Lehigh researchers are developing new computational methods to accelerate the design of materials critical to energy, sustainability, and advanced manufacturing.
Supported by the Department of Energy’s Office of Advanced Scientific Computing Research, the project brings together Akwum Onwunta (pictured), an assistant professor of industrial and systems engineering, and Lifang He, an associate professor of computer science and engineering, with lead researcher Chinedu Ekuma, an associate professor of physics in Lehigh’s College of Arts and Sciences, and Bao Wang, an assistant professor of mathematics at the University of Utah.
The team is creating scientific machine learning (SciML) algorithms that can handle the enormous complexity of high-dimensional materials data. By combining “deep learning–assisted nonnegative matrix factorization” with diffusion models, their framework is designed to reveal hidden structure-property relationships and make accurate predictions about material behaviors. Unlike traditional “black box” models, this physics-informed approach emphasizes interpretability and addresses data scarcity, making the results more transparent and broadly applicable.
An open-source platform will share these tools with the wider research community, extending the project’s impact across disciplines.