PRNANANov 16, 2013

Explicit error bounds for Markov chain Monte Carlo

arXiv:1108.3201120 citationsh-index: 17
AI Analysis

This work offers rigorous error guarantees for MCMC practitioners, but the results are incremental as they extend known asymptotic bounds to explicit forms.

The paper provides explicit, non-asymptotic error bounds for Markov chain Monte Carlo methods, showing that integration with respect to log-concave densities and over convex bodies is polynomially tractable using specific algorithms.

We prove explicit, i.e. non-asymptotic, error bounds for Markov chain Monte Carlo methods. The problem is to compute the expectation of a function f with respect to a measure π. Different convergence properties of Markov chains imply different error bounds. For uniformly ergodic and reversible Markov chains we prove a lower and an upper error bound with respect to the L2 -norm of f . If there exists an L2 -spectral gap, which is a weaker convergence property than uniform ergodicity, then we show an upper error bound with respect to the Lp -norm of f for p > 2. Usually a burn-in period is an efficient way to tune the algorithm. We provide and justify a recipe how to choose the burn-in period. The error bounds are applied to the problem of the integration with respect to a possibly unnormalized density. More precise, we consider the integration with respect to log-concave densities and the integration over convex bodies. By the use of the Metropolis algorithm based on a ball walk and the hit-and-run algorithm it is shown that both problems are polynomial tractable.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes