Abstract:We consider the optimization problem of minimizing a functional defined over a family of probability distributions, where the objective functional is assumed to possess a variational form. Such a distributional optimization problem arises widely in machine learning and statistics, with Monte-Carlo sampling, variational inference, policy optimization, and generative adversarial network as examples. For this problem, we propose a novel particle-based algorithm, dubbed as variational transport, which approximately performs Wasserstein gradient descent over the manifold of probability distributions via iteratively pushing a set of particles. Specifically, we prove that moving along the geodesic in the direction of functional gradient with respect to the second-order Wasserstein distance is equivalent to applying a pushforward mapping to a probability distribution, which can be approximated accurately by pushing a set of particles. Specifically, in each iteration of variational transport, we first solve the variational problem associated with the objective functional using the particles, whose solution yields the Wasserstein gradient direction. Then we update the current distribution by pushing each particle along the direction specified by such a solution. By characterizing both the statistical error incurred in estimating the Wasserstein gradient and the progress of the optimization algorithm, we prove that when the objective function satisfies a functional version of the Polyak-Łojasiewicz (PL) (Polyak, 1963) and smoothness conditions, variational transport converges linearly to the global minimum of the objective functional up to a certain statistical error, which decays to zero sublinearly as the number of particles goes to infinity.

Variational interacting particle systems and Vlasov equations

Geometric particle-in-cell methods for Vlasov--Poisson equations with Maxwell--Boltzmann electrons

Variational problems related to some fractional kinetic equations

Existence of a minimizer to the particle number-Casimir functional for the Einstein-Vlasov system

A variational perspective on the dissipative Hamiltonian structure of the Vlasov-Fokker-Planck equation

Minimizers of constrained functionals involving a point interaction

Variational Discretizations for Hamiltonian Systems

Variational Transport: A Convergent Particle-BasedAlgorithm for Distributional Optimization

Dynamics And Variational Integrators Of Stochastic Hamiltonian Systems

Generalized Variational Principles, Global Weak Solutions and Behavior with Random Initial Data for Systems of Conservation Laws Arising in Adhesion Particle Dynamics

Gradient-based Monte Carlo methods for relaxation approximations of hyperbolic conservation laws

Variational Time Integration Approach for Smoothed Particle Hydrodynamics Simulation of Fluids

A Variational Perspective on High-Resolution ODEs

Variational Principles for Hamiltonian Systems

A variational approach to nonlinear and interacting diffusions

A space-time variational formulation for the many-body electronic Schr{ö}dinger evolution equation

Geometric integration of the Vlasov-Maxwell system with a variational particle-in-cell scheme

Uniform Approximation of 2D Navier-Stokes Equations with Vorticity Creation by Stochastic Interacting Particle Systems

Stochastic viscosity approximations of Hamilton-Jacobi equations and variance reduction

A variational approach to a fuzzy Boltzmann equation

Variational formulation of macro-particle plasma simulation algorithms