解大规模反对称矩阵内部特征问题的广义调和Lanczos方法

A generalized harmonic Lanczos method for solving large skew-symmetric eigenproblems

利用广义Lanczos算法,提出一种计算反对称矩阵内部特征值的广义调和Lanczos方法．这种方法只需要简单的三项递推关系式就可以将大规模特征问题转化为一个中小规模广义特征问题求解,因此计算量和存储量都很小．

This paper presents a harmonic generalized lanczos algorithm for computing a few interior eigenvahes and correspongding eigenvectors of skew - symmetric matrices. The new method is based on a generalized Lanczos algorithm. So it needs less computational and storage costs than harmonic Arnoldi method is popular for nonsymmtric interior eigenproblems. Numerical experiments are reported to show the efficiency of the method.