Dataset, models, and scripts for the prediction of electronic properties of nano-porous graphene with DFT and ML

• band_gap_training: configuration and property files for the training of the band gaps using ALIGNN (https://github.com/usnistgov/alignn) and DeeperGATGNN (https://github.com/usccolumbia/deeperGATGNN) codes
• band_structures: raw data for the band structures of the relaxed geometries in CP2K format
• database: database of band gaps and other quantities calculated for the NPG structures (see 00_README file in that folder for explanations of the quantities)
• forces_and_energies: forces and energies of all distorted NPG structures in extxyz format
• geometries: relaxed and distorted NPG and graphene geometries used for training of band gaps and interatomic potential (only a part of the structures was used for the latter)
• MACE_training: training and test data sets used in training of interatomic potential
• models: trained MACE and ALIGNN models; MACE model in torch format, ALIGNN model in zip format
• phDOS: raw data for the phonon DOS calculated using the relaxed structures
• scripts: python scripts used to perform geometry optimizations and molecular dynamics, phonon DOS calculations, and fitting of the temperature dependent band gaps

Versions for different codes:

• MACE: 0.3.4 (Python 3.9.19)
• ALIGNN: 2024.4.10 (Python 3.10.14)
• deeperGATGNN: 1.0 (Python 3.9.19)
• CP2K: 2024.1