Supporting Dataset for Benchmarking Universal Machine-Learned Interatomic Potentials for High-Temperature Metal-Organic Framework Chemistry

This dataset contains computational data supporting the research article <i>"Benchmarking Universal Machine-Learned Interatomic Potentials for High-Temperature Metal-Organic Framework Chemistry."</i> The repository includes ab initio molecular dynamics (AIMD) trajectories, CP2K input files, and analysis scripts used in the study.<b>File Format &amp; Access </b>The complete dataset is provided as a compressed <code>.tar.xz</code> archive. Users must extract this archive to access the directory structure and files described below.<b>Content Overview </b>Once extracted, the repository is organised into three primary directories covering analysis tooling, AIMD trajectory data, and simulation input files:<code>1-scripts/</code>: Contains a comprehensive suite of Python scripts for analysing DFT trajectories and evaluating model performance. This directory houses:Evaluation utilities for calculating Mean Absolute Error (MAE) across models (<code>evaluate.py</code>).Core analysis functions for computing carbon cluster size, molecule counts, ring statistics, and radial distribution functions (RDFs) (<code>analysis_functions.py</code>, <code>run_all.py</code>, <code>plot.py</code>).Structural analysis scripts for metal coordination numbers (<code>calc_coord_num.py</code>, <code>plot_cn.py</code>) and root-mean-square deviation (RMSD) calculations over time (<code>RMSD_calc.py</code>, <code>plot_RMSD.py</code>).Training set generation and sampling utilities (<code>gen_training_sets.py</code>, <code>sampling_functions.py</code>).RDF comparison plotting between crystalline references and final structures at each temperature (<code>first_last_RDF_plot.py</code>).Simulation files for running 1 ns quench simulations with orbital (<code>quench_files/</code>).<code>2-trajectories/</code>: Contains XYZ trajectory files for each AIMD simulation, with embedded energies, forces, and stress tensors. Trajectories are provided for nine metal-organic frameworks: CALF-20, MIP-206, MOF-5, MOF-10, NU-1000, UiO-66, UiO-66-NH2, UiO-67, and ZIF-8.<code>3-CP2K_inputs/</code>: Contains the CP2K input files used to run AIMD simulations of each MOF at 2000 K, with one input file per framework matching those listed in the trajectories directory.For a detailed breakdown of the file tree and specific script descriptions, please refer to the <code>README.md</code> included within the root of the extracted archive.