Classification of the Asymptotic Behaviour of Globally Stable Linear Differential Equations with Respect to State-independent Stochastic Perturbations

Abstract

In this paper we consider the global stability of solutions of an affine stochastic differential equation. The differential equation is a perturbed version of a globally stable linear autonomous equation with unique zero equilibrium where the diffusion coefficient is independent of the state. We find necessary and sufficient conditions on the rate of decay of the noise intensity for the solution of the equation to be globally asymptotically stable, stable but not asymptotically stable, and unstable, each with probability one. In the case of stable or bounded solutions, or when solutions are a.s. unstable asymptotically stable in mean square, it follows that the norm of the solution has zero liminf, by virtue of the fact that $\|X\|^2$ has zero pathwise average a.s. Sufficient conditions guaranteeing the different types of asymptotic behaviour which are more readily checked are developed. It is also shown that noise cannot stabilise solutions, and that the results can be extended in all regards to affine stochastic differential equations with periodic coefficients.