Aneesur Rahman Prize for Computational Physics Talk: Digital Alchemy, Machine Learning and Inverse Design for Self Assembly

From the Stone Age to the Silicon Age, the materials available to humankind define the world in which we live. The materials of tomorrow will be designed and engineered on demand, where and when they are needed, with precision and personalization. Computer simulation and machine learning both have critical roles to play in creating this future. Already, they allow — from a nearly infinite number of possibilities — the inverse design of nanoparticle building blocks optimized for self assembly into colloidal crystal structures with targeted properties. In this talk, we present a new thermodynamic computational approach to the inverse design of colloidal matter, and demonstrate its use in obtaining colloidal crystals with arbitrary complexity, engineered phase transitions, and target photonic properties. We show how machine learning can be used to autonomously identify crystal structures in hundreds of thousands of simulations, as well as to identify key alchemical attributes of particles that correlate with colloidal crystal structure.