Global optimization of atomic structure enhanced by machine learning

Global Optimization with First-principles Energy Expressions (GOFEE) is an efficient method for identifying low energy structures in computationally expensive landscapes such as the ones described by density functional theory (DFT), van der Waals-enabled DFT, or even methods beyond DFT. GOFEE relies on a machine learned surrogate model of energies and forces, trained on-the-fly, to explore configuration space, eliminating need relaxations all candidate using first-principles methods. In this paper we elaborate importance use Gaussian kernel two length scales Process Regression (GPR) model. We further role lower confidence bound relaxation selection structures. addition, present improvements method: 1) population generation now clustering low-energy evaluated lowest member each cluster making up population. 2) very final well-sampled basins landscape, exploitation steps, are performed continued paths within method, allow arbitrarily fine best structures, independently predictive resolution The versatility demonstrated applying it identify gas-phase fullerene-type 24-atom carbon clusters dome-shaped 18-atom supported Ir(111).