Automatic generation of complementary auxiliary basis sets for explicitly correlated methods. Issue 25 (19th July 2022)

• Record Type: Journal Article
• Authors: Semidalas, Emmanouil; Martin, Jan M. L.
• Is Part Of: Journal of computational chemistry. Volume 43:Issue 25(2022)
• Page Start: 1690
• Page End: 1700
• Publication Date: 2022-07-19
• Subjects: automatic fitting basis set generation -- complementary auxiliary basis set -- explicitly correlated methods; Chemistry -- Data processing -- Periodicals; 542.85
• DOI: 10.1002/jcc.26970
• Languages: English
• ISSNs: 0192-8651

Abstract

Abstract: Explicitly correlated calculations, aside from the orbital basis set, typically require three auxiliary basis sets: Coulomb‐exchange fitting (JK), resolution of the identity MP2 (RI‐MP2), and complementary auxiliary basis set (CABS). If unavailable for the orbital basis set and chemical elements of interest, the first two can be auto‐generated on the fly using existing algorithms, but not the third. In this paper, we present a quite simple algorithm named autoCABS; a Python implementation under a free software license is offered at Github. For the cc‐pVnZ‐F12 ( n = D, T, Q, 5), the W4‐08 thermochemical benchmark, and the HFREQ2014 set of harmonic frequencies, we demonstrate that autoCABS‐generated CABS basis sets are comparable in quality to purpose‐optimized OptRI basis sets from the literature, and that the quality difference becomes entirely negligible as n increases. Abstract : Explicitly correlated methods exhibit dramatically faster basis set convergence compared to orbital‐only methods. The lack of an appropriate complementary auxiliary basis set (CABS) can be an obstacle to application for less‐common chemical elements or orbital basis sets. We propose a simple automated "autoCABS" approach to generate CABSes from an arbitrary basis set; for test sets where purpose‐built CABSes are available, we show that they are comparable in accuracy and compactness to those generated by autoCABS.