Unsupervised topological learning for identification of atomic structures

TL;DR

This paper introduces an unsupervised topological data analysis approach to identify atomic structures in materials, enabling autonomous analysis of complex atomic-scale phenomena without prior knowledge.

Contribution

The paper presents a novel unsupervised learning method using TDA descriptors and Gaussian mixture models for atomic structure identification.

Findings

Successfully applied to Zr in crystalline and liquid states

Enabled autonomous detection of nucleation events

Facilitates deeper analysis of atomic-scale phenomena

Abstract

We propose an unsupervised learning methodology with descriptors based on Topological Data Analysis (TDA) concepts to describe the local structural properties of materials at the atomic scale. Based only on atomic positions and without a priori knowledge, our method allows for an autonomous identification of clusters of atomic structures through a Gaussian mixture model. We apply successfully this approach to the analysis of elemental Zr in the crystalline and liquid states as well as homogeneous nucleation events under deep undercooling conditions. This opens the way to deeper and autonomous study of complex phenomena in materials at the atomic scale.