How Do Initial Spins Affect Density Functional Theory Convergence for Magnetic High Entropy Alloys?

Machine‐learned interatomic potentials (MLIPs) have proven to be effective in accelerating catalytic simulations, achieving density functional theory (DFT) accuracy at a small fraction of the computational cost. To streamline the acquisition of representative DFT data, autonomous active learning workflows have been integrated with unmanned DFT calculations. As a result, DFT settings are typically standardized across various geometries selected for computation. However, magnetic systems often require more human intervention to determine the correct final magnetizations—the ones that correspond to the lowest electronic energy—leading to MLIPs being predominantly developed for non‐magnetic systems. Despite this, magnetic high‐entropy alloys (HEAs) are recognized for their promising catalytic properties, including CO 2 reduction, NH 3 decomposition and transport, as well as their desirable physical attributes such as high‐temperature strength and corrosion resistance. Given their importance, it is crucial to optimize the modeling of these materials. This study proposes a methodology for selecting appropriate initial atomic spin states for DFT optimization and single‐point calculations of representative HEA bulk, slab, and nanoparticle structures. The goal is to resolve a critical bottleneck in the development of magnetic HEA MLIPs with the automated and reliable generation of consistent DFT training data. By establishing a robust protocol for magnetic systems, this work enables the creation of high‐quality potential energy surfaces necessary for training next generation interatomic potentials.