From DFT to MD simulations: Short introduction and hands-on with classical and machine learning-based approaches

Short abstract: Materials design is a crucial part of the development of materials. There are instances where the experimental methods failed to explain the phenomena or the probing technique can not access the atomic scale. In such cases, the insight from the simulations is crucial for the material’s design. Among the simulation tools, the first principle calculations and molecular dynamics (MD) are the most popular ones. With the use of high-performance computers (HPC) in the mix, these methods are now very promising to contribute to materials design. Furthermore, machine-learning(ML)-based interatomic potentials can enable simulations of extended systems with an accuracy that is largely comparable to DFT, but with a computational cost that is linear orders of magnitude. However, using HPC with these methods can be tricky. In this workshop, we will teach you the basics with an interactive environment to overcome the initial barriers to using those methods. Hence, we will go through various properties of the materials with DFT, ab initio molecular dynamics and classical molecular dynamics (using ML interatomic potential) methods. In addition, we will briefly cover an approache to train and identify the structure from atomistic data using machine learning.