Orthogonal expansions and multilevel algorithms
In [Appl. Comput. Harmon. Anal. 2, No. 3, 201-218 (1995; Zbl 0845.42014)], \textit{Y. W. Koh, S. L. Lee} and \textit{H. H. Tan} showed that multiresolutions and wavelet bases of \(L^2[0,2\pi)\) can be constructed using the so-called orthogonal splines. The Fourier series with respect to the orthogonal system \(\{e^{inx}\), \(n \in \mathcal Z\}\) played a central role in the construction, and the system \(\{e^{inx}\), \(n \in \mathcal Z\}\) was used to define the \textit{orthogonal splines}. The paper under review studies the weighted Hilbert spaces \(L_{\alpha}^2(I)\), where \(I\) is an interval which may be bounded or unbounded. The role of the Fourier series with respect to \(\{e^{inx}\), \(n \in \mathcal Z\}\) is assumed by the Fourier series with respect to a system of normalized orthogonal polynomials \(\{p_n\), \(n \in \mathcal Z^+\}\) associated with the measure \(\alpha\). Using an analogue of the orthogonal splines defined by the system \(\{p_n\), \(n \in \mathcal Z^+\}\), the authors present a general approach for the construction of a multiresolution and the corresponding orthonormal basis of \(L_{\alpha}^2(I)\). This alternative basis leads to algorithms for the evaluation of the coefficients of expansion of a function. These algorithms are analogous to the decomposition and reconstruction algorithms for wavelet analysis and synthesis in \(L^2(\mathcal R)\). This paper contains proofs of only some of the theorems stated. According to the authors, full details can be found in a preprint entitled ``Multiresolutions and orthogonal polynomials. For the entire collection see [Zbl 0896.00026].
This page was built for publication: Orthogonal expansions and multilevel algorithms
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4213446)
