Variance reduction for diffusions
From MaRDI portal
Abstract: The most common way to sample from a probability distribution is to use Monte-Carlo methods. For distributions on a continuous state space, one can find diffusions with the target distribution as equilibrium measure, so that the state of the diffusion after a long time provides a good sample from the desired distribution. There exist many diffusions with a common equilibrium, and one would naturally like to choose those that make the convergence to equilibrium faster. One way to do this is to consider a reversible diffusion, and add to it an antisymmetric drift which preserves the invariant measure. It has been proven that, in general, the irreversible algorithm performs better than the reversible one, in that the spectral gap is larger. In the present work, asymptotic variance is used as the criterion to compare these algorithms. We first provide a general comparison result, and then apply it to the specific cases of a diffusion on , or on a compact Riemannian manifold. We prove that, in general, adding an antisymmetric drift to a reversible diffusion reduces the asymptotic variance. We also provide some extensions of this result concerning strict inequality, the worst-case analysis, and the behavior of the asymptotic variance when the drift goes to infinity.
Recommendations
- Variance reduction using nonreversible Langevin samplers
- Irreversible Langevin samplers and variance reduction: a large deviations approach
- A note on the asymptotic variance of drift accelerated diffusions
- Variance Reduction for Simulated Diffusions
- Optimal variance reduction for Markov chain Monte Carlo
Cites work
- scientific article; zbMATH DE number 3719745 (Why is no real title available?)
- scientific article; zbMATH DE number 43057 (Why is no real title available?)
- scientific article; zbMATH DE number 3483022 (Why is no real title available?)
- scientific article; zbMATH DE number 3622441 (Why is no real title available?)
- scientific article; zbMATH DE number 739283 (Why is no real title available?)
- scientific article; zbMATH DE number 1515832 (Why is no real title available?)
- scientific article; zbMATH DE number 1560252 (Why is no real title available?)
- scientific article; zbMATH DE number 876688 (Why is no real title available?)
- scientific article; zbMATH DE number 3244852 (Why is no real title available?)
- Accelerating Brownian motion on \(N\)-torus
- Accelerating Gaussian diffusions
- Accelerating diffusions
- Accelerating reversible Markov chains
- Efficiency of finite state space Monte Carlo Markov chains
- Fluctuations in Markov processes. Time symmetry and martingale approximation.
- Heat kernel and analysis on manifolds
- Irreversible Langevin samplers and variance reduction: a large deviations approach
- Monte Carlo sampling methods using Markov chains and their applications
- On the Ornstein-Uhlenbeck operator in 𝐿² spaces with respect to invariant measures
- On the functional central limit theorem and the law of the iterated logarithm for Markov processes
- On the optimal transition matrix for Markov chain Monte Carlo sampling
- 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.
- The behavior of the spectral gap under growing drift
- \(L^p\)-regularity for elliptic operators with unbounded coefficients
Cited in
(24)- Optimal variance reduction for Markov chain Monte Carlo
- A simple variance reduction method with applications to finance and queueing theory
- Acceleration of convergence to equilibrium in Markov chains by breaking detailed balance
- Variance reduction for additive functionals of Markov chains via martingale representations
- Reducing the variance by smoothing
- Robust Variance Reduction for Random Walk Methods
- Accelerating planar Ornstein-Uhlenbeck diffusion with suitable drift
- Variance reduction by smoothing revisited
- Irreversible Langevin samplers and variance reduction: a large deviations approach
- Variational formulas for asymptotic variance of general discrete-time Markov chains
- Large deviations of empirical measures of diffusions in weighted topologies
- A note on the asymptotic variance of drift accelerated diffusions
- Variance reduction for discretised diffusions via regression
- Some remark on the asymptotic variance in a drift accelerated diffusion
- Accelerating Gaussian diffusions
- Improving the convergence of reversible samplers
- Variance reduction for irreversible Langevin samplers and diffusion on graphs
- Variational principles for asymptotic variance of general Markov processes
- Using perturbed underdamped Langevin dynamics to efficiently sample from probability distributions
- On the convergence time of some non-reversible Markov chain Monte Carlo methods
- scientific article; zbMATH DE number 5490661 (Why is no real title available?)
- Variance reduction using nonreversible Langevin samplers
- A note on variance reduction
- Dirichlet eigenvalue problems of irreversible Langevin diffusion
This page was built for publication: Variance reduction for diffusions
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q491924)
