id: 05176011
dt: j
an: 05176011
au: Lee, Young-Ju; Wu, Jinbiao; Xu, Jinchao; Zikatanov, Ludmil
ti: On the convergence of iterative methods for semidefinite linear systems.
so: SIAM J. Matrix Anal. Appl. 28, No. 3, 634-641 (2006).
py: 2006
pu: Society for Industrial and Applied Mathematics (SIAM), Philadelphia, PA
la: EN
cc: 
ut: iterative methods; regular splittings; large semidefinite linear systems;
    convergence
ci: Zbl 0135.37503
li: doi:10.1137/050644197
ab: Large semidefinite linear systems arise for instance from the
    discretization of pure Neumann problems. A sufficient condition for
    convergence of iterative methods of the form $M x^l=N x^{l-1}+b$ was
    given by {\it H. B. Keller} [J. Soc. Ind. Appl. Math., Ser. B Numer.
    Anal. 2, 281‒290 (1965; Zbl 0135.37503)]. Here the condition is
    generalized. The matrix $M$ need only be invertible on the range of the
    given matrix and not on the complete $n$-space.
rv: Dietrich Braess (Bochum)