AM Seminar: Machine Learning in Molecular Simulations: From Free Energy to Vibrational Spectroscopy

Presenter: Marcos Calegari Andrade, Assistant Professor, Chemistry and Biochemistry, UC Santa Cruz

Description: In this talk, I will demonstrate how neural networks can represent the high-dimensional potential energy surfaces of many-body systems. By achieving the accuracy of first-principles quantum calculations at a fraction of the computational cost, these models enable atomistic simulations of condensed matter at unprecedented scales. I will discuss how this approach provides a detailed lens into chemical reaction dynamics under experimentally relevant conditions and facilitates the direct calculation of observables, such as vibrational spectra, effectively bridging the gap between theoretical simulation and experimental observation.

About the speaker: Marcos Calegari Andrade is an Assistant Professor in the Department of Chemistry and Biochemistry at the University of California, Santa Cruz. He earned his PhD from Princeton University, where he developed machine learning models to simulate the chemistry and vibrational spectroscopy of condensed-phase systems. He later joined the Quantum Simulations Group at Lawrence Livermore National Laboratory, applying deep neural network models to fundamental challenges in climate and energy security. His current research at UCSC focuses on the application of machine learning to molecular simulations, with a particular emphasis on chemical reaction mechanisms, vibrational spectroscopy, and the development of automated simulation frameworks.

This seminar is hosted by Applied Mathematics