ACM Transactions on Mathematical Software (TOMS)
On the computation of few eigenvalues of positive definite Hamiltonian matrices
Future Generation Computer Systems
On the computation of few eigenvalues of positive definite Hamiltonian matrices
Future Generation Computer Systems
Hi-index | 0.00 |
This paper addresses some numerical issues that arise in computing a basis for the stable invariant subspace of a Hamiltonian matrix. Such a basis is required in solving the algebraic Riccati equation using the well-known method due to Laub. Two algorithms based on certain properties of Hamiltonian matrices are proposed as viable alternatives to the conventional approach.