Penalty-free, accelerated and scalable training and optimization

George Mason University received a $813,750 Project Grant from the Air Force Research Laboratory under the Air Force Defense Research Sciences Program (CFDA 12.800) awarded on October 1, 2025, with an ultimate completion date of September 30, 2028. The project, titled "Penalty-Free, Accelerated and Scalable Training and Optimization," supports fundamental and applied scientific research designed to maintain technological superiority for the U.S. Air Force and Space Force.

The primary deliverables include the development of penalty-free algorithms for nonsmooth, nonconvex optimization problems, led by Sandia National Laboratories (sub-awarded on April 23, 2026). Specific technical contributions encompass the development and implementation of inexact proximal-regularized trust-region methods for risk-averse optimization, mechanical contact problems, and elasto-plastic torsion applications. Additionally, the project will develop parallel-in-time computational capabilities to accelerate algorithm performance and provide software for electromagnetic systems design. Deliverables will be disseminated through peer-reviewed journal publications, invited conference presentations, and high-performance open-source software, representing a collaborative effort between George Mason University and Sandia National Laboratories to advance optimization methodologies applicable to Department of the Air Force mission requirements.