Stochastic optimization on continuous domains with finite-time guarantees by Markov chain Monte Carlo methods

TL;DR

This paper establishes finite-time performance bounds for Markov chain Monte Carlo algorithms in solving stochastic optimization problems over continuous spaces, providing theoretical guarantees and comparisons with existing methods.

Contribution

It introduces finite-time bounds for MCMC algorithms in continuous stochastic optimization and compares them with other state-of-the-art methods.

Findings

MCMC algorithms have quantifiable finite-time convergence guarantees.

The paper provides bounds that improve understanding of MCMC efficiency.

Comparative analysis shows advantages over some existing methods.

Abstract

We introduce bounds on the finite-time performance of Markov chain Monte Carlo algorithms in approaching the global solution of stochastic optimization problems over continuous domains. A comparison with other state-of-the-art methods having finite-time guarantees for solving stochastic programming problems is included.