# Lightweight Geometric Deep Learning for Molecular Modelling in Catalyst Discovery

arXiv: https://arxiv.org/abs/2404.10003 · HTML: https://mlsanity.com/paper/2404.10003/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 10 of its 13 references that are in the map.
- First posted on arXiv 5 Apr 2024.

## Builds on (most cited first, 10 of 10)

- [A Comprehensive Survey on Graph Neural Networks](https://mlsanity.com/paper/1901.00596/) (2019)
- [SchNet: A continuous-filter convolutional neural network for modeling quantum interactions](https://mlsanity.com/paper/1706.08566/) (2017)
- [GemNet: Universal Directional Graph Neural Networks for Molecules](https://mlsanity.com/paper/2106.08903/) (2021)
- [Fast and Uncertainty-Aware Directional Message Passing for Non-Equilibrium Molecules](https://mlsanity.com/paper/2011.14115/) (2020)
- [The Open Catalyst 2020 (OC20) Dataset and Community Challenges](https://mlsanity.com/paper/2010.09990/) (2020)
- [Geometric Deep Learning on Molecular Representations](https://mlsanity.com/paper/2107.12375/) (2021)
- [An Introduction to Electrocatalyst Design using Machine Learning for Renewable Energy Storage](https://mlsanity.com/paper/2010.09435/) (2020)
- [Towards Training Billion Parameter Graph Neural Networks for Atomic Simulations](https://mlsanity.com/paper/2203.09697/) (2022)
- [PQK: Model Compression via Pruning, Quantization, and Knowledge Distillation](https://mlsanity.com/paper/2106.14681/) (2021)
- [Graph Neural Networks for Molecules](https://mlsanity.com/paper/2209.05582/) (2022)

## Abstract (arXiv)

New technology for energy storage is necessary for the large-scale adoption of renewable energy sources like wind and solar. The ability to discover suitable catalysts is crucial for making energy storage more cost-effective and scalable. The Open Catalyst Project aims to apply advances in graph neural networks (GNNs) to accelerate progress in catalyst discovery, replacing Density Functional Theory-based (DFT) approaches that are computationally burdensome. Current approaches involve scaling GNNs to over 1 billion parameters, pushing the problem out of reach for a vast majority of machine learning practitioner around the world. This study aims to evaluate the performance and insights gained from using more lightweight approaches for this task that are more approachable for smaller teams to encourage participation from individuals from diverse backgrounds. By implementing robust design patterns like geometric and symmetric message passing, we were able to train a GNN model that reached a MAE of 0.0748 in predicting the per-atom forces of adsorbate-surface interactions, rivaling established model architectures like SchNet and DimeNet++ while using only a fraction of trainable parameters.