Computational prediction of phase-stability skyrmion maps, internal magnetic configuration, and size of magnetic skyrmions in confined magnetic nanostructures

Abstract

In this paper, we present a computational study predicting the phase-stability skyrmion maps, internal magnetic configuration, and radii of magnetic skyrmions in rectangular magnetic nanostructures, using machine learning (ML) algorithms. The rectangular magnetic nanostructures have a fixed length of 128 nm and variable widths ranging from 56 and 128 nm. The study considers different values of perpendicular magnetic anisotropy and the Dzyaloshinskii-Moriya interaction constants. Artificial neural networks (ANNs) and Generative Adversarial Networks (GANs) were successfully employed to predict phase-stability skyrmion maps, internal magnetization images, and magnetization profiles along the z-axes for circular magnetic skyrmions. These predictions were validated through simulations using the micromagnetic Mumax3 program, demonstrating the success of the machine learning approach despite the complexity of the magnetic interactions. The results of this work highlight the potential of machine learning algorithms in advancing the study of magnetic skyrmions in confined magnetic nanostructures by accurately predicting a wide range of scenarios in a significant short time.