We present the foundational steps toward the extension of the many-body molecular dynamics (MB-MD) methodology to systems containing general molecular species comprised of several atoms. The central features of this approach are 1) a many-body representation of the interaction energy, 2) flexible, permutationally-invariant polynomials capable of describing complex quantum effects, and 3) classical expressions for electrostatic, dispersion, and many-body induction interactions. All parameters present in the model are fit to high-level electronic structure calculations, yielding an analytical potential energy function (PEF) and gradient of the PEF. As such, special focus will be placed on the generation of suitable/representative training sets for these relatively high-dimensional systems, as well as the nature of the optimization/fitting problem in the context of fitting PEFs, including the relationship between the chosen complexity of the PEF and its accuracy. The strengths and limitations of the approach will be illustrated by way of representative examples, including hydrocarbons and (hydrated) molecular ions relevant to a wide array of environmental, biological, and industrial processes.

From <i>ab initio</i> data to high-dimensional potential energy surfaces: Nuts and bolts of generating a general many-body potential energy function

2019 Fall National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.

From ab initio data to high-dimensional potential energy surfaces: Nuts and bolts of generating a general many-body potential energy function