Bayesian methods are uniquely well-suited for uncertainty analysis of measurements in the physical sciences because typically there is a large amount of auxiliary information that is difficult to include in classical models of the measurement process, but which Bayesian models can incorporate very easily. Furthermore, the metrological guide to uncertainty evaluation, the GUM, has recently been interpreted in a Bayesian manner. This research program focuses on both foundational and practical issues that arise in Bayesian uncertainty analysis. Some of the foundational issues concern the statistical interpretation of metrological terminology and of uncertainty assessments. The practical issues include computation for Bayesian inference using Markov Chain Monte Carlo to propagate information in graphical models for measurement situations. Work is also in progress on Bayesian models for uncertainty assessment for measurements with high-dimensional and functional data, for measurements on complex systems, and for virtual measurements.
Bayesian metrology; Complex systems; High-dimensional data; MCMCD; Metrics; Uncertainty assessment;