Scalable Electron Correlation Methods. 4. Parallel Explicitly Correlated Local Coupled Cluster with Pair Natural Orbitals (PNO-LCCSD-F12)

Abstract

We present an efficient explicitly correlated pair natural orbital local coupled cluster (PNO-LCCSD-F12) method. The method is an extension of our previously reported PNO-LCCSD approach ( Schwilk et al., J. Chem. Theory Comput. 2017 , 13 , 3650 - 3675 ). Near linear scaling with the molecular size is achieved by using pair, domain, and projection approximations, local density fitting and local resolution of the identity, and by exploiting the sparsity of the local molecular orbitals as well as of the projected atomic orbitals. The effect of the various domain approximations is tested for a wide range of chemical reactions and intermolecular interactions. In accordance with previous findings, it is demonstrated that the F12 terms significantly reduce the domain errors. The convergence of the reaction and interaction energies with respect to the parameters that determine the domain sizes and pair approximations is extensively tested. The results obtained with our default thresholds agree within a few tenths of a kcal mol with the ones computed with very tight options. For cases where canonical calculations are still feasible, the relative energies of local and canonical calculations agree within similar error bounds. The PNO-LCCSD-F12 method needs only 25-40% more computer time than a corresponding PNO-LCCSD calculation while greatly improving the accuracy. Our program is well parallelized and capable of computing accurate correlation energies for molecules with more than 150 atoms using augmented triple-ζ basis sets and more than 5000 basis functions. Using several nodes of a small computer cluster, such calculations can be carried out within a few hours.

Calculation of Ligand Dissociation Energies in Large Transition-Metal Complexes

“…In general, single-reference meth- Table 1: Multi-configurational Z s(1) diagnostic and single-configurational D 1 diagnostic from Ref. 29 for all molecules involved in reactions in the WCCR10 set. The reactants and products (charged and neutral fragments, respectively) are displayed in Figure 1.…”

“…In particular, we discuss the suitability of single-reference approaches and then consider DLPNO-CCSD(T) ligand dissociation energies. Very recently, Ma et al 29 presented PNO-LCCSD-F12 WCCR10 ligand dissociation energies, to which we compare our DLPNO-CCSD results. We then re-assess several density functionals with a focus on dispersion interactions with reference to DLPNO-CCSD(T).…”

Coupled‐cluster singles, doubles and perturbative triples with density fitting approximation for massively parallel heterogeneous platforms

“…Even more crucial has been the recent development of CCSD(T) with reduced complexity, for example, methods based on pair‐natural orbital (PNOs), which allows treatment of systems with hundreds of atoms; systems of this size are far beyond the limits of the conventional‐scaling CCSD(T) on a classical computer. Efficient reduced‐scaling explicitly correlated CCSD(T) has been demonstrated by several groups, including ours.…”

“…[9][10][11][12] Even more crucial has been the recent development of CCSD(T) with reduced complexity, for example, methods based on pair-natural orbital (PNOs), [13][14][15] which allows treatment of systems with hundreds of atoms; systems of this size are far beyond the limits of the conventional-scaling CCSD(T) on a classical computer. Efficient reduced-scaling explicitly correlated CCSD(T) has been demonstrated by several groups, [16][17][18][19][20][21][22] including ours.Although the capability of conventional CCSD(T) has been surpassed by its reduced-scaling variants, the full-scaling (canonical) CCSD(T) is still highly desirable for assessing the impact of numerical approximations responsible for the complexity reduction. With this goal in mind we developed a reference scalable canonical CCSD(T) implementation capable of treating molecules as large as presently feasible.…”

Machine learning the computational cost of quantum chemistry

“…perf allows profiling of the kernel and user code at run time with little CPU overhead and can give FLOP counts with reasonable accuracy. FLOP count is an adequate measure of the computational cost when the program execution is CPU bound by numerical operations, which is given in the PNO-LCCSD(T)-F12 implementation [48][49][50]70 in Molpro.…”