Error bounds for function approximation using generated sets

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cools, Ronald, Nuyens, Dirk, Wilkes, Laurence
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909598244405248
author Cools, Ronald
Nuyens, Dirk
Wilkes, Laurence
author_facet Cools, Ronald
Nuyens, Dirk
Wilkes, Laurence
contents This paper explores the use of "generated sets" $\{ \{ k \boldsymbolζ \} : k = 1, \ldots, n \}$ for function approximation in reproducing kernel Hilbert spaces which consist of multi-dimensional functions with an absolutely convergent Fourier series. The algorithm is a least squares algorithm that samples the function at the points of a generated set. We show that there exist $\boldsymbolζ \in [0,1]^d$ for which the worst-case $L_2$ error has the optimal order of convergence if the space has polynomially converging approximation numbers. In fact, this holds for a significant portion of the generators. Additionally we show that a restriction to rational generators is possible with a slight increase of the bound. Furthermore, we specialise the results to the weighted Korobov space, where we derive a bound applicable to low values of sample points, and state tractability results.
format Preprint
id arxiv_https___arxiv_org_abs_2505_00440
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Error bounds for function approximation using generated sets
Cools, Ronald
Nuyens, Dirk
Wilkes, Laurence
Numerical Analysis
65D15, 65T40
This paper explores the use of "generated sets" $\{ \{ k \boldsymbolζ \} : k = 1, \ldots, n \}$ for function approximation in reproducing kernel Hilbert spaces which consist of multi-dimensional functions with an absolutely convergent Fourier series. The algorithm is a least squares algorithm that samples the function at the points of a generated set. We show that there exist $\boldsymbolζ \in [0,1]^d$ for which the worst-case $L_2$ error has the optimal order of convergence if the space has polynomially converging approximation numbers. In fact, this holds for a significant portion of the generators. Additionally we show that a restriction to rational generators is possible with a slight increase of the bound. Furthermore, we specialise the results to the weighted Korobov space, where we derive a bound applicable to low values of sample points, and state tractability results.
title Error bounds for function approximation using generated sets
topic Numerical Analysis
65D15, 65T40
url https://arxiv.org/abs/2505.00440