On the strong divergence of Hilbert transform approximations and a problem of Ul'yanov (Q5965178)

Let \(Hf \) represent the Hilbert transform of a continuous function \(f\). This work contributes to the understanding of the approximation of the operator \(H\) by a sequence of operators \(H_N\) each of which is based on a finite number of samples. The predominant defining feature for an operator, \(T\), to be based on a finite number of samples is that there exists a finite set, \(S\), with the property that the values of a function \(g\) (which is in the domain of \(T\)) on \(S\) completely determine \(Tg\). It is known that for each such sequence \(H_N\) there is an \(f\) such that \(H_Nf\) does not converge uniformly to \(Hf\). That is, all such sequences of approximating operators diverge weakly. This work contributes two results that support the conjecture that, in fact, \(H_N\) strongly diverges. One result shows that even the Fejer means of these approximating sequences, \(H_N\), diverge. The other shows that, given two positive integers \(k\) and \(m\), there is an \(f\) such that the \(\| H_N f \| > k\) for more than \(m\) consecutive values of \(N\). The paper also contains applications to other related problems.