Jeffrey B. Neaton is a professor at University of California, Berkeley and a senior leader at Lawrence Berkeley National Laboratory. His research uses computational physics to study and predict the behavior of complex materials. He has led major scientific facilities, published over 300 papers, and received honors including a U.S. Department of Energy Early Career Award and fellowship in the American Physical Society.

This talk explores how advances in first-principles computational methods and machine learning are transforming the discovery of materials for energy applications. The first part presents new ab initio approaches for predicting photophysical properties of complex semiconductors such as halide perovskites, zintl absorbers, and metal oxides, including temperature-dependent exciton binding energies and dissociation dynamics, with strong comparison to experiment. The second part discusses physics-informed machine learning models that incorporate symmetry, interactions, and chemical intuition to study porous materials for selective small-molecule adsorption, revealing cooperative bonding patterns difficult to detect with conventional methods. Together, these examples show how combining theory and data-driven tools can deepen physical understanding, guide experiments, and accelerate the design of next-generation energy materials.