Papers
Topics
Authors
Recent
Search
2000 character limit reached

The Randomized Midpoint Method for Log-Concave Sampling

Published 12 Sep 2019 in cs.LG, cs.DS, math.PR, and stat.ML | (1909.05503v1)

Abstract: Sampling from log-concave distributions is a well researched problem that has many applications in statistics and machine learning. We study the distributions of the form p∗∝exp⁡(−f(x))p^{*}\propto\exp(-f(x)), where f:Rd→Rf:\mathbb{R}^{d}\rightarrow\mathbb{R} has an LL-Lipschitz gradient and is mm-strongly convex. In our paper, we propose a Markov chain Monte Carlo (MCMC) algorithm based on the underdamped Langevin diffusion (ULD). It can achieve ϵ⋅D\epsilon\cdot D error (in 2-Wasserstein distance) in O~(κ7/6/ϵ1/3+κ/ϵ2/3)\tilde{O}\left(\kappa^{7/6}/\epsilon^{1/3}+\kappa/\epsilon^{2/3}\right) steps, where D=defdmD\overset{\mathrm{def}}{=}\sqrt{\frac{d}{m}} is the effective diameter of the problem and κ=defLm\kappa\overset{\mathrm{def}}{=}\frac{L}{m} is the condition number. Our algorithm performs significantly faster than the previously best known algorithm for solving this problem, which requires O~(κ1.5/ϵ)\tilde{O}\left(\kappa^{1.5}/\epsilon\right) steps. Moreover, our algorithm can be easily parallelized to require only O(κlog⁡1ϵ)O(\kappa\log\frac{1}{\epsilon}) parallel steps. To solve the sampling problem, we propose a new framework to discretize stochastic differential equations. We apply this framework to discretize and simulate ULD, which converges to the target distribution p∗p^{*}. The framework can be used to solve not only the log-concave sampling problem, but any problem that involves simulating (stochastic) differential equations.

Authors (2)
Citations (109)

Summary

No one has generated a summary of this paper yet.

Paper to Video (Beta)

No one has generated a video about this paper yet.

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Continue Learning

We haven't generated follow-up questions for this paper yet.

Tweets

Sign up for free to view the 1 tweet with 0 likes about this paper.