1: \begin{abstract}

2: We consider algebraic iterative reconstruction methods with applications

3: in image reconstruction.

4: In particular, we are concerned with methods based on an unmatched

5: projector/backprojector pair; i.e., the backprojector is not the exact adjoint

6: or transpose of the forward projector.

7: Such situations are common in large-scale computed tomography, and we

8: consider the common situation where the

9: method does not converge due to the nonsymmetry of the iteration matrix.

10: We propose a modified algorithm that incorporates a small shift

11: parameter, and we give the conditions that guarantee convergence of this method

12: to a fixed point of a slightly perturbed problem.

13: We also give perturbation bounds for this fixed point.

14: Moreover, we discuss how to use Krylov subspace methods to

15: efficiently estimate the leftmost eigenvalue of a certain matrix

16: to select a proper shift parameter.

17: The modified algorithm is illustrated with test problems from

18: computed tomography.

19: \end{abstract}

20: