scientific article; zbMATH DE number 4046397

From MaRDI portal
Publication:3784635





An orthonormal basis of wavelets (``ondelettes) is defined to be an orthonormal basis of \(L^ 2({\mathbb{R}}^ d)\) of the form \(2^{dj/2}\psi_ m(2^ jx-\gamma)\), \(x\in {\mathbb{R}}^ d\), \(j\in {\mathbb{Z}}\), \(1\leq m<2^ d(m\in {\mathbb{N}})\), \(\gamma\in \Gamma\) where \(\Gamma\) is a lattice in \({\mathbb{R}}^ d\) (i.e. \(\Gamma\) is a discrete subgroup of \({\mathbb{R}}^ d\) with compact quotient \({\mathbb{R}}^ d/\Gamma)\) and the \(\psi_ m's\) belong to the class of functions \(\psi:{\mathbb{R}}^ d\to {\mathbb{C}}\) such that, for some \(C>0\), \[ | \psi (x)| \leq C(1+| x|)^{-d-1},| \partial /\partial x_ j\psi (x)| \leq C(1+| x|)^{-d-1},1\leq j\leq d,\quad x\in {\mathbb{R}}^ d \] and \(\int_{{\mathbb{R}}^ d}\psi (x)dx=0\). The author points out the usefulness of these special orthonormal bases and outlines a way of obtaining many of them (perhaps all -this remains an open question) via what he calls an ``analyse graduée dyadique of \(L^ 2({\mathbb{R}}^ d)\); this latter is an increasing family of closed subspaces of \(L^ 2({\mathbb{R}}^ d)\), \(V_ j\), \(j\in {\mathbb{Z}}\), satisfying certain requirements (some reminiscent of those in time series analysis). The author points out the fascinating relationship between ``analyse graduée, martingale theory and classical methods of numerical analysis (like those involving splines and finite elements). The existence of orthonormal bases of wavelets is not at all obvious; on the other hand, they seem to be a powerful tool in many different problems. Articles referred to in the paper and others to come should be of interest for further understanding.




Cited in
(59)








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 Q3784635)