The forward-backward envelope for sampling with the overdamped Langevin algorithm
From MaRDI portal
Abstract: In this paper, we analyse a proximal method based on the idea of forward-backward splitting for sampling from distributions with densities that are not necessarily smooth. In particular, we study the non-asymptotic properties of the Euler-Maruyama discretization of the Langevin equation, where the forward-backward envelope is used to deal with the non-smooth part of the dynamics. An advantage of this envelope, when compared to widely-used Moreu-Yoshida one and the MYULA algorithm, is that it maintains the MAP estimator of the original non-smooth distribution. We also study a number of numerical experiments that corroborate that support our theoretical findings.
Recommendations
- Using perturbed underdamped Langevin dynamics to efficiently sample from probability distributions
- Interacting Langevin diffusions: gradient structure and ensemble Kalman sampler
- Weak backward error analysis for overdamped Langevin processes
- Unadjusted Langevin algorithm for sampling a mixture of weakly smooth potentials
- Irreversible Langevin samplers and variance reduction: a large deviations approach
- Sampling from non-smooth distributions through Langevin diffusion
- On Irreversible Metropolis Sampling Related to Langevin Dynamics
- Conditional path sampling of SDEs and the Langevin MCMC method
- Langevin diffusions and the Metropolis-adjusted Langevin algorithm
- Advanced MCMC methods for sampling on diffusion pathspace
Cites work
- An algorithm for total variation minimization and applications
- Analysis of Langevin Monte Carlo via convex optimization
- Computing ergodic limits for Langevin equations
- Convergence of numerical time-averaging and stationary measures via Poisson equations
- Convex optimization theory.
- Coupling and convergence for Hamiltonian Monte Carlo
- Efficient Bayesian computation by proximal Markov chain Monte Carlo: when Langevin meets Moreau
- Exponential convergence of Langevin distributions and their discrete approximations
- First-order methods in optimization
- Forward-backward quasi-Newton methods for nonsmooth optimization problems
- Handbook of Markov Chain Monte Carlo
- High order numerical approximation of the invariant measure of ergodic SDEs
- High-dimensional Bayesian inference via the unadjusted Langevin algorithm
- scientific article; zbMATH DE number 1113187 (Why is no real title available?)
- Nonasymptotic bounds for sampling algorithms without log-concavity
- Nonasymptotic convergence analysis for the unadjusted Langevin algorithm
- Nonlinear total variation based noise removal algorithms
- On weighted parallel volumes
- Optimal Transport
- Proximal Markov chain Monte Carlo algorithms
- Quantitative contraction rates for Markov chains on general state spaces
- Sampling from a log-concave distribution with projected Langevin Monte Carlo
- Sampling from non-smooth distributions through Langevin diffusion
- Smooth minimization of non-smooth functions
- Theoretical Guarantees for Approximate Sampling from Smooth and Log-Concave Densities
- User-friendly guarantees for the Langevin Monte Carlo with inaccurate gradient
Cited in
(4)
This page was built for publication: The forward-backward envelope for sampling with the overdamped Langevin algorithm
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6173566)
