Sharp non-asymptotic concentration inequalities for the approximation of the invariant distribution of a diffusion |
| |
Institution: | 1. Laboratoire Jacques-Louis Lions, Sorbonne Universités, Paris, France;2. Faculty of Mathematics, Ruhr University Bochum, Germany;3. Institut de Recherche Mathématiques de Rennes, Université de Rennes 1, Rennes, France;4. Mathematics Department, Université de Liège, Liège, Belgium |
| |
Abstract: | Let be an ergodic diffusion with invariant distribution . Consider the empirical measure where is an Euler scheme with decreasing steps which approximates . Given a test function , we obtain sharp concentration inequalities for which improve the results in Honoré et al. (2019). Our hypotheses on the test function cover many real applications: either is supposed to be a coboundary of the infinitesimal generator of the diffusion, or is supposed to be Lipschitz. |
| |
Keywords: | Invariant distribution Diffusion processes Inhomogeneous Markov chains Non-asymptotic concentration inequalities |
本文献已被 ScienceDirect 等数据库收录! |
|