Unsaturable methods for solving severely ill-posed problems (Q1037268)

The authors understand the paper as a contribution to the regularization theory for linear ill-posed operator equations \(Ax=y\), where the forward operator \(A:X \to Y\) is a compact linear operator mapping between the two Hilbert spaces \(X\) and \(Y\). In particular, with focus on order optimal methods for severely ill-posed situations the interplay of Tikhonov regularization and discretization in combination with the discrepancy principle is under consideration. Complementing the former theory which discussed Hölder convergence rates with saturation effects only there had been published a seminal paper by \textit{S. Pereverzev} and \textit{E. Schock} [Numer.~Funct.~Anal.~Optimization 21, No.~7--8, 901--916 (2000; Zbl 0974.65056)] that established the fact that no saturation occurs in case of severely ill-posed problems. This paper aims at logarithmic source conditions and consequently logarithmic convergence rates tries to extend those results to the case of noisy operators \(A_h\) instead of \(A\) in addition to noisy right-hand sides \(y^\delta \in Y\) with \(\|y-y^\delta\| \leq \delta\). Unfortunately, the paper is difficult to read since the main results in Theorem~4.1 and 4.2 are formulated in a rather technical manner an are not interpreted sufficiently well. Numerical case studies illustrate the theoretical study.