Automatic eigenvalue method in micromagnetic and atomistic simulations

IF 2.5 3区材料科学 Q3 MATERIALS SCIENCE, MULTIDISCIPLINARY

Journal of Magnetism and Magnetic Materials Pub Date : 2025-03-24 DOI:10.1016/j.jmmm.2025.172989

Xintao Fan , Shichao Zhang , Weiwei Wang , Lingyao Kong , Youmin Guo , Yizhou Liu , Haifeng Du

{"title":"Automatic eigenvalue method in micromagnetic and atomistic simulations","authors":"Xintao Fan , Shichao Zhang , Weiwei Wang , Lingyao Kong , Youmin Guo , Yizhou Liu , Haifeng Du","doi":"10.1016/j.jmmm.2025.172989","DOIUrl":null,"url":null,"abstract":"<div><div>Magnetic spin textures, such as domain walls, skyrmions, and hopfions, exhibit dynamic properties that are essential for spintronic devices. A powerful method for understanding their normal modes and spin excitations is the eigenvalue approach, which relies on linearizing the Landau–Lifshitz–Gilbert equation. However, the nonuniform magnetization and nonlinear energy terms greatly increase the complexity of this linearization process, preventing eigenvalue-based methods from becoming standard in modern micromagnetic simulation tools. In this work, we present two novel approaches to implement the eigenvalue method: standard automatic differentiation (AD) and symbolic dual numbers. Both methods are integrated into the micromagnetic simulation package MicroMagnetic.jl. These approaches unify the treatment of linear and nonlinear terms, eliminating the need for manual derivations and enabling scalable, accurate analyses of excitation modes in both micromagnetic and atomistic models.</div></div>","PeriodicalId":366,"journal":{"name":"Journal of Magnetism and Magnetic Materials","volume":"622 ","pages":"Article 172989"},"PeriodicalIF":2.5000,"publicationDate":"2025-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Magnetism and Magnetic Materials","FirstCategoryId":"88","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0304885325002203","RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"MATERIALS SCIENCE, MULTIDISCIPLINARY","Score":null,"Total":0}

引用次数: 0

Abstract

Magnetic spin textures, such as domain walls, skyrmions, and hopfions, exhibit dynamic properties that are essential for spintronic devices. A powerful method for understanding their normal modes and spin excitations is the eigenvalue approach, which relies on linearizing the Landau–Lifshitz–Gilbert equation. However, the nonuniform magnetization and nonlinear energy terms greatly increase the complexity of this linearization process, preventing eigenvalue-based methods from becoming standard in modern micromagnetic simulation tools. In this work, we present two novel approaches to implement the eigenvalue method: standard automatic differentiation (AD) and symbolic dual numbers. Both methods are integrated into the micromagnetic simulation package MicroMagnetic.jl. These approaches unify the treatment of linear and nonlinear terms, eliminating the need for manual derivations and enabling scalable, accurate analyses of excitation modes in both micromagnetic and atomistic models.

查看原文本刊更多论文

磁性自旋纹理（如畴壁、skyrmions 和 hopfions）表现出的动态特性对自旋电子器件至关重要。了解它们的法向模式和自旋激发的有力方法是特征值方法，该方法依赖于线性化 Landau-Lifshitz-Gilbert 方程。然而，非均匀磁化和非线性能量项大大增加了线性化过程的复杂性，使得基于特征值的方法无法成为现代微磁仿真工具的标准。在这项工作中，我们提出了实现特征值方法的两种新方法：标准自动微分（AD）和符号二元数。这两种方法都集成到了微磁仿真软件包 MicroMagnetic.jl。这些方法统一了线性和非线性项的处理，无需手动推导，可对微磁和原子模型中的激励模式进行可扩展的精确分析。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Journal of Magnetism and Magnetic Materials 物理-材料科学：综合

CiteScore

5.30

自引率

11.10%

发文量

1149

审稿时长

59 days

期刊介绍： The Journal of Magnetism and Magnetic Materials provides an important forum for the disclosure and discussion of original contributions covering the whole spectrum of topics, from basic magnetism to the technology and applications of magnetic materials. The journal encourages greater interaction between the basic and applied sub-disciplines of magnetism with comprehensive review articles, in addition to full-length contributions. In addition, other categories of contributions are welcome, including Critical Focused issues, Current Perspectives and Outreach to the General Public. Main Categories: Full-length articles: Technically original research documents that report results of value to the communities that comprise the journal audience. The link between chemical, structural and microstructural properties on the one hand and magnetic properties on the other hand are encouraged. In addition to general topics covering all areas of magnetism and magnetic materials, the full-length articles also include three sub-sections, focusing on Nanomagnetism, Spintronics and Applications. The sub-section on Nanomagnetism contains articles on magnetic nanoparticles, nanowires, thin films, 2D materials and other nanoscale magnetic materials and their applications. The sub-section on Spintronics contains articles on magnetoresistance, magnetoimpedance, magneto-optical phenomena, Micro-Electro-Mechanical Systems (MEMS), and other topics related to spin current control and magneto-transport phenomena. The sub-section on Applications display papers that focus on applications of magnetic materials. The applications need to show a connection to magnetism. Review articles: Review articles organize, clarify, and summarize existing major works in the areas covered by the Journal and provide comprehensive citations to the full spectrum of relevant literature.