Review and benchmark paper on sparse polynomial chaos expansions published
A paper about sparse polynomial chaos expansions by N. Lüthen, S. Marelli and B. Sudret has recently been published in the SIAM/ASA Journal of Uncertainty Quantification.
Sparse polynomial chaos expansions (PCE) are a powerful and widely used surrogate modelling technique. Our paper gives an extensive overview of the various sparse PCE methods developed within the last decades, and compares the most promising ones on a number of benchmark examples. We find that the choice of solver and sampling scheme can make a difference of several orders of magnitude in the resulting approximation error. The results differ by model dimensionality and experimental design size.
The publication can be found external page here and the associated report on our internal archive here.