Roberts, Gareth O.; Tweedie, Richard L. Exponential convergence of Langevin distributions and their discrete approximations. (English) Zbl 0870.60027 Bernoulli 2, No. 4, 341-363 (1996). Summary: We consider a continuous-time method of approximating a given distribution \(\pi\) using the Langevin diffusion \(d{\mathbf L}_t= d{\mathbf W}_t+{1\over 2}\nabla\log\pi({\mathbf L}_t)dt\). We find conditions under which this diffusion converges exponentially quickly to \(\pi\) or does not: in one dimension, these are essentially that for distributions with exponential tails of the form \(\pi(x)\propto\exp(-\gamma|x|^\beta)\), \(0<\beta<\infty\), exponential convergence occurs if and only if \(\beta\geq 1\). We then consider conditions under which the discrete approximations to the diffusion converge. We first show that even when the diffusion itself converges, naive discretizations need not do so. We then consider a ‘Metropolis-adjusted’ version of the algorithm, and find conditions under which this also converges at an exponential rate: perhaps surprisingly, even the Metropolized version need not converge exponentially fast even if the diffusion does. We briefly discuss a truncated form of the algorithm which, in practice, should avoid the difficulties of the other forms. Cited in 2 ReviewsCited in 173 Documents MSC: 60F99 Limit theorems in probability theory 60J60 Diffusion processes 65C10 Random number generation in numerical analysis Keywords:diffusions; discrete approximations; geometric ergodicity; Hastings algorithms; irreducible Markov processes; Langevin models; Markov chain Monte Carlo; Metropolis algorithms; posterior distributions PDF BibTeX XML Cite \textit{G. O. Roberts} and \textit{R. L. Tweedie}, Bernoulli 2, No. 4, 341--363 (1996; Zbl 0870.60027) Full Text: DOI Link OpenURL