A matrix decomposition-reduction procedure for the pole-zero calculation of transfer functions

This paper is concerned with eigenvalue approaches for the pole-zero calculations of the transfer function of a linear timeinvariant network. Combinatorial and sparse-matrix algorithms are fully used to increase numerical accuracy and computational speed in the two-sets-of-eigenvalues approach. Some matrix decompositionreduction algorithms are presented to simplify and stabilize the numerical eigenvalue-finding procedures.