RPA(D) and HRPA(D): Two new models for calculations of NMR indirect nuclear spin–spin coupling constants

Abstract

In this article, the RPA(D) and HRPA(D) models for the calculation of linear response functions are presented. The performance of the new RPA(D) and HRPA(D) models is compared to the performance of the established RPA, HRPA, and SOPPA models in calculations of indirect nuclear spin–spin coupling constants using the CCSD model as a reference. The doubles correction offers a significant improvement on both the RPA and HRPA models; however, the improvement is more dramatic in the case of the RPA model. For all coupling types investigated in this study, the results obtained using the HRPA(D) model are comparable in accuracy to those given by the SOPPA model, while requiring between 30% and 90% of the calculation time needed for SOPPA. The RPA(D) model, while of slightly lower accuracy compared to the CCSD model than HRPA(D), offered calculation times of only approximately 25% of those required for SOPPA for all the investigated molecules. © 2018 Wiley Periodicals, Inc.

Noniterative Doubles Corrections to the Random Phase and Higher Random Phase Approximations: Singlet and Triplet Excitation Energies

“…On the other hand the savings in computation time are much larger and more consistent with the RPA(D) method, as also observed in Ref. . The implementation of triplet excitation energies with the doubles corrected methods enabled the further implementation of nuclear spin–spin coupling constants which is also presented in Ref.…”

“…The implementation of triplet excitation energies with the doubles corrected methods enabled the further implementation of nuclear spin–spin coupling constants which is also presented in Ref. .…”

“…The benchmark set of Thiel et al, or a subset hereof, has been used by various authors to study vertical excitation energies using semiempirical methods, various ADC methods, the PNO‐CC2 method as well as a recent implementation of multiconfigurational short‐range DFT . The applicability of the RPA(D) as well as the new HRPA(D) method to linear response properties has recently been demonstrated by Schnack‐Petersen et al in the specific case of NMR indirect nuclear spin–spin coupling constants. In this study, we explore now the reasons for the good performance of RPA(D) and HRPA(D) for coupling constants by investigating the performance of these methods in the calculation of electronic excitation energies.…”

The Second-Order-Polarization-Propagator-Approximation (SOPPA) in a four-component spinor basis

“…PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002389 (SOPPA) 5 has proven very useful, as it yields results in good agreement with experiment, while being computationally less demanding than the high-accuracy Coupled Cluster (CC) methods 6 . For larger systems of more than roughly 30 atoms or 800 basis functions, however, even SOPPA can be rather time consuming, wherefore cheaper, but also less reliable alternatives are available such as RPA 7,8 , RPA(D) [9][10][11] , HRPA 12 and HRPA(D) 9,10 . In addition to the abovementioned methods, a range of TDDFT methods are also available 13 .…”

Benchmarking anisotropic polarizabilities for 14 (hetero)‐aromatic molecules at RPA, RPA(D), HRPA, HRPA(D), SOPPA, SOPPA(CC2), SOPPA(CCSD), CC2, CCSD and CC3 levels

“…The second study [28] extended the first study and tested the two precursors to SOPPA, the random phase approximation (RPA) and the higher order random phase approximation (HRPA) as well as the two new, doubles corrected methods, RPA(D) and HRPA(D). These two newer methods were originally derived for the calculation of excitation energies [43–45] and NMR spin–spin coupling constants [46] and were in the previous study [28] extended to the calculation of polarizabilities. Surprisingly it was found that the relatively more simple model, RPA, performed closer to CC3 than SOPPA for both static and dynamic polarizabilities.…”