Multi-parameter Tikhonov regularization and model function approach to the damped Morozov principle for choosing regularization parameters (Q2428101): Difference between revisions

The multi-parameter regularization of Tikhonov type for linear ill-posed operator equations in a Hilbert space setting is under consideration. The first part of the paper extends the analysis around the model function approach for iterative choices of the regularization parameter along the lines of the original paper presented by \textit{K. Kunisch} and \textit{J. Zou} [Inverse Probl. 14, No. 5, 1247--1264 (1998; Zbl 0917.65053)] to the case of two regularization parameters. It proceeds as the original paper and considers in this context Morozov's discrepancy principle and also the damped version of this principle. The second part of the paper discusses a couple of new ansatz functions called, for example, hyperpolic, exponential and logarithmic model functions and formulates adapted algorithms for their handling. In a concluding section the algorithms concerning the damped Morozov principle are applied to four examples of integral equations of the first kind. Three of the test problems are one-dimensional. The fourth test problem refers to a two-dimensional image deblurring situation. A large series of tables shows the performance of the different algorithms and the dependence of the results on the damping exponents.