Regularity of multivariate vector subdivision schemes (Q1776171)

Subdivision schemes are computational means for generating recursively discrete functions defined on denser and denser grids in \({\mathbb R}^s\). At each step of the subdivision recursion the new values are obtained simply by local averaging of the previously computed values on the coarser grid. The averaging coefficients form the so-called refinement mask. If they are real numbers the scheme is said to be a scalar subdivision scheme acting on scalar sequences. If the averaging coefficients are matrices, the scheme is called either a matrix subdivision scheme or a vector subdivision scheme as it now maps vector valued sequences to vector valued sequences. Vector subdivision schemes play an important role in the convergence and regularity analysis even of scalar multivariated subdivisions schemes, in the analysis of hermite type subdivision schemes, and in the context of multi-wavelets. The authors give a characterization of the convergence of multivariated subdivision schemes and derive sufficient conditions for a refinable function to possess a certain order of differentiability. This is done classifying the subdivision schemes with respect to the dimension of a certain finite dimensional subspace depending on the matrices that form the mask of the subdivision scheme as introduced by \textit{C.A. Micchelli} and \textit{T. Sauer} [Adv. Comput. Math 7, 455-545 (1997; Zbl 0902.65095); Math Z, 229, 621-674 (1998; Zbl 0930.65007)]. Based on this dimension, the authors use a suitable difference operator to pass to a difference scheme whose mask consists of larger matrices. The conditions on the mask of the original scheme ensuring the existence of the difference scheme allow for an algebraic interpretation that generalizes the ''zero at \(-1\)'' property from the univariate case. The concept of the restricted spectral radius, introduced and investigated by the authors, enable to characterize the convergence of the original subdivision scheme in terms of the spectral properties of the difference scheme. Finally, the authors show that the convergence of the difference scheme implies that the original scheme is convergent to a smoother limit function.