Sampling inequalities for infinitely smooth functions, with applications to interpolation and machine learning (Q849069): Difference between revisions

Optimal algebraic convergence orders for kernel based smoothed interpolation methods in a finitely smooth setting have been derived in [\textit{H. Egger} and \textit{A. Neubauer}, Numer. Math. 101, No. 4, 643--662 (2005; Zbl 1080.65043)]. In this paper sampling inequalities are derived for infinitely smooth functions where the convergence orders turn out to depend exponentially on the fill distance. The main applications in the paper deal with reconstruction problems in Hilbert spaces. In particular, the theory presented here reproduces the well-known error estimates for the standard interpolation problem in the native Hilbert spaces of the inverse multiquadrics and Gaussian kernels on cubes [\textit{H. Wendland}, Scattered data approximation. Cambridge Monographs on Applied and Computational Mathematics 17. Cambridge: Cambridge University Press (2005; Zbl 1075.65021)].