Machine-learning enabled optimization of atomic structures using atoms with fractional existence

TL;DR

This paper presents a novel global optimization method for atomic structures that incorporates atoms with fractional existence, enabling efficient exploration of energy landscapes and overcoming barriers in atomic configuration space.

Contribution

It introduces a Gaussian process-based approach utilizing fractional existence variables for improved atomic structure optimization, especially in complex systems.

Findings

Fractional existence variables accelerate optimization of large clusters.

The method effectively explores energy barriers in atomic configurations.

Application to copper clusters demonstrates improved efficiency.

Abstract

We introduce a method for global optimization of the structure of atomic systems that uses additional atoms with fractional existence. The method allows for movement of atoms over long distances bypassing energy barriers encountered in the conventional position space. The method is based on Gaussian processes, where the extrapolation to fractional existence is performed with a vectorial fingerprint. The method is applied to clusters and two-dimensional systems, where the fractional existence variables are optimized while keeping the atomic positions fixed on a lattice. Simultaneous optimization of atomic coordinates and existence variables is demonstrated on copper clusters of varying size. The existence variables are shown to speed up the global optimization of large and particularly difficult-to-optimize clusters.