Ceramic and glass materials play critical roles in many aspects of today’s ecosystems, from transportation and architecture to energy storage and medicine, to name a few. While these materials have been used for thousands of years, there is still much we do not understand about them, and thus there are numerous opportunities to improve their design and application.

Data-driven computer simulations have enabled rapid research and development of new materials with improved properties by reducing the need for numerous experimental samples. But mathematically modeling the relationship between a glass composition and its properties remains a challenge for several reasons, such as lack of knowledge concerning structural units for certain glass compositions and understanding how model uncertainties relate to experimental noise.