High-Order Langevin Monte Carlo Algorithms

Abstract

Langevin algorithms are popular Markov chain Monte Carlo (MCMC) methods for large-scale sampling problems that often arise in data science. We propose Monte Carlo algorithms based on the discretizations of $P$-th order Langevin dynamics for any $P\ge 3$. Our design of $P$-th order Langevin Monte Carlo (LMC) algorithms is by combining splitting and accurate integration methods. We obtain Wasserstein convergence guarantees for sampling from distributions with log-concave and smooth densities. Specifically, the mixing time of the $P$-th order LMC algorithm scales as $O\left(d^{\frac{1}{R}}/\hat{I}{\mu }^{\frac{1}{2R}}\right)$ for $R=4\cdot 1_{\{P=3\}}+(2P-1)\cdot 1_{\{P\ge 4\}}$, which has a better dependence on the dimension $d$ and the accuracy level $\hat{I}\mu$ as $P$ grows. Numerical experiments illustrate the efficiency of our proposed algorithms.

73 pages, 3 figures, 1 table