Increasing modeling detail is not necessarily correlated with increasing predictive ability. Setting modeling and numerical discretization errors aside, the more detailed a model gets, the larger the number of parameters required to accurately specify its initial/boundary conditions, constitutive laws, external forcing, object geometries, etc. To be predictive, we need to quantify this uncertainty by combining our prior physical knowledge with noisy experimental data obtained from various heterogeneous sources. Once we have quantified this uncertainty, all we need to do is propagate it through the model and obtain predictive error bars for any quantity of interest. What kinds of uncertainty do we encounter in physical models? How is uncertainty described mathematically? What can go wrong if uncertainty is ignored?
Dr. Ilias Bilionis is an Assistant Professor at the School of Mechanical Engineering, Purdue University. His research is motivated by energy and material science applications and it focuses on the development of generic methodologies for design and optimization under uncertainty, reliability analysis, model calibration and learning models out of data. Prior to his appointment at Purdue he was a Postdoctoral Researcher at the Mathematics and Computer Science Division (MCS), Argonne National Laboratory. He received his PhD in Applied Mathematics from Cornell University in 2013 and his Diploma in Applied Mathematics from the National Technical University of Athens in 2008.
Cite this work
Researchers should cite this work as follows:
129 Burton Morgan, Purdue University, West Lafayette, IN