Concentration of the Langevin Algorithm's Stationary Distribution

TL;DR

This paper establishes concentration properties for the stationary distribution of the Langevin Algorithm with discretization, showing it is sub-exponential or sub-Gaussian depending on convexity, with bounds that are essentially tight.

Contribution

It provides the first concentration results for the stationary distribution of the discretized Langevin Algorithm, extending classical properties from the continuous case.

Findings

or any positive stepsize ,  ext{ is sub-exponential if convex, sub-Gaussian if strongly convex.

ound that concentration bounds are essentially tight.

inite-iteration and inexact gradient estimates also satisfy these concentration bounds.

Abstract

A canonical algorithm for log-concave sampling is the Langevin Algorithm, aka the Langevin Diffusion run with some discretization stepsize $\eta > 0$. This discretization leads the Langevin Algorithm to have a stationary distribution $\pi_{\eta}$ which differs from the stationary distribution $\pi$ of the Langevin Diffusion, and it is an important challenge to understand whether the well-known properties of $\pi$ extend to $\pi_{\eta}$. In particular, while concentration properties such as isoperimetry and rapidly decaying tails are classically known for $\pi$, the analogous properties for $\pi_{\eta}$ are open questions with algorithmic implications. This note provides a first step in this direction by establishing concentration results for $\pi_{\eta}$ that mirror classical results for $\pi$. Specifically, we show that for any nontrivial stepsize $\eta > 0$, $\pi_{\eta}$ is sub-exponential (respectively, sub-Gaussian) when the potential is convex (respectively, strongly convex). Moreover, the concentration bounds we show are essentially tight. We also show that these concentration bounds extend to all iterates along the trajectory of the Langevin Algorithm, and to inexact implementations which use sub-Gaussian estimates of the gradient. Key to our analysis is the use of a rotation-invariant moment generating function (aka Bessel function) to study the stationary dynamics of the Langevin Algorithm. This technique may be of independent interest because it enables directly analyzing the discrete-time stationary distribution $\pi_{\eta}$ without going through the continuous-time stationary distribution $\pi$ as an intermediary.