Orbital stability index for stochastic systems (Q4518328)

Suppose a stochastic differential equation has an invariant orbit, which is a closed smooth curve such that every solution starting on this curve remains on the curve. An ``orthogonal linearisation'' along the orbit of such a system is calculated explicitly. Almost sure and \(p\)th mean Lyapunov exponents, and the stability index (the second zero of the \(p\)th mean exponent) for the orthogonal linearisation are introduced, and their properties are stated. Then large deviations estimates for the convergence (or divergence, resp.) of solutions of the original nonlinear equation to the invariant orbit are derived, where the rate function's exponent is the stability index. This generalises results known for equilibrium points of stochastic differential equations. An additional problem arises here from the fact that nondegeneracy conditions need not be satisfied, e.g., if the invariant orbit is a periodic solution without diffusion. For systems on the plane explicit calculations of the Lyapunov exponents are given, and this is applied to obtain expansions of the exponents for planar systems with small noise.