Monolayer Using Deep Potentials*

In nanometer clusters (NCs), the specific arrangement of metal atoms
determines the unique size-dependent functionalities of the NCs and
hence their potential applications. Here we employ a combined approach
utilizing a genetic algorithm and a deep potential (DP)
to determine the lowest energy configurations of the NCs including
metastable structures. The DP is trained using density functional
theory calculations. We discuss the strengths and shortcomings of the
DP approach compared to a standard method utilizing density
functional theory calculations.