ClasSOMfier: A neural network for cluster analysis and detection of lattice defects

TL;DR

ClasSOMfier is a fast, unsupervised neural network-based software for classifying atoms into clusters and detecting lattice defects without prior knowledge of atomic environments.

Contribution

It introduces a Kohonen network implementation for lattice defect detection that operates without pre-defined atomic environment patterns.

Findings

Efficiently classifies atoms into clusters and detects defects.

Operates without prior knowledge of lattice structures.

Provides a user-friendly interface in Python.

Abstract

ClasSOMfier is a software package to classify atoms into a given number of disconnected groups (or clusters) and detect lattice defects, such as vacancies, interstitials, dislocations, voids and grain boundaries. Each cluster is formed by atoms whose atomic environment can be described by a common pattern. Unlike many methods available in the literature, where these patterns are given in advance and are associated with known lattice structures (i.e. fcc, bcc or hcp), this code implements a Kohonen network, which is based on unsupervised learning and where no information about the atomic environment has to be given in advance. ClasSOMfier accelerates the application of machine learning for cluster analysis by providing an efficient and fast code in Fortran with a user-friendly interface in Python.