CMU Campus
Center for                           Nonlinear Analysis
CNA Home People Seminars Publications Workshops and Conferences CNA Working Groups CNA Comments Form Summer Schools Summer Undergraduate Institute PIRE Cooperation Graduate Topics Courses SIAM Chapter Seminar Positions Contact
Publication 23-CNA-004

Speeding Up Langevin Dynamics By Mixing

Alexander Christie
Department of Mathematics
Penn State University
State College, PA

Yuanyuan Feng
School of Mathematical Sciences
Shanghai Key Laboratory of PMMP, East China
Normal University, Shanghai, 200241, P.R. China

Gautam Iyer
Department of Mathematical Sciences
Carnegie Mellon University
Pittsburgh, PA 15213

Alexei Novikov
Department of Mathematics
Pennsylvania State University
State College PA 16802

Abstract: We study an overdamped Langevin equation on the d-dimensional torus with stationary distribution proportional to p = e-U/k. When U has multiple wells the mixing time of the associated process is exponentially large (of size eO(1/k)). We add a drift to the Langevin dynamics (without changing the stationary distribution) and obtain quantitative estimates on the mixing time. We show that an exponentially mixing drift can be rescaled to make the mixing time of the Langevin system arbitrarily small. For numerical purposes, it is useful to keep the size of the imposed drift small, and we show that the smallest allowable rescaling ensures that the mixing time is O(d/k2), which is an order of magnitude smaller than eO(1/k).

We provide one construction of an exponentially mixing drift, although with rate constants whose k-dependence is unknown. Heuristics (from discrete time) suggest that k-dependence of the mixing rate is such that the imposed drift is of size O(d/k3). The large amplitude of the imposed drift increases the numerical complexity, and thus we expect this method will be most useful in the initial phase of Monte Carlo methods to rapidly explore the state space.

Get the paper in its entirety as  23-CNA-004.pdf

«   Back to CNA Publications