Order-Reduction of Fields-Level Models with Affine and Non-Affine Parameters by Interpolation of Subspaces

Model-order reduction provides an appealing approach to solving parametric large-scale models stemming from spatial discretization methods. The high-dimensional model at the fields-level is replaced by a surrogate model that is fast to evaluate, at a controllable level of error. This paper presents an interpolation-based order-reduction method for systems with non-affine parameters. The main novelty is the construction of the parameter-dependent projection matrix. For a given error level, the suggested approach reduces the number of instantiations of the fields-level model compared to state-of-the-art methods.