Cardinal interpolation with Gaussian kernels

From MaRDI portal



Abstract: In this paper, interpolation by scaled multi-integer translates of Gaussian kernels is studied. The main result establishes Lp Sobolev error estimates and shows that the error is controlled by the Lp multiplier norm of a Fourier multiplier closely related to the cardinal interpolant, and comparable to the Hilbert transform. Consequently, its multiplier norm is bounded independent of the grid spacing when 1<p<infty, and involves a logarithmic term when p=1 or infty.


Error estimates in \(L_p\)-norms are at the core of numerical analysis of approximation and interpolation methods. This type of work has a long and interesting history in the context of radial basis functions. To some extend, much of the work on radial basis functions emerged from the analysis of interpolation of Gauß-kernels. This is now enlarged by additional error estimates of approximants by multi-integer translates of such kernels. The error estimates the authors derive include all \(p\) between and including one and infinity and have only in the extreme cases the typical, additional \(\log h\) factors.




Cited in
(28)








This page was built for publication: Cardinal interpolation with Gaussian kernels

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q421205)