Max Ludwig Hodapp

Professor: Alexander Shapeev

Quantum-atomistic-continuum multiscale modeling through machine learning and atomistic-to-continuum coupling

Machine-learning interatomic potentials, developed in the group of A. Shapeev, is a methodology of upscaling a computationally expensive quantum-mechanical model to an efficient atomistic model. This is sufficient to model crystal structure, however, critical properties of materials are often defined by the microstructure (e.g., an ideal iron crystal can be reversibly stretched by 10%, but a typical steel can be stretched by only 0.1-0.2% because of defects). The project, hence, is aimed at developed a methodology to combine the machine-learning interatomic potentials with a state-of-the-art algorithm of simulating defects.