Abstract

A novel hybrid correlation energy (HyCE) approach is proposed that determines the total correlation energy via distinct computation of its internal and external components. This approach evolved from two related studies. First, rigorous assessment of the accuracies and size extensivities of a number of electron correlation methods, that include perturbation theory (PT2), coupled-cluster (CC), configuration interaction (CI), and coupled electron pair approximation (CEPA), shows that the CEPA(0) variant of the latter and triples-corrected CC methods consistently perform very similarly. These findings were obtained by comparison to near full CI results for four small molecules and by charting recovered correlation energies for six steadily growing chain systems. Second, by generating valence virtual orbitals (VVOs) and utilizing the CEPA(0) method, we were able to partition total correlation energies into internal (or nondynamic) and external (or dynamic) parts for the aforementioned six chain systems and a benchmark test bed of 36 molecules. When using triple-ζ basis sets it was found that per orbital internal correlation energies were appreciably larger than per orbital external energies and that the former showed far more chemical variation than the latter. Additionally, accumulations of external correlation energies were seen to proceed smoothly, and somewhat linearly, as the virtual space is gradually increased. Combination of these two studies led to development of the HyCE approach, whereby the internal and external correlation energies are determined separately by CEPA(0)/VVO and PT2/external calculations, respectively. When applied to the six chain systems and the 36-molecule benchmark test set it was found that HyCE energies followed closely those of triples-corrected CC and CEPA(0) while easily outperforming MP2 and CCSD. The success of the HyCE approach is more notable when considering that its cost is only slightly more than MP2 and significantly cheaper than the CC approaches.

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.