Comparison of the clock, stochastic cutoff, and Tomita Monte Carlo methods in simulating the dipolar triangular...

Magnetic nanostructures find application in diverse technological domains, and their behavior is significantly influenced by long-range dipolar interactions. However, simulating these systems using the traditional Metropolis Monte Carlo method poses high computational demand. Several methods, including the clock, stochastic cutoff, and Tomita approaches, can reduce the computational burden of simulating 2D systems with dipolar interactions. Although these three methods rely on distinct theoretical concepts, they all achieve complexity reduction by a common strategy. Instead of calculating the energy difference between a spin and all its neighbors, they evaluate the energy difference with only a limited number of randomly chosen neighbors. This is achieved through methods like the dynamic thinning and Fukui-Todo techniques. In this article, we compared the performance of the clock, SCO, Tomita, and Metropolis methods near the critical point of the dipolar triangular lattice to identify the most suitable algorithm for this type of simulation. Our findings show that while these methods are less suitable for simulating this system in their untuned implementation, incorporating the boxing near-neighbors method and overrelaxation moves makes them significantly more efficient and better suited than the Metropolis method with overrelaxation.