Abstract:Atomistic simulations hold significant value in clarifying crucial phenomena such as phase transitions and energy transport in materials science. Their success stems from the presence of potential energy functions capable of accurately depicting the relationship between system energy and lattice changes. In magnetic materials, two atomic scale degrees of freedom come into play: the lattice and the spin. However, accurately tracing the simultaneous evolution of both lattice and spin in magnetic materials at an atomic scale is a substantial challenge. This is largely due to the complexity involved in depicting the interaction energy precisely, and its influence on lattice and spin-driving forces, such as atomic force and magnetic torque, which continues to be a daunting task in computational science. Addressing this deficit, we present DeepSPIN, a versatile approach that generates high-precision predictive models of energy, atomic forces, and magnetic torque in magnetic systems. This is achieved by integrating first-principles calculations of magnetic excited states with deep learning techniques via active learning. We thoroughly explore the methodology, accuracy, and scalability of our proposed model in this paper. Our technique adeptly connects first-principles computations and atomic-scale simulations of magnetic materials. This synergy presents opportunities to utilize these calculations in devising and tackling theoretical and practical obstacles concerning magnetic materials.

A deep-learning method for studying magnetic superstructures

Deep-learning electronic-structure calculation of magnetic superstructures

Accelerating the electronic-structure calculation of magnetic systems by equivariant neural networks

Learning Spin Liquids on a Honeycomb Lattice with Artificial Neural Networks.

Deep Learning Illuminates Spin and Lattice Interaction in Magnetic Materials

Super-resolution of magnetic systems using deep learning

Magnetic Hamiltonian Parameter Estimation Using Deep Learning Techniques

Equivariant Neural Network Force Fields for Magnetic Materials

Machine learning magnetism classifiers from atomic coordinates

Complex Spin Hamiltonian Represented by an Artificial Neural Network

Application of deep learning methods to the study of magnetic phenomena

Towards Large-Scale and Spatiotemporally Resolved Diagnosis of Electronic Density of States by Deep Learning

General time-reversal equivariant neural network potential for magnetic materials

Spin-Dependent Graph Neural Network Potential for Magnetic Materials

A deep equivariant neural network approach for efficient hybrid density functional calculations

Machine Learning Magnetic Parameters from Spin Configurations

Deep-learning density functional theory Hamiltonian for efficient ab initio electronic-structure calculation

Deep learning density functional theory Hamiltonian in real space

Deep learning methods for Hamiltonian parameter estimation and magnetic domain image generation in twisted van der Waals magnets

Large diversity of magnetic phases in two-dimensional magnets with spin-orbit coupling and superconductivity