Good Practices in Database Generation for Benchmarking Density Functional Theory

Abstract

The hundreds of density functional theory (DFT) methods developed over the past three decades are often referred to as the “zoo” of DFT approximations. In line with this terminology, the numerous DFT benchmark studies might be considered the “safari” of DFT evaluation efforts, reflecting their abundance, diversity, and wide range of application and methodological aspects. These benchmarks have played a critical role in establishing DFT as the dominant approach in quantum chemical applications and remain essential for selecting an appropriate DFT method for specific chemical properties (e.g., reaction energy, barrier height, or noncovalent interaction energy) and systems (e.g., organic, inorganic, or organometallic). DFT benchmark studies are a vital tool for both DFT users in method selection and DFT developers in method design and parameterization. This review provides best-practice guidance on key methodological aspects of DFT benchmarking, such as the quality of benchmark reference values, dataset size, reference geometries, basis sets, statistical analysis, and electronic availability of the benchmark data. Additionally, we present a flowchart to assist users in systematically choosing these methodological aspects, thereby enhancing the reliability and reproducibility of DFT benchmarking studies.