On necessary conditions for the qualified convergence of methods for solving linear ill-posed problems (Q1599410)

The iterative regularizing algorithms for linear operator equation in Hilbert space are studied. Let \(A\) be a linear, positive and self-conjugate operator in a Hilbert space. In this space the equation \[ Au=f, \] is considered where the right of hand-side \(f\) is from the range of operator \(A\). Suppose that the range of the operator \(A\) is non-closed. Then the considered problem is non-stable. For the solution of this problem iterative regularized procedures are studied, which are convergent under the so-called well known condition ``source representation of initial approximation''. The authors show that this necessary condition is like the sufficient condition for the convergence of the studied iterative algorithms. The problem with noise data is also considered.