<ul><li>Researchers are extending the concept of foundation models to scientific machine learning and computational science.</li><li>There is a lack of a universally accepted definition of foundation models in computational science, leading to confusion.</li><li>A position paper proposes a formal definition of foundation models in computational science based on generality, reusability, and scalability, in line with traditional foundational methods.</li><li>Introduction of the Data-Driven Finite Element Method (DD-FEM) combines classical FEM with data-driven learning, addressing scalability, adaptability, and physics consistency challenges in computational science.</li></ul>

Defining Foundation Models for Computational Science: A Call for Clarity and Rigor

Discover more