Parallelization of a dynamic Monte Carlo algorithm: A partially rejection-free conservative approach (Q1819039)

This paper deals with a simulation method of a magnetization switching in a kinetic Ising model on a distributed-memory parallel computer. The parallel \(n\)-fold way algorithm proposed by \textit{B. D. Lubachevsky} [J. Comput. Phys. 75, 103-122 (1988; Zbl 0642.65001)] has been applied to the isotropic, two-dimensional Ising model, and compared its performance to the parallel Metropolis algorithm. After defining the model and summarizing the standard serial Metropolis and rejection-free serial algorithms, the authors outline the basic conservative approach for parallel discrete-event simulation applied to Ising spins, and describe the parallel Metropolis and \(n\)-fold way algorithms. Then, some details of the implementation are given and their performances are analyzed. Finally, some examples of ongoing and future applications to model magnetic and ferromagnetic systems are provided.