Application of Local Information Entropy in Cluster Monte Carlo Algorithms

Abstract

The chapter refers to a modification of the so-called adding probability used in cluster Monte Carlo algorithms. The modification is based on the fact that in real systems, different properties can influence its clusterization. Finally, an additional factor related to property disorder was introduced into the adding probability, which leads to more effective free energy minimization during MC iteration. As a measure of the disorder, we proposed to use a local information entropy. The proposed approach was tested and compared with the classical methods, showing its high efficiency in simulations of multiphase magnetic systems where magnetic anisotropy was used as the property influencing the system clusterization.