Uncertainty quantification of multiscale models

Uncertainty quantification (UQ) is a scientific field, which supports decision making using computational models that involve uncertainties. This PhD thesis focuses on UQ algorithms for multiscale models. We have proposed semi-intrusive multiscale UQ methods, in which the single scale components are considered as black-boxes that can be substituted and which coupling may be modified. The efficiency of the semi-intrusive methods is demonstrated by applying them to a multiscale model of in-stent restenosis. Additionally, an application of sensitivity analysis for efficient UQ for multiscale models is discussed. Thus, in the proposed method, sensitivity analysis is applied to a computationally cheap single scale model in order to reduce the dimensionality of the multiscale model input.