Menu
Predict material properties of crystal structures with CGCNN models.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Prepare, validate, and submit UMA/fairchem machine-learning interatomic potential calculations on Genkai with the established PJM launcher and UMA virtual environment. Use for MLIP or UMA structure relaxation, molecular dynamics, energy/force inference, GPU calculations, PJM submission, restart preparation, or locating and reporting calculation outputs.
Read surface-research PDFs or JSON records and extract structured reaction, material, and modeling information for downstream agent workflows.
Build realistic oxide-surface candidates: oxygen-vacancy landscapes, adsorbate coverage landscapes, and metal-cluster-on-surface starting structures using ASE, Optuna, and optional UMA/FAIRChem calculators.
Filter and normalize paperread surface-extraction outputs into Agent-ready modeling inputs, including facet equivalence, loaded nanoparticle species, material class, reaction type, and executable modeling task mapping.
Prepare, stage, and summarize LOBSTER bonding-analysis calculations from completed VASP, CP2K, or Quantum ESPRESSO outputs.
Skill for executing ABACUS DFT materials calculations. Help users set up, run, and analyze ABACUS calculations.
| name | cgcnn-predictor |
| description | Predict material properties of crystal structures with CGCNN models. |
| metadata | {"tools":[],"dependent_skills":[]} |
Use this skill to predict crystal properties with a local CGCNN setup, including regression targets such as formation energy, band gap, Fermi energy, or elastic properties, as well as classification tasks such as metal versus semiconductor.
If the help command runs without import errors, the environment is ready.
Prepare the dataset directory as:
dataset_dir/
├── id_prop.csv
├── atom_init.json
├── id0.cif
├── id1.cif
├── ...
Requirements:
id_prop.csv must have two columns: crystal ID and target value.cif suffixatom_init.json should usually be copied from data/sample-regression/atom_init.json under CGCNN_ROOTThis dataset layout is required by the upstream CGCNN prediction workflow.
Common checkpoints in pre-trained/ include:
formation-energy-per-atom.pth.tar: formation energy, unit eV/atomfinal-energy-per-atom.pth.tar: absolute energy, unit eV/atomband-gap.pth.tar: band gap, unit eVefermi.pth.tar: Fermi energy, unit eV/atombulk-moduli.pth.tar: bulk modulus, unit log(GPa)shear-moduli.pth.tar: shear modulus, unit log(GPa)poisson-ratio.pth.tar: Poisson ratio, dimensionlesssemi-metal-classification.pth.tar: classification model for metal vs semiconductor, no physical unitRun prediction from cgcnn dir:
python predict.py <model_path> <dataset_dir>
Examples:
source "${cgcnn_env}/bin/activate"
cd "${CGCNN_ROOT}"
python predict.py pre-trained/formation-energy-per-atom.pth.tar dataset_dir
The upstream CGCNN run writes test_results.csv under ${CGCNN_ROOT}. After prediction, always copy that file into the dataset_dir. For classification tasks, predicted values are probabilities between 0 and 1.
If the predicted property is bulk modulus or shear modulus, convert the prediction to the standard unit GPa after inference.
test_results.csv