The parametrized \(SR\) algorithm for Hamiltonian matrices
Electronic transactions on numerical analysis, Tome 26 (2007), pp. 121-145
The heart of the implicitly restarted symplectic Lanczos method for Hamiltonian matrices consists of the algorithm, a structure-preserving algorithm for computing the spectrum of Hamiltonian matrices. The
§
symplectic Lanczos method projects the large, sparse Hamiltonian matrix onto a small, dense $\ddot \copyright \ddot \copyright \ddot $ Hamiltonian -Hessenberg matrix , . This Hamiltonian matrix is uniquely determined by $\ddot !\copyright \ddot "\ddot $ # parameters. Using these # parameters, one step of the algorithm can be carried out in %' %' ###(###
| $###$ |
| $###$ |
| $ )10243 arithmetic operations (compared to arithmetic operations when working on the actual Hamiltonian matrix).$ |
Classification :
65F15
Keywords: Hamiltonian matrix, eigenvalue problem, algorithm $$###(###$$
Keywords: Hamiltonian matrix, eigenvalue problem, algorithm $$###(###$$
@article{ETNA_2007__26__a19,
author = {Fa{\ss}bender, H.},
title = {The parametrized {\(SR\)} algorithm for {Hamiltonian} matrices},
journal = {Electronic transactions on numerical analysis},
pages = {121--145},
year = {2007},
volume = {26},
zbl = {1171.65375},
language = {en},
url = {http://geodesic.mathdoc.fr/item/ETNA_2007__26__a19/}
}
Faßbender, H. The parametrized \(SR\) algorithm for Hamiltonian matrices. Electronic transactions on numerical analysis, Tome 26 (2007), pp. 121-145. http://geodesic.mathdoc.fr/item/ETNA_2007__26__a19/