Friday, March 16
Physical Sciences A, room 107
Using computational models to help make scientific inferences is becoming increasingly commonplace in the physical and engineering sciences. Of key importance is quantifying the uncertainty in such predictions. This is complicated by the fact that the models are never perfect representations of reality. This talk will give a brief overview of some relevant issues from the fields of validation and uncertainty quantification for computational models. The talk will then go on to describe some approaches for assessing uncertainty in model-based predictions. Examples from cosmology and climate will be considered.
Dave Higdon is a member of the Statistical Sciences Group at Los Alamos National Laboratory. He is an expert in Bayesian statistical modeling of environmental and physical systems. He has also led numerous programmatic efforts at LANL in the quantification of margins and uncertainties and uncertainty quantification. His recent research has focused on simulation-aided inference in which physical observations are combined with computer simulation models for prediction and inference. His research interests include space-time modeling; inverse problems in physics, hydrology and tomography; inference based on combining deterministic and stochastic models; multiscale models; parallel processing in posterior exploration; statistical modeling in physical, environmental and biological sciences; Monte Carlo and simulation based methods.