Abstract by Derek Hensley
Physics and Astronomy
Mapping Grain Boundary Metastable States with Unsupervised Learning
Grain Boundaries (GBs) influence most of the physical properties observed in metals. While exploring the entire GB space is infeasible due to their complexity, I look to simplify this by mapping the metastable states of specific GB subsets. I accomplished this by applying unsupervised machine learning techniques to cluster these subsets. From this analysis, I found that a subset of 1797 GBs could be reduced to around 20 clusters, and another subset of 508 down to 20 clusters as well. With these methods, the GB space can be greatly simplified so as to be much easier to research and explore.