This repository contains the data, scripts, and Jupyter notebooks associated with the manuscript "Resolving the body-order paradox of machine learning interatomic potentials" by Chong et al. (arXiv:2509.14146).

It is organized into the following:

== analysis_notebooks ==
This folder contains notebooks that were used to extract the model-specific body-order energy and force decomposition results, and also the scripts to generate the associated figures in the manuscript and its SI.

== evaluation_xyzs == 
This folder contains all the evalutions of the 8-mers and m-mers that have been done with the models that have been trained within the study.

== training_scripts ==
This folder contains the actual training scripts that have been used for the MLIP training for the study.

== training_xyzs ==
This folder contains the ASE-compatible extended xyz files that served as training, validation, and test sets for the MLIP training and evaluation.

== QMC_analysis_results ==
This folder contains the results of the QMC calculations performed for our study, the details of which appear in the Supporting Information of the manuscript above.