This research investigates how to improve the structure-property analysis of materials using machine learning interatomic potentials.
The main objective is to enhance predictions of transport properties, reactivity, and mechanical behaviour beyond current electronic structure methods. The work addresses key challenges in developing efficient, durable, and environmentally sustainable materials. For example, the project will develop new tools for predictive synthesis to accelerate material discovery and reduce experimental costs. Methods include generating training datasets with density functional theory, machine learning interatomic potentials, such as graph neural networks, and applying them in molecular dynamics simulations.
