On the possibility of generalizing ``correction'' theorems (Q1277504)

Let \(I\) be a set of indices with a finite number of elements, and let \(\Omega\) be a family of its subsets. Further, let \(\{\varphi_\alpha\}_{\alpha \in I}\) be an orthonormal system of functions on a measure space \((X,\Sigma,\mu)\), where \(\mu(X)=1\). For an orthogonal expansion \[ f(x) \sim \sum_{\alpha \in I} c_\alpha \varphi_\alpha(x), \] let the \(\Omega \)-majorant of partial sums be defined as \[ S_\Omega^*f(x)= \sup_{\Lambda \in \Omega} \left|\sum_{\alpha \in \Lambda} c_\alpha \varphi_\alpha(x) \right|,. \] In this note, which is a continuation of the paper [\textit{B. S. Kashin}, Mat. Zametki 46, No. 6, 67-74 (1989; Zbl 0736.42006)], the author imposes some suitable conditions on \(\Omega\) and \(\{\varphi_\alpha\}\) and explores the possibility to modify a bounded function \(f\) on a small set in such a way that the ``corrected'' function \(\widetilde{f}\) has a bounded \(\Omega\)-majorant, \[ \|S_\Omega^*\widetilde{f}\|_{L^\infty} < \infty . \] The results are applicable also to the study of majorants of rectangular partial sums of multiple orthogonal series.