| Description |
Venkat Kapil (University of Cambridge) Abstract:
Simulating complex materials, particularly interfacial and confined systems, is challenging due to the quantum mechanics of electrons and nuclei. In this talk, I will present progress in developing efficient and accurate methods for incorporating electronic and quantum effects. I will cover highly data-efficient approaches for creating machine learning potentials using only a few tens of structures [1] and equivariant models of electronic properties, such as polarization and polarizability tensors for spectroscopy [2]. Finally, I will discuss the development of secondary potentials that account for quantum nuclear corrections to Born-Oppenheimer potentials and simulate approximate quantum dynamics at a classical cost [3]. These developments make full quantum simulations computationally viable.
[1] Kaur, Della Pia, ..., & Kapil (2024). Faraday Discuss. [2] Kapil, Kovács, Csányi, & Michaelides (2023). Faraday Discuss. [3] Musil, Zaporozhets, ..., & Kapil. (2022) J. Chem. Phys.
Zoom registration link:
https://zoom.us/meeting/register/tJMkdeihpzsrHtBYYljZS-OwT83Uo-nFYfFG |
CMSP Lecture (Atomistic Simulation Seminar Series): Machine learning for full quantum simulations
Go to day