Abstract:Sampling a target probability distribution with an unknown normalization constant is a fundamental challenge in computational science and engineering. Recent work shows that algorithms derived by considering gradient flows in the space of probability measures open up new avenues for algorithm development. This paper makes three contributions to this sampling approach by scrutinizing the design components of such gradient flows. Any instantiation of a gradient flow for sampling needs an energy functional and a metric to determine the flow, as well as numerical approximations of the flow to derive algorithms. Our first contribution is to show that the Kullback-Leibler divergence, as an energy functional, has the unique property (among all f-divergences) that gradient flows resulting from it do not depend on the normalization constant of the target distribution. Our second contribution is to study the choice of metric from the perspective of invariance. The Fisher-Rao metric is known as the unique choice (up to scaling) that is diffeomorphism invariant. As a computationally tractable alternative, we introduce a relaxed, affine invariance property for the metrics and gradient flows. In particular, we construct various affine invariant Wasserstein and Stein gradient flows. Affine invariant gradient flows are shown to behave more favorably than their non-affine-invariant counterparts when sampling highly anisotropic distributions, in theory and by using particle methods. Our third contribution is to study, and develop efficient algorithms based on Gaussian approximations of the gradient flows; this leads to an alternative to particle methods. We establish connections between various Gaussian approximate gradient flows, discuss their relation to gradient methods arising from parametric variational inference, and study their convergence properties both theoretically and numerically.

Brownian Meanders, Importance Sampling and Unbiased Simulation of Diffusion Extremes

Simulation of an Integro-Differential Equation and Application in Estimation of Ruin Probability with Mixed Fractional Brownian Motion

Importance Sampling for the Empirical Measure of Weakly Interacting Diffusions

Anomalous Fluctuations of Extremes in Many-Particle Diffusion

A Geometric Perspective on Diffusion Models

The Brownian motion, Wiener process, and Pu migration as examples of random walk

Diffusion Simulation via Green Function Evaluation

On importance sampling and independent Metropolis-Hastings with an unbounded weight function

Randomly Shifted Lattice Rules with Importance Sampling and Applications

Unbiased likelihood estimation of Wright-Fisher diffusion processes

Non-parametric estimation of Stochastic Differential Equations from stationary time-series

Efficient Importance Sampling for Rare Event Simulation with Applications

Importance sampling for jump processes and applications to finance

Achieving High Convergence Rates by Quasi-Monte Carlo and Importance Sampling for Unbounded Integrands

Computing Functionals of Multidimensional Diffusions via Monte Carlo Methods

New algorithms for sampling and diffusion models

Sampling via Gradient Flows in the Space of Probability Measures

Fluctuations in Wasserstein dynamics on Graphs

Convergence of a Metropolized Integrator for Stochastic Differential Equations with Variable Diffusion Coefficient

Multilevel Monte Carlo method for jump-diffusion SDEs

High-Dimensional Simulation Optimization Via Brownian Fields and Sparse Grids