Data-driven learning and prediction of inorganic crystal structures

Machine learning-based interatomic potentials, fitting energy landscapes “on the fly”, are emerging and promising tools for crystal structure prediction.