scientific article; zbMATH DE number 3944468

From MaRDI portal
Publication:3715575





[For the entire collection see Zbl 0575.00004.] The aim of the paper is to replace the sine-function in Shannon's sampling series by a general function \(\phi \in L^ 1({\mathbb{R}})\cap C({\mathbb{R}})\) and to investigate the approximation process obtained in this way if W, the sampling frequency, goes to infinity. Since this kind of approximation process can be considered as the discrete version of the singular integral of Fejér's type, the proofs partly follow classical lines. The main result is an equivalence theorem that connects degree of approximation, discrete moments of \(\phi\), and the Fourier transform of \(\phi\). However, it turns out that the conditions on \(\phi\) in the discrete case are much stronger than in the continuous case, since the condition on the Fourier-transform of \(\phi\) which has to hold only at one point in the classical case must be imposed for countably many points in the discrete case. The theorem of Dickmeis and Nessel is applied to show that the degree of convergence, given in the main theorem, is best possible. Some examples are also given, showing that the theory of generalized sampling series developed allows one to replace the sine-function in the Shannon sampling theorem with other functions especially time-limited functions which are more suitable for numerical evaluation.




Cited in
(98)








This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3715575)