GemNet: Universal Directional Graph Neural Networks for Molecules

Johannes Gasteiger, Florian Becker, Stephan Günnemann

Keywords: GNN, graph neural network, molecules, equivariance, point clouds, weight initialization, normalizer-free, message passing

TL;DR: We prove the universality of GNNs with geometric message passing for rotationally equivariant predictions, and propose the GemNet architecture for molecular dynamics predictions

Abstract: Effectively predicting molecular interactions has the potential to accelerate molecular dynamics by multiple orders of magnitude and thus revolutionize chemical simulations. Graph neural networks (GNNs) have recently shown great successes for this task, overtaking classical methods based on fixed molecular kernels. However, they still appear very limited from a theoretical perspective, since regular GNNs cannot distinguish certain types of graphs. In this work we close this gap between theory and practice. We show that GNNs with directed edge embeddings and two-hop message passing are indeed universal approximators for predictions that are invariant to translation, and equivariant to permutation and rotation. We then leverage these insights and multiple structural improvements to propose the geometric message passing neural network (GemNet). We demonstrate the benefits of the proposed changes in multiple ablation studies. GemNet outperforms previous models on the COLL, MD17, and OC20 datasets by 34%, 41%, and 20%, respectively, and performs especially well on the most challenging molecules. Our implementation is available online.

Code Of Conduct: I certify that all co-authors of this work have read and commit to adhering to the NeurIPS Statement on Ethics, Fairness, Inclusivity, and Code of Conduct.

Code: https://www.daml.in.tum.de/gemnet