摘要
文[1]中,Saad在矩阵只有实单特征值的情况下,证明了广义Lanczos方法是快速收敛的.本文将Saad这一结果推到一般复矩阵的情况.并给出了相应的误差界.
In this paper, we extended the error estimates of eigenvalue problem in [3] and the Convergent theorem in [1] to the Case in which matrices may have nonlinear elementary divisor. Bounds for the rates of Convergence were obtained.
出处
《黑龙江大学自然科学学报》
CAS
1991年第4期26-30,50,共6页
Journal of Natural Science of Heilongjiang University