Chemical Graph Transformer Neural Operator (CGTNO) is a neural operator for learning point cloud dynamics with specific application to molecular dynamics.

The notation in the paper generally corresponds to our comments, with the following caveats:

• Timesteps - P -> T

Please install with poetry install --with dev if you want type checking (i.e., for mypy).

Data sources

The same as used in EGNO: https://www.sgdml.org/

Recommended further reading:

1. EGNO: https://arxiv.org/pdf/2401.11037
   • General graph neural operator architecture
2. https://arxiv.org/pdf/2203.06442
   • Dataset setup in more detail
3. http://www.sgdml.org/
   • Dataset source
4. https://arxiv.org/pdf/2302.14376
   • Heterogenous attention - Gave inspiration for our approach
5. https://arxiv.org/pdf/2108.08481
   • Neural operator theory