Improvement of the eigenvalue-counting method based on the argument principle

Abstract

This note proposes an improved eigenvalue-counting method based on the argument principle by introducing Rombouts' algorithm, which is a stable, efficient, and accurate algorithm to calculate the coefficients of the characteristic polynomial of a general square matrix. In addition, this note comprehensively investigates the effectiveness of the proposed method from the viewpoint of the practical consideration by comparing with the modified Sturm sequence property-based counting method which has been recently developed and also known as the well-proven method. In this note, the operation counts in each method are carefully compared and numerically verified. According to analytical and numerical comparison, the argument principle-based counting method is much better than the modified Sturm sequence property-based method from a practical point of view, even if the latter method is much well-established one theoretically.