Approach to ergodicity in monte carlo simulations

Area of Science:

• Computational Physics
• Statistical Mechanics
• Chemical Physics

Background:

• Monte Carlo (MC) simulations are crucial for modeling complex systems, but understanding their convergence to the ergodic limit is challenging.
• Existing measures for simulation efficiency often lack a direct link to the fundamental ergodic properties of the simulation dynamics.

Purpose of the Study:

• To develop and validate a novel metric for assessing the ergodic behavior of Monte Carlo simulations.
• To analytically and numerically investigate the approach to the ergodic limit in various simulation regimes.
• To compare the relative efficiencies of different MC sampling schemes using the proposed metric.

Main Methods:

• Development of a stochastic model to define a metric for ergodic and nonergodic simulation analysis.
• Analytical derivation of the decay law governing the approach to ergodicity from the nonergodic regime.
• Numerical simulations of Lennard-Jones 13-particle clusters using Metropolis, J-walking, and parallel tempering MC methods.

Main Results:

• The defined metric successfully distinguishes between ergodic and nonergodic simulation regimes.
• The analytically inferred decay law accurately predicts the onset of ergodic behavior in simulations.
• The metric quantitatively compares the efficiencies of Metropolis, J-walking, and parallel tempering, aligning with model predictions for atomic clusters.

Conclusions:

• The developed metric provides a robust tool for monitoring and comparing MC simulation convergence.
• The analytical decay law is valuable for optimizing simulation parameters, particularly high-temperature settings in parallel tempering and J-walking.
• This work offers a pathway to more efficient and reliable Monte Carlo simulations in various scientific domains.