[Paper Review] Simple implementation of Langevin dynamics neglecting detailed balance condition

Masayuki Ohzeki, Akihisa Ichiki|arXiv (Cornell University)|Jul 1, 2013

Markov Chains and Monte Carlo Methods2 citations

TL;DR

This paper proposes a modified Langevin dynamics that intentionally violates the detailed balance condition by introducing an asymmetric component in the Fokker-Planck operator, accelerating relaxation to the target Gibbs-Boltzmann distribution. Numerical results show faster convergence and shorter correlation times, with validation via Nemoto-Sasa theory confirming enhanced sampling efficiency.

ABSTRACT

An improved method for driving a system into a desired distribution, for example, the Gibbs-Boltzmann distribution, is proposed, which makes use of an artificial relaxation process. The standard techniques for achieving the Gibbs-Boltzmann distribution involve numerical simulations under the detailed balance condition. In contrast, in the present study we formulate the Langevin dynamics, for which the corresponding Fokker-Planck operator includes an asymmetric component violating the detailed balance condition. This leads to shifts in the eigenvalues and results in the acceleration of the relaxation toward the steady state. The numerical implementation demonstrates faster convergence and shorter correlation time, and the technique of biased event sampling, Nemoto-Sasa theory, further highlights the efficacy of our method.

Motivation & Objective

• To accelerate convergence to the Gibbs-Boltzmann distribution in statistical sampling methods.
• To overcome limitations of standard Langevin dynamics constrained by detailed balance, which can slow relaxation.
• To develop a numerically efficient method that maintains accuracy while improving mixing and decorrelation.
• To validate the method using stochastic thermodynamics, particularly the Nemoto-Sasa theory, for non-equilibrium sampling.

Proposed method

• Introduces an artificial relaxation process that modifies the drift term in Langevin dynamics to break detailed balance.
• Constructs a Fokker-Planck operator with an asymmetric component, altering the eigenvalue spectrum to speed up relaxation.
• Employs numerical integration of the modified stochastic differential equation to simulate system evolution.
• Applies biased event sampling techniques to probe the efficiency of the non-equilibrium dynamics.
• Uses Nemoto-Sasa theory to quantify the non-equilibrium character and validate the enhanced sampling performance.

Experimental results

Research questions

• RQ1 Can breaking detailed balance in Langevin dynamics significantly reduce relaxation time to the target distribution?
• RQ2 How does the introduction of an asymmetric drift term affect the eigenvalue spectrum and convergence speed?
• RQ3 To what extent does the proposed method improve sampling efficiency compared to standard equilibrium Langevin dynamics?
• RQ4 Can the Nemoto-Sasa theory be used to quantify and validate the non-equilibrium sampling advantage?

Key findings

• The modified Langevin dynamics exhibits faster convergence to the steady-state Gibbs-Boltzmann distribution compared to standard methods.
• The asymmetric component in the Fokker-Planck operator shifts eigenvalues, accelerating relaxation dynamics.
• Numerical simulations confirm shorter correlation times, indicating improved sampling efficiency.
• Biased event sampling and Nemoto-Sasa theory provide theoretical support for the enhanced sampling performance.