Calibration, Validation, and Bayesian Approaches in Predictive Modeling and Simulation

The advent of ever larger computing resources has enabled computational scientists to begin considering tighter integration of experimental data with predictions made using mathematical models. Such integration includes Bayesian methods for calibrating parameters as probability density functions that incorporate uncertainty in experimental data, optimal design of experiments, and real-time assimilation of data. The treatment of data with modeling brings hosts of new challenges to the forefront: designing statistical methods that can leverage costly computational models, coping with vast amounts of experimental data both in algorithms and in hardware, new formulations for calibrations of PDEs when the parameter space is infinite dimensional, communicating and sharing large data sets and results of parameter calibrations, as well as many other issues. This minisymposium brings together researchers who are actively working on these issues directly or whose applications intersect these issues.