CrystalXRD project

An illustration that depicts the management of a crystal database.

Crystal Database (link)

The CryDB.db is a crystal database that contains 723,158 pieces of crystals:

• 154,718 crystals from the MP dataset, March 8, 2024
• 55,723 crystals from the Jarvis dataset, March 8, 2024
• 512,717 crystals from the COD dataset, May 8, 2024

demo.db

The demo.db shared here is a demo for illustrating the toolkit to manage the database.

Toolkit (see code.ipynb)

1. Show and Query Data

To show and query each data entry, use the CryDBkit package:

from CryDBkit import website

website.show('demo.db')

2. Graph Embedding of Each Data Entry

To perform graph embedding on each data entry, use the Crylearn package:

Python==3.9.19

from Crylearn import cry2graph
from ase.db import connect

database = connect('demo.db')
entry_id = 1

node_embedding, _, dis_matrix, XRDpattern = cry2graph.parser(database, entry_id).get(model='Simulation')

Parse the crystal by lattice cell. Each atom contained in the lattice is a node with a 106-dimensional embedding (N * 106). The distance between any pair of nodes is given in the distance matrix (N * N). XRDpattern is the simulated diffraction pattern of the crystal.

• node_embedding (np.ndarray): The node embeddings, 106-dimensional.
• distance_matrix (np.ndarray): The distance matrix in Cartesian coordinates.
• XRDpattern (np.ndarray): Global information about the graph, 3501-dimensional.