Normalizing constants of log-concave densities

DOI10.1214/18-EJS1411MaRDI QIDQ1746544zbMATH OpenOpenAlexWikidataFDO

Authors Nicolas Brosse, Alain Durmus, Eric Moulines

Publication date 25 April 2018

Published in Electronic Journal of Statistics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1707.00460, https://projecteuclid.org/euclid.ejs/1520240451

zbMATH Keywords

Bayes factornormalizing constantsannealed importance samplingunadjusted Langevin algorithm

Mathematics Subject Classification ID

Monte Carlo methods (65C05) Numerical analysis or methods applied to Markov chains (65C40) Sequential statistical analysis (62L10) Discrete-time Markov processes on general state spaces (60J05) (L^p)-limit theorems (60F25)

Abstract: We derive explicit bounds for the computation of normalizing constants $Z$ for log-concave densities $pi=exp(-U)/Z$ with respect to the Lebesgue measure on $mathbb{R}^d$. Our approach relies on a Gaussian annealing combined with recent and precise bounds on the Unadjusted Langevin Algorithm (High-dimensional Bayesian inference via the Unadjusted Langevin Algorithm, A. Durmus and E. Moulines). Polynomial bounds in the dimension $d$ are obtained with an exponent that depends on the assumptions made on $U$. The algorithm also provides a theoretically grounded choice of the annealing sequence of variances. A numerical experiment supports our findings. Results of independent interest on the mean squared error of the empirical average of locally Lipschitz functions are established.

Recommendations

• Theoretical Guarantees for Approximate Sampling from Smooth and Log-Concave Densities
• The geometry of logconcave functions and sampling algorithms
• Higher order Langevin Monte Carlo algorithm
• Sampling from a log-concave distribution with projected Langevin Monte Carlo
• Log-concave sampling: Metropolis-Hastings algorithms are fast

Cites work

• scientific article; zbMATH DE number 3144053 (Why is no real title available?)
• scientific article; zbMATH DE number 18983 (Why is no real title available?)
• scientific article; zbMATH DE number 1085980 (Why is no real title available?)
• scientific article; zbMATH DE number 2106098 (Why is no real title available?)
• A comparative study of Monte Carlo methods for efficient evaluation of marginal likelihood
• Approximation algorithms for the normalizing constant of Gibbs distributions
• Bayes model selection with path sampling: factor models and other examples
• Bypassing KLS: Gaussian cooling and an \(O^\ast(n^3)\) volume algorithm
• Concentration inequalities. A nonasymptotic theory of independence
• Curvature, concentration and error estimates for Markov chain Monte Carlo
• Error bounds and normalising constants for sequential Monte Carlo samplers in high dimensions
• Exponential convergence of Langevin distributions and their discrete approximations
• Fixed Precision MCMC Estimation by Median of Products of Averages
• Free energy computations. A mathematical perspective
• From microphysics to macrophysics. Methods and applications of statistical physics. II. Translated from the French by D. ter Haar.
• Geometry of isotropic convex bodies
• High-dimensional Bayesian inference via the unadjusted Langevin algorithm
• Improving power posterior estimation of statistical evidence
• Introductory lectures on convex optimization. A basic course.
• Markov chain Monte Carlo methods and the label switching problem in Bayesian mixture modeling
• Maximum-a-posteriori estimation with Bayesian confidence regions
• Measure theory and fine properties of functions
• Monte Carlo methods in Bayesian computation
• Multilevel Sequential Monte Carlo Samplers for Normalizing Constants
• Multilevel particle filters: normalizing constant estimation
• Nonasymptotic convergence analysis for the unadjusted Langevin algorithm
• Normalizing constants of log-concave densities
• Optimal Transport
• Random generation of combinatorial structures from a uniform distribution
• Sequential Monte Carlo Samplers
• Simulating normalizing constants: From importance sampling to bridge sampling to path sampling
• Stability of Markovian processes III: Foster–Lyapunov criteria for continuous-time processes
• Stochastic comparisons of order statistics from heterogeneous populations, with applications in reliability
• The Bayesian Choice
• Theoretical Guarantees for Approximate Sampling from Smooth and Log-Concave Densities
• Tuning tempered transitions

Cited in

(9)

• Fourier transform MCMC, heavy-tailed distributions, and geometric ergodicity
• An Invitation to Sequential Monte Carlo Samplers
• Normalizing constants of log-concave densities
• Hamiltonian Monte Carlo with energy conserving subsampling
• Multilevel estimation of normalization constants using ensemble Kalman-Bucy filters
• User-friendly guarantees for the Langevin Monte Carlo with inaccurate gradient
• Normalizing constants of a distribution which belongs to the domain of attraction of the Gumbel distribution
• Geometry-informed irreversible perturbations for accelerated convergence of Langevin dynamics
• Unadjusted Langevin algorithm for sampling a mixture of weakly smooth potentials

Describes a project that uses

Uses Software

• R

This page was built for publication: Normalizing constants of log-concave densities

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1746544)