求解二次特征值问题的添加精化向量的直接法

A direct projection method with refined vector for quadratic eigenvalue problems

基于残量范数极小的原则,提出了一种在迭代反位移的Arnoldi方法基础上进行改进的新算法,该算法在数值实验方面体现了其优越性。

Based on a residual norm minimizing scheme, a variant of the iterated shift-and-invert Amoldi algorithm is proposed. The numerical experimental results indicate that the new algorithm is of better performance.