On the computation of few eigenvalues of positive definite Hamiltonian matrices
On the computation of few eigenvalues of positive definite Hamiltonian matrices
复制标题
正定哈密顿矩阵少数特征值的计算
DOI：
10.1016/j.future.2004.11.027
复制
发表时间：
2006
期刊：
影响因子：
--
通讯作者：
P. Amodio
中科院分区：
文献类型：
--
作者：
P. Amodio
给定一个S对称正定的哈密顿矩阵H=JS，分析了用辛Lanczos算法将H=JS变换成一个半大小的对称正定三对角矩阵。通过两种有效的重新启动过程，将该算法用于计算h的几个极值特征值，并通过数值算例对所提方法进行比较。
Given a Hamiltonian matrix H=JS with S symmetric and positive definite, we analyze a symplectic Lanczos algorithm to transform −H2in a symmetric and positive definite tridiagonal matrix of half size. By means of two effective restarted procedures, this algorithm is then used to compute few extreme eigenvalues of H. Numerical examples are also reported to compare the presented techniques.