About this Abstract |
| Meeting |
2024 TMS Annual Meeting & Exhibition
|
| Symposium
|
Computational Thermodynamics and Kinetics
|
| Presentation Title |
Prediction of Material Properties by Integrating Molecular Dynamics and Machine Learning Approaches |
| Author(s) |
Yasushi Shibuta |
| On-Site Speaker (Planned) |
Yasushi Shibuta |
| Abstract Scope |
In recent years, spatial scale that can be handled by numerical simulations has been greatly extended due to the dramatic improvement of computing environment. As simulations become larger in scale, post-analysis, storage, and reuse of the huge amount of data output from molecular dynamics (MD) simulations are becoming a new challenge. Therefore, we have been investigating new approaches for efficient prediction of materials properties by integrating MD and machine learning (ML) methods. For example, a graph convolutional networks (GCN)-based ML model is constructed to predict physical properties of metallic materials. Moreover, we developed a new method based on a deep generative model to predict the microstructure that cannot be reproduced on the time scale of MD simulations. In the presentation, these latest studies will be introduced. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, Modeling and Simulation, Machine Learning |