Measuring sample quality with diffusions
From MaRDI portal
Publication:2286455
Abstract: Stein's method for measuring convergence to a continuous target distribution relies on an operator characterizing the target and Stein factor bounds on the solutions of an associated differential equation. While such operators and bounds are readily available for a diversity of univariate targets, few multivariate targets have been analyzed. We introduce a new class of characterizing operators based on Ito diffusions and develop explicit multivariate Stein factor bounds for any target with a fast-coupling Ito diffusion. As example applications, we develop computable and convergence-determining diffusion Stein discrepancies for log-concave, heavy-tailed, and multimodal targets and use these quality measures to select the hyperparameters of biased Markov chain Monte Carlo (MCMC) samplers, compare random and deterministic quadrature rules, and quantify bias-variance tradeoffs in approximate MCMC. Our results establish a near-linear relationship between diffusion Stein discrepancies and Wasserstein distances, improving upon past work even for strongly log-concave targets. The exposed relationship between Stein factors and Markov process coupling may be of independent interest.
Recommendations
- An account of truthmaking
- Multivariate Stein factors for a class of strongly log-concave distributions
- Stein's density method for multivariate continuous distributions
- Convergence rates for a class of estimators based on Stein's method
- Langevin diffusions and the Metropolis-adjusted Langevin algorithm
Cites work
- A framework for computing the greedy spanner
- A multivariate CLT for decomposable random vectors with finite second moments
- Accelerating Gaussian diffusions
- Applications of Malliavin calculus to Monte Carlo methods in finance
- Bayesian and Non-Bayesian Analysis of the Regression Model with Multivariate Student-t Error Terms
- Bayesian data analysis.
- Bayesian Robust Multivariate Linear Regression With Incomplete Data
- Calculation of the Wasserstein Distance Between Probability Distributions on the Line
- Computing in operations research using Julia
- Conditional path sampling of SDEs and the Langevin MCMC method
- Consistency and fluctuations for stochastic gradient Langevin dynamics
- Control functionals for Monte Carlo integration
- Convection-diffusion equation with space-time ergodic random flow
- Convergence of numerical time-averaging and stationary measures via Poisson equations
- Correction to: Multivariate approximations in Wasserstein distance by Stein's method and Bismut's formula
- Curvature, concentration and error estimates for Markov chain Monte Carlo
- Dirichlet approximation of equilibrium distributions in Cannings models with mutation
- Erratum: ``Multivariate Stein factors for a class of strongly log-concave distributions
- Exploration of the (non-)asymptotic bias and variance of stochastic gradient Langevin dynamics
- Exponential convergence of Langevin distributions and their discrete approximations
- Fast Construction of Nets in Low-Dimensional Metrics and Their Applications
- Gibbs Sampler Convergence Criteria
- Graph spanners
- Handbook of Markov Chain Monte Carlo
- scientific article; zbMATH DE number 1713116 (Why is no real title available?)
- scientific article; zbMATH DE number 3951715 (Why is no real title available?)
- scientific article; zbMATH DE number 4080508 (Why is no real title available?)
- scientific article; zbMATH DE number 3505981 (Why is no real title available?)
- scientific article; zbMATH DE number 1229396 (Why is no real title available?)
- scientific article; zbMATH DE number 3438144 (Why is no real title available?)
- scientific article; zbMATH DE number 1448982 (Why is no real title available?)
- scientific article; zbMATH DE number 3215021 (Why is no real title available?)
- Integral Probability Metrics and Their Generating Classes of Functions
- Interpolation theory
- Irreversible Langevin samplers and variance reduction: a large deviations approach
- Kolmogorov equations in infinite dimensions: well-posedness and regularity of solutions, with applications to stochastic generalized Burgers equations
- Kolmogorov operators in spaces of continuous functions and equations for measures
- Langevin diffusions and Metropolis-Hastings algorithms
- Measuring sample quality with diffusions
- Multivariate linear regression with non-normal errors: a solution based on mixture models
- Multivariate normal approximation using exchangeable pairs
- Multivariate normal approximation using Stein's method and Malliavin calculus
- Multivariate normal approximation with Stein's method of exchangeable pairs under a general linearity condition
- Nonnormal approximation by Stein's method of exchangeable pairs with application to the Curie-Weiss model
- Normal Approximation by Stein’s Method
- On Stein's method for multivariate normal approximation
- On the Poisson equation and diffusion approximation. I
- On the rate of convergence in the multivariate CLT
- One-Parameter Semigroups for Linear Evolution Equations
- Rates of convergence in normal approximation under moment conditions via new bounds on solutions of the Stein equation
- Reflection couplings and contraction rates for diffusions
- Riemann manifold Langevin and Hamiltonian Monte Carlo methods. With discussion and authors' reply
- Robust Statistics
- Second order PDE's in finite and infinite dimension
- Semi log-concave Markov diffusions
- Spectral gaps for a Metropolis-Hastings algorithm in infinite dimensions
- Stein's method for comparison of univariate distributions
- Stein's method for diffusion approximations
- Stochastic differential equations. An introduction with applications.
- Stochastic processes and applications. Diffusion processes, the Fokker-Planck and Langevin equations
- Stochastic stability of differential equations. With contributions by G. N. Milstein and M. B. Nevelson
- The Fokker-Planck equation. Methods of solution and applications.
- The Whitney extension problem and Lipschitz selections of set-valued mappings in jet-spaces
- Time-reversible diffusions
- Variance reduction using nonreversible Langevin samplers
- Étude de quelques algèbres tayloriennes
Cited in
(36)- Convergence rates for a class of estimators based on Stein's method
- Geometry-informed irreversible perturbations for accelerated convergence of Langevin dynamics
- Performance analysis of greedy algorithms for minimising a maximum mean discrepancy
- Improved bounds for discretization of Langevin diffusions: near-optimal rates without convexity
- First-order covariance inequalities via Stein's method
- Stein's method for the Poisson-Dirichlet distribution and the Ewens sampling formula, with applications to Wright-Fisher models
- On Stein's method for multivariate self-decomposable laws
- Measuring sample quality with diffusions
- Multivariate approximations in Wasserstein distance by Stein's method and Bismut's formula
- On Stein's method for multivariate self-decomposable laws with finite first moment
- Nonasymptotic bounds for sampling algorithms without log-concavity
- A Riemann-Stein kernel method
- Stein's method meets computational statistics: a review of some recent developments
- An account of truthmaking
- Regulated quality diffusion revisited
- Multivariate Stein factors for a class of strongly log-concave distributions
- Stability estimates for invariant measures of diffusion processes, with applications to stability of moment measures and Stein kernels
- Stein’s method and the distribution of the product of zero mean correlated normal random variables
- The prelimit generator comparison approach of Stein's method
- Using coupling methods to estimate sample quality of stochastic differential equations
- scientific article; zbMATH DE number 7164690 (Why is no real title available?)
- Dirichlet approximation of equilibrium distributions in Cannings models with mutation
- Stochastic gradient Markov chain Monte Carlo
- Comments on: ``Tests for multivariate normality -- a critical review with emphasis on weighted \(L^2\)-statistics
- Convergence of Position-Dependent MALA with Application to Conditional Simulation in GLMMs
- Online MCMC Thinning with Kernelized Stein Discrepancy
- Scalable Control Variates for Monte Carlo Methods Via Stochastic Optimization
- Stein's density method for multivariate continuous distributions
- Transportation onto log-Lipschitz perturbations
- Bounding Wasserstein Distance with Couplings
- Some notes on quantitative generalized CLTs with self-decomposable limiting laws by spectral methods
- Controlling moments with kernel Stein discrepancies
- Gradient estimates for semigroups associated with stochastic differential equations driven by cylindrical Lévy processes
- Analysis of Langevin Monte Carlo from Poincaré to log-Sobolev
- Regularized Stein variational gradient flow
- Analysing quality with generalized kinetic methods
This page was built for publication: Measuring sample quality with diffusions
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2286455)
