On the refinement matrix mask of interpolating Hermite splines (Q2005987): Difference between revisions

Splines and in particular Hermite splines are a useful tool for the approximation of functions by interpolation or quasi-interpolation (e.g. by the Schoenberg operator) partly because they can be computed recursively. For this, subdivision is important (more precisely, vector subdivision schemes) and for that in turn the refinement equation that defines the subdivision process. Refinement equations and especially the refinement (matrix) masks are considered in this article. Its main focus is on finite mask sequences for Hermite splines. It is of special interest to compute the entries of the refinement mask conveniently and the authors of the paper address to that question. They do not only include equally spaced points but allow for dilation factors too without making the computation any more complicated. The basis of the work is formed by the remarkable observation that just the straight univariate cardinal B-splines (also due to Schoenberg, where only simple knots are employed) are needed for the construction which makes the computation even more clear and useful.