Robustness test of the spacegroupMining model for determining space groups from atomic pair distribution function data

TL;DR

This study evaluates the robustness of a CNN-based model for predicting crystal space groups from atomic pair distribution functions across varying experimental parameters, confirming consistent accuracy.

Contribution

It demonstrates that the spacegroupMining model maintains high accuracy despite changes in key experimental parameters, highlighting its robustness.

Findings

Top-1 and top-6 accuracy are robust across different experimental parameters.

Model performance remains stable with variations in $r_ ext{max}$, $Q_ ext{max}$, $Q_ ext{damp}$, and atomic displacement parameters.

Supports the model's applicability in diverse experimental conditions.

Abstract

Machine learning models based on convolutional neural networks have been used for predicting space groups of crystal structures from their atomic pair distribution function (PDF). However, the PDFs used to train the model are calculated using a fixed set of parameters that reflect specific experimental conditions, and the accuracy of the model when given PDFs generated with different choices of these parameters is unknown. In this paper, we report that the results of the top-1 accuracy and top-6 accuracy are robust when applied to PDFs of different choices of experimental parameters $r_\text{max}$, $Q_\text{max}$, $Q_\text{damp}$ and atomic displacement parameters.