AI4Science at Microsoft: Catalyst Design & Material Discovery

This lecture introduces AI4Science at Microsoft, a team focusing on machine learning, quantum physics, computational chemistry, molecular biology, and fluid dynamics to address societal challenges. The talk delves into accelerating catalyst design through scalable simulation methods, exploring reaction mechanisms, and automating reaction network exploration. It covers challenges in exploring high-dimensional spaces, designing efficient simulation methods, and generative modeling of materials using Clifford algebra. The lecture showcases the application of Clifford convolution and Fourier transform in PDE modeling, with promising results in weather modeling, Maxwell's equations, and shallow water simulations.