qai222 / crystalgraph Goto Github PK

Finite graphs for describing bond topology of crystal structures.

For more information about this representation, see the section of Net and quotient graph in j.patter.2022.100588.

1. Install pymatgen: conda install -y numpy scipy matplotlib; pip install pymatgen
2. To visualize graph with parallel edges you would need graphviz.
3. Install this package by pip install crystalgraph.

Some examples can be found in the folder examples, specifically:

1. qg_base.ipython describes basic usage of quotient graph class.
2. qg_cif.ipython convert CIFs to quotient graphs, and contract atom nodes to building units.

• basic QG classes
• QG visulaization functions
• QG features, such as dimensionality
• enumeration of LQGs from a UQG
• QG embedding functions
• string representation

1. Allow parallel edges and self-loops. I don't know how to do this in SMILES. For SELFIES it seems we can just add two characters with Q=1 and Q=0, i.e. parallel edges are rings of size 2 and self-loops are rings of size 1.
2. For inorganic structures we can extend alphabet to include building units.
3. From CIF to string, bond type has to be determined for each edge. This is largely solved for organics but I'm not sure there is a robust method for inorganic structures, especially when hydrogen atoms are missing.
4. Strictly speaking, finite graph is derived for a specific cell. That means two cells representing the same crystal could give two different finite graphs. So we need to either start from the most reduced cell, or find a way to reduce the finite graph (or the string representation itself).