Information-geometric Markov chain Monte Carlo methods using diffusions

DOI10.3390/E16063074zbMATH Open1338.65018arXiv1403.7957OpenAlexW3102676823MaRDI QIDQ296467FDOQ296467

Authors: Samuel Livingstone, M. Girolami 

Publication date: 15 June 2016

Published in: Entropy (Search for Journal in Brave)

Abstract: Recent work incorporating geometric ideas in Markov chain Monte Carlo is reviewed in order to highlight these advances and their possible application in a range of domains beyond Statistics. A full exposition of Markov chains and their use in Monte Carlo simulation for Statistical inference and molecular dynamics is provided, with particular emphasis on methods based on Langevin diffusions. After this geometric concepts in Markov chain Monte Carlo are introduced. A full derivation of the Langevin diffusion on a Riemannian manifold is given, together with a discussion of appropriate Riemannian metric choice for different problems. A survey of applications is provided, and some open questions are discussed.

Full work available at URL: https://arxiv.org/abs/1403.7957

Recommendations

• Riemann manifold Langevin and Hamiltonian Monte Carlo methods. With discussion and authors' reply
• Geodesic Monte Carlo on embedded manifolds
• Markov chain Monte Carlo and numerical differential equations
• Langevin diffusions and the Metropolis-adjusted Langevin algorithm
• Zero variance differential geometric Markov chain Monte Carlo algorithms

zbMATH Keywords

Markov chain Monte CarloBayesian inferencemachine learningcomputational statisticsinformation geometrystatistical mechanicsdiffusions

Mathematics Subject Classification ID

Computational methods in Markov chains (60J22) Statistical aspects of information-theoretic topics (62B10) Directional data; spatial statistics (62H11) Numerical analysis or methods applied to Markov chains (65C40) Applications of Brownian motions and diffusion theory (population genetics, absorption problems, etc.) (60J70) Discrete-time Markov processes on general state spaces (60J05) Diffusion processes and stochastic analysis on manifolds (58J65)

Cites Work

• Computing the nearest correlation matrix--a problem from finance
• Inference from iterative simulation using multiple sequences
• Exponential convergence of Langevin distributions and their discrete approximations
• Markov chains for exploring posterior distributions. (With discussion)
• Geodesic Monte Carlo on embedded manifolds
• Title not available (Why is that?)
• Title not available (Why is that?)
• On Choosing and Bounding Probability Metrics
• Langevin diffusions and Metropolis-Hastings algorithms
• Optimal scaling for various Metropolis-Hastings algorithms.
• Geometric ergodicity of Metropolis algorithms
• Stochastic analysis on manifolds
• Geometric convergence and central limit theorems for multidimensional Hastings and Metropolis algorithms
• Time-reversible diffusions
• On the Markov chain central limit theorem
• Central limit theorem for additive functionals of reversible Markov processes and applications to simple exclusions
• Necessary conditions for geometric and polynomial ergodicity of random-walk-type Markov chains
• Rates of convergence of the Hastings and Metropolis algorithms
• An invariant form for the prior probability in estimation problems
• The geometry of Hessian structures
• Markov chain Monte Carlo inference for Markov jump processes via the linear noise approximation
• Stability of Markovian processes III: Foster–Lyapunov criteria for continuous-time processes
• On the local geometry of mixture models
• The imbedding problem for Riemannian manifolds
• Preferred point geometry and statistical manifolds
• The Role of Differential Geometry in Statistical Theory
• Title not available (Why is that?)
• Title not available (Why is that?)
• Applications of Differential Geometry to Econometrics
• A stochastic Newton MCMC method for large-scale statistical inverse problems with application to seismic inversion
• MCMC methods for functions: modifying old algorithms to make them faster
• Honest exploration of intractable probability distributions via Markov chain Monte Carlo.
• Advanced MCMC methods for sampling on diffusion pathspace
• Proposals which speed up function-space MCMC
• Langevin diffusions and the Metropolis-adjusted Langevin algorithm
• Title not available (Why is that?)
• Optimal scaling of the random walk Metropolis: general criteria for the 0.234 acceptance rule
• Geometric ergodicity of Metropolis-Hastings algorithms for conditional simulation in generalized linear mixed models
• Optimal scaling of the random walk Metropolis on elliptically symmetric unimodal targets
• The random walk Metropolis: linking theory and practice through a case study
• Optimal scaling for random walk Metropolis on spherically constrained target densities

Cited In (15)

• Zero variance differential geometric Markov chain Monte Carlo algorithms
• Markov Kernels Local Aggregation for Noise Vanishing Distribution Sampling
• Monte Carlo Information-Geometric Structures
• Adaptive step size selection for Hessian-based manifold Langevin samplers
• On the geometric ergodicity of Hamiltonian Monte Carlo
• Asymptotic analysis of the random walk metropolis algorithm on ridged densities
• A fresh Take on ‘Barker Dynamics’ for MCMC
• Geodesic Monte Carlo on embedded manifolds
• Computing with Fisher geodesics and extended exponential families
• Geometric Integration of Measure-Preserving Flows for Sampling
• On the Conditional Mutual Information in the Gaussian–Markov Structured Grids
• Affine invariant interacting Langevin dynamics for Bayesian inference
• Convergence of Position-Dependent MALA with Application to Conditional Simulation in GLMMs
• Geometric adaptive Monte Carlo in random environment
• Geometry-informed irreversible perturbations for accelerated convergence of Langevin dynamics

Uses Software

• R
• CODA

This page was built for publication: Information-geometric Markov chain Monte Carlo methods using diffusions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q296467)