The Pólya algorithm in sequence spaces (Q2566251)

Let \(K\) be a finite-dimensional linear subspace of the sequence space \({l_1}\). Let \({h\in l_p}\) and \({h\notin K}\). Let \(h_p\) denote the best \(l_p\)-approximation of \(h\) from \(K\). It is well known that \(h_p\) exists and is unique. In general, the unicity of the best uniform approximation of \(h\) is not guarantied. However, a unique ``strict uniform approximation'' \(h^*_{\infty}\) can be defined [see \textit{M. Marano}, Approximation Theory Appl. 6, No. 1, 99--109 (1990; Zbl 0729.41030)]. The remarkable property of the strict uniform approximation is that \({h_p\to h^*_{\infty}}\) as \({p\to +\infty}\). This way of getting the best uniform approximation is called the Pólya algorithm. The aim of the paper under review is to establish the rate of convergence of \({\| h_p-h^*_{\infty}\| _\infty}\) as \({p\to +\infty}\) by proving that there are constants \({L_1,L_2>0}\) and \({a\in [0,1)}\) such that \[ L_1 a^p\leq p \| h_p-h^*_{\infty}\| _\infty\leq L_2 a^p \] for \(p\) large enough.