Phase space contraction of degenerately damped random splittings
David P. Herzog, Jonathan C. Mattingly
TL;DR
The paper develops a general probabilistic Lyapunov framework for partially dissipative systems and applies it to randomly split models of Lorenz '96 and Galerkin truncations of the 2D Euler equations. By combining subconservation of an energy-like observable with partial dissipation in a controlled dissipative region and quantified entrance probabilities, the authors prove the existence of stationary measures for the randomly split dynamics and derive rates of return to compact sets (exponential for Lorenz '96 and stretched-exponential for Euler). The main technical contributions include a set of energy-transfer lemmas, a mechanism to bound entrance probabilities to damping regions that is uniform in energy, and a detailed analysis of triadic interactions via random splittings. These results show that minimal damping, when geometrically arranged and injected through random splittings, suffices to stabilize high-dimensional systems and generate statistically steady states, providing insights into energy cascade and dissipation in simplified fluid models. The framework lays groundwork for future work combining random switching with forcing and extending to broader classes of partially damped, high-dimensional systems.
Abstract
When studying out-of-equilibrium systems, one often excites the dynamics in some degrees of freedom while removing the excitation in others through damping. In order for the system to converge to a statistical steady state, the dynamics must transfer the energy from the excited modes to the dissipative directions. The precise mechanisms underlying this transfer are of particular interest and are the topic of this paper. We explore a class of randomly switched models introduced in [2,3] and provide some of the first results showing that minimal damping is sufficient to stabilize the system in a fluids model.
