Abstract
While CCSD(T) is often considered the ``gold standard'' of computational chemistry, the scaling of its computational cost as N^7 limits is applicability for large and complex molecular systems. In this work, we apply the density-based many-body expansion [Int. J. Quantum Chem. 2020, 120, e26228] in combination with CCSD(T). The accuracy of this approach is assessed for neutral, protonated, and deprotonated water hexamers as well as (H2O)16 and (H2O)17 clusters. For the neutral water clusters, we find that already with a density-based two-body expansion, we are able to approximate the supermolecular CCSD(T) energies within chemical accuracy (4 kJ/mol). This surpasses the accuracy that is achieved with a conventional, energy-based three-body expansion. We show that this accuracy can be maintained even when approximating the electron densities using Hartree-Fock instead of using coupled-cluster densities. The density-based many-body expansion thus offers a simple, resource efficient, and highly parallelizable approach that makes CCSD(T)-quality calculations feasible where they would otherwise be prohibitively expensive.
Supplementary materials
Title
Supporting Information
Description
Total interaction energies obtained with the eb-MBE and db-MBE as well as with supermolecular calculations for all considered water clusters.
Actions