskii dynamics. In a relativistic \(O(N)\)-symmetric scalar quantum field theory in \(3+1\) dimensions, nonequilibrium evolution solved with a 2PI \(1/N\) expansion to next-to-leading order shows that the infrared inverse-particle cascade with \(\kappa_N\approx 5\) persists beyond the weak-coupling limit and remains visible at couplings of order one. The infrared sector is largely insensitive to whether the initial state is overoccupied or contains a macroscopic field expectation value, whereas ultraviolet transport depends on the presence or absence of that background field [1607.02160].

In longitudinally expanding non-Abelian plasmas, NTFPs organize the prescaling dynamics of the gluon distribution. A Fokker-Planck collision kernel yields the fixed-point exponents \(\alpha_*=-2/3\), \(\beta_*=0\), and \(\gamma_*=1/3\), reproducing the bottom-up scaling of the expanding plasma. The derived stability equations show that perturbations decay with negative eigenvalues, so the fixed point is a true attractor of the simplified kinetic evolution rather than a formal scaling ansatz only [2203.02299].

Strong coupling does not remove the phenomenon. In a holographic \((2+1)\)-dimensional superfluid at finite temperature and chemical potential, large ensembles of vortices and antivortices evolve into a universal late-time regime characterized by
\[
n(k,t)\sim k^{-\zeta},\qquad 4.1\lesssim \zeta \lesssim 4.3,
\]
together with
\[
l_+(t)\sim t^{1/2},\qquad l_c(t)\sim t^{0.5\text{--}0.6}.
\]
The authors interpret this regime as an NTFP of the boundary theory and as a stationary or partially stable point of the dual classical bulk equations of motion [1410.3472].

Fermionic nonequilibrium matter admits stationary analogues as well. For a spatially homogeneous, isotropic dilute ultracold Fermi gas, a kinetic equation derived from next-to-leading-order Schwinger-Keldysh/Kadanoff-Baym theory has Kolmogorov-type non-thermal fixed points in the quasiparticle limit, with particle- and energy-cascade exponents
\[
\kappa_w=\frac{3d-z-3}{\ell},\qquad
\kappa_w=\frac{3d-3}{\ell}.
\]
Beyond the quasiparticle approximation, the same framework yields additional stationary states, including an off-shell energy-flux solution
\[
\kappa_w=\frac{1}{2}\left(-11+3d+4z+4\eta\right)
\]
and strong-coupling large-\(N\) exponents
\[
\kappa_s=\frac{d+z-1}{\ell},\qquad
\kappa_s=\frac{d+2z-1}{\ell}.
\]
This extends the NTFP concept from bosonic wave and defect dynamics to dilute fermionic matter [2508.18821].

## 6. Basin structure, prescaling, and unresolved issues

A central unresolved issue is classification. Equilibrium universality classes are largely organized by symmetry and dimensionality, whereas far-from-equilibrium universality also depends on initial conditions, defect content, and the geometry of transport. The observation of two distinct NTFPs for the same microscopic symmetry in a quasi-one-dimensional spinor Bose gas makes this point explicit: the fixed-point data themselves are insufficient without the basin structure that selects them [2306.16497].

Prescaling sharpens this picture. In prescaling, the scaling function has already approached its fixed-point form, while the exponents \(\alpha(\tau)\), \(\beta(\tau)\), and \(\gamma(\tau)\) still drift slowly. Position-space observables may show the universal exponents earlier than full momentum-space collapse. This behavior has been discussed both in expanding kinetic theory and in multicomponent Bose gases, where low-order terms in short-distance correlation expansions can lock onto fixed-point exponents before higher-order terms do so [1810.08143, 2203.02299].

Another active development concerns the direct approach to the fixed point. Quasinormal-mode analyses propose that deviations from the self-similar scaling state decay as a tower of power laws rather than exponentials, reflecting the time-dependent scaling background. Hydrodynamic analyses add that NTFPs can support shear and sound modes with time-dependent transport coefficients. These results do not make the NTFP thermal; they show instead that some of the structures familiar from near-equilibrium dynamics reappear around manifestly nonequilibrium scaling backgrounds [2502.01622, 2504.18754].

Several common misconceptions can therefore be excluded. First, an NTFP is not a synonym for any transient power law: in the holographic superfluid, an intermediate Kolmogorov regime depends on the initial vortex ensemble and is non-universal, whereas the later \(k^{-4.1\text{--}4.3}\) regime with algebraic length-scale growth is the universal attractor [1410.3472]. Second, microscopic symmetry does not by itself fix the universality class, as shown by multiple NTFPs with identical symmetry but different initial conditions [2306.16497]. Third, “Gaussian” fixed point can mean Gaussian with respect to effective low-energy variables, not with respect to the underlying microscopic fields [1807.10228].

The present state of the subject is therefore twofold. On the one hand, NTFPs have become a unifying language for universal self-similar dynamics in quantum fluids, spinor condensates, relativistic field theories, plasmas, holographic systems, and dilute Fermi gases. On the other hand, a systematic classification of all such attractors, their basins, crossover structure, and stability properties remains incomplete and is explicitly identified as a task for future nonequilibrium many-body theory [2304.12464].

Source: https://www.emergentmind.com/topics/non-thermal-fixed-point