Optimal Lagrange interpolation by quartic \(C^1\) splines on triangulations (Q929937): Difference between revisions

Let \(\mathcal{S}^r_q(\triangle)\) be the space of bivariate splines of degree \(q\) and smoothness \(r\) on a given triangulation \(\triangle\). The corresponding Lagrange interpolation problem consists in the construction of sets \(\{z_1,\dots,z_d\}\) such that for any given data \(f = \{f_1,\dots,f_d\}\) one can find a unique \(s \in \mathcal{S}^r_q(\triangle)\) such that \(s(z_i)=f_i\), \(i=1,\dots,d\), where \(d=\dim(\mathcal{S}^r_q(\triangle))\). The paper pesents a local Lagrange interpolation method with optimal approximation order for \(\mathcal{S}^1_4(\widetilde{\triangle})\), where \(\widetilde{\triangle}\) is a modification of \(\triangle\). The new triangulation \(\widetilde{\triangle}\) is obtained by decomposing \(\triangle\) into pairs of neighboring triangles and by adding ``diagonals'' to some of these pairs. It is said that the approximation order for \(\mathcal{S}^1_4(\widetilde{\triangle})\) is optimal in the sense that the estimate \(\| f-s\| \leq K\cdot h^5\) holds true. Here \(h\) is the maximal diameter of the triangles in \(\widetilde{\triangle}\) and \(K\) only depends on the smallest angle of \(\triangle\). The main result of the paper shows how a Lagrange interpolation set for \(\mathcal{S}^1_4(\widetilde{\triangle})\) can be found which is local and stable. The process to obtain an interpolation method which posseses optimal approximation order is based on a simple refinement.