About this Abstract |
Meeting |
2024 TMS Annual Meeting & Exhibition
|
Symposium
|
Algorithm Development in Materials Science and Engineering
|
Presentation Title |
Applications of Persistent Homology for Microstructure Quantification |
Author(s) |
Simon Mason, Stephen Niezgoda, Dennis Dimiduk |
On-Site Speaker (Planned) |
Simon Mason |
Abstract Scope |
In addition to more commonly used state and geometric microstructure descriptors, persistent homology-based topological analysis of microstructure summarizes the changing connectivity of a system across multiple length-scales. These persistence summaries are invariant to translation, rotation, shear, and magnification, describing the global structure of local information. Persistent homology can be applied to many material systems, including multi-phase and polycrystalline microstructures. Due to its robust construction, it can analyze experimental data without the need for prior processing or image segmentation. Additionally, the flexibility of construction allows for a wide range of persistence spaces to be explored, including descriptive metrics or orientation-informed filtration. Quantitative analysis of persistence summaries can be used to develop statistical goodness-of-fit tests for describing the distribution of topological features within a microstructure system and differentiating between similar systems. These principles can be applied to microstructure classification, manifold exploration, and machine learning loss functions. |
Proceedings Inclusion? |
Planned: |
Keywords |
Computational Materials Science & Engineering, Characterization, |