Symmetry-electronic fingerprints reveal competing magnetic phases in two-dimensional materials
Researchers have developed a new machine learning representation called the symmetry-electronic fingerprint (SEF) to better predict magnetic properties in two-dimensional materials. Unlike previous methods that focused on chemical environments, SEF incorporates crystallographic symmetry and electronic structure. This approach allows for accurate classification of magnetic ordering and regression of magnetic moments and anisotropy energies, distinguishing between different magnetic mechanisms like Stoner ferromagnetism and superexchange. Notably, the SEF's model uncertainty can identify materials where these magnetic phases compete, indicating potential for complex magnetic behaviors. AI
IMPACT This new representation could accelerate the discovery and design of novel magnetic materials for spintronics and quantum technologies.