More Stable Template Localization for an Incremental Focal-Point Approach—Implementation and Application to the Intramolecular Decomposition of Tris-perfluoro-tert-butoxyalane

Abstract

We present a new implementation of the template localization for the fully automated and parallizable incremental method, which excludes failures of this important step within the domain-specific basis set approach and thus ensures a higher reliability of the scheme, preserving its very high accuracy. Furthermore, we combine our method with an efficient focal-point ansatz to reach the complete basis set limit and carefully assess this approach for the first time with regard to reaction energies. For a test set of 51 reactions the incremental focal-point method with a basis set of moderate size provides a very high accuracy with respect to the complete basis set limit. That way, we are finally able to apply the scheme as a benchmark method (e.g., for density functionals) in the context of a relevant chemical topic, the intramolecular decomposition of tris-perfluoro-tert-butoxyalane (43 heavy atoms, 352 electrons).

Combining Accuracy and Efficiency: An Incremental Focal-Point Method Based on Pair Natural Orbitals

“…Next, Figure gives a comparison of our PNO-based incremental focal-point approach with standard incremental methods, assessed in a previous work . When using the def2-TZVP basis with truncation values of 10 –8 or the cc-pVDZ basis with 10 –7 , our scheme with Laplace triples provides nearly the same distribution as the corresponding conventional incremental method, that is, without PNOs.…”

“…Different basis sets (DZ = cc-pVDZ, TZ = def2-TZVP) and truncation values (10 –7 , 10 –8 ), as well as conv 2 were used. [a] Comparison to conventional focal-point and explicitly correlated incremental methods from refs and .…”

“…The incremental scheme was applied only to the latter ones, and the resulting relative energies of 51 corresponding reactions were assessed. These 81 absolute and 51 relative energies have already been considered in previous papers. , …”

“…As the initial domain partitioning, which provides all orbitals to be correlated for a specific increment, is done in a small SV basis, the corresponding correlated orbitals in the domain-specific basis set must be identified for every single incremental calculation. Therefore, we map the orbitals in a domain-specific basis to those in the SV basis by a template localization, for which a new implementation has been presented recently …”

“…Therefore, we map the orbitals in a domain-specific basis to those in the SV basis by a template localization, for which a new implementation has been presented recently. 131 The introduction of the domain-specific basis set causes an error. It has already been shown that this error can be reduced significantly by an MP2-based correction scheme 132,133 according to…”