In this webinar we will introduce the Open Catalyst Project. The Open Catalyst Project provides datasets and pre-trained machine learned models that can be used to supplement or enhance the use of density functional theory in studies of catalysts and materials. There are three common ways to use OCP in these tasks we will explore in the workshop. The first is the direct use of pre-trained models for predictions of a variety of tasks including predictions of energy, forces, or adsorption energies from structures. This approach is useful when exploring systems that are very similar to the dataset the models have been trained on. The second approach is to use these models to accelerate DFT calculations, especially for geometry optimization. Finally, the third approach is to build your own fine-tuned MLP by training on your own dataset of DFT calculations. We will work through examples in each category, and talk about how to assess the results.
Speakers: Prof. John Kitchin (Carnegie Mellon University) and Dr. Zachary Ulissi (Meta’s Fundamental AI Research Lab)
To ask your questions for the Q&A session, Slido link: [ Ссылка ]
