Generalization of Metropolis and heat-bath sampling for Monte Carlo simulations

Area of Science:

• Computational physics and statistical mechanics
• Numerical methods and algorithms

Background:

• The Monte Carlo method is widely used for complex system simulations.
• Ensuring convergence to a target equilibrium distribution is crucial for accurate results.
• Existing methods like Metropolis can be computationally intensive.

Purpose of the Study:

• To introduce a new general sampling methodology for Monte Carlo simulations.
• To guarantee convergence to a specified equilibrium distribution function.
• To offer an alternative to existing methods with potential for unconditional acceptance.

Main Methods:

• Developing a sampling strategy based on local state probabilities.
• Utilizing the square root of the desired global probability density function for local sampling.
• Deriving the method's validity from the Chapman-Kolmogorov equation.

Main Results:

• A general sampling methodology for Monte Carlo applications is presented.
• The method is proven to converge to a specified equilibrium distribution.
• Demonstrated potential for unconditional acceptance of trial moves, differing from Metropolis.
• Illustrated utility through a prototypical numerical experiment.

Conclusions:

• The proposed method offers a robust and potentially more efficient alternative for Monte Carlo simulations.
• It provides a theoretical guarantee of convergence to equilibrium distributions.
• Applicable to a wide range of problems requiring accurate distribution sampling.