Model Error Assessment in Computational Physical Models

While parametric uncertainty quantification is performed routinely in computational science, model error assessment is still largely lacking and faces significant challenges that are yet to be resolved. For example, it is unclear how to properly disambiguate data noise from model errors. Recently there has been growing interest and some development towards the quantification of model errors, i.e. the inherent discrepancies from the truth that cannot be reduced within the model assumptions.
 

This minisymposium focuses on state-of-the-art and novel methods for quantification of model errors. It highlights challenges and introduces a range of available methods to the computational mechanics community, hoping to contribute significantly to improving predictive fidelity of computational physical models across a range of disciplines.