Accurate and efficient protocols for high-throughput first-principles materials simulations

Abstract: Advancements in theoretical and algorithmic approaches, workflow engines, and an ever-increasing computational power have enabled a novel paradigm for materials discovery through first-principles high-throughput simulations. A major challenge in these efforts is to automate the selection of parameters used by simulation codes to deliver numerical precision and computational efficiency. Here, we propose a rigorous methodology to assess the quality of self-consistent DFT calculations with respect to smearing and $k$-point sampling across a wide range of crystalline materials. For this goal, we develop criteria to reliably estimate average errors on total energies, forces, and other properties as a function of the desired computational efficiency, while consistently controlling $k$-point sampling errors. The present results provide automated protocols (named standard solid-state protocols or SSSP) for selecting optimized parameters based on different choices of precision and efficiency tradeoffs. These are available through open-source tools that range from interactive input generators for DFT codes to high-throughput workflows.