Parallelizing Wang-Landau algorithm in the field: the micromagnetic ensemble

TL;DR

This paper introduces a parallelization method for the Wang-Landau algorithm using micromagnetic ensembles, enabling larger system simulations in magnetic field studies on standard hardware.

Contribution

It presents a novel parallelization approach for the Wang-Landau algorithm leveraging micromagnetic ensembles, significantly increasing system size capabilities.

Findings

Enables simulation of larger 2D magnetic systems (up to 256x256).

Parallelization reduces computational time and resource requirements.

Applicable to studies involving field-dependent magnetic behavior.

Abstract

It is shown in this work how the Wang-Landau algorithm can be parallelized through the concept of the micromagnetic ensemble, when the Hamiltonian contains both spin interaction and the external field terms, and thus energy-magnetization plane is used for characterizing the density of states. Within this framework random walk is performed on mutually independent micromagnetic lines, and can thus be paralellized on a computer grid, without need for shared memory among individual processes. This approach pushes forward significantly the size of the systems that may be addressed on current computer hardware (from currently reported $42\times 42$ to at least $256\times 256$ in the case of two dimensional systems), and should turn out important for diverse studies where field dependent behavior is essential.