Answers are grounded in this paper and cite their sources.
A method for automatic generation of flexible-monomer three-body (3B) nonadditive contribution to intermolecular force fields (FFs) has been developed. It is an extension of the method developed earlier for two-body contributions. Together with one-body contributions and a polarization model accounting for four- and higher-body contributions, these components form a complete FF that can be used to simulate molecular clusters and condensed phases. The method was used to develop 3B potential energy surfaces for (H2O)3, (HF)3, and HF–(H2O)2, enabling simulations of water, hydrogen fluoride, and some mixtures of the two. The fitting errors of these 3B surfaces are generally smaller than those of the published analytic expansion fits and on par with the best neural network model for the water trimer. The properties of small clusters predicted by the FFs developed here are more accurate than those given by the published benchmarks from single-point ab initio calculations for such clusters, and we had to compute more accurate benchmarks of this type to evaluate the quality of our FFs.
No takes yet. Share an insight, caveat, or question.
Jing et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: