Markov chain Monte Ca rlo methods have become standard tools to sample f rom complex high-dimensional probability measures. Many available techniques rely on discrete-time r eversible Markov chains whose transition kernels b uilt up over the Metropolis-Hastings algorithm. In our recent work\, we investigate an alternative a pproach\, the Bouncy Particle Sampler (BPS) where the target distribution of interest is explored us ing a continuous-time\, non reversible Markov proc ess. In this alternative approach\, a particle mov es along straight lines continuously around the sp ace and\, when facing a high energy barrier\, it i s not rejected but its path is modified by bouncin g against this barrier. The resulting non-reversib le Markov process provides a rejection-free Markov chain Monte Carlo sampling scheme. This method\, inspired from recent work in the molecular simulat ion literature\, is shown to be a valid\, efficien t sampling scheme applicable to a wide range of Ba yesian problems. We present several additional or iginal methodological extensions and establish var ious theoretical properties of these procedures. W e demonstrate experimentally the efficiency of the se algorithms on a variety of Bayesian inference p roblems. LOCATION:Seminar Room 1\, Newton Institute CONTACT:INI IT END:VEVENT END:VCALENDAR