Gaussian Processes for Structural Analysis

Gaussian processes provide a Bayesian framework to analyze experimental scattering data while performing rigorous uncertainty quantification and propagation (UQ/P). Our team uses state-of-the-art Gaussian process approaches, including spectral and non-stationary kernel design, to design next generation scattering analysis tools. This framework allows us to throw all available experimental data in the hopper while preserving established physical knowledge accumulated through generations of scientific innovation and discovery.