Equivalence theorem on weighted simultaneous \(L_ p\)-approximation by the method of Kantorovič operators (Q1335030): Difference between revisions

The intention of this paper is to give new direct and converse results on weighted simultaneous \(L_ p\)-approximation by the method of the classical Kantorovich operators defined by \[ K_ n (f;x):= (n+1) \sum_{k=0}^ n p_{n,k}(x) \int_{k/(n+1)}^{(k+1)/(n+1)} f(t)dt, \qquad x,t\in [0,1], \quad n\in\mathbb{N}, \] with \(p_{n,k} (x)= {n\choose k} x^ k (1-x)^{n-k}\), using the weighted Ditzian and Totik modulus of smoothness. With a direct technique the author proves for integrable functions the complete equivalence characterization for weighted simultaneous approximation by this method \((K_ n)_{n\in \mathbb{N}}\). Smoothness of the functions to be approximated will be measured by weighted second and third order Ditzian-Totik moduli of smoothness respectively with the special weight \(\varphi^{2r}\) and the step-weight \(\varphi\). One of the central results is also given by an interesting modified Voronovskaya theorem, which uses the third weighted modulus of smoothness. This crucial estimate enables us to prove a converse result, which includes the saturation class for the weighted simultaneous \(L_ p\)-approximation by the method \((K_ n)_{n \in\mathbb{N}}\). With so-called order functions \(\psi\), \(\psi(k,t)\leq C k^ 2\psi(t)\) for example \(\psi(t):= t^ \alpha\), \(\alpha\in (0,2]\) and \(\psi(t):= t^ 2|\ln t|^ \beta\), \(\beta\in\mathbb{R}\) one can get the following main result: Let \(f\in L_ p[0,1]\) with \(f^{(2r})\in L_ p[0,1]\), \(1\leq p<\infty\), \(r\in \mathbb{N}\), \(\varphi(x)= \sqrt {x(1-x)}\), then the following statements are equivalent: \[ \begin{alignedat}{2} \|\varphi^{2r} (K_ n f- f)^{(2r)} \|_ p &= {\mathcal O}( \psi(n^{-1/2})+ {\textstyle {1\over n}}) &\qquad &(n\to\infty)\tag{i}\\ \omega_ \varphi^ 2 (f^{(2r)}; t)_{\varphi^{2r} ,p} &= {\mathcal O}( \psi(t)+ t^ 2) &\qquad &(t\to 0).\tag{ii}\end{alignedat} \] This paper shows a complete characterization for the global rate of weighted simultaneous \(L_ p\)-approximation by this method \((K_ n)_{n\in\mathbb{N}}\).