Magnetic Structures Database from Symmetry-aided High-Throughput Calculations

Magnetic structures, which play a central role in determining their physical properties, are known for only very limited compounds. Traditional theoretical approaches to predicting magnetic structures predominantly rely on first-principles calculations. A key challenge of these methods is their requirement for initial magnetic configurations as inputs, which theoretically possess infinite possibilities. In this work, we introduce a strategy based on irreducible representation basis vectors that effectively narrows down the vast space of potential magnetic configurations to a finite set, typically comprising around 20 candidates per material. Despite this significant reduction, the compact input sets generated by our method already encompass the experimental magnetic structures for 253 out of 302 benchmark materials (83.8%) from the MAGNDATA database. These materials have propagation vectors q = 0 and unit cells containing up to 40 atoms, all within the Landau framework. Subsequent first-principles calculations correctly identify the magnetic structure in 198 of these cases. We further apply our highly efficient method to 8,422 stoichiometric transition-metal compounds with fewer than 30 atoms per unit cell in the Inorganic Crystal Structure Database, and establish a magnetic structure database containing 2,906 magnetic materials. To demonstrate its utility, we use this database for the systematic exploration of magnetic topological phases and altermagnets, identifying 1,070 and 392 materials, respectively.