Accelerated Jarzynski Estimator with Deterministic Virtual Trajectories

Abstract

The Jarzynski estimator is a powerful tool that uses nonequilibrium statistical physics to numerically obtain partition functions of probability distributions. The estimator reconstructs partition functions with trajectories of the simulated Langevin dynamics through the Jarzynski equality. However, the original estimator suffers from slow convergence because it depends on rare trajectories of stochastic dynamics. In this paper, we present a method to significantly accelerate the convergence by introducing deterministic virtual trajectories generated in augmented state space under the Hamiltonian dynamics. We theoretically show that our approach achieves second-order acceleration compared to a naive estimator with the Langevin dynamics and zero variance estimation on harmonic potentials. We also present numerical experiments on three multimodal distributions and a practical example where the proposed method outperforms the conventional method, and provide theoretical explanations.

References in corpus (17)

• A nonequilibrium equality for free energy differences
• The Entropy Production Fluctuation Theorem and the Nonequilibrium Work Relation for Free Energy Differences
• The Nonequilibrium Thermodynamics of Small Systems
• High-precision test of Landauer's principle in a feedback trap
• Optimal finite-time processes in stochastic thermodynamics
• Rare events and the convergence of exponentially averaged work values
• Dissipation and lag in irreversible processes
• Markov Chain Monte Carlo Method without Detailed Balance
• Escorted Free Energy Simulations: Improving Convergence by Reducing Dissipation
• Double wells, scalar fields and quantum phase transitions in ions traps
• An illustrative example of the relationship between dissipation and relative entropy
• Work distribution functions in polymer stretching experiments
• Quantum annealing with Jarzynski equality
• Work needed to drive a thermodynamic system between two distributions
• Dynamical computation of the density of states and Bayes factors using nonequilibrium importance sampling
• Equilibrium sampling by re-weighting non-equilibrium simulation trajectories
• Jarzynski matrix equality: calculating free energy difference by non-equilibrium simulations with an arbitrary initial distribution