Equiconvergence of some complex interpolatory polynomials (Q1825313): Difference between revisions

Let \(D_{\rho}:=\{z|\) \(| z| <\rho \}\) and let \(A_{\rho}\) denote the class of functions analytic in \(D_{\rho}\) but not in \(\bar D_{\rho}\). Let \(\alpha\), \(\beta\) \((\alpha \neq \beta)\in D_{\rho}\) and let \(L_ n(z,\alpha,f)\), \(L_ m(z,\beta,f)\) denote the Lagrange interpolants to \(f\in A_{\rho}\) in the zeros of \(z^ n-\alpha^ n\) and \(z^ m-\beta^ m\) respectively. If \(m=rn+q\), \(0<q<n\), \textit{Lou Yuanren} [Approximation Theory Appl. 2, No.3, 19-32 (1986; Zbl 0614.41005)] considers the difference \[ \Delta_{n,m}(z,f):=L_ m(z,\beta,L_ n(z,\alpha,f))-L_ n(z,\alpha,L_ m(z,\beta,f)) \] and shows that this difference converges to zero in a region bigger than \(D_{\rho}\) as \(n\to \infty\) under mild growth conditions on q. The object of this note is to obtain analogous results when the Lagrange interpolants are replaced by Hermite interpolants or approximants in the sense of mean square. This provides possibilites of extending Walsh's equiconvergence theorem in several directions.