Equivariant Many-body Message Passing Interatomic Potentials for Magnetic Materials

TL;DR

This paper presents an equivariant graph neural network that explicitly models atomic magnetic moments, enabling accurate, transferable predictions of magnetic properties in materials beyond traditional methods.

Contribution

It introduces a novel message-passing model that incorporates magnetic moments and spin-orbit coupling, improving accuracy and transferability in magnetic materials modeling.

Findings

Achieves near DFT accuracy with high data efficiency.

Successfully models magnetic phenomena and transformations.

Enables high-throughput discovery of magnetic materials.

Abstract

Magnetism governs key properties of materials used in energy, data storage, and spintronic technologies, yet its complex coupling to lattice and electronic degrees of freedom challenges conventional first-principles approaches. We introduce an equivariant message-passing graph neural network that embeds atomic magnetic moments as explicit degrees of freedom, enabling the learning of magnetic interactions beyond collinear approximations. The model learns physically consistent and transferable representations of magnetic behaviour and can incorporate spin-orbit coupling, achieving near density-functional-theory accuracy with strong data efficiency across diverse magnetic systems by fine-tuning from a pre-trained model. Applications to structural transformations, finite-temperature magnetic phenomena, and materials screening for strongly spin-orbit coupled materials demonstrate transferable magnetic behaviour, establishing a practical foundation for data-driven, high-throughput discovery of complex magnetic materials.