Aspects of nonnormality for iterative methods (Q1765891)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Aspects of nonnormality for iterative methods
scientific article

    Statements

    Aspects of nonnormality for iterative methods (English)
    0 references
    0 references
    23 February 2005
    0 references
    The paper deals with the study of various aspects of nonnormality of matrices arising from the existence of some algorithms. First, the binormal matrices are introduced and the dimension of the set of binormal matrices is computed. A related circulant matrix structure is highlighted, and the polynomial normality for matrices are defined. The author discusses the solution of linear systems involving nonnormal matrices. It is proved that a linear system involving a binormal matrix can be solved by executing an optimal 3-term recurrence for normal matrices. Since polynomial normality of particular degree remains invariant under unitary similarity transformations, the unitary orbit of binormal matrices and polynomially normal matrices of moderate degree, are considered. Measures of nonnormality related to iterative methods are discussed and three algorithms for computing the minimal normal polynomial of a matrix are presented.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    nonnormal matrix
    0 references
    binormal matrix
    0 references
    polynomially normal operator
    0 references
    unitary orbit
    0 references
    involution
    0 references
    iterative methods for linear systems
    0 references
    Ritz values
    0 references
    measure of nonnormality
    0 references
    circulant matrix
    0 references
    algorithms
    0 references
    minimal normal polynomial
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references