Accelerating Langevin Sampling with Birth-death

Abstract

A fundamental problem in Bayesian inference and statistical machine learning is to efficiently sample from multimodal distributions. Due to metastability, multimodal distributions are difficult to sample using standard Markov chain Monte Carlo methods. We propose a new sampling algorithm based on a birth-death mechanism to accelerate the mixing of Langevin diffusion. Our algorithm is motivated by its mean field partial differential equation (PDE), which is a Fokker-Planck equation supplemented by a nonlocal birth-death term. This PDE can be viewed as a gradient flow of the Kullback-Leibler divergence with respect to the Wasserstein-Fisher-Rao metric. We prove that under some assumptions the asymptotic convergence rate of the nonlocal PDE is independent of the potential barrier, in contrast to the exponential dependence in the case of the Langevin diffusion. We illustrate the efficiency of the birth-death accelerated Langevin method through several analytical examples and numerical experiments.

References in corpus (12)

• Escaping free-energy minima
• An efficient, multiple range random walk algorithm to calculate the density of states
• Simulated Tempering: A New Monte Carlo Scheme
• Strong convergence of an explicit numerical method for SDEs with nonglobally Lipschitz continuous coefficients
• Nonequilibrium candidate Monte Carlo: A new tool for efficient equilibrium simulation
• Stein Variational Gradient Descent as Gradient Flow
• Log-concave sampling: Metropolis-Hastings algorithms are fast
• Non-asymptotic convergence analysis for the Unadjusted Langevin Algorithm
• Poincaré and logarithmic Sobolev inequalities by decomposition of the energy landscape
• Rapid Mixing of Hamiltonian Monte Carlo on Strongly Log-Concave Distributions
• A JKO splitting scheme for Kantorovich-Fisher-Rao gradient flows
• Global convergence of neuron birth-death dynamics

Cited by in corpus (9)

• Ensemble Kalman Inversion: mean-field limit and convergence analysis
• On the geometry of Stein variational gradient descent
• Ensemble Kalman Inversion for nonlinear problems: weights, consistency, and variance bounds
• Birth-death dynamics for sampling: Global convergence, approximations and their asymptotics
• Information Newton's flow: second-order optimization method in probability space
• Adaptive Monte Carlo augmented with normalizing flows
• Unbalanced Sobolev Descent
• Fast and Smooth Interpolation on Wasserstein Space
• A Wasserstein Minimum Velocity Approach to Learning Unnormalized Models