Large deviation results for a class of Markov chains arising from population genetics (Q1825508): Difference between revisions

Consider stationary Markov chains \((X^ N_ n)_{n\in {\mathbb{N}}}\) on a bounded subset of \({\mathbb{R}}^ d\) depending on a parameter \(N\in {\mathbb{N}}\) with \(E_ x(X^ N_ 1)=:f_ N(x)\). Let \(x_ 0\) be a common stable fixed point of the functions \(f_ N\) and D an open neighborhood of \(x_ 0\). Assume that N \(cov_ x(X^ N_ 1)\) is bounded away from 0 and \(\infty\) uniformly in \(N\in {\mathbb{N}}\) and \(x\in D.\) Under various regularity assumptions it is proved that the asymptotics (N\(\to \infty)\) of the mean exit times from D and the invariant measures are governed by a quasi-potential as in {\S} 4.4 of \textit{M. I. Frejdlin} and \textit{A. D. Venttsel'}, Random perturbations of dynamical systems (1984; Zbl 0522.60055). The results apply to the Wright-Fisher model in populations genetics.