A new factorization technique of the matrix mask of univariate refinable functions (Q5927650): Difference between revisions

Let \(\Phi \subset L^2({\mathbb R})\) denote a finite set and \(T(\Phi)\) the smallest closed vector subspace of \(L^2({\mathbb R})\) that contains the integral translations of all elements \(\phi \in \Phi\). Then \(\Phi\) is called a refinable vector when the vector space \(T(\Phi)\) is invariant under the dilation operators of parameter \({1 \over n}\) where the reciprocal is defined by a natural number \(n > 1\). The purpose of the paper is the introduction of a special factorization algorithm for refinable vectors that generalizes the univariate case as closely as possible. The univariate case consists in the representation of a compactly supported refinable function by a convolution product of a spline function with a compactly supported distribution [cf. \textit{G. Plonka}, Constructive Approximation 13, No. 2, 221-244 (1997; Zbl 0870.41015)]. The usefulness of the factorization property for the study of the smoothness of multiwavelets is demonstrated via the processing of its mask symbol.