Foundations Models for PDEs

The goal of this effort is to build a foundation model for specific classes of partial differential equations (PDEs). PDEs are the the natural language of physical systems and describe a wide variety of physical phenomena. The aim is to develop and integrate new machine learning architectures with a focus on enhancing transfer across different physical regimes, incorporate multi-modal data (e.g. partial field measurements and integral quantities) to improve data efficiency, model fidelity and robustness.