Computational materials design with machine learning and atomistic simulations
|Hour||15:30 › 16:30|
Rafael Gomez-Bombarelli is the Jeffrey Cheah Career Development Professor in MIT's Department of Materials Science and Engineering. His work aims to fuse machine learning and atomistic simulations for designing materials and their transformations. Through collaborations at MIT and beyond, Rafael’s group develops new practical materials such as catalysts, therapeutic peptides, organic electronics or electrolytes for batteries.
Rafael joined MIT in 2018 after earning BS, MS, and PhD (2011) degrees in chemistry from Universidad de Salamanca (Spain) and carrying postdoctoral work at Heriot-Watt, Harvard and at Kyulux North America. His machine learning work has received faculty awards from the Dreyfus Foundation and Google. Rafael was a co-founder of Calculario, a Harvard spinout company, and served as Chief Learning Officer of ZebiAI, a drug discovery startup.
|Category||Conferences - Seminars|
Designing new materials is vital for addressing pressing societal challenges in health, energy, and sustainability. The computational techniques of atomistic simulation and machine learning (ML) offer an avenue to rapidly invent new materials and navigate this enormous space. By populating the continuum between physics-based simulations and machine learning, the Learning Matter Lab seeks to enable rapid, computation-first design of materials that accelerate the materials discovery cycle.
Atomistic simulations, using techniques from quantum mechanics or statistical mechanics can predict the properties of hypothetical materials, and by engineering high-throughput simulation pipelines, Gomez-Bombarelli can evaluate millions of candidates and find compositions or structures that optimize a given property. Simulations are, nevertheless, relatively costly, and may lack accuracy compared to experiment. This is where the synergy with ML enables a new paradigm: surrogate models bypass simulations by interpolating among pre-existing calculations at a fraction of the cost, while embedding physics-based priors in ML ensures robustness and transferability.
- General public
- Andres M Bran, Jeff Guo, Kevin Maik Jablonka, Philippe Schwaller, Puck van Gerwen