Crystal Generative Framework Based on Wyckoff Generative Adversarial Network

In this study, we present the Crystal Generative Framework based on the Wyckoff Generative Adversarial Network (CGWGAN).

All templates with 3-4 asymmetric units generated in our work are available as open-source resources in the CGWGAN datasets.

Python Implementation

from ase.db import connect

database = connect('open.db')
entry_id = 1  # The crystal index 
atoms = database.get_atoms(id=entry_id)

# Chemical symbols
symbols = atoms.get_chemical_symbols()
# Volume
latt_vol = atoms.get_volume()
# Fractional positions
positions = atoms.get_scaled_positions()
# etc...

Operating and Displaying the DB File

# Install CryDBkit
pip install CryDBkit

from CryDBkit import website

website.show('open.db')

@misc{caobin_2024,
    author = {Cao Bin},
    title = {CGWGAN (Revision 3415d7a)},
    year = 2024,
    url = {https://huggingface.co/datasets/caobin/CGWGAN},
    doi = {10.57967/hf/3002},
    publisher = {Hugging Face}
}