Sequences with equi-distributed extreme points in uniform polynomial approximation (Q596818)

Let \(E\subset {\mathbb{C}}\) be a compact set whose complement \(\Omega:=\overline{\mathbb{C}}\setminus E\) is connected, and \(f\) be a continuous function on \(E\), and analytic in the interior of \(E\). Let \(\mathcal{A}_n(f)\) be the extremal set of best polynomial approximation to \(f\) on \(E\), and \({\mathcal{F}}_n(M)\) be a Fekete point set of \(n\) points of \(M\subset {\mathbb{C}}\). If \(B\subset {\mathbb{C}}\) is a finite set, the normalized counting measure of \(B\) is denoted by \(\tau(B)\). The paper under review deals with the distribution of extremal point sets for complex approximation by polynomials on \(E\). The main result is Theorem 1: Let \(\Lambda=(n_j)_{j\in {\mathbb{N}}}\) be a sequence of \({\mathbb{N}}\) with \(\lim_j n_{j+1}/n_j=1\). Then it can be chosen a subsequence \(\Lambda'\subseteq \Lambda\), such that \[ \tau({\mathcal{F}}_{n+2}({\mathcal{A}}_n(f)))\overset{*}{}\mu_E,\;\text{ as }\;{n} \infty,\;n\in \Lambda', \] where \(\mu_E\) is the equilibrium distribution of \(E\) supported on \(\partial E\) and \(\overset{*}{}\) denotes {*}-weak convergence of the measures. Let \(\nu_1\) and \(\nu_2\) be two measures on the set \(E\) which is either a Jordan curve or an arc. The discrepancy of \(\nu_1\) and \(\nu_2\) is defined by \[ D[\nu_1-\nu_2]:=\sup\{| (\nu_1-\nu_2)({\mathbf{J}})| : {\mathbf{J}}\text{ is a subarc of }\partial E\}. \] The paper also shows how the {*}-weak convergence of Theorem 1 can be sharpened by estimating the discrepancy between the measures \(\tau({\mathcal{F}}_{n+2}({\mathcal{A}}_n(f)))\) and \(\mu_E\) (Theorems 2 and 3).