A general framework of compactly supported splines and wavelets (Q1195885)

Let \(\dots\subset V_{-1}\subset V_ 0\subset V_ 1\subset\dots\) be an arbitrary sequence of closed subspaces. The authors study the family \(\phi\) of \(L^ 2\)-functions \(\phi\) with the properties \(V_ k=\text{clos}_{L^ 2}\{\phi_{kj}: j\in\mathbb{Z}\}\), all \(k\), where \(\phi_{kj}(x)=\phi(2^ k x-j)\); \(\phi\) has a finite two-scaled relation, i.e., \(\phi(x)=\sum^{N_ \phi}_{n=0} p^ \phi_ n \phi(2x-n)\), \(x\in\mathbb{R}\), with some finite \(N_ \phi\). The function \(\varphi\in\Phi\) with minimal support is called a generalized \(B\)-spline. Basic properties of the functions \(\phi\in\Phi\) and the \(B\)-spline \(\varphi\) are established. \(\varphi\) is completely characterized and an algorithm for determining \(\varphi\) from any \(\phi\in\Phi\) is enclosed. The wavelet spaces \(\{W_ k\}\), corresponding to \(\{V_ k\}\), are introduced. They are generated by a simple \(L^ 2\)-function \(\eta\). If \(\eta\) has at least exponential decay as \(| x|\to \infty\), then \(\eta\) is called a wavelet. The collection of all wavelets is denoted by \(\Psi\). The authors characterize the minimal supported wavelet \(\psi\) from \(\Psi\) and call it \(B\)-wavelet. The ``\(B\)-wavelet'' part \((\varphi,\psi)\) provides the most economical and efficient spline representations and wavelet decompositions of \(L^ 2\) functions from \(\{V_ k\}\) and \(\{W_ k\}\). A dual relation between the basis of both \(\{\varphi(\cdot-j):\;j\in\mathbb{Z}\}\) and \(\{\psi(\cdot-j): j\in\mathbb{Z}\}\) is given. A set of formulas for computing the sequences for the two-scale relations and the decomposition relations are established. The authors derive conditions on \(\{V_ k\}\) that guarantee the properties of generalized linear phase and linear phase for \(\phi\) and \(\eta\).