| Title | : | ML4Science Seminar: Heather J. Kulik (MIT) |
| Duration | : | 01:09:56 |
| Viewed | : | 0 |
| Published | : | 13-05-2020 |
| Source | : | Youtube |
Molecular design blueprints: materials and catalysts from new simulation and machine learning tools
Many compelling functional materials and highly selective catalysts have been discovered that are defined by their metal-organic bonding. The rational design of de novo transition metal complexes however remains challenging. First-principles (i.e., with density functional theory, or DFT) high-throughput screening is a promising approach but is hampered by high computational cost, particularly in the brute force screening of large numbers of materials. In this talk, I will outline our efforts over the past few years to accelerate the design of inorganic complexes for catalysis and materials science applications. I will describe our software and machine learning (ML) models that both simplify and accelerate the screening of new materials. I will talk about our recent efforts in autonomous computational chemistry by developing ML models as decision engines capable of predicting when calculations will fail and when single-reference methods are insufficient. I will describe how we have paired ML models with multiobjective optimization strategies, robust uncertainty quantification, and highly parallel accelerated computing to discover optimal materials in weeks instead of decades. Finally, I will describe how this powerful toolkit has advanced our understanding of a range of metal-organic complexes from functional spin crossover materials to open-shell transition metal catalysts not just by enabling the rapid screening of millions of candidate molecules but by providing a new approach to interpreting and uncovering design rules.
SHARE TO YOUR FRIENDS
Scan me
