Large-scale dual regularized total least squares
Electronic transactions on numerical analysis, Tome 42 (2014), pp. 13-40
The total least squares (TLS) method is a successful approach for linear problems when not only the right-hand side but also the system matrix is contaminated by some noise. For ill-posed TLS problems, regularization is necessary to stabilize the computed solution. In this paper we present a new approach for computing an approximate solution of the dual regularized large-scale total least squares problem. An iterative method is proposed which solves a convergent sequence of projected linear systems and thereby builds up a highly suitable search space. The focus is on an efficient implementation with particular emphasis on the reuse of information.
Classification :
65F15, 65F22, 65F30
Keywords: total least squares, regularization, ill-posedness, generalized eigenproblem
Keywords: total least squares, regularization, ill-posedness, generalized eigenproblem
@article{ETNA_2014__42__a8,
author = {Lampe, J\"org and Voss, Heinrich},
title = {Large-scale dual regularized total least squares},
journal = {Electronic transactions on numerical analysis},
pages = {13--40},
year = {2014},
volume = {42},
zbl = {1295.65043},
language = {en},
url = {http://geodesic.mathdoc.fr/item/ETNA_2014__42__a8/}
}
Lampe, Jörg; Voss, Heinrich. Large-scale dual regularized total least squares. Electronic transactions on numerical analysis, Tome 42 (2014), pp. 13-40. http://geodesic.mathdoc.fr/item/ETNA_2014__42__a8/