Where are the nodes of ''good'' interpolation polynomials on the real line? (Q1976279)

To every Freud-type weight on \(\mathbb{R}\) and every positive integer \(n\), a positive number \(a_n\), the so-called Mhaskar-Rahmanov-Saff number, is associated so that \[ \sup_{x\in\mathbb{R}} \bigl|q(x)\bigr|= \sup_{|x|\leq a_n}\bigl|q(x)\bigr| \] holds for all weighted polynomials of degree \(\leq n\). In this note, the author shows that the interval where nodes \(x_0<x_1< \cdots <x_n\) for Lagrange interpolation are placed is restricted by \(a_n\) and the corresponding Lebesgue constant \(\lambda_n\), more precisely \[ \max_{0\leq k\leq n}|x_k|\leq a_n\left(1+c \left({\log \lambda_n \over n}\right)^{2/3} \right) \] for some constant \(c\) depending only on the weight.