Answers are grounded in this paper and cite their sources.
It is demonstrated that there is still scope for improvement in the quality of conventional, semiempirical hybrid exchange-correlation functionals in density-functional theory. A new functional, denoted B97-3, is determined from a fit to eight chemical properties (316 data points). For a series of 25 chemical assessments (850 data points) including 17 assessments and 10 chemical properties absent from the fitting data, B97-3 provides the lowest or joint-lowest mean absolute error on 15 occasions, compared to 6, 5, and 4 occasions for B3LYP, PBE0, and B97-2, respectively [A. D. Becke, J. Chem. Phys. 98, 5648 (1993); M. Ernzerhof and G. E. Scuseria, J. Chem. Phys. 110, 5029 (1999); C. Adamo and V. Barone, J. Chem. Phys. 110, 6158 (1999); P. J. Wilson, T. J. Bradley, and D. J. Tozer, J. Chem. Phys. 115, 9233 (2001)]. Mean absolute errors from B97-3 are, on average, 21%, 18%, and 12% smaller than from these three functionals. The most notable improvements are obtained for classical reaction barriers, where the error reductions are 60%, 54%, and 27%.
No takes yet. Share an insight, caveat, or question.
Keal et al. (2005) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: