Residual, restarting and Richardson iteration for the matrix exponential

Mikhail A. Bochev

    Research output: Book/ReportReportProfessional

    525 Downloads (Pure)


    A well-known problem in computing some matrix functions iteratively is a lack of a clear, commonly accepted residual notion. An important matrix function for which this is the case is the matrix exponential. Assume, the matrix exponential of a given matrix times a given vector has to be computed. We interpret the sought after vector as a value of a vector function satisfying the linear system of ordinary differential equations (ODE), whose coefficients form the given matrix. The residual is then defined with respect to the initial-value problem for this ODE system. The residual introduced in this way can be seen as a backward error. We show how the residual can efficiently be computed within several iterative methods for the matrix exponential. This completely resolves the question of reliable stopping criteria for these methods. Furthermore, we show that the residual concept can be used to construct new residual-based iterative methods. In particular, a variant of the Richardson method for the new residual appears to provide an efficient way to restart Krylov subspace methods for evaluating the matrix exponential.
    Original languageUndefined
    Place of PublicationEnschede
    PublisherUniversity of Twente
    Number of pages20
    Publication statusPublished - Nov 2010

    Publication series

    NameMemorandum / Department of Applied Mathematics
    PublisherDepartment of Applied Mathematics, University of Twente
    ISSN (Print)1874-4850
    ISSN (Electronic)1874-4850


    • MSC-65F30
    • MSC-65F60
    • MSC-65N22
    • MSC-65F10
    • EWI-18832
    • Stopping criterion
    • Residual
    • MSC-65L05
    • Restarting
    • Richardson iteration
    • Matrix exponential
    • Matrix cosine
    • Backward stability
    • Chebyshev polynomials
    • IR-74752
    • METIS-271140
    • Krylov subspace methods

    Cite this