1st TIMES Workshop joint with ETSF

Machine-Learning Potentials in Computational Chemistry

7 Apr 2025, 15:40

40m

Bell Lecture Theatre (Main Physics Building)

Talk Session 1

Speaker

David Wilkins (Queen's University Belfast)

Description

In recent years, machine learning (ML) approaches to circumvent quantum-mechanical calculations have become increasingly popular, allowing atomistic simulations with the accuracy of electronic structure theory to be run in a fraction of the time needed for ab inition molecular dynamics (AIMD).

I will describe the background and basics of these approaches, explaining the theory behind ML predictions of the interatomic forces and how to get started on training an ML potential.