📝 SummarXiv

Abstract

Spontaneous stochasticity refers to the emergence of intrinsic randomness in deterministic systems under singular limits, a phenomenon conjectured to be fundamental in turbulence. Armstrong and Vicol recently constructed a deterministic, divergence-free multiscale vector field arbitrarily close to a weak Euler solution, proving that a passive scalar transported by this field exhibits anomalous dissipation and lacks a selection principle in the vanishing diffusivity limit. We show that this advection-diffusion PDE also selects a non-Dirac measure in the space of weak solutions in the inviscid limit, thereby exhibiting Eulerian spontaneous stochasticity. We further provide numerical evidence of Lagrangian spontaneous stochasticity, together with a numerical illustration of the Obukhov-Corrsin conjecture for this system. We formulate a general framework for spontaneous stochasticity in arbitrary finite dimensional systems under arbitrary regularizations, distinguishing two regimes: weak, where different probability measures may arise along subsequences of inviscid limits, and strong, where the limit measure is unique. The advection diffusion system of Armstrong and Vicol lies in the strong regime. We prove that the set of selected measures is compact and equals the closed convex hull of Dirac measures. Moreover, for any non-Dirac measure supported on the set of nonunique solutions of the inviscid system, there exists a regularization that produces strong spontaneous stochasticity. Finally, we relate this framework to renormalization-group methods \`a la Feigenbaum and examine how the underlying dynamical system influences the inviscid limit. The discussion is complemented by elementary finite-dimensional examples illustrating a variety of cases.