Optimal weighted least-squares methods
The SMAI Journal of computational mathematics · 01.10.2017 · 3 · 181-203
DOI: 10.5802/smai-jcm.24 ↗Kurz gefasst
Fachartikel in der Zeitschrift The SMAI Journal of computational mathematics, online erschienen am 01.10.2017. Verfasst von Albert Cohen und 1 weiteren Autor:innen. Die Arbeit wurde bisher 171-mal zitiert (Crossref) und nennt 14 Literaturangaben.
Abstract (Originalsprache)
We consider the problem of reconstructing an unknown bounded function u defined on a domain X ⊂ ℝ d from noiseless or noisy samples of u at n points ( x i ) i = 1 , ⋯ , n . We measure the reconstruction error in a norm L 2 ( X , d ρ ) for some given probability measure d ρ . Given a linear space V m with dim ( V m ) = m ≤ n , we study in general terms the weighted least-squares approximations from the spaces V m based on independent random samples. It is well known that least-squares approximations can be inaccurate and unstable when m is too close to n , even in the noiseless case. Recent results from [6, 7] have shown the interest of using weighted least squares for reducing the number n of samples that is needed to achieve an accuracy comparable to that of best approximation in V m , compared to standard least squares as studied in [4]. The contribution of the present paper is twofold. From the theoretical perspective, we establish results in expectation and in probability for weighted least squares in general approximation spaces V m . These results show that for an optimal choice of sampling measure d μ and weight w , which depends on the space V m and on the measure d ρ , stability and optimal accuracy are achieved under the mild condition that n scales linearly with m up to an additional logarithmic factor. In contrast to [4], the present analysis covers cases where the function u and its approximants from V m are unbounded, which might occur for instance in the relevant case where X = ℝ d and d ρ is the Gaussian measure. From the numerical perspective, we propose a sampling method which allows one to generate independent and identically distributed samples from the optimal measure d μ . This method becomes of interest in the multivariate setting where d μ is generally not of tensor product type. We illustrate this for particular examples of approximation spaces V m of polynomial type, where the domain X is allowed to be unbounded and high or even infinite dimensional, motivated by certain applications to parametric and stochastic PDEs.
Zitieren
APA
Cohen, A., & Migliorati, G. (2017). Optimal weighted least-squares methods. The SMAI Journal of computational mathematics, 3, 181-203. https://doi.org/10.5802/smai-jcm.24
Vancouver
Cohen A, Migliorati G. Optimal weighted least-squares methods. The SMAI Journal of computational mathematics. 2017;3:181-203. doi:10.5802/smai-jcm.24
BibTeX
@article{Cohen2017,
author = {Cohen, Albert and Migliorati, Giovanni},
title = {Optimal weighted least-squares methods},
journal = {The SMAI Journal of computational mathematics},
year = {2017},
volume = {3},
pages = {181--203},
doi = {10.5802/smai-jcm.24}
}
EUVIMED