With advances in high-performance computing hardware and algorithms, computer simulation has become indispensable for scientific discovery and engineering design. However, tackling today’s grand challenges—such as the development of sustainable transportation systems, the design of personalized medicines, and the study of dark matter and dark energy—requires significant advances in our ability to simulate complex physical phenomena. Such scientific and safety-critical challenges demand reliable predictions with accuracy guarantees so that simulation can be confidently used in a truly predictive capacity and can inform crucial decisions with significant societal impacts. Synthesizing physics-based models derived by first principles in science and engineering with real-world data presents a promising paradigm for achieving predictive capabilities beyond what either alone can offer.

Bridging the Gap: From Computational Physics, to Physics-informed Machine Learning, to Data-driven Scientific Discovery, a DSI Emerging Data Science Program, is dedicated to bringing  together experts in numerical simulation and data science to explore the intersection and bridge the gap between physics-based models based on first principles in sciences/engineering and data-driven models based on machine learning techniques. The goal is to provide a collaborative platform that drives research towards truly predictive simulation capabilities that guide scientists and engineers making crucial decisions with high societal impact, ranging from sustainable design to medicine to astrophysics. The program is structured around four themes: data-informed computational physics; physics-informed machine learning; data-driven discovery of physics; and their mathematical foundations. The cross-disciplinary team comprises four co-leads and 16 members from nine departments with complementary expertise and experiences.