Materials Fingerprinting Classification

Significant progress in many classes of materials could be made with the availability experimentally-derived large datasets composed atomic identities and three-dimensional coordinates. Methods for visualizing local structure, such as atom probe tomography (APT), which routinely generate comprised millions atoms, are an important step realizing this goal. However, state-of-the-art APT instruments noisy sparse that provide information about elemental type, but obscure structures, thus limiting their subsequent value discovery. The application a fingerprinting process, machine learning algorithm coupled topological data analysis, provides avenue by here-to-fore unprecedented structural can extracted from dataset. As proof concept, material fingerprint is applied to high-entropy alloy containing body-centered cubic (BCC) face-centered (FCC) crystal structures. A configuration centered on arbitrary assigned descriptor, it characterized BCC or FCC lattice near perfect accuracy, despite inherent noise This successful identification crucial first development algorithms extract more nuanced information, chemical ordering, existing complex materials. Program Title: Materials Fingerprinting CPC Library link program files: https://doi.org/10.17632/2fhch3x85m.1 Developer's repository link: https://github.com/maroulaslab/Materials-Fingerprinting Licensing provisions: GPLv3 Programming language: Python Supplementary material: user manual examples provided source code GitHub repository. Nature problem: Atom sub-nanometer resolution material, due sparsity introduced structure cannot presently determined resulting data. Solution method: Our library presents topologically informed methodology classify We create persistence diagrams small neighborhoods at each use summary statistics novel metric space features classification algorithm.