Optimal recovery of twice differentiable functions based on symmetric splines (Q1759359)

Given the values and gradients of a function on a finite knot set in \(\mathbb{R}^d\), \(d\in \mathbb{N}\), the authors introduce local symmetric spline recovery methods and study properties of these symmetric splines. It is shown that under certain not too restrictive assumptions, there do not exist smooth optimal symmetric spline methods for the problem at hand. For the recovery methods defined by the authors, general error estimates are derived and some explicit constructions of optimal bivariate interpolating quartic and quintic splines over regular triangulations are presented.