A model function method in regularized total least squares

Abstract

In this article, we investigate the dual regularized total least squares (dual RTLS) from a computational aspect. More precisely, we propose a strategy for finding two regularization parameters in the resulting equation of dual RTLS. This strategy is based on an extension of the idea of model function originally proposed by Kunisch, Ito and Zou for a realization of the discrepancy principle in the standard one-parameter Tikhonov regularization. For dual RTLS we derive a model function of two variables and show its reliability using standard numerical tests.

Keywords:

• ill-posed problems
• regularized total least squares
• model function method

AMS Subject Classifications::

• 65J20
• 65M30

Acknowledgements

S. Lu and S.V. Pereverzev are supported by the Austrian Fonds Zur Förderung der Wissenschaftlichen Forschung (FWF), Grant P20235-N18.