POD-DEIM Model Order Reduction for Large Nonlinear Problems (Stephan Rave, Universität Münster, Germany)

30.04.2019 14:00

Proper Orthogonal Decomposition (POD) is a widely used technique for the construction of low-dimensional approximation spaces from high-dimensional input data. In combination with the Discrete Empirical Interpolation Method (DEIM) highly efficient reduced order models can be obtained for many classes of application problems. For large-scale models and an increasing amount of snapshot data, however, computing a POD-DEIM model quickly becomes prohibitively expensive.

In this talk I will present a general, reliable and easy to implement approach to compute an approximate POD based on arbitrary tree hierarchies of worker nodes where each worker computes a POD of only a small amount of input vectors. The tree hierarchy can be freely adapted to optimally suit the available computational resources. In particular, this hierarchical approximate POD (HAPOD) allows for both, simple parallelization with low communication overhead, as well as live sequential POD computation under constrained memory resources. Extending this approach by a simultaneous computation of the DEIM interpolation space, the memory and time requirements for the generation of POD-DEIM reduced order models can be significantly reduced.

Lieu

Room 623, Séminaire d'analyse numérique

entrée libre