Integrating Deep-Learning-Based Magnetic Model and Non-Collinear Spin-Constrained Method: Methodology, Implementation and Application

Abstract: We propose a non-collinear spin-constrained method that generates training data for deep-learning-based magnetic model, which provides a powerful tool for studying complex magnetic phenomena at the atomic scale. First, we propose a projection method for atomic magnetic moments by applying a radial truncation to the numerical atomic orbitals. We then implement a Lagrange multiplier method that can yield the magnetic torques of atoms by constraining the magnitude and direction of atomic magnetic moments. The method is implemented in ABACUS with both plane wave basis and numerical atomic orbital basis. We benchmark the iron (Fe) systems with the new method and analyze differences from calculations with the plane wave basis and numerical atomic orbitals basis in describing magnetic energy barriers. Based on more than 30,000 first-principles data with the information of magnetic torque, we train a deep-learning-based magnetic model DeePSPIN for the Fe system. By utilizing the model in large-scale molecular dynamics simulations, we successfully predict Curie temperatures of $\alpha$-Fe close to experimental values.