Unsupervised learning of representative local atomic arrangements in molecular dynamics data

TL;DR

This paper introduces an unsupervised approach combining dimensionality reduction and clustering to analyze local atomic arrangements in molecular dynamics data, enabling detailed characterization of molecular environments.

Contribution

It presents a novel method integrating UMAP and HDBSCAN for unsupervised analysis of local atomic environments in MD data, reducing data complexity and revealing structural isomer families.

Findings

Successfully characterized cation coordination in electrolytes

Efficiently identified molecular formula families within MD data

Enhanced understanding of local atomic arrangements

Abstract

Molecular dynamics (MD) simulations present a data-mining challenge, given that they can generate a considerable amount of data but often rely on limited or biased human interpretation to examine their information content. By not asking the right questions of MD data we may miss critical information hidden within it. We combine dimensionality reduction (UMAP) and unsupervised hierarchical clustering (HDBSCAN) to quantitatively characterize prevalent coordination environments of chemical species within MD data. By focusing on local coordination, we significantly reduce the amount of data to be analyzed by extracting all distinct molecular formulas within a given coordination sphere. We then efficiently combine UMAP and HDBSCAN with alignment or shape-matching algorithms to partition these formulas into structural isomer families indicating their relative populations. The method was employed to reveal details of cation coordination in electrolytes based on molecular liquids.