On the Lebesgue constants for interpolation of analytic functions (Q2640724): Difference between revisions

Continuing ideas from an earlier paper [Mat. Zametki 12, 465-476 (1972; Zbl 0248.30033)] the author considers for a given system \(\tau\) of points \(t_ 1,...,t_ n\) in the interval [-1,1] the interpolation process \(S(t)=\sum_{j}f(t_ j)D_ j(t)\) which is optimal in the class of functions f analytic and bounded in the disc \(\{| z| <1/\sqrt{kw}\}\). Lagrangean interpolation is included as the special case \(k=0\). The basis functions \(D_ j\) and an optimal system Z of interpolation points can be explicitly determined. Let D(t) be the vector function with components \(D_ j(t)\) and \(\| \cdot \|_ p\) the p- norm in \({\mathbb{C}}^ n\). The Lebesgue constants of this interpolation process are defined by \[ L_ n(\tau,p,k):=\max \{\| D(t)\|_ p,\quad t\in [-1,1]\}. \] The author derives estimates \(L_ n(Z,p,k)\leq \sqrt{2}\) for \(p\geq 2\) and gives explicit although complicated representations for \(L_ n(Z,1,k)\).