摘要
本文提出一种能求解实非对称矩阵广义特征值问题的正交共轭子空间迭代法,算例表明,该法收敛速度快,且能求解复特征值、非正实特征值、非常接近的特征根或重根,不会发生丢根的现象。
This paper presents the orthogonal conjugate subspace iteration method which can solve the generalized eigenvalue problems of real unsymmetrical matrix. The numerical examples show that by using this method convergence is very quickly and very close eigenvalues or multiple roots can be gotten without missing roots.
关键词
正交共轭
子空间迭代
非对称子空间
orthogonal conjugate subspace iteration
complex double-step QR algorithm
unsymmetrical subspace