Weighted approximation by Baskakov-type operators (Q653964)

The paper deals with the following sequence of Baskakov-type operators defined for natural \(n\) by \[ V_nf(x)=\sum_{k=0}^{\infty}P_{n,k}(x)v_{n,k}(f), \] where \[ P_{n,k}(x)={n+k-1\choose k}x^k(1+x)^{-n-k},\;k\in {\mathbb N}_0={\mathbb N}\cup \{0\}, \] \[ v_{n,0}(f)=f(0);\;\;v_{n,k}(f)=(n+1)\int _{0}^{\infty}P_{n+2,k-1}(y)f(y)dy,\;k\in {\mathbb N}, \] and \(f\) is Lebesgue measurable on \((0,\infty )\) with a finite limit \(f(0)\) at \(0\). The paper shows that operators \(V_n\) are related to Baskakov operators in the same way as the well-known Goodman-Sharma operators are related to the classical Bernstein polynomials. Specifically, the main goal is the characterization of the uniform weighted approximation error \(\| w(f -V_nf)\| \) for weight functions given by \[ w(x)=w(\gamma_0,\gamma_{\infty };x)=\left({x \over 1+x}\right)^{\gamma_0}(1+x)^{\gamma_{\infty }} \] for \(x\in [0,\infty )\) and real \(\gamma_0\), \(\gamma_{\infty }\). The main result of the paper is stated below. First, the required notation is written. By \(C[0,\infty )\) the authors denote the space of all functions continuous on \([0,\infty )\). The functions from \(C[0,\infty )\) are not expected to be bounded or uniformly continuous. By \(L_{\infty} [0,\infty )\) we denote the space of all Lebesgue measurable and essentially bounded in \([0,\infty )\) functions equipped with the uniform norm \(\| \cdot \| \). For a weight function \(w\) we set \(C(w) = \{f\in C[0,\infty ): wf\in L_{\infty}[0,\infty )\}\) and \[ W^2(w\psi )=\{g,g'\in AC_{\text{loc}}(0,\infty ): w\psi D^2g\in L_{\infty}[0,\infty )\}, \] where \(AC_{\text{loc}}(0,\infty )\) consists of the functions which are absolutely continuous in \([a,b]\) for every \([a,b]\subset (0,\infty)\). Set \(C_0(w)=\{f\in C(w):f(0)=0\}\). Similarly, by \(W_0^2(w\psi )\) we denote the subspace of \(W^2(w\psi )\) of functions \(g\) satisfying the additional boundary condition \[ \lim_{x\rightarrow 0+0}\psi (x)D^2g(x)=0. \] The weighted approximation error of \(V_n\) will be compared with the \(K\)-functional between the weighted spaces \(C(w)\) and \(W^2(w\psi )\), which for every \(f\in C(w)+W^2(w\psi )\) and \(t>0\) is defined by \[ K_w(f,t)= \inf \{\| w(f-g)\| +t\| w\psi D^2g\| :g\in W^2(w\psi ),f-g\in C(w)\}. \] Theorem. Let \(w=w(\gamma_0, \gamma_{\infty })\) as above with \(\gamma_0, \gamma_{\infty }\in [-1,0]\). Then for every \(f\in C(w)+W^2(w\psi )\) and every \(n\in {\mathbb N}\), \(n\geq 4\), we have \[ \| w(f-V_nf)\| \leq 2K_w(f,1/(2n))\leq 13.7\| w(f-V_nf)\| . \]