On universality for the kinetic wave equation
Abstract.
On compact Riemannian manifolds with chaotic geometries, specifically those exhibiting the random wave model conjectured by Berry, we derive heuristically a homogeneous kinetic wave equation that is universal for all such manifolds.
1. Introduction
1.1. The kinetic wave equation
Classical references on the theory of the kinetic wave equation and its development can be found in the textbooks [17, 27, 39]. Although this theory is highly versatile and is applicable to any nonlinear dispersive equation, we shall focus on the model case of the nonlinear Schrödinger equation posed on a Riemannian manifold :
(NLS) |
Our attention is directed towards the weakly nonlinear setting, where the strength of nonlinearity, denoted as , is small. However, the sign of holds no significance in our analysis. We are specifically interested in the long-time dynamics of the correlation function , with representing a weakly interacting random wave ensemble. Assuming the phase randomization condition of the waves, it has been conjectured that the dynamics obey a certain kinetic wave equation at the kinetic time order (or the Van Hove limit [37]) .
Take the torus , where , for example. The phase randomization condition mandates the Fourier coefficients
to be uncorrelated. For initially prepared solutions — these are solutions of the form where decays sufficiently fast and are independent and i.i.d. random phases — this condition is proven to propagate in time (see e.g., [10]). In this context, the correlation function is entirely determined by the energy spectra .
Let us introduce the slow time variable with
(1.1) |
representing the kinetic order. It has been demonstrated (see e.g., [9, 11, 7, 8]) that, taking the infinite volume limit and the weak nonlinearity limit simultaneously while adhering to certain scaling laws results in , where satisfies the kinetic wave equation:
(KW) |
with initial data . The collision operator is expressed as
where , and the kernel is given by
(1.2) |
The Kolmogorov–Zakharov (KZ) spectrum, a stationary solution of the kinetic wave equation (KW), holds particular significance as it is anticipated to depict turbulent cascades. Consequently, it often constitutes the primary focus of interest in various applications. Experimental validations, both physical and numerical, of equation (KW) predominantly rely on the observation of the KZ spectrum, see for instance reviews [16, 28, 30] and recent developments [13, 21, 26, 41].
Remarkably, a series of mathematical works over recent years [7, 8, 9, 11, 12, 19, 35] culminated in a rigorous derivation of (KW). This convergence of efforts has significantly advanced our understanding of the kinetic wave equation, providing a solid mathematical foundation for its theoretical underpinnings.
1.2. The random wave model
Known derivations, be they heuristic or rigorous, of the kinetic wave equation always take as their starting point Hamiltonian equations set in Euclidean geometry (full space or its quotients), the only exception being [14], which considers a model based on random matrices. This seems a considerable limitation on the validity of the kinetic wave equation; nevertheless, the KZ spectrum seems to have broad relevance, and it is natural to wonder whether an avatar of (KW) might hold in more general geometries.
We will focus in this article on chaotic geometries; these are compact manifolds, with or without boundary (even though we will perform our analysis only on manifolds without boundaries), for which the geodesic flow is chaotic. Typically, chaotic billiards and compact hyperbolic surfaces fall under this category. In this setting, Berry [2, 3] (see also Voros [38]) conjectured that, in the semiclassical limit , a corresponding eigenfunction “appears to be a (real) Gaussian random function exhibiting … oscillations … (that) are statistically isotropic with the autocorrelation … given by a Bessel function.”. This is the random wave model (RWM).
To date, different mathematical formulations of Berry’s conjecture exist, see e.g., [32, 40, 29, 31, 6, 1, 23, 34]. To illustrate the ideas, let us follow Bourgain [6] (see also [23]) who states that, in the semiclassical limit , the ensemble of functions , where is the restriction of to a geodesic ball centered at with radius much larger than the Planck scale , resembles the monochromatic Gaussian random wave on . Recall that, this is an isotropic Gaussian random field on whose spectrum is given by
where is the area of . Equivalently, this means that the correlation function of is a radial function given by the Fourier transform
(1.3) |
where is any vector with . It is know that can be computable in terms of Bessel functions (see e.g., [36, Ch. VIII, Eq. (25)]): , where and is the Bessel function of the first kind of order .
Following Berry, to make sense of this approximation, one needs to define the (multi-point) correlations of the ensemble by taking local averages over domains much larger than the Planck scale . We postpone till §3 a discussion of the properties of the RWM in greater detail, but immediately provide some references pointing towards its validity. Rigorous results on the RWM are scant [6, 23], but physical [22, 33] and numerical [20, 25] evidence is available, in particular since the connection of the RWM to the number of nodal domains was established [4].
In the present article, we propose a heuristic derivation of a kinetic wave equation, which is valid under Berry’s conjecture.
1.3. Main result: kinetic limit on domains with chaotic geometries
We will now state precisely our main (formal) result: the solutions of (NLS) on a domain with chaotic geometry converge to a universal kinetic wave equation.
Consider a compact manifold with chaotic geometry and without boundary, for example a compact hyperbolic surface. Let be eigenvalues (counting multiplicity) of and let be the corresponding eigenfunctions that are orthogonal. We further assume that the eigenfunctions are real-valued (even though this condition is not necessary for our results) and are normalized with . For , on the dilated manifold , one has eigenvalues and eigenfunctions , with .
We set (NLS) on , and provide it with prepared initial data
(1.4) |
where are i.i.d. random variables, uniformly distributed on , and is a smooth, rapidly decaying function. Our normalization corresponds to and up to logarithmic corrections. The solution to (NLS) can be expanded in the orthonormal basis as
(1.5) |
Under the simultaneous infinite volume limit and the weak nonlinearity limit , we assert that as . Here solves the kinetic wave equation
(KWR) |
with initial data . Identifying , the collision operator is expressed as
where the kernel is defined by
and the interaction integral is defined by
As we shall see in §3.3, this integral quantifies the interactions among four eigenfunctions, serving the same role as in (1.2).
Detailed computations leading to the derivation of (KWR) are provided in §4. We shall also see in §2.2 that the equation (KWR) also describes the dynamics of radial solutions to (KW). Therefore, we have named the kinetic equation (KWR) due to the dual interpretation of the letter “R” not only as the “random wave model” but also as “radial”.
2. Discussion of the result
2.1. Range of validity of the singular limit
The derivation of (KWR) in §4 relies on the convergence of the sum over frequencies to the integral defining the collision operator, see §4.5 for the details. To ensure the validity of this sum-to-integral limit, the points (with fixed) must be equidistributed in a neighborhood around the resonant manifold defined by the equation (with fixed).
Here, all frequencies can be confined to be of size due to the rapid decay of . Furthermore, we can assume heuristically that the frequencies are random, uniformly distributed and independent. Thus, it suffices for there to be an infinite number of points in the vicinity of the resonant manifold. Considering the volume of this neighborhood, which is of order , and the number of points in a unit volume, which is of order , we obtain the condition , or equivalently .
Another condition that is necessary to ensure the convergence of the function to the Dirac , see also §4. This convergence occurs as , i.e., when .
In summary, the kinetic limit is deemed valid in the limit where
2.2. From (KW) to (KWR)
2.3. Universality of turbulent spectra
We showed that the kinetic wave equation is universal for domains with chaotic geometry. Since the KZ spectra arise as stationary solutions of (KWR), they are also universal over this class of domains.
Note that the ideas used in the present article are applicable to any isotropic equation, and thus that universality of turbulent spectra should extend to this class of equations.
2.4. Questions
To summarize the findings of the present article, the kinetic equation (KWR) is valid on domains with chaotic geometry, but also on tori under an isotropy assumption (which we do not attempt to formulate more precisely).
It is tempting to ask if the class of domains for which (KWR) is valid might not be even larger. Amongst Riemannian manifolds, the sphere often exhibits the most irregular behavior in terms of equidistribution of eigenvalues over , or equidistribution of eigenfunctions over the manifold. Therefore, the sphere seems to be the most natural example for which (KWR) might not apply.
One can also ask the same question beyond the framework of Riemannian manifolds: what can be said of confining potentials? Or even of graphs in the proper limit?
3. The random wave model and its implications
In this section, we discuss the random wave model, review and mathematically formulate Berry’s conjecture, and obtain several consequences of it. We shall henceforth denote for two functions of variable , when for and . We shall also denote when .
3.1. Berry’s conjecture
Conjecture 3.1 (Berry).
When , the ensemble of random variables with respect to the probability measure on :
where is a geodesic ball of radius , resembles an isometric Gaussian random field on with variance , given by (1.3).
When the geometry is locally Euclidean (e.g., in the interior of a chaotic billiard in ), this means that for any sequence of vectors in , the -tuple of random variables defined by is approximately a joint Gaussian distribution with respect to . This Gaussianity is characterized via the following Isserlis–Wick type formula for multi-point correlations:
(3.1) |
where denote the expectation with respect to , ranges among all partitions of into disjoint doubletons, and the two-point correlations are given by
On a non-Euclidean manifold (which is compact and without boundary), we need to be (smooth) vector fields. In this situation, the random variables are defined by , where is the exponential map at . We conjecture that, the formula (3.1) still holds in this setting, with the two-point correlation given by
(3.2) |
We believe that this formulation is essentially equivalent to the one given by Sodin in [34]. Particularly, when is even and , the estimate (3.1) yields
the right hand side being the number of all possible ways of partitioning a set of elements into disjoint doubletons. Taking , one recovers the quantum unique ergodicity at scale . We refer to works [24, 18] for investigations of the quantum unique ergodicity at small scales on tori.
With the natural conjecture that and being uncorrelated with respect to when , Berry’s conjecture can be extended to the ensemble . In this case, under the assumptions that and , the estimate (3.1) becomes
(3.3) |
where is defined as in (3.2). Particularly, taking in (3.3) yields
(3.4) |
As another consequence, if , if and , then (3.2) implies that
(3.5) |
3.2. Weak formulation
For our derivation of the kinetic wave equation, as we shall see, a weak formula of Berry’s conjecture is already sufficient. We start from (3.5) with the belief that local averages in yields a similar estimate:
(3.6) |
where is a probability measure on lifting :
We can easily deduce (3.6) from (3.5) as follows. Let be the center of the geodesic ball . For , define the vector field by , where is the parallel transport from to along the (radial) geodesic (with respect to the Levi–Civita connection). Is is known that is a linear isometry, so . Identifying , we have
Similar analysis applies to multi-point correlations: letting be distinct indices, so that no pairings in between them exist, then
(3.7) |
3.3. Interaction integrals
The principle term of the multi-point correlation (3.3) vanishes when no pairings exists. This happens particularly when are distinct. However, we can still compute its sub-principal term, when , using (3.7). In fact, using the spherical coordinates,
(3.8) | ||||
We shall call an interaction integral. It clearly has the scaling invariance:
(3.9) |
One then expects estimates similar to (3.8) to hold over subsets much larger than the plank scale , particularly when . Precisely, letting denote the expectation with respect to , one expects that, when are distinct, the multi-point correlation is determined, up to a sign, by the square root of the corresponding interaction integral:
(3.10) |
To establish (3.10), it is essential to ensure that distant points make negligible contributions to the correlation. This is quantified in the following bound, which is suggested by the decay of , but probably not optimal:
3.4. Phase randomization
As stated by [39, §2.1.2], due to the dispersion of the wave system, even initially correlated oscillations with different wavenumbers undergo phase randomization as time progresses. Thus, in describing a free wave field it would be natural to average over the ensemble of chaotic (random) phases, i.e., to use the random phase approximation. This assumption is closely related to the molecular chaos assumption in kinetic theory [5, 15]. Let us therefore choose i.i.d. random variables , uniformly distributed on , and put . The correlation formulas can be stated in terms of . The two-point correlation now states that
Similar analysis as in the previous subsection yields that: if are distinct, then
Consequently, for some random phase (in fact ), there holds
4. Derivation of the kinetic wave equation
In this section, we derive the kinetic wave equation (KWR). For this purpose, let us fix to be a compact manifold with chaotic geometry and without boundary, so that the analysis in §3 applies.
4.1. Fourier series expansion
Let the eigenvalues and eigenfunctions be chosen as in §1.3. Recall that, by Weyl’s law, we have , where is the volume of the unit ball in . Therefore, setting , one obtains the asymptotic behavior for eigenvalues:
(4.1) |
4.2. Wick ordering
Denote for simplicity. We have
This is a consequence of the conservation of mass and the following estimate which holds true for all and is proved using the Weyl law (4.1) and the convergence of Riemann sums:
(4.3) |
4.3. Perturbative expansion
Using the Picard iteration method and Duhamel’s principle, we obtain an asymptotic expansion , where the first few iterates are given by
where we denote , and , , for simplicity.
4.4. Energy density
We look for the asymptotics of the energy density . By the asymptotic expansion, we write
4.5. Kinetic limit
References
- [1] M. Abert, N. Bergeron, and E. L. Masson. Eigenfunctions and random waves in the benjamini-schramm limit. arXiv preprint arXiv:1810.05601, 2018.
- [2] M. V. Berry. Regular and irregular semiclassical wavefunctions. Journal of Physics A: Mathematical and General, 10(12):2083, 1977.
- [3] M. V. Berry. Semiclassical mechanics of regular and irregular motion. Les Houches lecture series, 36:171–271, 1983.
- [4] E. Bogomolny and C. Schmit. Percolation model for nodal domains of chaotic wave functions. Physical Review Letters, 88(11):114102, 2002.
- [5] L. Boltzmann. Weitere Studien über das Wärmegleichgewicht unter Gasmolekülen. (Wien), 1872.
- [6] J. Bourgain. On toral eigenfunctions and the random wave model. Israel Journal of Mathematics, 201(2):611–630, 2014.
- [7] C. Collot and P. Germain. On the derivation of the homogeneous kinetic wave equation. arXiv preprint arXiv:1912.10368, 2019.
- [8] C. Collot and P. Germain. Derivation of the homogeneous kinetic wave equation: longer time scales. arXiv preprint arXiv:2007.03508, 2020.
- [9] Y. Deng and Z. Hani. On the derivation of the wave kinetic equation for nls. In Forum of Mathematics, Pi, volume 9, page e6. Cambridge University Press, 2021.
- [10] Y. Deng and Z. Hani. Propagation of chaos and the higher order statistics in the wave kinetic theory. arXiv preprint arXiv:2110.04565, 2021.
- [11] Y. Deng and Z. Hani. Full derivation of the wave kinetic equation. Inventiones mathematicae, pages 1–182, 2023.
- [12] Y. Deng and Z. Hani. Long time justification of wave turbulence theory. arXiv preprint arXiv:2311.10082, 2023.
- [13] R. S. Dù and O. Bühler. The impact of frequency bandwidth on a one-dimensional model for dispersive wave turbulence. Journal of Nonlinear Science, 33(5):81, 2023.
- [14] G. Dubach, P. Germain, and B. Harrop-Griffiths. On the derivation of the homogeneous kinetic wave equation for a nonlinear random matrix model. arXiv preprint arXiv:2203.13748, 2022.
- [15] I. Gallagher, L. Saint-Raymond, and B. Texier. From Newton to Boltzmann: hard spheres and short-range potentials. Zurich Lectures in Advanced Mathematics. European Mathematical Society (EMS), Zürich, 2013.
- [16] S. Galtier. Wave turbulence: The case of capillary waves. Geophysical & Astrophysical Fluid Dynamics, 115(3):234–257, 2021.
- [17] S. Galtier. Physics of Wave Turbulence. Cambridge University Press, 2022.
- [18] A. Granville and I. Wigman. Planck-scale mass equidistribution of toral laplace eigenfunctions. Communications in Mathematical Physics, 355:767–802, 2017.
- [19] Z. Hani, J. Shatah, and H. Zhu. Inhomogeneous turbulence for wick nls. arXiv preprint arXiv:2309.12037, 2023.
- [20] D. A. Hejhal and B. N. Rackner. On the topography of maass waveforms for psl (2, z). Exp. Math., 1(4):275–305, 1992.
- [21] A. Hrabski and Y. Pan. Verification of wave turbulence theory in the kinetic limit. arXiv preprint arXiv:2311.10846, 2023.
- [22] O. Hul, N. Savytskyy, O. Tymoshchuk, S. Bauch, and L. Sirko. Investigation of nodal domains in the chaotic microwave ray-splitting rough billiard. Physical Review E, 72(6):066212, 2005.
- [23] M. Ingremeau. Local weak limits of laplace eigenfunctions. Tunisian Journal of Mathematics, 3(3):481–515, 2021.
- [24] S. Lester and Z. Rudnick. Small scale equidistribution of eigenfunctions on the torus. Communications in Mathematical Physics, 350:279–300, 2017.
- [25] S. W. McDonald and A. N. Kaufman. Wave chaos in the stadium: Statistical properties of short-wave solutions of the helmholtz equation. Physical review A, 37(8):3067, 1988.
- [26] N. Navon, C. Eigen, J. Zhang, R. Lopes, A. L. Gaunt, K. Fujimoto, M. Tsubota, R. P. Smith, and Z. Hadzibabic. Synthetic dissipation and cascade fluxes in a turbulent quantum gas. Science, 366(6463):382–385, 2019.
- [27] S. Nazarenko. Wave turbulence, volume 825. Springer Science & Business Media, 2011.
- [28] S. Nazarenko and S. Lukaschuk. Wave turbulence on water surface. Annual Review of Condensed Matter Physics, 7:61–88, 2016.
- [29] F. Nazarov and M. Sodin. Random complex zeroes and random nodal lines. In Proceedings of the International Congress of Mathematicians 2010 (ICM 2010) (In 4 Volumes) Vol. I: Plenary Lectures and Ceremonies Vols. II–IV: Invited Lectures, pages 1450–1484. World Scientific, 2010.
- [30] A. C. Newell and B. Rumpf. Wave turbulence: a story far from over. In Advances in wave turbulence, pages 1–51. World Scientific, 2013.
- [31] S. Nonnenmacher. Anatomy of quantum chaotic eigenstates. In Chaos: Poincaré Seminar 2010, pages 193–238. Springer, 2013.
- [32] Z. Rudnick and P. Sarnak. The behaviour of eigenstates of arithmetic hyperbolic manifolds. Communications in Mathematical Physics, 161:195–213, 1994.
- [33] N. Savytskyy, O. Hul, and L. Sirko. Experimental investigation of nodal domains in the chaotic microwave rough billiard. Physical Review E, 70(5):056209, 2004.
- [34] M. Sodin, V. Sidoravicius, and S. Smirnov. Lectures on random nodal portraits. Probability and statistical physics in St. Petersburg, 91:395–422, 2016.
- [35] G. Staffilani and M.-B. Tran. On the wave turbulence theory for a stochastic kdv type equation. arXiv preprint arXiv:2106.09819, 2021.
- [36] E. M. Stein and T. S. Murphy. Harmonic analysis: real-variable methods, orthogonality, and oscillatory integrals, volume 3. Princeton University Press, 1993.
- [37] L. Van Hove. Quantum-mechanical perturbations giving rise to a statistical transport equation. Physica, 21:517–540, 1955.
- [38] A. Voros. Semi-classical ergodicity of quantum eigenstates in the wigner representation. In Stochastic Behavior in Classical and Quantum Hamiltonian Systems: Volta Memorial Conference, Como, 1977, pages 326–333. Springer, 2005.
- [39] V. E. Zakharov, V. S. L’vov, and G. Falkovich. Kolmogorov spectra of turbulence I: Wave turbulence. Springer Science & Business Media, 2012.
- [40] S. Zelditch. Recent developments in mathematical quantum chaos. Current developments in mathematics, 2009, 2009:115–205, 2010.
- [41] Y. Zhu, B. Semisalov, G. Krstulovic, and S. Nazarenko. Testing wave turbulence theory for the gross-pitaevskii system. Physical Review E, 106(1):014205, 2022.