Semi-intrusive Uncertainty Propagation for multiscale models

Anna Nikishova and Alfons G. Hoekstra have published an article on semi-intrusive Uncertainty Propagation for multiscale models in the Journal of Computational Science: In this work, an efficient way to estimate uncertainty in the results of multiscale models is discussed. The proposed methods are based on the idea of the Multiscale Modeling and Simulation Framework, where single scale models are connected using a coupling method. This allows us to consider the multiscale models as a collection of single scale black box models and, therefore, propose different methods to improve the efficiency of the uncertainty estimation. The methods are tested on two application based on reaction-diffusion dynamics.