Iterative implementation of the adaptive regularization yields optimality (Q2574682)

Consider a linear ill-posed equation \(Tx=y\), where \(T: X \to Y\) is a bounded linear operator and \(X\), \(Y\) are Hilbert spaces. Suppose that instead of \(y\), an approximation \(y_{\delta}\) and the error level \(\delta\) are given and \(\| y_{\delta}-y \| \leq \delta\). The authors establish the convergence of the iterative process \((H^2+\alpha_n I) x_n^{\delta}=\alpha_n x_{n-1}^{\delta}+H T^* y_{\delta}\), where \(H=T^* T\) and \(\alpha_n=\alpha_0 \xi^{n-1}\), \(\xi \in (0,1)\).