University of Cambridge > > Lennard-Jones Centre > Adding functionality to machine-learning potentials: going beyond accuracy and speed

Adding functionality to machine-learning potentials: going beyond accuracy and speed

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Eszter Varga-Umbrich.

Machine learning potentials (MLPs) have emerged in recent years as powerful tools for atomistic materials modeling. The pace of development of these MLPs has been formidable, and nowadays new or modified frameworks appear in the literature monthly. To some degree, the developments have focused on winning the race for speed and accuracy, somewhat sidelining the fundamental issues related to the locality (or “shortsightedness” of MLPs). At the same time, the flexibility of ML models allows us to combine MLPs, i.e., the modeling of the potential energy surface, with observables that can be directly compared with experiment, like spectroscopic measurements. In this presentation I will discuss work that we have done on these “augmented” MLP simulations, and how they can help us understand and predict the structure and properties of materials beyond the accurate (and fast) description of bonded interatomic interactions.

This talk is part of the Lennard-Jones Centre series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2024, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity