Next Generation Materials Modelling

Introduction

Data-driven materials modelling is gaining ground in the field of materials prediction and characterization, but does this imply that fundamental theories and computer simulations based on scientific knowledge are made redundant?

Well, even in a data-driven world, the data bases needed to train machine learned models are often rooted in traditional atomistic simulations. Another example is mesoscopic models that often build on electronic and atomistic information derived from quantum mechanical and force-field calculations, as such may be easier and cheaper to perform than to undertake complex experiments. Furthermore, physics/chemistry based models can provide unique information to assist in the interpretation of experimental measurements – insight that data-driven approaches cannot give.

Or can they?

In this session we will discuss perspectives and visions for materials modelling: Is the model accuracy the largest limitation in materials modelling? Will there still be a role for fundamental theories and science-based models in materials modelling? Is multiscale modelling moving forward? How different are the needs and expectations from academic and industrial users? Will the applications determine the next generation methods?

Session 1 is related to EMMC Focus Area "Advancements in Modelling and Characterisation"