“Polymer design in the era of machine learning”

Advances in molecular modeling algorithms, optimization strategies, and machine learning techniques, are rapidly changing the way in which computational tools can be used to design polymeric materials systems or to interpret experimental data. In this presentation I will discuss these ideas by relying on three examples taken from our own research. In the first, I will discuss the automated creation of data bases, and the development of new graph based neural network strategies to represent polymeric structures. I will also discuss how such a framework can then be used to predict the properties of polymers and to design new materials with target properties. In the second, I will discuss how multiscale modeling and machine learning can be used to engineer the structure and, ultimately, the rheology of polymeric materials. In a third example, I will discuss how machine learning can be used to learn or extract physical principles from large data sets, particularly in the context of long-time structural relaxation in ordered polymeric materials.