An improved Metropolis algorithm for hard core systems

TL;DR

This paper introduces an enhanced Metropolis algorithm that updates chains of particles collectively, significantly reducing autocorrelation times near phase transitions in hard sphere models across dimensions.

Contribution

The paper proposes a novel collective update scheme for the Metropolis algorithm applicable to hard core systems in any dimension, improving sampling efficiency.

Findings

Reduces autocorrelation times near transition points in 2D hard sphere models

Applicable to arbitrary dimensions for hard core systems

Enhances efficiency of Monte Carlo simulations in dense systems

Abstract

We present an improved Metropolis algorithm for arbitrary hard core systems in any dimensions. In the new updating scheme the conventional Metropolis step of a single particle is replaced by a collective step of a chain of particles. For the two-dimensional hard sphere model we show that this algorithm essentially reduces autocorrelation times in vicinity of the transition point.