The Open Catalyst 2020 dataset of computational catalyst simulations has enabled a new generation of large generalizable machine learning models. While still far from chemical accuracy in all materials, these models can already be leveraged to accelerate day-to-day tasks in computational catalysis workflows. I will highlight three recent examples from my group that leverage these pretrained models: to accelerate DFT simulations with on-the-fly active learning, to improve adsorbate placement, and to fit smaller computational datasets through transfer learning. All examples are open-source resources available to the broader community.

Leveraging large pre-trained machine learning potentials for computational catalysis

Fall 2022 abstracts, posters, and presentations will launch in mid-September.

ACS Fall 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.