A new algorithm for a recursive construction of the minimal interpolation space (Q730886)

Given \(N+1\) distinct points \(x_0, \dots, x_ N\) in \(\mathbb R^d\), \(d\geq 1\), there exists a subspace \(V_N\) of polynomials of degree \(\leq N\) in \(d\) variables such that the Lagrange interpolation problem in \(V_N\) is poised (uniquely solvable), although if \(d>1\) this subspace is not unique. An interpolation algorithm has been introduced by \textit{D. Barrera, D. Sbibih} and \textit{A. Serghini} [J. Comput. Appl. Math. 216, No.~2, 435--450 (2008; Zbl 1148.41002)] allowing to construct recursively a minimal degree interpolation space \(V_N\). Motivated by this work, the authors propose a new algorithm, which permits to construct recursively and iteratively the subspaces \(V_j\) such that the Lagrange interpolation problem at nodes \(x_0, \dots, x_j\) is poised in \(V_j\), with \(j=0, 1,\dots, N\). It is proved also that each \(V_j\) is the minimal interpolation space with respect to its nodes. The confluent (or osculatory) case, when some nodes coincide, is also considered. Numerical tests illustrate the algorithm.