Role of current fluctuations in nonreversible samplers

Abstract

It is known that the distribution of nonreversible Markov processes breaking the detailed balance condition converges faster to the stationary distribution compared to reversible processes having the same stationary distribution. This is used in practice to accelerate Markov chain Monte Carlo algorithms that sample the Gibbs distribution by adding nonreversible transitions or non-gradient drift terms. The breaking of detailed balance also accelerates the convergence of empirical estimators to their ergodic expectation in the long-time limit. Here, we give a physical interpretation of this second form of acceleration in terms of currents associated with the fluctuations of empirical estimators using the level 2.5 of large deviations, which characterises the likelihood of density and current fluctuations in Markov processes. Focusing on diffusion processes, we show that there is accelerated convergence because estimator fluctuations arise in general with current fluctuations, leading to an added large deviation cost compared to the reversible case, which shows no current. We study the current fluctuation most likely to arise in conjunction with a given estimator fluctuation and provide bounds on the acceleration, based on approximations of this current. We illustrate these results for the Ornstein-Uhlenbeck process in two dimensions and the Brownian motion on the circle.

arXiv commentv1: 14 pages, 2 figures. v2: minor corrections, close to published version

References in corpus (38)

• Stochastic thermodynamics, fluctuation theorems, and molecular machines
• The large deviation approach to statistical mechanics
• Dissipation bounds all steady-state current fluctuations
• Macroscopic fluctuation theory
• Nonequilibrium Markov processes conditioned on large deviations
• Universal bounds on current fluctuations
• Large deviations and ensembles of trajectories in stochastic models
• Nonequilibrium microcanonical and canonical ensembles and their equivalence
• Quantifying dissipation using fluctuating currents
• Introduction to dynamical large deviations of Markov processes
• Non equilibrium current fluctuations in stochastic lattice gases
• Inferring dissipation from current fluctuations
• Markov Chain Monte Carlo Method without Detailed Balance
• Variational and optimal control representations of conditioned and driven processes
• Irreversible Monte Carlo Algorithms for Efficient Sampling
• Optimal non-reversible linear drift for the convergence to equilibrium of a diffusion
• Variance Reduction using Nonreversible Langevin Samplers
• Accelerating diffusions
• Effective interactions and large deviations in stochastic processes
• A formal view on 2.5 large deviations and fluctuation relations
• Current large deviations for driven periodic diffusions
• Thermodynamics of currents in nonequilibrium diffusive systems: theory and simulation
• Non-reversible Metropolis-Hastings
• Irreversible Langevin samplers and variance reduction: a large deviation approach
• Perturbative calculation of quasi-potential in non-equilibrium diffusions: a mean-field example
• Violation of detailed balance accelerates relaxation
• Improving the convergence of reversible samplers
• Level 2 large deviation functionals for systems with and without detailed balance
• Canonical structure and orthogonality of forces and currents in irreversible Markov chains
• Acceleration of convergence to equilibrium in Markov chains by breaking detailed balance
• Large deviations for Markov processes with stochastic resetting : analysis via the empirical density and flows or via excursions between resets
• Eigenvalue analysis of an irreversible random walk with skew detailed balance conditions
• Process interpretation of current entropic bounds
• Large Deviations for the density and the current in Non-Equilibrium-Steady-States on disordered rings
• Central limit theorems for additive functionals of ergodic Markov diffusions processes
• Revisiting classical and quantum disordered systems from the unifying perspective of large deviations
• Using Perturbed Underdamped Langevin Dynamics to Efficiently Sample from Probability Distributions
• Stochastic Optimal Control as Non-equilibrium Statistical Mechanics: Calculus of Variations over Density and Current

Cited by in corpus (16)

• Dynamical large deviations of linear diffusions
• Sampling efficiency of transverse forces in dense liquids
• Methods and Conversations in (Post)Modern Thermodynamics
• Large deviations for the Skew-Detailed-Balance Lifted-Markov processes to sample the equilibrium distribution of the Curie-Weiss model
• Jump-Drift and Jump-Diffusion Processes : Large Deviations for the density, the current and the jump-flow and for the excursions between jumps
• Graph-combinatorial approach for large deviations of Markov chains
• Asymmetric Thermal Relaxation in Driven Systems: Rotations go Opposite Ways
• Irreversible Boltzmann samplers in dense liquids: weak-coupling approximation and mode-coupling theory
• Delocalization-localization dynamical phase transition of random walks on graphs
• Markov trajectories : Microcanonical Ensembles based on empirical observables as compared to Canonical Ensembles based on Markov generators
• Inhomogeneous asymmetric exclusion processes between two reservoirs : large deviations for the local empirical observables in the Mean-Field approximation
• Control of probability flow in Markov chain Monte Carlo -- Nonreversibility and lifting
• On correlations and fluctuations of time-averaged densities and currents with general time-dependence
• Feynman-Kac theory of time-integrated functionals: Itô versus functional calculus
• Exponential Change of Relaxation Rate by Quenched Disorder
• Stochastic dynamics without detailed balance condition connecting simple gradient method and Hamiltonian Monte Carlo