Lévy-Driven Hydrodynamic SPDEs

• Lévy-driven hydrodynamic SPDEs are evolution equations that combine nonlinear fluid dynamics with Lévy noise, capturing both continuous and jump perturbations.
• They employ advanced infinite-dimensional stochastic calculus, variational techniques, and Galerkin methods to establish existence, uniqueness, and regularity.
• These models underpin canonical fluid dynamics equations like Navier-Stokes, MHD, and Burgers, offering insights into turbulent scaling laws and rare-event analysis.

Lévy-driven hydrodynamic stochastic partial differential equations (SPDEs) represent a broad class of evolution equations central to the mathematical theory of stochastic turbulence, nonequilibrium statistical mechanics, and the rigorous analysis of random perturbations in fluid systems. These SPDEs arise by coupling nonlinear hydrodynamic deterministic dynamics—such as those in the Navier-Stokes, magnetohydrodynamics (MHD), and Boussinesq equations—with Lévy noise, which encapsulates both continuous Gaussian fluctuations and discontinuous, jump-type stochastic forcing. The mathematical analysis of such systems combines infinite-dimensional stochastic calculus, functional analysis, and advanced PDE techniques to investigate well-posedness, large deviations, scaling laws in turbulence, and the structure of solutions across different regimes of viscosity and noise.

1. Formulation of Lévy-driven Hydrodynamic SPDEs

The standard abstract setup employs a Gelfand triple $V \subset H \cong H^* \subset V^*$, where $H$ is a separable Hilbert space corresponding to the energy space of the hydrodynamic model, and $V$ a typically more regular subspace (e.g., mean-zero divergence-free Sobolev spaces for incompressible flow). The dynamics are governed by stochastic evolution equations of the general form: $$dX(t) = A(X(t))\,dt + B(X(t))\,dW(t) + \int_Z f(X(t^-),z)\,\widetilde N(dt,dz) + \int_Z g(X(t^-),z)\,N(dt,dz)$$ where:

• $A$ is the (possibly nonlinear, locally monotone or coercive) drift capturing dissipation and deterministic transport,
• $B$ is the diffusion coefficient for cylindrical Wiener noise $W$,
• $f$, $g$ are the small and large jump kernels, accounting for the action of the Lévy noise,
• $\widetilde{N}$ is the compensated Poisson random measure, and $H$0 the original Poisson measure associated with the jump process.

Concretely, this framework encompasses: the 2D or 3D Navier-Stokes equations, stochastic Burgers equations, MHD, Boussinesq (convection) equations, and shell models of turbulence, with state spaces and operators specialized accordingly (Yuan et al., 2021, Peng et al., 2020, Brzeźniak et al., 2011, Motyl, 2013).

2. Existence, Uniqueness, and Regularity Theory

Well-posedness of Lévy-driven hydrodynamic SPDEs relies on the structure of the drift and noise coefficients:

• Coercivity and local monotonicity of $H$1, e.g., corresponding to linear dissipation operators (Stokes operator for Navier-Stokes, Laplacian-plus-coupling for MHD/Boussinesq),
• Skew-symmetry, antisymmetry, and (local) Lipschitz/trilinear bounds for the bilinear nonlinearities $H$2,
• Sharp Lipschitz and polynomial growth conditions for noise coefficients $H$3, $H$4, and the diffusion $H$5.

The main results, as developed in (Brzeźniak et al., 2011) and sharpened in (Peng et al., 2020), guarantee:

• Existence and uniqueness of strong solutions (in the PDE sense), i.e., $H$6-valued càdlàg adapted processes satisfying the integral version of the SPDE for all test vectors in $H$7, with minimal additional moment assumptions on the Lévy measure and jump kernels.
• Regularity results, including energy estimates in $H$8 and spatial Sobolev norms, are obtained via Itô’s formula (extended to handle jumps via compensated Poisson integrals) and Gronwall-type inequalities.

Further, for domains lacking compact Sobolev embeddings—such as unbounded spatial regions—Martingale solution theory and specialized tightness/compactness criteria are deployed, as in (Motyl, 2013), extending the framework to settings where classical variational approaches are unavailable.

3. Scaling Laws and Statistical Quantities in Stochastic Turbulence

A distinctive feature of Lévy-driven hydrodynamic SPDEs is the ability to rigorously analyze statistical turbulent quantities under the influence of jump noise. For the one-dimensional stochastic Burgers equation with cylindrical Lévy space-time white noise (Yuan et al., 2021), the following statistical observables are studied:

• Moment estimates: For solutions $H$9 in Sobolev spaces $V$0, moments scale with viscosity as $V$1.
• Structure functions: The $V$2th-order structure function at scale $V$3, $V$4, exhibits bifurcated scaling
  • Dissipation range: $V$5 for $V$6, $V$7 for $V$8.
  • Inertial range: $V$9 for $$dX(t) = A(X(t))\,dt + B(X(t))\,dW(t) + \int_Z f(X(t^-),z)\,\widetilde N(dt,dz) + \int_Z g(X(t^-),z)\,N(dt,dz)$$0.
• Energy spectra: The Fourier energy spectrum, $$dX(t) = A(X(t))\,dt + B(X(t))\,dW(t) + \int_Z f(X(t^-),z)\,\widetilde N(dt,dz) + \int_Z g(X(t^-),z)\,N(dt,dz)$$1, follows the scaling $$dX(t) = A(X(t))\,dt + B(X(t))\,dW(t) + \int_Z f(X(t^-),z)\,\widetilde N(dt,dz) + \int_Z g(X(t^-),z)\,N(dt,dz)$$2 in the inertial range.

In the inviscid limit ($$dX(t) = A(X(t))\,dt + B(X(t))\,dW(t) + \int_Z f(X(t^-),z)\,\widetilde N(dt,dz) + \int_Z g(X(t^-),z)\,N(dt,dz)$$3), these scaling laws persist globally, exemplifying "strong stochastic turbulence"—Kolmogorov-type scaling survives the vanishing viscosity (no dissipative cutoff) (Yuan et al., 2021).

4. Moderate and Large Deviation Principles

Deviation principles quantify the probability of rare events or excursions in the solution trajectories due to noise. For abstract nonlinear hydrodynamical SPDEs driven by multiplicative Lévy noise (Li et al., 11 Feb 2025):

• The moderate deviation principle (MDP) is formulated for the rescaled deviation process $$dX(t) = A(X(t))\,dt + B(X(t))\,dW(t) + \int_Z f(X(t^-),z)\,\widetilde N(dt,dz) + \int_Z g(X(t^-),z)\,N(dt,dz)$$4, where $$dX(t) = A(X(t))\,dt + B(X(t))\,dW(t) + \int_Z f(X(t^-),z)\,\widetilde N(dt,dz) + \int_Z g(X(t^-),z)\,N(dt,dz)$$5 solves the SPDE with small noise strength $$dX(t) = A(X(t))\,dt + B(X(t))\,dW(t) + \int_Z f(X(t^-),z)\,\widetilde N(dt,dz) + \int_Z g(X(t^-),z)\,N(dt,dz)$$6, and $$dX(t) = A(X(t))\,dt + B(X(t))\,dW(t) + \int_Z f(X(t^-),z)\,\widetilde N(dt,dz) + \int_Z g(X(t^-),z)\,N(dt,dz)$$7 the deterministic skeleton equation.
• Under precise scaling conditions and regularity for the coefficients, MDP holds on the canonical path-space $$dX(t) = A(X(t))\,dt + B(X(t))\,dW(t) + \int_Z f(X(t^-),z)\,\widetilde N(dt,dz) + \int_Z g(X(t^-),z)\,N(dt,dz)$$8 with explicit good rate function involving controlled skeleton equations.
• The proof leverages the weak convergence method (Budhiraja–Dupuis criterion), a priori uniform estimates, and finite-dimensional projections, notably dispensing with the compact embedding restriction previously required (Li et al., 11 Feb 2025).

5. Analytic and Probabilistic Methods

The development of the theory employs:

• Galerkin (spectral) approximation and projection methods: Finite-dimensional truncations underpin existence proofs and facilitate sharp energy estimates essential for passing to the infinite-dimensional limit, including for non-compact/unbounded domains (Motyl, 2013, Peng et al., 2020).
• Itô–Lévy calculus: Extending Itô’s formula to handle both the martingale part (compensated Poisson integral) and finite-variation part (pure jump integral), supporting control of higher moments and the derivation of functional inequalities (Brzeźniak et al., 2011).
• Energy/entropy inequality techniques: Applied to the kinetic/moment functionals over time, establishing deterministic and probabilistic bounds, and crucially tracking dependence on viscosity and noise parameters (Yuan et al., 2021).
• Compactness, tightness, and Skorokhod representation: Multilayered functional-analytic strategy for extracting convergent subsequences of solutions, particularly on product spaces and non-metrizable Fréchet-type spaces (Motyl, 2013).
• Monotonicity and local monotonicity: Utilized for uniqueness and regularity, as in the Brzeźniak–Liu–Zhu variational approach (Brzeźniak et al., 2011).

6. Concrete Models and Applications

The abstract theory specializes readily to canonical fluid models:

• 2D and 3D Navier-Stokes equations: Energy estimates, skew-symmetric nonlinearities, and scaling laws are underpinned by the Helmholtz projection and the Stokes operator formulation (Brzeźniak et al., 2011, Peng et al., 2020, Motyl, 2013).
• MHD and Boussinesq systems: Coupled systems (velocity/magnetic/temperature fields) are handled by product Hilbert space constructions and block-coupling of the operators, confirming that standard hydrodynamic nonlinearities respect the core analytical assumptions (Yuan et al., 2021, Peng et al., 2020, Motyl, 2013).
• Shell models of turbulence: Infinite-dimensional complex-valued ODEs on ℓ², driven by nonlocal Lévy noise, mapped directly to the theoretical framework via diagonal operators and nearest-neighbor bilinear forms (Li et al., 11 Feb 2025, Peng et al., 2020).
• Stochastic Burgers equation: One-dimensional forced dynamics with cylindrical Lévy space-time noise, allowing complete characterization of turbulent scaling and inviscid limits (Yuan et al., 2021).

7. Removal of Structural Constraints and Current Directions

Recent advances have eliminated structural assumptions found in earlier literature:

• Removal of the compact embedding condition on the Gelfand triple, achieved through a priori estimates uniform in the projection dimension and projections onto finite-dimensional templates (Li et al., 11 Feb 2025).
• Minimization of noise coefficient restrictions—classical $$dX(t) = A(X(t))\,dt + B(X(t))\,dW(t) + \int_Z f(X(t^-),z)\,\widetilde N(dt,dz) + \int_Z g(X(t^-),z)\,N(dt,dz)$$9- and “small-$A$0” Lipschitz/growth bounds suffice for existence and uniqueness, with no supplementary $A$1-moment or small-jump controls required (Peng et al., 2020).
• Unified treatment of Gaussian and Lévy noise, allowing for continuous and discontinuous stochastic perturbations in the same analytical framework (Brzeźniak et al., 2011).
• Applicability to a wider class of domains—including unbounded spatial regions—by systematically deploying compactness criteria and path-space topology constructions (Motyl, 2013).

This ongoing refinement of analytic and probabilistic tools positions Lévy-driven hydrodynamic SPDEs as a foundational topic for rigorous investigation of turbulence, rare-event dynamics, and statistical properties of stochastically forced fluid systems.

References:

(Yuan et al., 2021, Peng et al., 2020, Li et al., 11 Feb 2025, Brzeźniak et al., 2011, Motyl, 2013)