Atomistic Simulations with Machine Learning: From Dynamics to Vibrational Spectroscopy

Atomistic simulations with machine learning: From dynamics to vibrational spectroscopy

Author(s): Dr. Philipp Schienbein, Prof. Dr. Jörg Behler 

Publication: Bunsen-Magazin 2026, 6, 139-141

Publisher: Deutsche Bunsen-Gesellschaft für physikalische Chemie e.V., Frankfurt

Language: English

DOI: 10.26125/ahhp-k814

Abstract: We highlight advances in atomistic simulations with machine learning. Four generations of machine learning potentials progress from small systems and local environments to long-range interactions and charge transfer. Beyond energies and forces, learned electronic properties, including atomic polar tensors (APTs), enable vibrational spectroscopy and electric-field simulations, opening a route to atomistic modeling of pump–probe experiments.

Cite this:  P. Schienbein, J. Behler, Bunsen-Magazin 2026, 6, 139-141, DOI: 10.26125/ahhp-k814

Get the full article (pdf)