Premium
Stepwise basis set selection
Author(s) -
Li M. W.,
Zimmerman P. M.
Publication year - 2018
Publication title -
journal of computational chemistry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.907
H-Index - 188
eISSN - 1096-987X
pISSN - 0192-8651
DOI - 10.1002/jcc.25363
Subject(s) - basis (linear algebra) , basis set , set (abstract data type) , forcing (mathematics) , atom (system on chip) , basis function , computer science , selection (genetic algorithm) , quantum , algorithm , exploit , quality (philosophy) , mathematical optimization , mathematics , computational chemistry , chemistry , physics , quantum mechanics , artificial intelligence , density functional theory , geometry , mathematical analysis , computer security , embedded system , programming language
The computational cost of quantum chemical methods grows rapidly with increasing level of theory and basis set size. At increasing costs, higher accuracies can be reached, forcing a compromise between cost and accuracy for most molecular systems. Heats of reaction, however, are mostly determined by a subset of atoms that experience significant bonding and/or electronic changes. To exploit this fact, the Stepwise Basis Builder (SBB) algorithm selectively adds basis functions to reactive atoms and maintains small basis sets on spectator atoms. This article introduces the SBB algorithm and how it chooses a basis for each atom, predicts calculation errors, and uses these predicted errors to reach target levels of accuracy. Benchmarks show SBB heats of reaction and activation barriers converge to values consistent with higher‐quality calculations using a greatly reduced number of basis functions. © 2018 Wiley Periodicals, Inc.