On the approximation of vector-valued functions by samples

MaRDI QIDQ6432339zbMATH OpenFDO

Authors Daniel Kressner, Tingting Ni, André Uschmajew

Publication date 6 April 2023

Abstract: Given a Hilbert space $mathcalH$ and a finite measure space $Omega$, the approximation of a vector-valued function $f:OmegaomathcalH$ by a $k$-dimensional subspace $mathcalUsubsetmathcalH$ plays an important role in dimension reduction techniques, such as reduced basis methods for solving parameter-dependent partial differential equations. For functions in the Lebesgue--Bochner space $L^2(Omega;mathcalH)$, the best possible subspace approximation error $d_k^{(2)}$ is characterized by the singular values of $f$. However, for practical reasons, $mathcalU$ is often restricted to be spanned by point samples of $f$. We show that this restriction only has a mild impact on the attainable error; there always exist $k$ samples such that the resulting error is not larger than $sqrtk+1cdot{d}_k^{(2)}$. Our work extends existing results by Binev at al. (SIAM J. Math. Anal., 43(3):1457--1472, 2011) on approximation in supremum norm and by Deshpande et al. (Theory Comput., 2:225--247, 2006) on column subset selection for matrices.

This page was built for publication: On the approximation of vector-valued functions by samples

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6432339)