English

A registration method for model order reduction: data compression and geometry reduction

Numerical Analysis 2019-11-12 v2 Numerical Analysis

Abstract

We propose a general --- i.e., independent of the underlying equation --- registration method for parameterized Model Order Reduction. Given the spatial domain ΩRd\Omega \subset \mathbb{R}^d and a set of snapshots {uk}k=1ntrain\{ u^k \}_{k=1}^{n_{\rm train}} over Ω\Omega associated with ntrainn_{\rm train} values of the model parameters μ1,,μntrainP\mu^1,\ldots, \mu^{n_{\rm train}} \in \mathcal{P}, the algorithm returns a parameter-dependent bijective mapping Φ:Ω×PRd\boldsymbol{\Phi}: \Omega \times \mathcal{P} \to \mathbb{R}^d: the mapping is designed to make the mapped manifold {uμΦμ:μP}\{ u_{\mu} \circ \boldsymbol{\Phi}_{\mu}: \, \mu \in \mathcal{P} \} more suited for linear compression methods. We apply the registration procedure, in combination with a linear compression method, to devise low-dimensional representations of solution manifolds with slowly-decaying Kolmogorov NN-widths; we also consider the application to problems in parameterized geometries. We present a theoretical result to show the mathematical rigor of the registration procedure. We further present numerical results for several two-dimensional problems, to empirically demonstrate the effectivity of our proposal.

Keywords

Cite

@article{arxiv.1906.11008,
  title  = {A registration method for model order reduction: data compression and geometry reduction},
  author = {Tommaso Taddei},
  journal= {arXiv preprint arXiv:1906.11008},
  year   = {2019}
}
R2 v1 2026-06-23T10:04:04.606Z