On the Lebesgue constant for the Xu interpolation formula (Q2502872): Difference between revisions

The numerical experiments of \textit{L. Bos, M. Caliari, S. De Marchi, M. Vianello} [Computing 76, No. 3--4, 311--324 (2006; Zbl 1087.65009)] gave a good evidence that the Lebesgue constant of Xu-like interpolation has growth of the order \((\log n)^2\). The purpose of this paper is to prove that this is indeed the case. From this, the authors conclude that the points studied by Xu are an excellent point for practical polynomial interpolation. Moreover this result also gives strong evidence that the minimal rate of growth for the Lebesgue constant for interpolation of polynomials of total degree \(n\) on a square is the order \((\log n)^2\). This indicates a fundamental difference between a square and a disk, where the minimal growth is of order \((n)^{(1/2)}\). They also remark that there has recently been introduced a set of points in the square, the so-called Padua points, that are another Chebyshev-like family, and for which numerical experiments indicate that the Lebesgue constant has this minimal \(O((\log n)^2)\) growth.