Explicitly Correlated Methods within the ccCA Methodology

Abstract

The prediction of energetic properties within "chemical accuracy" (1 kcal mol(-1) from well-established experiment) can be a major challenge in computational quantum chemistry due to the computational requirements (computer time, memory, and disk space) needed to achieve this level of accuracy. Methodologies such as coupled cluster with single, double, and perturbative triple excitations (CCSD(T)) combined with very large basis sets are often required to reach this level of accuracy. Unfortunately, such calculations quickly become cost prohibitive as system size increases. Our group has developed an ab initio composite method, the correlation consistent Composite Approach (ccCA), which enables such accuracy to be possible, on average, but at reduced computational cost as compared with CCSD(T) in combination with a large basis set. While ccCA has proven quite useful, computational bottlenecks still occur. In this study, the means to reduce the computational cost of ccCA without compromising accuracy by utilizing explicitly correlated methods within ccCA have been considered, and an alternative formulation is described.

W4‐17: A diverse and high‐confidence dataset of atomization energies for benchmarking high‐level electronic structure methods

“…In particular, we will consider composite methods that (i) attempt to approximate the CCSD(T) energy in conjunction with a triple-f-quality basis set, such as the popular Gaussian-n [31,[53][54][55][56] and CBS [57,58] family of methods, and (ii) attempt to approximate the CCSD(T) energy closer to the infinite basis-set limit, such as the Weizman-n theories (W1, W2, W1-F12, W2-F12), [38,59,60] the modified Wn methods (W1X-1, W1X-2, and W2X), [61][62][63] and the correlation-consistent Composite Approach (ccCA) methods. [32,[64][65][66][67] The reference values are all-electron, relativistic, DBOC-inclusive TAEs at the bottom of the well, that is, excluding the zero-point vibrational energy (ZPVE). The reference values used for evaluating the CCSD(T) composite procedures and DHDFT methods are given in Table S1 of the Supporting Information.…”

Toward a W4-F12 approach: Can explicitly correlated and orbital-based ab initio CCSD(T) limits be reconciled?

“…[56][57][58] Alas, the use of explicitly correlated methods for high-accuracy computational thermochemistry met with mixed success in the work of the present authors 15,24,59,60 and others. 61,62 While F12 methods enable rapidly reaching the vicinity of the basis set limit, approaching more closely in a consistent way proved a greater challenge. For instance, as will also be seen in this paper, basis set convergence of CCSD-F12 TAEs from F12 methods can be oscillatory or even (anomalously) monotonically decreasing, unlike the monotonically increasing behavior in conventional calculations.…”

The MC-DFT approach including the SCS-MP2 energies to the new minnesota-type functionals

“…As a further note to the current approach, the empirical basis set extrapolation in the MC-DFT method focused entirely on the electronic energies, which was also true in other basis set extrapolation methods [33,34] and in various composite or multicoefficient methods such as Gaussian-n, [9,35] G3S, [8] complete basis set (CBS), [36,37] Wn, [38,39] ccCA, [40,41] HEAT, [42,43] MCG3-DFT, [14] and multi-level with scaled energy-DFT (MLSE-DFT). [10] Extrapolation of the basis set effects has been known to be most crucial and most successful for predicting accurate bond energies, and, thus, it was applied primarily on the study of thermochemistry and chemical kinetics.…”