Understanding the thermal properties of amorphous solids using machine-learning-based interatomic potentials

There is much to be done in the field of disordered systems, and machine learning has to help – exciting times ahead!

There is no point in petending otherwise: “Machine Learning” is so much hyped it’s just not true. However, it turns out these clever (?) algorithms can help us with a number of things. For instance, building interatomic potentials for complex systems (for which electronic structure calculations would cost too much) enabling large scale molecular dynamics simulations. To what end? Many. In here, we focus on what can be done with respect to the thermal properties of amorphous systems. If it sounds cumbersome, think about your window glasses basking in the sun – and there’s more, I promise. Have a look at the paper online, or just grab a .pdf.