Ab initio-Berechnung von Moleküleigenschaften in der orbital-optimierten Random-Phase-Näherung
Zusammenfassung der Projektergebnisse
This project established analytical electronic structure methods for geometry optimization and molecular properties on the level of the random phase approximation (RPA). The implemented RI-RPA yields the currently lowest scaling of the operation count, O(N4log N), and mass storage, O(N3), with system size N for energy and gradient calculations and the computational cost is comparable to that of RI-MP2 methods. The method’s results suggest that structures obtained with RPA are significantly more accurate than MP2 structures, especially for smallgap systems. Thus, RI-RPA emerges as a valuable tool for predicting minima and transition state structures of open-shell d- and f-element compounds, particularly when semi-local functionals produce wildly different results. Being robust, non-empirical, and relatively inexpensive, RI-RPA possesses many desirable features of a general purpose method that can be applied indiscriminately to a wide range of different systems and chemical environments. The vast improvement over MP2 for small-gap systems depends on the use of a noninteracting KS reference rather than HF 41 in RPA calculations. This is in line with previous observations 42 that HF-based direct RPA performs poorly for energy differences. The remaining dependence on the semi-local functional used for the KS calculation is mostly small compared to typical RPA errors, but could be reduced further by orbital optimization.
Projektbezogene Publikationen (Auswahl)
- “Analytical First-Order Molecular Properties and Forces within the Adiabatic Connection Random Phase Approximation”. J. Chem. Theory Comput. 10, 180-194 (2014)
Asbjorn M. Burow, Jefferson E. Bates, Filipp Furche, and Henk Eshuis
(Siehe online unter https://doi.org/10.1021/ct4008553)
