Effect of the CYBER 205 on the choice of method for solving the eigenvalue problem \((A-\lambda M)x=0\) (Q1097638)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Effect of the CYBER 205 on the choice of method for solving the eigenvalue problem \((A-\lambda M)x=0\)
scientific article

    Statements

    Effect of the CYBER 205 on the choice of method for solving the eigenvalue problem \((A-\lambda M)x=0\) (English)
    0 references
    0 references
    0 references
    0 references
    1986
    0 references
    For the eigenvalue problem \(Ax=\lambda Mx\), A, B large, sparse, symmetric matrices, two methods, subspace iteration and Lanczos method, are compared when running on typical examples from structural dynamic analysis (order of A, B up to 8000) on a Cyber 205. A fixed number of eigenpairs is calculated. As on serial computers it turns out on this vector computer that the Lanczos algorithm is considerably faster. However, on problems with substantial overhead in reading/writing, a block Lanczos method is preferable.
    0 references
    0 references
    vectorization
    0 references
    comparison of methods
    0 references
    eigenvalue problem
    0 references
    large, sparse, symmetric matrices
    0 references
    subspace iteration
    0 references
    Lanczos method
    0 references
    vector computer
    0 references
    0 references
    0 references