A class of limit theorems for singular diffusions (Q1182745)

This article establishes functional forms of the central limit theorem and law of the iterated logarithm for singular diffusions on \(R^ k\). The diffusions satisfy Itô's equation for suitable smooth integrands and an \(l\)-dimensional Brownian motion, and are required to be stable in distribution, i.e. for each starting point, the transition probabilities converge weakly to an invariant probability measure. The main interest of the paper is in the more difficult cases which lack full rank (e.g. \(l<k\)). Section 2 provides a verifiable condition (in terms of drift and diffusion coefficients) under which the transition probability is stable in distribution. In fact, under the given conditions there is exponentially quick convergence in a bounded Lipschitz norm to the invariant measure. Under strengthened conditions the paper proves a functional central limit theorem in the sense that for \(f\) in a class of functions, \(n^{-1/2}\int^{nt} f(X(s)) ds\) converges weakly to a Brownian motion \(B(t)\). Here \(X\) is the diffusion and \(f\) is assumed to be centerd and to belong to a subclass belonging to the range of the infinitesimal generator. A functional law of the iterated logarithm is also proved. Both of these results use martingale central limit theorems.