Representing molecule-surface interactions with symmetry-adapted neural networks.

Jörg Behler1, Sönke Lorenz, Karsten Reuter

  • 1Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, D-14195 Berlin, Germany.

Summary

This study introduces symmetry-adapted neural networks (NNs) for accurately mapping molecule-surface interactions. These NNs precisely account for surface symmetry, improving potential-energy surface (PES) calculations for systems like oxygen on aluminum.