Machine Learning with bond information for local structure optimizations in surface science

TL;DR

This paper introduces an anisotropic kernel in Gaussian process regression to improve local structure optimization in surface science, achieving faster results and reduced computational costs.

Contribution

The work presents a novel anisotropic kernel tailored for bond information, enhancing machine learning-based optimization of adsorption systems.

Findings

Speed-up of up to two times compared to standard methods

Limited memory approach reduces energy and force calculations

Effective across various atomic system types

Abstract

Local optimization of adsorption systems inherently involves different scales: within the substrate, within the molecule, and between molecule and substrate. In this work, we show how the explicit modeling of the different character of the bonds in these systems improves the performance of machine learning methods for optimization. We introduce an anisotropic kernel in the Gaussian process regression framework that guides the search for the local minimum, and we show its overall good performance across different types of atomic systems. The method shows a speed-up of up to a factor two compared with the fastest standard optimization methods on adsorption systems. Additionally, we show that a limited memory approach is not only beneficial in terms of overall computational resources, but can result in a further reduction of energy and force calculations.