On error growth in the Bartels-Golub and Fletcher-Matthews algorithms for updating matrix factorizations (Q1096336)

From MaRDI portal
scientific article
Language Label Description Also known as
English
On error growth in the Bartels-Golub and Fletcher-Matthews algorithms for updating matrix factorizations
scientific article

    Statements

    On error growth in the Bartels-Golub and Fletcher-Matthews algorithms for updating matrix factorizations (English)
    0 references
    0 references
    1987
    0 references
    Matrix updating techniques play a large role in numerical optimization. The Bartels-Golub and Fletcher-Matthews methods are two possible updates for the LU-factorization when columns are exchanged. In this paper, the stability of these updates is considered. It is shown that serious loss of accuracy can occur in both cases. Artificially constructed examples are used to illustrate exponential error growth.
    0 references
    matrix update
    0 references
    rank-1 change
    0 references
    active-set strategy
    0 references
    Bartels-Golub method
    0 references
    numerical examples
    0 references
    Fletcher-Matthews methods
    0 references
    LU-factorization
    0 references
    stability
    0 references
    exponential error growth
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references