Improving the convergence of reversible samplers
From MaRDI portal
Publication:330615
Abstract: In Monte-Carlo methods the Markov processes used to sample a given target distribution usually satisfy detailed balance, i.e. they are time-reversible. However, relatively recent results have demonstrated that appropriate reversible and irreversible perturbations can accelerate convergence to equilibrium. In this paper we present some general design principles which apply to general Markov processes. Working with the generator of Markov processes, we prove that for some of the most commonly used performance criteria, i.e., spectral gap, asymptotic variance and large deviation functionals, sampling is improved for appropriate reversible and irreversible perturbations of some initially given reversible sampler. Moreover we provide specific constructions for such reversible and irreversible perturbations for various commonly used Markov processes, such as Markov chains and diffusions. In the case of diffusions, we make the discussion more specific using the large deviations rate function as a measure of performance.
Recommendations
Cites work
- scientific article; zbMATH DE number 3127846 (Why is no real title available?)
- scientific article; zbMATH DE number 472933 (Why is no real title available?)
- scientific article; zbMATH DE number 1560252 (Why is no real title available?)
- A note on Metropolis-Hastings kernels for general state spaces
- Accelerating Gaussian diffusions
- Accelerating diffusions
- Accelerating reversible Markov chains
- An extension of Peskun and Tierney orderings to continuous time Markov chains
- Analysis of a nonreversible Markov chain sampler.
- Asymptotic evaluation of certain markov process expectations for large time, I
- Efficiency of finite state space Monte Carlo Markov chains
- General state space Markov chains and MCMC algorithms
- Irreversible Langevin samplers and variance reduction: a large deviations approach
- Irreversible Monte Carlo algorithms for efficient sampling
- Large deviations
- Markov Chains and Stochastic Stability
- Monotonicity of the dynamical activity
- Non-reversible Metropolis-Hastings
- On Large Deviations from the Invariant Measure
- On the convergence of the Markov chain simulation method
- On the infinite swapping limit for parallel tempering
- On the large deviation rate function for the empirical measures of reversible jump Markov processes
- On the spectral analysis of second-order Markov chains
- Optimal non-reversible linear drift for the convergence to equilibrium of a diffusion
- Optimal spectral structure of reversible stochastic matrices, Monte Carlo methods and the simulation of Markov random fields
- Optimum Monte-Carlo sampling using Markov chains
- Ordering and improving the performance of Monte Carlo Markov chains.
- Rates of convergence of the Hastings and Metropolis algorithms
- Riemann manifold Langevin and Hamiltonian Monte Carlo methods. With discussion and authors' reply
- Stochastic simulation: Algorithms and analysis
- The behavior of the spectral gap under growing drift
- Variance reduction for diffusions
- Variance reduction for irreversible Langevin samplers and diffusion on graphs
- Variance reduction using nonreversible Langevin samplers
Cited in
(24)- Large deviations for the skew-detailed-balance lifted-Markov processes to sample the equilibrium distribution of the Curie–Weiss model
- Gibbsian stationary non-equilibrium states
- Irreversible Monte Carlo algorithms for efficient sampling
- Optimal variance reduction for Markov chain Monte Carlo
- Is there an analog of Nesterov acceleration for gradient-based MCMC?
- Non-reversible metastable diffusions with Gibbs invariant measure. I: Eyring-Kramers formula
- Acceleration of convergence to equilibrium in Markov chains by breaking detailed balance
- Peskun-Tierney ordering for Markovian Monte Carlo: beyond the reversible scenario
- Characterizing limits and opportunities in speeding up Markov chain mixing
- Analysis of multiscale integrators for multiple attractors and irreversible Langevin samplers
- Geometry-informed irreversible perturbations for accelerated convergence of Langevin dynamics
- Irreversible Langevin samplers and variance reduction: a large deviations approach
- Accelerated convergence to equilibrium and reduced asymptotic variance for Langevin dynamics using Stratonovich perturbations
- Limit theorems for the zig-zag process
- Non-reversible metastable diffusions with Gibbs invariant measure. II: Markov chain convergence
- Optimal non-reversible linear drift for the convergence to equilibrium of a diffusion
- Large deviations of empirical measures of diffusions in weighted topologies
- Irreversible samplers from jump and continuous Markov processes
- Constructing sampling schemes via coupling: Markov semigroups and optimal transport
- On Irreversible Metropolis Sampling Related to Langevin Dynamics
- Nonreversible sampling schemes on submanifolds
- Variance reduction for irreversible Langevin samplers and diffusion on graphs
- A large deviation principle for the empirical measures of Metropolis-Hastings chains
- On multiple acceleration of reversible Markov chain
This page was built for publication: Improving the convergence of reversible samplers
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q330615)
