Benchmark for <i>Ab Initio</i> Prediction of Magnetic Structures Based on Cluster-Multipole Theory

The cluster multipole (CMP) expansion for magnetic structures provides a scheme to systematically generate candidate specifically including noncollinear configurations adapted the crystal symmetry of given material. A comparison with experimental data collected on MAGNDATA shows that most stable in nature are linear combinations only few CMPs. Furthermore, high-throughput calculation all is performed framework spin-density functional theory (SDFT). We benchmark predictive power CMP+SDFT $2935$ calculations, which show (i) CMP administers an exhaustive list structures, (ii) can narrow down possible handful computed configurations, and (iii) SDFT reproduces accuracy $\pm0.5\,\mu_\text{B}$. For subset impact on-site Coulomb repulsion $U$ investigated by means $1545$ CMP+SDFT+U calculations revealing no further improvement power.