Mean convergence of truncated Hermite interpolation on an arbitrary system of nodes (Q1272198)

Let \(X\) be a triangular table of interpolation nodes from \([-1,1]\) whose \(n\)-th row is \({\mathbf x}_n=\{x_{nk}\}_{k=1}^n\). Let \(L_n(X,f,x)\) be the Lagrange interpolating polynomial (degree \(n-1\)) for \(f\) in the points \({\mathbf x}_n\). Turán posed the following problem: given \(p>1\), what is the necessary and sufficient condition for \(\lim_{n\to\infty}\int_{-1}^1| f(x)-L_n(X,f,x)| ^pdx=0\) for every \(f\in C[-1,1]\). This paper gives an answer to a more general form of this problem. It considers truncated Hermite interpolating polynomials of the form \(L_{nmr}(X,f,x)=\sum_{j=0}^r\sum_{k=1}^n f^{(j)}(x_{nk})A_{jk}(x)\) with \(0\leq r \leq m-1\), \(f\in C^r[-1,1]\). The \(A_{jk}\) are polynomials of degree \(nm-1\) which satisfy \(A_{jk}^{(p)}(x_{nq})=\delta_{jp}\delta_{kq}\), \(j,p=0,\ldots,m-1\), \(k,q=1,\ldots,n\). Note \(L_n=L_{n10}\) is the usual Lagrange interpolant and \(L_{n20}\) is the classical Hermite-Fejér interpolant. Moreover, Lebesgue's measure in the Turán problem is replaced by an arbitrary positive measure on \([-1,1]\). Special cases of these results were published in earlier papers of the author.