2020/01/06 by Joshua J. Gabriel, Faical Yannick C. Congo, Gabriel, Joshua J. +11
Chemistry · Computer Science · Materials Science · #Advanced Physical and Chemical Molecular Interactions #Computational Drug Discovery Methods #Computational Physics (physics.comp-ph) #FOS: Physical sciences #Machine Learning in Materials Science #Materials Science (cond-mat.mtrl-sci)
paper · pdf · doi:10.48550/arxiv.2001.01851
openalex publication_date 2020/01/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Many computational databases emerged over the last five years that report\nmaterial properties calculated with density functional theory. The properties\nin these databases are commonly calculated to a precision that is set by choice\nof the basis set and the k-point density for the Brillouin zone integration. We\ndetermine how the precision of properties obtained from the Birch equation of\nstate for 29 transition metals and aluminum in the three common structures --\nfcc, bcc, and hcp -- correlate with the k-point density and the precision of\nthe energy. We show that the precision of the equilibrium volume, bulk modulus,\nand the pressure derivative of the bulk modulus correlate comparably well with\nthe k-point density and the precision of the energy, following an approximate\npower law. We recommend the k-point density as the convergence parameter\nbecause it is computationally efficient, easy to use as a direct input\nparameter, and correlates with property precision at least as well as the\nenergy precision. We predict that common k-point density choices in high\nthroughput DFT databases result in precision for the volume of 0.1%, the bulk\nmodulus of 1%, and the pressure derivative of 10%.\n