Convergence of the Riemannian Langevin Algorithm

Authors Khashayar Gatmiry, Santosh S. Vempala

Publication date 22 April 2022

Manifolds of metrics (especially Riemannian) (58D17)

Abstract: We study the Riemannian Langevin Algorithm for the problem of sampling from a distribution with density

u

with respect to the natural measure on a manifold with metric

g

. We assume that the target density satisfies a log-Sobolev inequality with respect to the metric and prove that the manifold generalization of the Unadjusted Langevin Algorithm converges rapidly to

u

for Hessian manifolds. This allows us to reduce the problem of sampling non-smooth (constrained) densities in

to sampling smooth densities over appropriate manifolds, while needing access only to the gradient of the log-density, and this, in turn, to sampling from the natural Brownian motion on the manifold. Our main analytic tools are (1) an extension of self-concordance to manifolds, and (2) a stochastic approach to bounding smoothness on manifolds. A special case of our approach is sampling isoperimetric densities restricted to polytopes by using the metric defined by the logarithmic barrier.

This page was built for publication: Convergence of the Riemannian Langevin Algorithm

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6397231)