Adaptive sampling and modal expansions in pattern-forming systems (Q2048199)

The authors have worked on proper orthogonal decomposition (POD) and in [\textit{M. L. Rapún} et al., J. Comput. Phys. 335, 1--20 (2017; Zbl 1376.65074)], they have already developed a collocated POD, denoted by LUPOD, in which optimal sets of snapshots and points are sampled via an incomplete LU decomposition. In the current work, the extend their method and construct a collocated LUPOD-based reduced order models in which adaptation to the local dynamics of a given system is carried out and combined with low-dimensional modeling. Their approach yields reliable approximations of the dynamical systems, is well-suited to descibe localized spatio-temporal compelxity and provides fast computations. The performance of their algorithm in respect to robustness, accuracy and computational efficiency are tested and illustrated for some relevant pattern-forming systems, in both one and two space dimensions.