No abstract

Wassercluster Im Forschungsartikel auf S. 11496 berichten M. Havenith et al. das erste vollständige Spektrum des Wassertrimers im Frequenzbereich von 70 bis 620 cm−1. Quantenchemische Rechnungen ermöglichten die Zuordnung der experimentell beobachteten Banden.

Polarizable force fields are central to computational chemistry and biochemistry;however, they are largely empirical or semi-empirical. Here we propose and demonstrate a Δ-Machine Learning approach to correct such force fields to the gold standard coupled cluster level of theory. This is done explicitly for the TTM2.1-F water potential. Specifically we correct this potential up to 4-b interactions. The new potential achieves unprecedented accuracy for eight low-lying isomers of the water hexamer and 20-mer clusters. Rigorous diffusion Monte Carlo (DMC) calculations of the dissociation energies of the water dimer and trimer using the Δ-corrected potential are in excellent agreement with experiment. In addition, the new potential is robust for unconstrained DMC calculations for the hexamer, as well as VSCF/VCI calculations of the IR spectrum of the book and ring isomers, which agree well with experiment. We show for 256 monomers that the Δ-corrected potential is negligibly more computationally intensive than TTM2.1 for energies and gradients. Thus, the many-body Δ-Machine Learning approach provides a new scheme to significantly elevate the accuracy of conventional force field in a computationally efficient manner.

Water Clusters In their Research Article on page 11399, M. Havenith et al. report the first comprehensive spectrum of the water trimer in the frequency region from 70 to 620 cm−1. Quantum calculations allowed assigning the experimentally observed bands.