This is a collection of benchmark models that I have used for uncertainty quantification (UQ) purposes. They are parametric models, the parameters of which can be easily recasted as random variables (RVs). They have been predominantly used to test own UQ software, as well as to perform UQ studies and develop surrogate models with UQLab, SPINTERP, Chaospy, OpenTURNS, and the MATLAB-Sparse-Grids-Kit. UQ studies using the models provided in this repository are available in the following works:
@article{loukrezis2019assessing, author = {Dimitrios Loukrezis and Ulrich Römer and Herbert De Gersem}, title = {Assessing the performance of Leja and Clenshaw-Curtis collocation for computational electromagnetics with random input data}, journal = {International Journal for Uncertainty Quantification}, issn = {2152-5080}, year = {2019}, volume = {9}, number = {1}, pages = {33--57} }
@article{loukrezis2019approximation, author = {{Loukrezis}, Dimitrios and {De Gersem}, Herbert}, title = "{Approximation and Uncertainty Quantification of Stochastic Systems with Arbitrary Input Distributions using Weighted Leja Interpolation}", journal = {arXiv e-prints}, year = "2019", eid = {arXiv:1904.07709} }
@phdthesis{loukrezis2019adaptive, author = {Dimitrios Loukrezis}, title = {Adaptive Approximations for High-Dimensional Uncertainty Quantification in Stochastic Parametric Electromagnetic Field Simulations}, school = {TU Darmstadt}, year = {2019} }