betaFeedback

WorksPhysics and Astronomy

Kinetic Fluctuation-Dissipation Relation in collisionless plasmas: a correlated noise source

Abstract

We here focus on small-scale dynamics of collisionless plasmas by employing a generalized kinetic Langevin framework driven by a stochastic time-correlated forcing. Our investigation captures the steady-state Fluctuation-Dissipation Relation (FDR) where the energy injection at low frequencies is due to the time-correlated forcing and the small-scale energy conversion occurs via Landau damping, the latter –though formally reversible– acting as an effective dissipation mechanism. We analytically extract the resulting asymptotic spectral scalings across multiple physical regimes and characterize the velocity-space cascade by decomposing the plasma distribution function with Hermite polynomials. The framework is then generalized to a multi-scale forcing scenario in which correlation rates are distributed according to a power-law distribution. This extension provides closed-form analytical expressions that may serve to interpret spacecraft in-situ observations. All analytical predictions are successfully validated through direct numerical integrations of the underlying kinetic equations.

Introduction

The issue of energy conversion and dissipation in nearly collisionless plasmas is a fundamental, long-standing problem in space and astrophysical plasma physics[1, 2, 3]. In these environments, such as the solar wind and the Earth’s magnetosphere, viscous-like dissipation induced by collisional effects is thought to be negligible, as collisional frequencies are way smaller than typical frequencies associated with collisionless phenomena. Therefore, energy conversion at kinetic scales is mediated by non-linear wave-particle interactions and generally occurs in the entire phase space, frequently driving non-thermal plasma distribution functions.

A common feature of space and astrophysical plasmas is that they are manifestly turbulent[4]. Nonlinear mechanisms transfer energy from large, energy-containing scales to small, kinetic scales at which energy conversion and, eventually, dissipation occur. Moreover, temporal and spatial coherent structures such as vortices and sheets emerge across scales as a result of intermittency. The fundamental turbulent nature of nearly-collisionless space plasmas has paved the way to investigate these systems in the framework of statistical physics of non-equilibrium complex systems[5, 6, 7, 8, 9, 10, 11]. In this context, the Fluctuation-Dissipation Relation (FDR) represents a powerful theoretical tool to describe the steady-state balance between the energy injected into the system at large scales and the internal energy conversion and dissipation mechanisms.

Previous theoretical models[12] have made significant progress by treating the turbulent energy injection as a white-noise source within a kinetic Langevin framework. However, this approach intrinsically leaves a knowledge gap, as also evidenced by recent studies based on space plasmas observations[13, 14]. Indeed, by definition, turbulence is characterized by finite spatial and temporal correlations[15, 16, 17], and assuming instantaneous decorrelation (white noise) may oversimplify the turbulent plasma dynamics. Hence, treating turbulent forcing with time-correlated noise is important to accurately model large-scale forcing and, in turn, interpret the plasma behavior and spectral signatures[13, 14].

In this work, we generalize the kinetic framework based on the Langevin phenomenology introduced in Ref.[12], by considering a damped-forced collisionless electrostatic plasma model where plasma response is driven by a stochastic forcing term characterized by a finite correlation time, which mimics the property of plasma turbulence[13]. In this framework, the continuous power injected from larger scales is counterbalanced by collisional energy dissipation. However, unlike collisional plasmas, collisions are assumed to be negligible, and energy conversion is mediated by the collisionless wave-particle interaction. The use of a finite correlation time for the stochastic forcing allows us to model the non-Markovian nature of turbulent energy injection, providing a more accurate representation of energy transfer in collisionless plasmas. We derive the steady-state Fluctuation-Dissipation Relation and analyze the spectral properties of the system across different physical regimes. Furthermore, we extend the framework to a multi-scale forcing scenario, where multiple correlation rates are distributed according to a power-law distribution, enabling the description of a broader range of turbulent behaviors. sionless energy conversion mechanisms, as Landau damping. The forcing drives a stationary state where these kinetic processes effectively mediate the energy conversion and, despite their reversible nature, act as an effective dissipation, intended here as a energy transfer from electric field to particle distribution function. This balance ultimately drives the plasma toward a steady-state equilibrium governed by the FDR. We analytically derive the generalized FDR for the system response and extract its asymptotic spectral scalings across several interesting physical regimes. Moreover, to establish a direct connection between theoretical modeling and in-situ spacecraft observations, we extend our analysis to a multi-scale forcing scenario: the stochastic source is modeled as a superposition of noises with different correlation times governed by a power-law distribution[13]. All of our analytical predictions are validated against direct numerical integrations of the underlying kinetic equations.

The paper is organized as follows. In Section II we introduce the kinetic analogue of the Langevin equation, provide the details of the generalized FDR derivation, and explore its asymptotic limits and Hermite spectrum. Then, the numerical simulations validating the theoretical scalings are illustrated in Section III, Section IV extends the framework to the multi-scale forcing where the correlation times follow a power-law distribution, and we summarize our conclusions in Section V.

This work is dedicated to the memory of Professor Vincenzo Carbone, mentor and friend for many of us, who sadly passed away about two years ago. His pioneering paper[13] and later discussions with several of us undoubtedly contributed to set the theoretical foundation of the analyses shown here.

Kinetic analogue of the Langevin equation

We are interested in describing the evolution of a quasi-neutral plasma, where one species is treated within a kinetic Vlasov framework and another one is a background neutralizing flow. To establish a connection with the phenomenology of the Langevin equation, a damped-forced scenario must be considered. In the model analyzed here, some of the properties of large- and ion-scale plasma turbulence are collected into a stochastic source term on the right-hand side of the Vlasov equation. This forcing represents the power injected into the system from larger scales and takes the form of a noise with a finite correlation time[13]. Conversely, dissipation is included in our description solely through the Landau damping mechanism. Although Landau damping is not a true collisional dissipation owing to its reversibility, it effectively mediates the transfer of energy from field to plasma particles. Nevertheless, by restricting the analysis on the wavenumbers excited through the forcing, Landau damping successfully mimics macroscopic dissipation. Indeed, it continuously drains energy from fluctuations, counter-

balancing the stochastic injection and ultimately driving the plasma toward a steady-state equilibrium governed by an FDR.

Following the analysis in Kanekar et al. [12], we define a kinetic analogue of the Langevin equation by considering the system

∂g∂t+v∂g∂x+vF0∂φ∂x=χF0(1)\frac{\partial g}{\partial t}+v\frac{\partial g}{\partial x}+vF_0\frac{\partial\varphi}{\partial x}=\chi F_0 \tag*{(1)}
φ=α∫dv g.(2)\varphi=\alpha\int dv\,g. \tag*{(2)}

Both equations are presented in dimensionless form. Specifically, the spatial coordinate xx is normalized by an arbitrary length LL such that it ranges within the interval [0,Lbox][0,L_{\mathrm{box}}], the velocity vv is normalized by the thermal speed vth=kBT/mv_{\mathrm{th}}=\sqrt{k_B T/m} (where kBk_B is the Boltzmann constant, TT is the plasma temperature, and mm is the mass of the species under consideration) and time is normalized by L/vthL/v_{\mathrm{th}}.

Eq. (1) is a linearized Vlasov equation —where we introduced a noise source term whose properties are described below— describing the evolution of the plasma species in the kinetic regime. In particular, g(x,v,t)g(x,v,t) is the perturbation to the distribution function f(x,v,t)f(x,v,t) with respect to the Maxwellian equilibrium F0(v)F_0(v): f(x,v,t)=F0(v)+g(x,v,t)f(x,v,t)=F_0(v)+g(x,v,t) with g/F0≪1g/F_0\ll1 and F0(v)=n0e−v2/2/2πF_0(v)=n_0e^{-v^2/2}/\sqrt{2\pi} (n0=1n_0=1 in normalized units). The stochastic forcing reads as χ(t,x)=a0q(x)s(t)\chi(t,x)=a_0q(x)s(t) in such a way that the power injected into the system scales as ∝a02\propto a_0^2. The spatial part of the forcing is defined as:

q(x)=∑n=1Nm2Nmsin⁡(nk0x+ϕn),(3)q(x)=\sum_{n=1}^{N_m}\sqrt{\frac{2}{N_m}}\sin(nk_0x+\phi_n), \tag*{(3)}

where NmN_m is the number of modes, k0=2π/Lboxk_0=2\pi/L_{\mathrm{box}} is the fundamental harmonic, ϕn\phi_n are random phases. It holds that ∫dx q(x)=0\int dx\,q(x)=0 and ∫dx q2(x)=1\int dx\,q^2(x)=1. The temporal part s(t)s(t) satisfies

⟨s(t)s(t′)⟩=e−λ∣t−t′∣(4)\langle s(t)s(t')\rangle=e^{-\lambda|t-t'|} \tag*{(4)}

where ⟨⋅⟩\langle\cdot\rangle represents an ensemble average. The function s(t)s(t) encodes the correlated noise driving the turbulence in the kinetic system[13]. The parameter λ=1/τc\lambda=1/\tau_c, being τc\tau_c the correlation time, dictates the nature of the forcing term. Specifically, its ratio to the characteristic timescales of the problem (such as the wave frequency or the Landau damping rate) establishes whether the system responds to a highly-correlated forcing or operates in a low-correlation regime. The phenomenology associated with a white-noise source as in Ref.[12] is consistently recovered in the limit λ→∞\lambda\to\infty.

Eq. (2) links the perturbation in the plasma distribution function with the electrostatic potential φ\varphi. As discussed in Ref.[12], different physical scenarios can be reproduced by tuning the parameter α\alpha. For instance, with α>0\alpha>0, the system (1) can represent a Vlasov-Poisson plasma composed by kinetic electrons and motionless ions, describing electronic Langmuir waves. In this case, when Eq. (1b) is transformed in Fourier space, the parameter α\alpha is expressed as α=1/(k2λD2)\alpha= 1/(k^2\lambda_D^2), being λD\lambda_D the Debye length and kk the wavenumber. We remark that the normalized electric potential φ\varphi must be defined in terms of the physical one ϕ\phi as φ=qϕ/kBT\varphi= q\phi/k_{\mathrm{B}}T, where qq is the signed charge of the kinetic species. Therefore, when electrons are considered, the signs of the terms in Eq. (1a) and Eq. (1b) are consistent with a positive sign for α\alpha. The case α>0\alpha> 0 encompasses also ion-acoustic waves. Here gg represents the perturbation to the ion distribution functions, while electrons have an instantaneous Boltzmann response, i.e. ne=n0eφn_e = n_0 e^\varphi where nen_e and n0n_0 are the electron and background density, respectively. In this case, α=ZTe/Ti\alpha= ZT_e/T_i with ZZ the ions atomic number, and Eq. (1b) can be either derived from the statement of quasi-neutrality or from the Poisson equation in the limit ZTe/Ti≫k2λD2ZT_e/T_i \gg k^2\lambda_D^2. Conversely, setting α<0\alpha< 0 yields a system that represents slow-MHD compressive modes in the high-β\beta (see, [12] for further details). The allowed interval of variation of the parameter α\alpha is (−1,+∞)(-1,+\infty). The singular case α=−1\alpha= -1 represents an undamped and unforced system, for which the analogy with Langevin phenomenology is no longer feasible. On the other hand, when the α<−1\alpha< -1 case is considered, the solutions of the dispersion relation correspond to instabilities (not shown here). In this regime, the exponential growth of the perturbations prevents the system from reaching the steady-state equilibrium required to establish an FDR; therefore, this scenario lies outside the scope of the present work.

In the following, we summarize the derivation of the FDR for the system (1), by employing the approach presented in [12] and extending it to the case of a correlated noise forcing. Applying a Fourier transform[18] in time and space to Eq. (1a) provides:

gk,ω=−vF0v−ω/kφk,ω−iF0k(v−ω/k)χk,ω.(5)g_{k,\omega}=-\frac{vF_0}{v-\omega/k}\varphi_{k,\omega}-\frac{iF_0}{k(v-\omega/k)}\chi_{k,\omega}. \tag*{(5)}

which, inserted back into Eq. (1b) gives a closed expression for the electric potential, namely

φk,ω=−iχk,ωkZ(ζ)Dα(ζ).(6)\varphi_{k,\omega}=-\frac{i\chi_{k,\omega}}{k}\frac{Z(\zeta)}{D_\alpha(\zeta)}. \tag*{(6)}

Here, we have introduced the plasma dispersion function and the dielectric function[19]:

Z(ζ)=∫dv F0v−ζ(7)Z(\zeta)=\int dv\,\frac{F_0}{v-\zeta} \tag*{(7)}
Dα(ζ)=1α+∫dv vF0v−ζ=1+1α+ζZ(ζ)(8)D_\alpha(\zeta)=\frac{1}{\alpha}+\int dv\,\frac{vF_0}{v-\zeta}=1+\frac{1}{\alpha}+\zeta Z(\zeta) \tag*{(8)}

where ζ=ω/k\zeta= \omega/k is the wave phase velocity. The dispersion relation of the collective plasma modes is provided by the pairs (ω,k)∈C×R(\omega,k)\in\mathbb{C}\times\mathbb{R} satisfying Dα(ω/k)=0D_\alpha(\omega/k)=0. In order to evaluate the steady-state spectrum of the electric perturbations, we back-transform Eq. (6) to the time domain, compute its square modulus, and take the ensemble average:

⟨∣φk∣2⟩=1(2πk)2∫dω∫dω′ ⟨χk,ωχk,ω′∗⟩×Z(ω/k)Dα(ω/k)Z∗(ω′/k)Dα∗(ω′/k)e−i(ω−ω′)t.(9)\langle|\varphi_k|^2\rangle=\frac{1}{(2\pi k)^2}\int d\omega\int d\omega'\,\langle\chi_{k,\omega}\chi^*_{k,\omega'}\rangle\times\frac{Z(\omega/k)}{D_\alpha(\omega/k)}\frac{Z^*(\omega'/k)}{D^*_\alpha(\omega'/k)}e^{-i(\omega-\omega')t}. \tag*{(9)}

The noise auto-correlation ⟨χk,ωχk,ω′∗⟩\langle\chi_{k,\omega}\chi^*_{k,\omega'}\rangle can be evaluated via Fourier analysis and exploiting the forcing temporal property in Eq. (4), resulting in

⟨χk,ωχk,ω′∗⟩=2πa02∣qk∣2δ(ω−ω′)2λλ2+ω2.(10)\langle\chi_{k,\omega}\chi^*_{k,\omega'}\rangle=2\pi a_0^2|q_k|^2\delta(\omega-\omega')\frac{2\lambda}{\lambda^2+\omega^2}. \tag*{(10)}

Inserting this expression in Eq. (9) provides the FDR for the correlated noise source case:

⟨∣φk∣2⟩=a02∣qk∣2πk∫−∞∞dζ λλ2+k2ζ2∣Z(ζ)Dα(ζ)∣2.(11)\langle|\varphi_k|^2\rangle=\frac{a_0^2|q_k|^2}{\pi k}\int_{-\infty}^{\infty}d\zeta\,\frac{\lambda}{\lambda^2+k^2\zeta^2}\left|\frac{Z(\zeta)}{D_\alpha(\zeta)}\right|^2. \tag*{(11)}

We remark that, by taking the limit λ→∞\lambda\to\infty of Eq. (11), we recover the FDR relation derived in [12] for the white-noise case, that reads as:

⟨∣φk∣2⟩=ϵk2πk∫−∞∞dζ ∣Z(ζ)Dα(ζ)∣2.(12)\langle|\varphi_k|^2\rangle=\frac{\epsilon_k}{2\pi k}\int_{-\infty}^{\infty}d\zeta\,\left|\frac{Z(\zeta)}{D_\alpha(\zeta)}\right|^2. \tag*{(12)}

[12] defines the noise power spectrum ⟨χ(t)χ(t′)⟩=ϵkδ(t−t′)\langle\chi(t)\chi(t')\rangle=\epsilon_k\delta(t-t'), whereas in our case we assume ⟨χ(t)χ(t′)⟩=a02∣qk∣2e−λ∣t−t′∣\langle\chi(t)\chi(t')\rangle=a_0^2|q_k|^2e^{-\lambda|t-t'|}. In order to find the equivalence between these two different expressions, we calculate, for the two cases, the total energy injected into the system for long times. Equating them, we obtain:

∫−∞∞dt ϵkδ(t)=∫−∞∞dt a02∣qk∣2e−λ∣t∣,(13)\int_{-\infty}^{\infty}dt\,\epsilon_k\delta(t)=\int_{-\infty}^{\infty}dt\,a_0^2|q_k|^2e^{-\lambda|t|}, \tag*{(13)}

which implies ϵk=2a02∣qk∣2λ−1\epsilon_k=2a_0^2|q_k|^2\lambda^{-1}. The same result can be achieved by noting that δ(t)=lim⁡λ→∞λ2e−λ∣t∣\delta(t)=\lim_{\lambda\to\infty}\frac{\lambda}{2}e^{-\lambda|t|}, where the limit is understood in the distributional sense.

Asymptotic scaling

It is worth calculating asymptotic scalings of Eq. (11) as these provide interesting insights of the underlying physics in some limit cases. In particular, we will analyze the asymptotic behavior of the spectrum in the limits α→−1\alpha\to-1 (vanishing phase velocity and weakly damped compressive waves) and α→∞\alpha\to\infty (short wavelength weakly damped fast Langmuir waves). As we will show, our findings in the α→−1\alpha\to-1 and α→∞\alpha\to\infty cases reproduce the white-noise phenomenology in the limit of vanishing correlation time, namely λ→∞\lambda\to\infty. Furthermore, we will also consider the short-wavelength limit k→∞k\to\infty, regardless of the choice of α\alpha. This limit is equivalent to the highly-correlated regime λ→0\lambda\to0 since the relevant ratio characterizing this scenario is λ/k\lambda/k.

Weakly-damped non-oscillatory limit (α→−1\alpha\to-1)

When the α→−1\alpha\to-1 limit is considered, the following approximations hold

Z(ζ)≈iπ2,Dα(ζ)≈iπ2(ζ+iγLk),(14)Z(\zeta) \approx i\sqrt{\frac{\pi}{2}}, \qquad D_{\alpha}(\zeta) \approx i\sqrt{\frac{\pi}{2}}\left(\zeta+i\frac{\gamma_{\mathrm{L}}}{k}\right), \tag*{(14)}

where the Landau damping rate is given by γL=2/π(1+α)k\gamma_{\mathrm{L}}=\sqrt{2/\pi}(1+\alpha)k. The dispersion relation, calculated from Dα(ω/k)=0D_{\alpha}(\omega/k)=0, yields a purely imaginary solution, ω=−iγL\omega=-i\gamma_{\mathrm{L}}, describing a weakly-damped, non-oscillatory solution. In this regime, the FDR takes the approximate form

⟨∣φk∣2⟩≈a02∣qk∣2πk3∫−∞∞dζ 1ζ2+λ2k21ζ2+γL2k2,(15)\left\langle\lvert\varphi_k\rvert^2\right\rangle\approx\frac{a_0^2\lvert q_k\rvert^2}{\pi k^3}\int_{-\infty}^{\infty}\mathrm{d}\zeta\,\frac{1}{\zeta^2+\frac{\lambda^2}{k^2}}\frac{1}{\zeta^2+\frac{\gamma_{\mathrm{L}}^2}{k^2}}, \tag*{(15)}

which, through the residue theorem, reads as:

⟨∣φk∣2⟩≈a02∣qk∣2γL(γL+λ).(16)\left\langle\lvert\varphi_k\rvert^2\right\rangle\approx\frac{a_0^2\lvert q_k\rvert^2}{\gamma_{\mathrm{L}}(\gamma_{\mathrm{L}}+\lambda)}. \tag*{(16)}

Depending on the relative magnitude of the two characteristic rates governing this scenario, namely the damping rate γL\gamma_{\mathrm{L}} and the correlation frequency λ\lambda, we can identify two distinct behaviors in the asymptotic spectrum. First, when λ≪γL\lambda\ll\gamma_{\mathrm{L}}, we observe that ⟨∣φk∣2⟩∝k−2\left\langle\lvert\varphi_k\rvert^2\right\rangle\propto k^{-2}. In this regime, the damping time is much shorter than the correlation time; this reproduces a highly correlated system where the Landau damping effectively acts on a noise that varies slowly compared to the decay timescale. In the opposite limit, λ≫γL\lambda\gg\gamma_{\mathrm{L}}, the noise is practically uncorrelated on the damping timescale, yielding a spectral scaling ⟨∣φk∣2⟩∝k−1\left\langle\lvert\varphi_k\rvert^2\right\rangle\propto k^{-1}, reproducing the white-noise case[12].

Weakly-damped fast-wave limit (α→∞\alpha\to\infty)

In this limit the following approximation holds:

Z(ζ)≈−1ζ−1ζ3+iπ2e−ζ2/2,Dα(x)≈1α−1ζ2+iζπ2e−ζ2/2.(17)\begin{aligned} Z(\zeta) &\approx-\frac{1}{\zeta}-\frac{1}{\zeta^3}+i\sqrt{\frac{\pi}{2}}e^{-\zeta^2/2},\\ D_{\alpha}(x) &\approx\frac{1}{\alpha}-\frac{1}{\zeta^2}+i\zeta\sqrt{\frac{\pi}{2}}e^{-\zeta^2/2}. \tag*{(17)} \end{aligned}

The collective plasma modes obey the dispersion relation

ω=±αk−iγL,γL=π8α2e−α/2k,(18)\omega=\pm\sqrt{\alpha}k-i\gamma_{\mathrm{L}}, \qquad\gamma_{\mathrm{L}}=\sqrt{\frac{\pi}{8}}\alpha^2e^{-\alpha/2}k, \tag*{(18)}

describing Langmuir waves with phase velocity much greater than the thermal speed. In this regime, Landau damping is exponentially small, allowing these modes to persist over long timescales within the plasma. Under these conditions the plasma response is well approximated by the sum of two narrow Lorentzians:

∣Z(ζ)Dα(ζ)∣2≈α24(1(ζ−α)2+γL2k2+1(ζ+α)2+γL2k2).(19)\left\lvert\frac{Z(\zeta)}{D_{\alpha}(\zeta)}\right\rvert^2\approx\frac{\alpha^2}{4}\left(\frac{1}{(\zeta-\sqrt{\alpha})^2+\frac{\gamma_{\mathrm{L}}^2}{k^2}}+\frac{1}{(\zeta+\sqrt{\alpha})^2+\frac{\gamma_{\mathrm{L}}^2}{k^2}}\right). \tag*{(19)}

By inserting last expression into the integral appearing in (11) and applying the residue theorem, we obtain two separate contributions, one arising from the plasma poles IP\mathcal{I}_{P} and the other related to the noise IN\mathcal{I}_{N}, reading

IP=λπα2k2γL(αk2+λ2).(20)\mathcal{I}_{P}=\frac{\lambda\pi\alpha^2 k}{2\gamma_{\mathrm{L}}(\alpha k^2+\lambda^2)}. \tag*{(20)}
IN=πα2k(αk2−λ2)2(αk2+λ2)2,(21)\mathcal{I}_{N}=\frac{\pi\alpha^2 k(\alpha k^2-\lambda^2)}{2(\alpha k^2+\lambda^2)^2}, \tag*{(21)}

whose ratio is:

IPIN=λ(αk2+λ2)γL(αk2−λ2).(22)\frac{\mathcal{I}_{P}}{\mathcal{I}_{N}}=\frac{\lambda(\alpha k^2+\lambda^2)}{\gamma_{\mathrm{L}}(\alpha k^2-\lambda^2)}. \tag*{(22)}

The plasma term is hence exponentially large with respect to the noise one (see (18)), therefore the latter can be safely neglected, yielding this simple expression for the asymptotic spectrum in the large-α\alpha limit:

⟨∣φk∣2⟩≈λa02∣qk∣2α22γL(αk2+λ2).(23)\left\langle\lvert\varphi_k\rvert^2\right\rangle\approx\frac{\lambda a_0^2\lvert q_k\rvert^2\alpha^2}{2\gamma_{\mathrm{L}}(\alpha k^2+\lambda^2)}. \tag*{(23)}

Two separate scenarios emerge depending on the relative magnitude of the correlation frequency λ\lambda and the wave frequency ωL=αk\omega_{\mathrm{L}}=\sqrt{\alpha}k. In the highly-correlated case, i.e. λ≪ωL\lambda\ll\omega_{\mathrm{L}}, or in other words, correlation time much larger than the Langmuir wave period, we expect a spectral scaling ⟨∣φk∣2⟩∝k−3\left\langle\lvert\varphi_k\rvert^2\right\rangle\propto k^{-3}. Conversely, in the white-noise limit λ≫ωL\lambda\gg\omega_{\mathrm{L}}, we recover the standard scaling outlined in[12], namely ⟨∣φk∣2⟩∝k−1\left\langle\lvert\varphi_k\rvert^2\right\rangle\propto k^{-1}.

Short-wavelength limit (k→∞k \to\infty)

As previously stated, the short-wavelength limit, performed without further assumptions on α\alpha, is of great importance in our analysis. The electrostatic kinetic framework developed here is inherently best suited to describe plasma turbulence at the smallest spatial scales, corresponding to the sub-ion and sub-electron regimes (i.e., beyond the spectral break) of solar wind turbulence. At such small scales the fluctuation spectrum becomes universally governed by the static (zero-frequency) susceptibility of the plasma. Given that the relevant dynamical parameter is the ratio λ/k\lambda/k, the limit k→∞k \to\infty intrinsically mimics the λ→0\lambda\to0 limit. Consequently, at the smallest spatial scales, the plasma always perceives the forcing as a highly-correlated noise, regardless of the actual finite value of the correlation rate λ\lambda.

The mathematical treatment of this limit is straightforward. By rewriting the FDR as

⟨∣φk∣2⟩=a02∣qk∣2πk3∫−∞∞dζ λλ2k2+ζ2∣Z(ζ)Dα(ζ)∣2,(24)\left\langle\lvert\varphi_k\rvert^2\right\rangle=\frac{a_0^2\lvert q_k\rvert^2}{\pi k^3}\int_{-\infty}^{\infty}\mathrm{d}\zeta\,\frac{\lambda}{\frac{\lambda^2}{k^2}+\zeta^2}\left\lvert\frac{Z(\zeta)}{D_{\alpha}(\zeta)}\right\rvert^2, \tag*{(24)}

we can perform a distributional limit from the Lorentzian to a Dirac delta. The approximate form for the FDR is readily obtained

⟨∣φk∣2⟩≈a02∣qk∣2k2∫−∞∞dζ δ(ζ)∣Z(ζ)Dα(ζ)∣2=πa02∣qk∣2α22k2(α+1)2.(25)\left\langle|\varphi_k|^2 \right\rangle\approx\frac{a_0^2|q_k|^2}{k^2}\int_{-\infty}^{\infty}d\zeta\,\delta(\zeta)\left|\frac{Z(\zeta)}{D_\alpha(\zeta)}\right|^2=\frac{\pi a_0^2|q_k|^2\alpha^2}{2k^2(\alpha+1)^2}. \tag*{(25)}

This mathematical outcome carries a precise physical meaning. At asymptotically small scales, the phase velocity of the perturbations vanishes (ω/k→0\omega/k \to0), and the plasma loses its ability to sustain collective wave modes. Consequently, dynamical kinetic effects are entirely suppressed, and the system responds solely through static screening, encoded in the α\alpha-dependent pre-factor. This implies that the transition to the k→∞k \to\infty regime effectively acts as a kinetic spectral break. At scales smaller than such a break, wave-particle interactions cease to actively shape the fluctuation spectrum. Since the electrostatic potential scales as ⟨∣φk∣2⟩∝k−2∣qk∣2\langle|\varphi_k|^2\rangle\propto k^{-2}|q_k|^2, the corresponding electric field fluctuations, obtained via ⟨∣Ek∣2⟩=k2⟨∣φk∣2⟩\langle|E_k|^2\rangle=k^2\langle|\varphi_k|^2\rangle, passively mirror the spatial injection spectrum, yielding ⟨∣Ek∣2⟩∝∣qk∣2\langle|E_k|^2\rangle\propto|q_k|^2. At these extreme scales, the electric field response becomes entirely uncoupled from the temporal correlation parameter λ\lambda.

Hermite decomposition

To elucidate the features of the turbulent cascade in velocity space, it is highly instructive to perform a Hermite decomposition of the perturbed distribution function g(x,v,t)g(x,v,t). We make use of the probabilist’s Hermite polynomials, defined as Hen(v)=(−1)nev2/2dndvne−v2/2\mathrm{He}_n(v)=(-1)^n e^{v^2/2}\frac{d^n}{dv^n}e^{-v^2/2}. The projection onto the Hermite basis and its corresponding inversion formula are given by

gm=∫−∞∞dv Hem(v)m!g(v)g_m=\int_{-\infty}^{\infty}dv\,\frac{\mathrm{He}_m(v)}{\sqrt{m!}}g(v)
g(v)=∑m=0∞Hem(v)F0(v)m!gm.(26)g(v)=\sum_{m=0}^{\infty}\frac{\mathrm{He}_m(v)F_0(v)}{\sqrt{m!}}g_m. \tag*{(26)}

By applying the Hermite transform to the Eq. (1a) and following the same steps that led to Eq. (11), we can derive the steady-state Hermite spectrum:

Cm,k≡⟨∣gm,k∣2⟩=a02∣qk∣2πkm!(α+1α)2∫−∞∞dζ λλ2+k2ζ2∣Z(m)(ζ)Dα(ζ)∣2.(27)C_{m,k}\equiv\left\langle|g_{m,k}|^2\right\rangle=\frac{a_0^2|q_k|^2}{\pi k m!}\left(\frac{\alpha+1}{\alpha}\right)^2\int_{-\infty}^{\infty}d\zeta\,\frac{\lambda}{\lambda^2+k^2\zeta^2}\left|\frac{Z^{(m)}(\zeta)}{D_\alpha(\zeta)}\right|^2. \tag*{(27)}

where we have defined

Z(m)≡dmZdζm=(−1)m∫−∞∞dv Hem(v)F0(v)v−ζ.(28)Z^{(m)}\equiv\frac{d^m Z}{d\zeta^m}=(-1)^m\int_{-\infty}^{\infty}dv\,\frac{\mathrm{He}_m(v)F_0(v)}{v-\zeta}. \tag*{(28)}

The asymptotic behavior of this spectrum for large mm is known to scale as Cm,k≈Ak(α,λ)/mC_{m,k}\approx A_k(\alpha,\lambda)/\sqrt{m} (see Ref.^{12} and references therein), where AkA_k is:

Ak(α,λ)=λa02∣qk∣22πk(α+1α)2∫−∞∞dζ e−ζ2/2(λ2+k2ζ2)∣Dα(ζ)∣2.(29)A_k(\alpha,\lambda)=\frac{\lambda a_0^2|q_k|^2}{\sqrt{2\pi}k}\left(\frac{\alpha+1}{\alpha}\right)^2\int_{-\infty}^{\infty}d\zeta\,\frac{e^{-\zeta^2/2}}{(\lambda^2+k^2\zeta^2)|D_\alpha(\zeta)|^2}. \tag*{(29)}

By implementing the same strategy proposed in Ref.^{12}, we evaluate the integral contained in the above formula:

Ak(α,λ)=a02∣qk∣2λk(α+1α)2(1Dα(iλ/k)−α1+α),(30)A_k(\alpha,\lambda)=\frac{a_0^2|q_k|^2}{\lambda k}\left(\frac{\alpha+1}{\alpha}\right)^2\left(\frac{1}{D_\alpha(i\lambda/k)}-\frac{\alpha}{1+\alpha}\right), \tag*{(30)}

where the plasma dispersion function evaluated on the imaginary axis is:

Dα(iλ/k)=1+1α−π2k2λkeλ2/(2k2)erfc⁡(λ2k).(31)D_\alpha(i\lambda/k)=1+\frac{1}{\alpha}-\sqrt{\frac{\pi}{2k^2}}\frac{\lambda}{k}e^{\lambda^2/(2k^2)}\operatorname{erfc}\left(\frac{\lambda}{\sqrt{2}k}\right). \tag*{(31)}

In order to demonstrate that we consistently recover the white-noise phenomenology in this limit, it is sufficient to recall that the complementary error function erfc⁡(y)\operatorname{erfc}(y) admits the following asymptotic expansion for large values of its argument:

erfc⁡(y)≈e−y2yπ(1−12y2).(32)\operatorname{erfc}(y)\approx\frac{e^{-y^2}}{y\sqrt{\pi}}\left(1-\frac{1}{2y^2}\right). \tag*{(32)}

By setting y=λ/(2k)y=\lambda/(\sqrt{2}k) and taking the limit λ→∞\lambda\to\infty, we obtain the asymptotic amplitude:

Ak(α,λ)≈2a02∣qk∣2λk(1+α).(33)A_k(\alpha,\lambda)\approx\frac{\sqrt{2}a_0^2|q_k|^2}{\lambda k}(1+\alpha). \tag*{(33)}

By recalling the formal equivalence for the injected power ϵk=2a02∣qk∣2/λ\epsilon_k=2a_0^2|q_k|^2/\lambda derived earlier, we immediately recognize that Eq. (33) yields Ak≈ϵk(1+α)/(2k)A_k\approx\epsilon_k(1+\alpha)/(\sqrt{2}k). This is precisely the cascade amplitude associated with a white-noise driven system.^{12} This confirms that our generalized model accurately recovers the classic white-noise phenomenology when the turbulent forcing decorrelates instantaneously.

It is important to remark that both the kk-space and Hermite steady-state spectra describe local, rather than global, equilibria, i.e. they accurately characterize the system at fixed kk and mm over a suitable timescale. In other words, energy is continuously injected in the system by the source term at m=0m=0 (the noise is Maxwellian in the velocity space) and transferred to higher Hermite modes via Landau damping, therefore equilibrium is reached at a given mm only after the energy cascade in Hermite space has had sufficient time to reach that particular mode.

Numerical simulations

In this section, we test the analytical predictions derived above against numerical simulations. We perform direct integrations of the equations system, Eqs. (1), by employing a spectral algorithm. We assume periodic boundary conditions on the spatial coordinate xx and evaluate spatial derivatives by means of a spectral approach. We set Lbox=4πL_{\mathrm{box}} = 4\pi and use a grid of Nx=256N_x = 256 evenly spaced points covering the spatial domain. The grid in the velocity space consists of Nv=2001N_v = 2001 points on the segment [−6,6][-6,6] (we recall that velocities are normalized to the thermal speed). We remark that no boundary conditions must be imposed on the velocity vv, for the fact that the latter enters the linearized system Eqs. (1) as a parameter, rather than a dynamical variable. In all simulations, we assume a forcing amplitude a0=10−3a_0 = 10^{-3} and a superposition of Nm=8N_m = 8 independent modes for the spatial part of the noise, while the phases ϕn\phi_n are randomly sampled for each run. Time integration is performed using a standard fourth-order Runge–Kutta scheme. The white-noise case results have been preliminarily reproduced in order to validate the performances of our numerical algorithm before applying it to the finite correlation time case.

We avoid introducing any numerical dissipation term as, for example, small collisional term or hyperviscosity for preventing eventual numerical instabilities. On one hand, this ensures that Landau damping is indeed the only energy conversion process at play in balancing the power supplied to the system by the random source, thus perfectly mimicking the analytical FDR derived above. On the other hand, this implies that other strategies are needed to prevent numerical artifacts. In this perspective, we set our maximum simulation time Tmax⁡T_{\max} and the velocity-space grid size Δv\Delta v in such a way that Tmax⁡<TrecT_{\max} < T_{\mathrm{rec}} is satisfied for all driven modes {k1,…,kNm}\{k_1,\ldots,k_{N_m}\}. Here, Trec=2π/(kΔv)T_{\mathrm{rec}} = 2\pi/(k\Delta v) is the free-streaming recurrence time[10], which marks the onset of artificial echoes due to the discreteness of the velocity grid. The time step Δt\Delta t is also properly adjusted for each case depending on the value of λ\lambda to ensure that the numerical solution appreciates the forcing decorrelation ∼1/λ\sim1/\lambda: Δt≪1/λ\Delta t \ll1/\lambda.

A final comment on the forcing term is in order. The correlated noise is generated through an Ornstein–Uhlenbeck process[20], which is inherently stochastic. For each case, the specific numerical realization of the noise is slightly affected by random fluctuations, and the convergence toward the expected analytical shape of the autocorrelation function is achieved only by ensemble averaging over an appropriate number of runs. As appreciated in Figure 1, five independent realizations of the numerical noise for a given set of the relevant parameters (λ\lambda, Tmax⁡T_{\max}, and dtdt) show slight discrepancies with respect to the theoretical behavior (red dashed line), while their average smoothly follows it (light blue points). Given this premise, it is clear that the numerical solution of Eqs. (1) will be affected by the same random fluctuations that characterize the source term. Consequently, all numerical verifications of the analytical FDRs reported in this section can only be achieved by averaging over a sufficient number of independent realizations. For each

Convergence of the numerically generated correlated noise to the theoretical curve

Figure 1. Convergence of the numerically generated correlated noise to the theoretical curve. Upper panel: independent realizations. Lower panel: ensemble average and theoretical curve.

realization, the forcing drives the growth of the electrostatic potential. In particular, the energy of the driven modes reaches a saturation regime at which we employ the FDR analysis, as detailed below. Note also that, for each realization, the invariant conservation is excellent: for example, the Boltzmann-Gibbs entropy is preserved up to 10−3%10^{-3}\%.

As a first testcase, we present the numerical verification of the FDR provided in Eq. (11). The procedure consists of: i) simulating Eqs. (1), ii) calculating, from the output of the numerical integration, the steady-state spectrum ⟨∣φk∣2⟩num\langle|\varphi_k|^2\rangle_{\mathrm{num}} where the average is computed on a time window of size Taver=250T_{\mathrm{aver}} = 250 in the saturation regime, and iii) ensemble-averaging the results over a certain number of trials NensN_{\mathrm{ens}}. The numerical verification of the FDR is then achieved if

⟨∣φk∣2⟩numa02∣qk∣2πk∫−∞∞dζλλ2+k2ζ2∣Z(ζ)Dα(ζ)∣2=1(34)\frac{\langle|\varphi_k|^2\rangle_{\mathrm{num}}}{\dfrac{a_0^2 |q_k|^2}{\pi k}\int_{-\infty}^{\infty} d\zeta\frac{\lambda}{\lambda^2+k^2\zeta^2}\left|\frac{Z(\zeta)}{D_\alpha(\zeta)}\right|^2}=1 \tag*{(34)}

holds for all driven kk. We remark again that, in each run, specific values for α\alpha and λ\lambda must be chosen and the denominator in the above equation is calculated separately from the numerical integration of Eqs. (1).

In Figure 2, we present the outputs of Nens=10N_{\mathrm{ens}} = 10 integrations of Eqs. (1) in which we set λ=10\lambda= 10, α=1\alpha= 1, and Δt=0.05\Delta t = 0.05. The plot clearly shows that the fluctuations, arising from the stochastic nature of the forcing, are symmetrically distributed around the expected theoretical behavior. On the other hand, by looking at the black crosses describing the mean obtained from the independent realizations, the mean deviation from the expected unity value is below 5% for the chosen ensemble size.

FDR: 10 realizations and average

Figure 2. Numerical verification of the FDR. The light blue dots indicate the 10 independent realizations, the black crosses represent the ensemble average.

Then, we present the numerical verification of the asymptotic scalings analytically derived in § 2 A. We start with the case α→−1\alpha\to-1, which we reproduce by setting α=−0.8\alpha=-0.8. Choosing this value for α\alpha provides a range of variation for the Landau damping rate between γLmin⁡=0.08\gamma_{L}^{\min}=0.08 and γLmax⁡=0.7\gamma_{L}^{\max}=0.7. The two regimes of high and low correlation are achieved by setting λ=0.01\lambda=0.01 and λ=20\lambda=20, respectively. In this limit, Landau damping vanishes linearly; however, for α=−0.8\alpha=-0.8, it still provides an effective mechanism for energy transfer towards the small scales. As displayed in Figure 3, the predicted theoretical behavior is well reproduced even when averaging over a small ensemble of runs Nens=5N_{\mathrm{ens}}=5.

Numerical verification of the asymptotic scaling in the $\alpha \to -1$ limit

Figure 3. Numerical verification of the asymptotic scaling in the α→−1\alpha\to-1 limit, simulated with α=−0.8\alpha=-0.8. The low correlation (LC) case, i.e. the regime λ≫γLmax⁡=0.7\lambda\gg\gamma_{L}^{\max}=0.7, is reproduced by setting λ=20\lambda=20. Conversely, the high correlation (HC) case, i.e. the regime λ≪γLmin⁡=0.08\lambda\ll\gamma_{L}^{\min}=0.08, is reproduced by setting λ=0.01\lambda=0.01.

Instead, in the α→∞\alpha\to\infty limit, the Landau damping rate vanishes exponentially. By setting α=10\alpha=10, the energy transfer across scales is very weak, and the integration of Eqs. (1) is more affected by numerical noise. Nevertheless, with Nens=20N_{\mathrm{ens}=20} realizations, a very good agreement with the theoretical predictions can still be achieved, as shown in Figure 4. Here, the cases of high and low correlation are given by λ=0.1≪ωLmin⁡≈1.6\lambda=0.1 \ll\omega_{L}^{\min}\approx1.6 and λ=150≫ωLmax⁡≈12.6\lambda=150 \gg\omega_{L}^{\max}\approx12.6, respectively.

Numerical verification of the asymptotic scaling in the $\alpha \to \infty$ limit

Figure 4. Numerical verification of the asymptotic scaling in the α→∞\alpha\to\infty limit, simulated with α=10\alpha=10. The low correlation (LC) case, i.e. the regime λ≫ωLmax⁡≈12.6\lambda\gg\omega_{L}^{\max}\approx12.6, is reproduced by setting λ=150\lambda=150. Conversely, the high correlation (HC) case, i.e. the regime λ≪ωLmin⁡≈1.6\lambda\ll\omega_{L}^{\min}\approx1.6, is reproduced by setting λ=0.1\lambda=0.1.

Finally, to properly investigate the short-wavelength limit (k→∞k \to\infty), we adopt a specifically designed numerical setup. First, we increase the number of driven modes to Nm=50N_m=50 and the spatial grid to Nx=512N_x=512 points to prevent aliasing at high wavenumbers. Resolving the fast dynamics associated with large kk values requires a high time resolution, and, to this purpose, we set Δt=0.001\Delta t=0.001, while fixing α=1\alpha=1 and λ=1\lambda=1. We also expand the physical domain to Lbox=40πL_{\mathrm{box}}=40\pi to ensure that only a specific subset of high-kk modes falls into the λ/k≪1\lambda/k \ll1 regime, allowing us to clearly observe where the asymptotic scaling ⟨∣φk∣2⟩∝k−2\langle|\varphi_k|^2\rangle\propto k^{-2} accurately captures their behavior. The results of this numerical validation are illustrated in Figure 5. The plot clearly demonstrates that the transition from the λ/k≫1\lambda/k \gg1 regime to the k−2k^{-2} spectral scaling occurs around the mode k20=20k0k_{20}=20k_0.

Numerical verification of the short-wavelength asymptotic scaling

Figure 5. Numerical verification of the short-wavelength asymptotic scaling (k→∞k \to\infty). By expanding the spatial domain to Lbox=40πL_{\mathrm{box}} = 40\pi and forcing Nm=50N_m = 50 modes, the transition from the λ/k≫1\lambda/k \gg1 regime to the short-scale limit (λ/k≪1\lambda/k \ll1) is captured within a single simulation. The spectrum smoothly transitions to the predicted k−2k^{-2} scaling around the mode k20k_{20}.

As a final step in this section, we provide a numerical validation of our analytical findings regarding the steady-state behavior of the Hermite spectrum Cm,kC_{m,k}. For this particular run, the system parameters are α=1\alpha=1 and λ=1\lambda=1. The numerical solution of Eqs. (1) was projected onto the Hermite basis up to mmax⁡=500m_{\max}=500 to ensure an adequate resolution of the kinetic dynamics. The corresponding Hermite spectrum for all the wavenumbers included in the forcing term is illustrated in Figure 6. The plot reveals a clear and robust agreement between the numerical integration (solid lines) and the theoretically predicted Ak/mA_k/\sqrt{m} asymptotic scaling (dashed lines) in terms of both the spectral slope and amplitude, the latter being calculated from Eq. (30), thus effectively demonstrating the capability of the model to capture fine-scale features in velocity space.

Numerical verification of the Hermite steady-state spectrum asymptotic scaling

Figure 6. Numerical verification of the Hermite steady-state spectrum asymptotic scaling for the all the driven wavenumbers. The numerical data align with the theoretically predicted Ak/mA_k/\sqrt{m} scaling (dashed lines).

Power-Law Distribution and Link with In-Situ Observations

A natural extension of the analysis provided in the previous sections is to treat the parameter λ\lambda as a stochastic variable itself. Indeed, as shown in [13, 14] by employing Cluster and MMS data, (i) plasma turbulence in near-Earth environments can be appropriately described by a Langevin system where the forcing is a correlated noise, and (ii) the correlation times are distributed according to a power law distribution:

dP(λ)=Cλ−μdλ(35)dP(\lambda) = C\lambda^{-\mu}d\lambda \tag*{(35)}

where λ\lambda lies in the interval [λmin⁡,λmax⁡][\lambda_{\min},\lambda_{\max}]. The normalization constant CC is such that ∫λmin⁡λmax⁡Cλ−μdλ=1\int_{\lambda_{\min}}^{\lambda_{\max}} C\lambda^{-\mu}d\lambda= 1,

C=1−μλmax⁡1−μ−λmin⁡1−μ.C = \frac{1-\mu}{\lambda_{\max}^{1-\mu}-\lambda_{\min}^{1-\mu}}.

Assuming a superposition of correlation rates to characterize the turbulence source is a physically well-grounded assumption. In solar wind turbulence, for example, the energy injection from fluid scales to kinetic ones can be understood as the result of multiple processes. Each of these occurs on a characteristic timescale, originating either from the chaotic superposition of convective motions on the solar surface or from the subsequent flow evolution during the propagation. Therefore, the spectral content of solar wind turbulence can be analyzed as the superposition of multiple contributions. By employing model-based fitting procedures, such as those introduced in [21], the most likely distribution of correlation times that reproduces the observed spectra can, in principle, be extracted from the data. Consequently, the observed spectral signatures can be interpreted as the collective response of the plasma to this multiscale stochastic forcing.

Deriving an appropriate version of the FDR for this case is immediate: the only modification with respect to the calculations presented in § 2 occurs when evaluating the noise autocorrelation ⟨χk,ωχk,ω′∗⟩\langle\chi_{k,\omega}\chi^{*}_{k,\omega'}\rangle in Eq. (10). Now λ\lambda is a stochastic variable and must be therefore averaged over its probability distribution to obtain a deterministic prediction:

⟨χk,ωχk,ω′∗⟩=4πCa02∣qk∣2δ(ω−ω′)∫λmin⁡λmax⁡dλ λ1−μλ2+ω2,(36)\left\langle\chi_{k,\omega}\chi^{*}_{k,\omega'}\right\rangle= 4\pi C a_0^2\lvert q_k\rvert^2\delta(\omega-\omega')\int_{\lambda_{\min}}^{\lambda_{\max}}d\lambda\,\frac{\lambda^{1-\mu}}{\lambda^2+\omega^2}, \tag*{(36)}

which, plugged into the electric potential spectrum, gives the kinetic FDR for a power-law distributed correlation time of the noise source:

⟨∣φk∣2⟩=Ca02∣qk∣2πk×∫λmin⁡λmax⁡dλ λ1−μ∫−∞∞dζ 1λ2+k2ζ2∣Z(ζ)Dα(ζ)∣2(37)\langle\lvert\varphi_k\rvert^2\rangle= \frac{C a_0^2\lvert q_k\rvert^2}{\pi k}\times\int_{\lambda_{\min}}^{\lambda_{\max}}d\lambda\,\lambda^{1-\mu}\int_{-\infty}^{\infty}d\zeta\,\frac{1}{\lambda^2+k^2\zeta^2}\left\lvert\frac{Z(\zeta)}{D_\alpha(\zeta)}\right\rvert^2 \tag*{(37)}

The asymptotic spectral scalings derived in § 2 A in the limits α→−1\alpha\to-1 and α→∞\alpha\to\infty given by Eq. (16) and Eq. (23) must be averaged over the power-law distribution, given their explicit dependence on λ\lambda (the asymptotic spectrum in the short-wavelength case is independent from λ\lambda, therefore the average and the single-λ\lambda FDR coincide). When the α→−1\alpha\to-1 limit is considered, we obtain:

⟨∣φk∣2⟩=Ca02∣qk∣2γL∫λmin⁡λmax⁡dλ λ−μγL+λ(38)\langle\lvert\varphi_k\rvert^2\rangle= \frac{C a_0^2\lvert q_k\rvert^2}{\gamma_L}\int_{\lambda_{\min}}^{\lambda_{\max}}d\lambda\,\frac{\lambda^{-\mu}}{\gamma_L+\lambda} \tag*{(38)}

and an explicit solution for the integral can be found in terms of the Gaussian hypergeometric function 2F1(a,b;c;z)_2F_1(a,b;c;z), giving

⟨∣φk∣2⟩=Ca02∣qk∣2γL2(1−μ)×[λ1−μ 2F1(1,1−μ;2−μ;−λγL)]λmin⁡λmax⁡.(39)\langle\lvert\varphi_k\rvert^2\rangle= \frac{C a_0^2\lvert q_k\rvert^2}{\gamma_L^2(1-\mu)}\times\left[\lambda^{1-\mu}\,{}_2F_1\left(1,1-\mu;2-\mu;-\frac{\lambda}{\gamma_L}\right)\right]_{\lambda_{\min}}^{\lambda_{\max}}. \tag*{(39)}

The above expression acquires a simplified form by assuming, as suggested in Ref.13, μ=3/2\mu= 3/2. In this specific case we obtain

⟨∣φk∣2⟩=2Ca02∣qk∣2γL2×[λmax⁡−λmin⁡λmin⁡λmax⁡+1γLarctan⁡(λmin⁡γL−λmax⁡γL1+λmax⁡λmin⁡γL)].(40)\left\langle|\varphi_k|^2 \right\rangle= \frac{2C a_0^2 |q_k|^2}{\gamma_L^2} \times\left[\frac{\sqrt{\lambda_{\max}}-\sqrt{\lambda_{\min}}}{\sqrt{\lambda_{\min}\lambda_{\max}}}+\frac{1}{\sqrt{\gamma_L}}\arctan\left(\frac{\sqrt{\frac{\lambda_{\min}}{\gamma_L}}-\sqrt{\frac{\lambda_{\max}}{\gamma_L}}}{1+\sqrt{\frac{\lambda_{\max}\lambda_{\min}}{\gamma_L}}}\right)\right]. \tag*{(40)}

In the limit α→∞\alpha\to\infty, the asymptotic spectrum averaged over the power-law distribution is:

⟨∣φk∣2⟩=Ca02∣qk∣2α22γL∫λmin⁡λmax⁡dλ λ1−μαk2+λ2.(41)\left\langle|\varphi_k|^2 \right\rangle= \frac{C a_0^2 |q_k|^2 \alpha^2}{2\gamma_L}\int_{\lambda_{\min}}^{\lambda_{\max}}d\lambda\,\frac{\lambda^{1-\mu}}{\alpha k^2+\lambda^2}. \tag*{(41)}

Also in this case, the integral in the above formula can be expressed in terms of the Gaussian hypergeometric function:

⟨∣φk∣2⟩=Ca02∣qk∣2α2k2γL(2−μ)[λ2−μ 2F1(1,1−μ2;2−μ2;−λ2αk2)]λmin⁡λmax⁡.(42)\left\langle|\varphi_k|^2 \right\rangle= \frac{C a_0^2 |q_k|^2\alpha}{2k^2\gamma_L(2-\mu)}\left[\lambda^{2-\mu}\,{}_2F_1\left(1,1-\frac{\mu}{2};2-\frac{\mu}{2};-\frac{\lambda^2}{\alpha k^2}\right)\right]_{\lambda_{\min}}^{\lambda_{\max}}. \tag*{(42)}

It is interesting to point out that, by setting again μ=3/2\mu= 3/2, a remarkably simple formula can be calculated:

⟨∣φk∣2⟩=Ca02∣qk∣2α222γLωL3/2×[arctanh⁡(2λωLλ+ωL)+arctan⁡(2λωLωL−λ)]λmin⁡λmax⁡.(43)\left\langle|\varphi_k|^2 \right\rangle= \frac{C a_0^2|q_k|^2\alpha^2}{2\sqrt{2}\gamma_L\omega_L^{3/2}}\times\left[\operatorname{arctanh}\left(\frac{\sqrt{2\lambda\omega_L}}{\lambda+\omega_L}\right)+\arctan\left(\frac{\sqrt{2\lambda\omega_L}}{\omega_L-\lambda}\right)\right]_{\lambda_{\min}}^{\lambda_{\max}}. \tag*{(43)}

where we recall that ωL=αk\omega_L = \sqrt{\alpha}k.

At this stage, it is worth noting the formal and physical analogy between the λ\lambda-averaged spectra obtained here, specifically in Eqs. (37), (38) and (41), and the mathematical framework described in Ref.21 in Eqs. (10) and (11) of his work. By evaluating integrals of the form ∫λmin⁡λmax⁡dλω2+λ2λβ\int_{\lambda_{\min}}^{\lambda_{\max}}\frac{d\lambda}{\omega^2+\lambda^2}\lambda^\beta, it is demonstrated that a continuous superposition of single exponential relaxation processes precisely generates an intermediate power-law spectral region. In a strikingly similar fashion, our kinetic model naturally produces the fundamental response spectra of the plasma (e.g., Lorentzian-like modes or highly damped transients). The superposition of these basis spectra, weighted by the power-law distribution of the stochastic correlation times λ\lambda, is precisely the mechanism that intrinsically generates these 1/fα1/f^\alpha macroscopic noise signatures.

The closed-form expressions derived above, such as Eqs. (40) and (43), can be employed for assessing whether

in-situ observations match with the derived theoretical expressions, thus giving the possibility to interpret space plasmas within the developed FDR framework. Indeed, analytical solutions can serve as theoretical functions to fit in-situ measurements at high frequency, bypassing the standard practice of piecewise linear fitting on log-log scales. Assuming the validity of the Taylor hypothesis to map the observed frequencies to spatial wavenumbers, a model-based fitting procedure, as performed in Refs.13,14, allows to extract the forcing parameter such as the exponent μ\mu and the memory cutoffs λmin⁡\lambda_{\min} and λmax⁡\lambda_{\max} from in-situ measurements.

Before transitioning to the observable quantities, it is however important to point out that if a flat spatial spectrum is assumed for the stochastic forcing, i.e., ∣qk∣2=const⁡|q_k|^2 = \operatorname{const}, the model predicts non-physical features for the electric field fluctuations, whose spectrum is given by ⟨∣Ek∣2⟩=k2⟨∣φk∣2⟩\left\langle|E_k|^2 \right\rangle= k^2\left\langle|\varphi_k|^2 \right\rangle. Specifically, multiplying the asymptotic potential scalings by k2k^2 would lead in some cases to flat or even growing electric field spectra at sub-ion scales, heavily contradicting standard spacecraft observations in the solar wind. This apparent shortcoming arises because a flat ∣qk∣2|q_k|^2 implies that turbulent energy is artificially injected with equal intensity at all spatial scales, thus bypassing the inertial cascade process. A natural solution to this issue is to assume that the power injection from the larger scales, modeled by the correlated noise, is itself the result of an ongoing turbulent cascade. Therefore, it is physically sound to characterize the spatial profile of the forcing with a spectral scaling of the form ∣qk∣2∝k−δ|q_k|^2 \propto k^{-\delta}, where δ\delta represents the spectral index of the incoming inertial cascade. Under this assumption, the collective plasma response described by Vlasov theory acts effectively as a low-pass filter over the background turbulence. Table 1 summarizes the asymptotic spectral scalings of the electric field, derived from the single-λ\lambda FDR (10) in Sec. II A by assuming a spatially correlated noise source with spectral index δ\delta.

Plasma RegimeCorrelation RegimeScaling
α→−1\alpha\to-1High correlation (λ≪γL\lambda\ll\gamma_L)k−δk^{-\delta}
Low correlation (λ≫γL\lambda\gg\gamma_L)k1−δk^{1-\delta}
α→∞\alpha\to\inftyHigh correlation (λ≪αk\lambda\ll\sqrt{\alpha}k)k1−δk^{1-\delta}
Low correlation (λ≫αk\lambda\gg\sqrt{\alpha}k)k1−δk^{1-\delta}
k→∞k \to\inftyIndependent of λ\lambdak−δk^{-\delta}

Table 1. TABLE I. Asymptotic spectral scalings for the electric field fluctuations ⟨∣Ek∣2⟩\left\langle|E_k|^2 \right\rangle, assuming an injection spectrum for the spatial forcing ∣qk∣2∝k−δ|q_k|^2 \propto k^{-\delta} for different plasma (α\alpha) and correlation regimes. The low correlation case corresponds to the white-noise forcing.

The strategy employed for the single-λ\lambda case can be straightforwardly extended to the scenario where the correlation rate λ\lambda is governed by a power-law distribution. By applying the exact same procedure, i.e multiplying the electrostatic potential spectrum by k2k^2 and incorporating the inertial cascade spectrum ∣qk∣2∝k−δ|q_k|^2 \propto k^{-\delta}, one can readily deduce the corresponding electric field scalings from the λ\lambda-averaged asymptotic limits derived previously. More generally, this framework is not restricted to asymptotic approximations. It can be applied directly to the full, non-asymptotic kinetic FDR averaged over the λ\lambda distribution given in Eq. (37), thereby providing a complete and exact theoretical prediction for the electric field spectrum ⟨∣Ek∣2⟩\langle|E_k|^2 \rangle across all relevant spatial scales.

To conclude this section we present the asymptotic behaviors of the electric field spectra in two specific cases. First, we consider the α→−1\alpha\to-1 case in which the characteristic timescale of the system, i.e. the Landau damping rate γL\gamma_L, vanishes in the limit. Therefore, for a fixed range [λmin⁡,λmax⁡][\lambda_{\min},\lambda_{\max}], the low-correlation regime λ≫γL\lambda\gg\gamma_L is, at some point, inevitable and much more likely to occur. Second, we consider the α→∞\alpha\to\infty limit in which the Langmuir frequency ωL=αk\omega_L = \sqrt{\alpha}k becomes infinitely large and the high-correlation regime λ≪ωL\lambda\ll\omega_L is the most likely configuration.

In Figure 7 we display the electric field spectra obtained from Eq. (37) in i) for α=−0.8\alpha=-0.8 and k∈[10−8,10−5]k \in[10^{-8},10^{-5}] and in ii) for α=200\alpha=200 and k∈[10−5,10−2]k \in[10^{-5},10^{-2}]. In both cases we set μ=3/2\mu=3/2, λmin⁡=10−3\lambda_{\min}=10^{-3}, λmax⁡=10−2\lambda_{\max}=10^{-2}, so that i) and ii) are representative of low and high correlation regimes, respectively. We fix the noise spectral index δ=5/3\delta=5/3, to model a Kolmogorov-like behavior for the energy cascade at large (fluid) scales.

Electric field asymptotic behavior for low- and high-correlation regimes

Figure 7. Asymptotic behavior of the electric field for a correlated noise spectral index δ=5/3\delta=5/3. Upper panel: the electric field spectral index in the α→−1\alpha\to-1 limit low correlation regime is −2/3-2/3. Lower panel: the electric field spectral index in the α→∞\alpha\to\infty limit high correlation regime is −8/3-8/3.

Under these assumptions the model predicts an asymptotic electric field spectral index equal to 2/32/3 in i) and 8/38/3 in ii) as it can be easily appreciated from Figure 7. We outline that these spectral slopes are qualitatively in agreement with recent observations in the turbulent Earth’s magnetosheath[22] in the ionic and electronic ranges, respectively. However, this correspondence should be taken with caution. Indeed, our model is very simplified as it focuses on electrostatic fluctuations and it is one-dimensional, and it does not explicitly include any microphysics plasma processes as the peculiar type of waves that may be present in the system. Further studies are necessary to better comprehend if this qualitative agreement is robust.

Conclusions

In this work, we have developed a generalized kinetic framework to model plasma turbulence driven by a stochastic source with a finite correlation time. By constructing a kinetic analogue of the Langevin equation, we derived an exact FDR that characterizes the steady-state balance between the energy injected by the correlated noise and the collisionless dissipation provided by Landau damping. This framework allows for the exploration of diverse physical scenarios, ranging from electronic Langmuir waves to slow-MHD compressive modes, by tuning the coupling parameter α\alpha.

We analytically extracted the asymptotic spectral scalings of the electrostatic potential in several limits of interest, including the weakly damped non-oscillatory limit (α→−1\alpha\to-1) and the weakly damped fast-wave limit (α→∞\alpha\to\infty). Furthermore, the investigation of the k→∞k \to\infty limit revealed that at sub-ion and sub-electron scales, kinetic dynamical effects are entirely suppressed, and the system responds purely through static screening. The turbulent cascade in velocity space was also formally characterized via a Hermite decomposition, confirming an m−1/2m^{-1/2} asymptotic behavior. All these analytical predictions successfully reproduce the white-noise phenomenology[12] when the vanishing correlation time limit λ→∞\lambda\to\infty is considered. Subsequently, we validated the asymptotic spectral scalings and the Hermite asymptotic spectrum through direct numerical integrations of the underlying collisionless Vlasov system.

To bridge the gap between theoretical modeling and in-situ spacecraft observations, we extended our analysis to a multiscale forcing scenario where the noise correlation rate λ\lambda is treated as a stochastic variable governed by a power-law distribution. This extension mathematically mirrors the superposition principles that generate 1/fα1/f^\alpha macroscopic noise signatures. By averaging the single-λ\lambda FDR over this distribution, we derived exact, closed-form expressions for the spectra. These analytical solutions are intended to serve as a library of theoretical fitting functions for observational data in the solar wind and magnetosphere, circumventing the limitations of standard piecewise linear fitting.

Finally, to ensure the physical consistency of the predicted electric field fluctuations at small scales, we incorporated an inertial cascade spectral index into the spatial forcing profile, assuming ∣qk∣2∝k−δ|q_k|^2 \propto k^{-\delta}. Under this hypothesis, the Vlasov plasma effectively acts as a kinetic low-pass filter over the background turbulence. This framework could provide a model-based method, agnostic with respect to the underlying kinetic plasma process at play, to interpret electric field spectrum at kinetic scales observed in space.

AUTHOR DECLARATIONS

Data availability statement

The data that supports the findings of this study are available from the corresponding author upon reasonable request.

Conflict of interest statement

The authors have no conflicts to disclose.

ACKNOWLEDGMENTS

FM, OP, and AL are supported by the FIS2 Starting Grant FIS-2023-00246 “PhAse-sPAce cOmplexity in turbulent nearly-reversible plasmas (PAPAO)” (CUP B53C24009610001) funded by the Italian Ministry of University and Research. SB and GC acknowledge support from the Space It Up! project funded by the Italian Space Agency, ASI, and the Ministry of University and Research, MUR, under Contract No. 2024-5-E.0-CUP n. I53D24000060005. LSV was supported by the Swedish National Space Agency (SNSA) Research Grant N. 2025-00229.

References

  1. [1]1 M. Wan, W. H. Matthaeus, V. Roytershteyn, H. Karimabadi, T. Parashar, P. Wu, and M. Shay. Intermittent dissipation and heating in 3d kinetic plasma turbulence. Phys. Rev. Lett., 114:175002, Apr 2015.pubmed.ncbi.nlm.nih.gov/25978241
  2. [2]2 A. Chasapis, Y. Yang, W. H. Matthaeus, T. N. Parashar, C. C. Haggerty, J. L. Burch, T. E. Moore, C. J. Pollock, J. Dorelli, D. J. Gershman, R. B. Torbert, and C. T. Russell. Energy Conversion and Collisionless Plasma Dissipation Channels in the Turbulent Magnetosheath Observed by the Magnetospheric Multiscale Mission. The Astrophysical Journal, 862(1):32, July 2018.DOI
  3. [3]Pezzi, O., Liang, H., Juno, J. L., Cassak, P. A., Vásconez, C. L., Sorriso-Valvo, L., Perrone, D., Servidio, S., Roytershteyn, V., TenBarge, J. M., & Matthaeus, W. H. (2021). Dissipation measures in weakly collisional plasmas. Mon. Not. R. Astron. Soc., 505(4), 4857–4873.
  4. [4]Bruno, R., & Carbone, V. (2013). The Solar Wind as a Turbulence Laboratory. Living Reviews in Solar Physics, 10(1), 2.
  5. [5]5 T. Chang. Self-organized criticality, multi-fractal spectra, sporadic localized reconnections and intermittent turbulence in the magnetotail. Physics of Plasmas, 6(11):4137–4145, 1999.DOI
  6. [6]6 A. J. Klimas, J. A. Valdivia, D. Vassiliadis, D. N. Baker, M. Hesse, and J. Takalo. Self-organized criticality in the substorm phenomenon and its relation to localized reconnection in the magnetospheric plasma sheet. Journal of Geophysical Research: Space Physics, 105(A8):18765–18780, 2000.DOI
  7. [7]7 G. Consolini and T. Chang. Magnetic Field Topology and Criticality in Geomagnetic Dynamics: Relevance to Substorm Phenomena. Space Storms and Space Weather, 95:309–321, January 2001.DOI
  8. [8]8 G. Consolini and T. Chang. Complexity, magnetic field topology, criticality, and metastability in magnetotail dynamics. Journal of Atmospheric and Solar-Terrestrial Physics, 64(5):541–549, 2002.DOI
  9. [9]9 H. K. Kiyani, S. C. Chapman, Yu. V. Khotyaintsev, M. W. Dunlop, and F. Sahraoui. Global Scale-Invariant Dissipation in Collisionless Plasma Turbulence. Phys. Rev. Lett., 103(7):075006, August 2009.pubmed.ncbi.nlm.nih.gov/19792654
  10. [10]10 O. Pezzi, E. Camporeale, and F. Valentini. Collisional effects on the numerical recurrence in Vlasov-Poisson simulations. Physics of Plasmas, 23(2):022103, February 2016.DOI
  11. [11]11 Vincent David, Sébastien Galtier, and Romain Meyrand. Monofractality in the Solar Wind at Electron Scales: Insights from Kinetic Alfvén Waves Turbulence. Phys. Rev. Lett., 132(8):085201, February 2024.DOI
  12. [12]12 A. Kanekar, A. A. Schekochihin, W. Dorland, and N. F. Loureiro. Fluctuation-dissipation relations for a plasma-kinetic Langevin equation. Journal of Plasma Physics, 81(1):305810104, January 2015.DOI
  13. [13]13 V. Carbone, D. Telloni, F. Lepreti, and A. Vecchio. High-frequency magnetic fluctuations in space plasmas and the role of electron landau damping. The Astrophysical Journal Letters, 924(2):L26, Jan 2022.DOI
  14. [14]14 Federica Chiappetta, Emiliya Yordanova, Zoltán Vörös, Fabio Lepreti, and Vincenzo Carbone. Energy conversion through a fluctuation–dissipation relation at kinetic scales in the earth’s magnetosheath. The Astrophysical Journal, 957(2):98, nov 2023.DOI
  15. [15]15 G. I. Taylor. The spectrum of turbulence. Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences, 164(919):476–490, 1938.DOI
  16. [16]16 W. H. Matthaeus and M. L. Goldstein. Measurement of the rugged invariants of magnetohydrodynamic turbulence in the solar wind. Journal of Geophysical Research, 87(A8):6011–6028, August 1982.DOI
  17. [17]Frisch, U. (1995). Turbulence. The legacy of A.N. Kolmogorov. In Cambridge University Press.
  18. [18]18 Here and in the following, we use k, by a slight abuse of notation, to denote the dimensionless wavenumber associated with the coordinate x in (1a).
  19. [19]Krall, N. A., & Trivelpiece, A. W. (1973). Principles of plasma physics. In McGraw-Hill.
  20. [20]20 G. E. Uhlenbeck and L. S. Ornstein. On the theory of the brownian motion. Phys. Rev., 36:823–841, Sep 1930.DOI
  21. [21]21 E. Milotti. Model-based fit procedure for power-law-like spectra. Journal of Computational Physics, 217(2):834–844, September 2006.DOI
  22. [22]22 Tien Vo, Robert Ergun, Alexandros Chasapis, Yi Qi, Neha Pathak, Harriet George, Giuseppe Consolini, Simone Benella, Daniele Belardinelli, and Trevor Bowen. Kinetic range of strong electric field turbulence associated with magnetotail reconnection. Phys. Rev. Lett., 136:105201, Mar 2026.DOI

Paper details

vlnc:W_9h6h8kwd
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