Workshop on Establishing Benchmarks for
Data-Driven Modeling of Physical Systems

Over the last decade the field of data-driven modeling for physical systems has sparked tremendous interest and undergone rapid growth. This has led to the development of many new algorithms that have shown initial promise. This rapid growth has made it imperative that a set of benchmarks be established so that:

• Developers of models can quickly access FAIR data for training and testing their algorithms.

• New models can be consistently compared against existing algorithms.

• Users/adopters of these algorithms can easily determine which algorithms are best suited for their needs.

With this as motivation, this workshop brings together experts in this field to establish a set of benchmarks. The initial goal will be to discuss appropriate benchmarks for:

• Operator learning.

• Inference of dynamical systems.

• Modeling in high-dimensional space.

Speakers are invited to present talks that

• Present novel ML algorithms in these areas and describe accompanying benchmark problems, or

• Focus predominantly on appropriate benchmark problem in these areas.

This workshop is sponsored by the Data-Driven Modeling Technical Thrust Area of the U. S. Association for Computational Mechanics.

Organizers:

Assad Oberai, University of Southern California

Alireza Doostan, University of Colorado Boulder

Paris Perdikaris, University of Pennsylvania

Jian-Xun Wang, Notre Dame University