Project Grant 2237616

This five-year, $410,524 National Science Foundation project grant supports research and educational outreach activities aimed at advancing efficient global optimization methods under uncertainty. Funded through NSF's Engineering Directorate under the CFDA 47.041 program, the grantee will develop novel Bayesian optimization algorithms that exploit known problem structures to overcome efficiency barriers in optimizing extremely expensive functions. Specific research goals include optimizing with hybrid physics-based and data-driven models given noisy datasets, constrained multi-fidelity modeling fusing heterogeneous data sources, and scaling to high-dimensional problems through non-myopic and graph-structured formulations. Additional deliverables comprise workshops, games and K-12 outreach to educate students on decision-making under uncertainty. The primary awardee is The Ohio State University, performing work in Columbus, Ohio through its Office of Sponsored Programs. If successful, the improved optimization methods have potential for broad application across engineering challenges involving grey-box modeling.