Efficient computation of large scale transfer function dominant zeros

This paper describes efficient algorithms for the computation of dominant zeros of large scale transfer functions. The transfer function zeros are demonstrated to be equal to the poles of a new inverse system, which is valid even for the strictly proper case. This is a new finding, which is important from practical as well as theoretical viewpoints. Hence, the dominant zeros can be computed as the dominant poles of the inverse transfer function by the recent efficient SADPA and SAMDP algorithms. The importance of computing dominant zeros and the performance of the algorithms are illustrated by examples from practical power system models.