ML for Materials (Harvard)

My current materials modeling research within the Kozinsky Lab at Harvard lies at the interface of machine learning, physics, chemistry, and analysis. The focus of the lab centers on computational materials modeling, discovery, and simulation efforts, with a synthesis of theoretical and applied frameworks. My efforts are centered on the following:

• Machine-learned exchange-correlation functionals for Density Functional Theory (DFT) methodologies.
• Method development for top-down differentiable learning of neural-network (NN) potentials.
• Asymmetric representations of spin-dependent neural-network potentials.