Geometric analysis for the Metropolis algorithm on Lipschitz domains

DOI10.1007/S00222-010-0303-6MaRDI QIDQ636824zbMATH OpenOpenAlexFDO

Authors Persi Diaconis, Gilles Lebeau, Laurent Michel

Publication date 30 August 2011

Published in Inventiones Mathematicae (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1007/s00222-010-0303-6

zbMATH Keywords

geometric analysisLipschitz domainsMetropolis algorithmMetropolis operator

Mathematics Subject Classification ID

Discrete-time Markov processes on general state spaces (60J05) Computational methods in Markov chains (60J22) Numerical analysis or methods applied to Markov chains (65C40)

Review text

The authors consider the convergence of the Metropolis algorithm to stationary distributions on bounded, connected Lipschitz domains in Euclidean space. To this end they employ a geometrical analysis and present tools that can be used to prove the convergence of the algorithm. The authors derive bounds on the rates of convergence of the Metropolis algorithm in terms of the spectral gap of the associated Metropolis operator. The behaviour of the spectral gap is analysed in more detail; a comparison, and Nash and Sobolev inequalities are presented. The general results are applied to the original application of the Metropolis algorithm: The random, non-overlapping placement in the \(d\)-dimensional unit square of \(N\) \(d\)-dimensional spheres with fixed radius \(\varepsilon\) such that \(N\varepsilon\) is sufficiently small. The resulting stationary distribution is the uniform distribution on the set of centres of all possible sphere placements. It is shown that the state space is a Lipschitz domain and that the general results hold in this example.

Recommendations

• Micro-local analysis for the Metropolis algorithm
• Metropolis
• Gibbs/Metropolis algorithms on a convex polytope
• On the rate of convergence of the Metropolis algorithm and Gibbs sampler by geometric bounds
• scientific article; zbMATH DE number 1263187

Cites work

• Comparison techniques for random walk on finite groups
• scientific article; zbMATH DE number 3622441 (Why is no real title available?)
• scientific article; zbMATH DE number 1069282 (Why is no real title available?)
• scientific article; zbMATH DE number 2208228 (Why is no real title available?)
• scientific article; zbMATH DE number 3259834 (Why is no real title available?)
• Micro-local analysis for the Metropolis algorithm
• Nash inequalities for finite Markov chains
• Random close packing of granular matter
• Semi-classical analysis of a random walk on a manifold
• Statistical mechanics: Algorithms and computations. With CD-ROM.

Cited in

(33)

• Micro-local analysis for the Metropolis algorithm
• Sparse locally-jammed disk packings
• Collective proposal distributions for nonlinear MCMC samplers: mean-field theory and fast implementation
• Sparse hard-disk packings and local Markov chains
• Correlation decay for hard spheres via Markov chains
• Another conversation with Persi Diaconis
• Long time behavior of stochastic hard ball systems
• PDMP characterisation of event-chain Monte Carlo algorithms for particle systems
• Birthday inequalities, repulsion, and hard spheres
• Asymptotic behaviors for nonlocal diffusion equations about the dispersal spread
• Metropolis
• The Markov chain Monte Carlo revolution
• Some things we've learned (about Markov chain Monte Carlo)
• scientific article; zbMATH DE number 1263187 (Why is no real title available?)
• Conductance bounds on the L² convergence rate of Metropolis algorithms on unbounded state spaces
• Gibbs/Metropolis algorithms on a convex polytope
• On the hard sphere model and sphere packings in high dimensions
• Spectral asymptotics for the Metropolis algorithm on singular domains
• Ornstein-Uhlenbeck pinball and the Poincaré inequality in a punctured domain
• Rates of decay in the classical Katznelson-Tzafriri theorem
• Spectral analysis of hypoelliptic random walks
• Random walk on surfaces with hyperbolic cusps
• Spectral gaps and error estimates for infinite-dimensional Metropolis-Hastings with non-Gaussian priors
• Multidimensional examples of the Metropolis algorithm
• Singular relaxation of a random walk in a box with a Metropolis Monte Carlo dynamics
• Harnack inequalities and Gaussian estimates for random walks on metric measure spaces
• The mathematics of mixing things up
• Archimedes' principle for Brownian liquid
• Metropolis Monte Carlo sampling: convergence, localization transition and optimality
• Poissonian pair correlations for dependent random variables
• Spectral analysis of a semiclassical random walk associated to a general confining potential
• Sampling from p-adic algebraic manifolds
• Convergence of Metropolis-type algorithms for a large canonical ensemble

This page was built for publication: Geometric analysis for the Metropolis algorithm on Lipschitz domains

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q636824)