Deep Learning Potential for Atomistic Simulation of Magnetic Materials

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. Addressing this deficit, we present DeltaSPIN, and DeepSPIN, a versatile framework that generates high-precision training data of energy, atomic forces, magnetic torque and therefore the high-accuracy predictive models for them in magnetic systems. This is achieved by integrating our in-house first-principles calculations of magnetic excited states with deep learning techniques via "pseodo atom" descriptor and active learning. We also demostrate that this model also has advantage in achieving the cooperative ground states including both spin and lattice. 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.