English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

By adopting a perspective informed by contemporary liquid state theory, weconsider how to train an artificial neural network potential to describeinhomogeneous, disordered systems. We find that neural network potentials basedon local representations of atomic environments are capable of describing someproperties of liquid-vapor interfaces, but typically fail for properties thatdepend on unbalanced long-ranged interactions which build up in the presence ofbroken translation symmetry. These same interactions cancel in thetranslationally invariant bulk, allowing local neural network potentials todescribe bulk properties correctly. By incorporating explicit models of theslowly-varying long-ranged interactions and training neural networks only onthe short ranged components, we can arrive at potentials that robustly recoverinterfacial properties. We find that local neural network models can sometimesapproximate a local molecular field potential to correct for the truncatedinteractions, but this behavior is variable and hard to learn. Generally, wefind that models with explicit electrostatics are easier to train and havehigher accuracy. We demonstrate this perspective in a simple model of anasymmetric dipolar fluid where the exact long-ranged interaction is known, andin an ab initio water model where it is approximated.

Learning intermolecular forces at liquid-vapor interfaces