| About this Abstract |
| Meeting |
6th World Congress on Integrated Computational Materials Engineering (ICME 2022)
|
| Symposium
|
6th World Congress on Integrated Computational Materials Engineering (ICME 2022)
|
| Presentation Title |
Phonon Based Universal Sampling Method for Machine Learning Interatomic Potentials |
| Author(s) |
Nathan Wilson, Xiaofeng Qian, Raymundo Arroyave |
| On-Site Speaker (Planned) |
Nathan Wilson |
| Abstract Scope |
Recent advances in machine learning have accelerated the development of interatomic force fields using first-principles density functional theory calculations. Currently, the predominant sampling method is obtained directly from ab initio molecular dynamics. Such a brute force approach requires multiple sets of ab initio calculations with many heating/cooling, equilibration, and production trajectories using small time steps, which becomes computationally expensive and limits the number of generated structures. In this talk, we will propose a phonon-based sampling method using high order phonons to resolve this bottleneck by efficiently sampling the structure space. The higher order phonons combined with the self-consistent phonon approach can be used to accurately sample structures at higher temperatures, including phonon-stabilized structures, while avoiding brute force sampling of the vast structural space. |
| Proceedings Inclusion? |
Definite: Other |