Space-Time Statistical Solutions of the Incompressible Euler Equations and Landau-Lifshitz Fluctuating Hydrodynamics

TL;DR

This paper rigorously connects the Landau-Lifschitz fluctuating hydrodynamics equations for incompressible fluids to the incompressible Euler equations in the infinite Reynolds limit, revealing the physical realization of non-uniqueness and stochasticity in turbulent flows.

Contribution

It provides a rigorous mathematical framework showing that solutions of fluctuating hydrodynamics converge to space-time statistical solutions of Euler equations, highlighting physical non-uniqueness.

Findings

Infinite Reynolds limit yields space-time statistical solutions of Euler equations.

Persistent randomness in the limit corresponds to nontrivial probability measures.

Supports the concept of spontaneous stochasticity in turbulence.

Abstract

We study rigorously the infinite Reynolds limit of the solutions of the Landau-Lifschitz equations of fluctuating hydrodynamics for an incompressible fluid on a $d$-dimensional torus for $d\geq 2.$ These equations, which model the effects of thermal fluctuations in fluids, are given a standard physical interpretation as a low-wavenumber ``effective field theory'', rather than as stochastic partial differential equations. We study in particular solutions which enjoy some a priori Besov regularity in space, as expected for initial data chosen from a driven, turbulent steady-state ensemble. The empirical basis for this regularity hypothesis, uniform in Reynolds number, is carefully discussed, including evidence from numerical simulations of turbulent flows that incorporate thermal fluctuations. Considering the initial-value problem for the Landau-Lifschitz equations, our main result is that the infinite Reynolds number limit is a space-time statistical solution of the incompressible Euler equations in the sense of Vishik \& Fursikov. Such solutions are described by a probability measure on space-time velocity fields whose realizations are weak or distributional solutions of the incompressible Euler equations, each with the same prescribed initial data. In conjunction with recent numerical evidence that these probability measures are non-trivial, our theorem shows that the mathematical non-uniqueness of the solutions of the Cauchy problem for Euler, as revealed by convex integration studies, may be physically realized. In support of the theory of spontaneous stochasticity, any persistent randomness in the infinite-Reynolds limit is associated to a nontrivial probability distribution over non-unique solutions of the ideal Euler equations.