About this Abstract |
Meeting |
2024 TMS Annual Meeting & Exhibition
|
Symposium
|
Algorithm Development in Materials Science and Engineering
|
Presentation Title |
Exascale Simulations Using Ultra-fast Force Field for Materials Discovery and Design |
Author(s) |
Richard G. Hennig, Ajinkya C. Hire, Jason B. Gibson, Hendrik Kraß, Ming Li, Pawan Prakash, Stephen R. Xie, Matthias Rupp |
On-Site Speaker (Planned) |
Richard G. Hennig |
Abstract Scope |
Accelerating the search for new materials and synthesis pathways and utilizing available data requires artificial intelligence (AI) workflows and machine learning techniques that identify favorable candidate structures, determine their stability, and rapidly predict phase diagrams for synthesis and materials properties for applications. This talk will discuss AI-based workflows that aim to identify novel metastable materials. To accelerate materials design and discovery, we develop the ultra-fast force field method available at https://github.com/uf3, which combines effective many-body potentials in a cubic B-spline basis with regularized linear regression to obtain machine-learning potentials that are physically interpretable, sufficiently accurate for applications, and as fast as the fastest traditional empirical potentials. We show that these potentials are ideally suited for exascale simulations of millions of atoms and are faster than state-of-the-art machine-learning potentials and discuss workflows that automate the model training, validation, and uncertainty quantification. |
Proceedings Inclusion? |
Planned: |
Keywords |
Machine Learning, Computational Materials Science & Engineering, Modeling and Simulation |