Special Session 102: 

Concentration inequalities and performance guarantees for hypocoercive MCMC samplers

Luc Rey-Bellet
University of Massachusetts Amherst
USA
Co-Author(s):    Jeremiah Birell
Abstract:
We prove a concentration inequality for time averages for hypocoercive dynamics including Langevin equation, bouncy particle samplers, and Hamiltonian Monte-Carlo. We also discuss how to obtain robustness bounds on steady states of hypocoercive dynamics by combining information theory inequalities and concentration inequalities.