Assessment of DLPNO-MP2 Approximations in Double-Hybrid DFT

Abstract

The unfavorable scaling (N 5) of the conventional second-order Møller–Plesset theory (MP2) typically prevents the application of double-hybrid (DH) density functionals to large systems with more than 100 atoms. A prominent approach to reduce the computational demand of electron correlation methods is the domain-based local pair natural orbital (DLPNO) approximation that is successfully used in the framework of DLPNO-CCSD­(T). Its extension to MP2 [Pinski P.; Riplinger, C.; Valeev, E. F.; Neese, F. J. Chem. Phys. 2015, 143, 034108.] paved the way for DLPNO-based DH (DLPNO-DH) methods. In this work, we assess the accuracy of the DLPNO-DH approximation compared to conventional DHs on a large number of 7925 data points for thermochemistry and 239 data points for structural features, including main-group and transition-metal systems. It is shown that DLPNO-DH-DFT can be applied successfully to perform energy calculations and geometry optimizations for large molecules at a drastically reduced computational cost. Furthermore, PNO space extrapolation is shown to be applicable, similar to its DLPNO-CCSD­(T) counterpart, to reduce the remaining error.

Calculation of electric field gradients with the exact two-component (X2C) quasi-relativistic method and its local approximations

“…The low-cost double-hybrid DFT method accelerated by the RIJCOSX technique is capable of producing highly accurate EFG values for medium-sized iron compounds and has a much better cost-effectiveness, than the large-scale DMRG-SCF and ICE-SCF methods, which display similar accuracy. Furthermore, a local linear scaling version of the double-hybrid DFT method with the domain-based local pair natural orbital (DLPNO) approximation is also in progress, 100 which can enable calculation of larger molecules with up to 100 atoms. Because numerous double-hybrid functionals have been proposed in the literature, DSD-PBEP86 might not be the best choice and, in the future, more double-hybrid functionals need to be tested in the calculation of the Mössbauer NQS parameters.…”

Analytical Gradient Using Cluster-in-Molecule RI-MP2 Method for the Geometry Optimizations of Large Systems

“…These methods can be classified into two categories. The first, often referred to as direct local correlation methods, treats the entire molecule at once by omitting insignificant contributions. − The second category is the cluster-based methods, obtaining the total correlation energy as the sum of contributions from smaller clusters containing a subset of occupied LMOs and their virtual orbitals. − Although both direct and cluster-based methods can achieve a linear scaling computational cost with the system size, the cluster-based methods require much less storage and thus are applicable to much larger systems. Resolution-of-the-identity (RI) − approximation, a technique employed to accelerate electronic structure calculations by approximating the four-index two-electron integrals with two- and three-index integrals, can be combined with local correlation methods to further reduce the computational costs.…”

Emerging conformational-analysis protocols from the RTCONF-16K reaction thermochemistry conformational benchmark set

“…The level of electron correlation and basis set can be, in principle, converged systematically following Pople's diagram. 1 Even the accuracy of the popular and cost-efficient Density Functional Theory (DFT) methods can be improved using the concept of Jacob's Ladder [2][3][4] and extensive benchmark studies, [5][6][7][8][9][10][11][12][13][14] albeit the systematic approach is replaced to some extent with empiricism. Furthermore, terms beyond the non-relativistic treatment can be systematically addressed by opening an additional dimension of complexity (Csaszar's cube).…”