Neural network force fields for simple metals and semiconductors: construction and application to the calculation of phonons and melting temperatures.

We present a practical procedure to obtain reliable and unbiased neural network based force fields for solids. Training and test sets are efficiently generated from global structural prediction runs, at the same time assuring the structural variety and importance of sampling the relevant regions of phase space. The neural networks are trained to yield not only good formation energies, but also accurate forces and stresses, which are the quantities of interest for molecular dynamics simulations. Finally, we construct, as an example, several force fields for both semiconducting and metallic elements, and prove their accuracy for a variety of structural and dynamical properties. These are then used to study the melting of bulk copper and gold.

Efficient Automatized Density-Functional Tight-Binding Parametrizations: Application to Group IV Elements.

Density-functional tight-binding methods stand out as a very good compromise between accuracy and computational efficiency. These methods rely on parameter sets that have to be determined and tabulated for every pair of chemical elements. We describe an efficient, and to a large extent automatic, procedure to build such parameter sets. This procedure includes the generation of unbiased training sets and subsequent optimization of the parameters using a pattern search method. As target for the optimization we ask that the formation energy and the forces on the atoms calculated within tight-binding reproduce the ones obtained using density-functional theory. We then use this approach to calculate parameter sets for group IV elements and their binaries. These turn out to yield substantially better results than previously available parameters, especially in what concerns energies and forces.