On an approximation processes in the space of analytical functions (Q707988): Difference between revisions

Let us recall that a sequence \(\{\alpha_n\}\subset\mathbb R\) is said to converge statistically to \(\alpha\in\mathbb R\) if for every \(\epsilon>0\), \(\lim_{n\to\infty}\frac{\{m\leq n:\;|\alpha_n-\alpha|>\epsilon\}}n=0\). Let \(A(D)\) denote the space of all analytic functions on a bounded simply connected domain \(D\) in \(\mathbb C\) endowed with the family of the norms \(\|f\|_{A(D),r}=\max_{|\phi(z)|\leq r<1}|f(z)|\), where \(\phi\) is any function mapping \(D\) conformally and injectively to the unit disk. A \(k\)-positive operator \(T\) acting from \(A(D)\) to \(A(D)\) is a linear operator that preserves the cone of functions in \(A(D)\) with nonnegative Taylor coefficients. The aim of this paper is to obtain Korovkin type theorems for \(k\)-positive operators with respect to the statistical convergence. The most general one of those proved by the authors reads as follows. Let \(\{g_k\geq 1\}\) be an increasing sequence with \(\limsup_{n\to\infty}g_k^{1/k}=1\) and let \(A_g(D)\) be the subspace of \(A(D)\) of functions with Taylor coefficients satisfying \(|f_k|\leq Mg_k,\;k=0,1,2\dots\) for some constant \(M=M(f)\). Let \(\{b_k\}\) be another increasing sequence of positive numbers with \(\limsup_{n\to\infty}b_k^{1/k}=1\) and let \(b_\nu(z)=\sum_{k=0}^\infty \frac{b_k^{\nu/2}}{1+b_k}g_k\phi^k(z)\). Let \(\{T_n\}\) be a sequence of \(k\)-positive operators from \(A(D)\) to \(A(D)\). Then the following conditions are equivalent: \(\text{(i)}\;st-\lim_{n\to \infty}\|T_nb_\nu(z)-b_\nu(z)\|_{A(D),r}=0,\;\nu=0,1,2;\) \(\text{(ii)}\;st-\lim_{n\to\infty}\|T_nf-f\|_{A(D),r}=0\;\;\text{for any function}\;f\in A_g(D).\)