Graph Neural Networks

Package for generating and inverse-designing 2D lattice materials. Represents lattices as heterogeneous graphs and utilizes message passing, automatic differentiation and surrogate gradients for the inverse design.

This repo is for the code relating to the paper: "GNN-EvoDevo: Bioinspired Generative Design via Evolutionary Development and Graph Neural Networks

Code repository for GAME-Net (Graph-based Adsorption on Metal Energy-neural Network), a graph neural network for predicting the adsorption energy of closed-shell molecules on metal surfaces.