Solving matrix polynomial equations arising in queueing problems (Q5956252): Difference between revisions

The matrix equation \(\sum^n_{i=0} A_i X^i= 0\) (\(A_i\) are \(m\times m\) matrices) is studied. It is important in the analysis of queueing problems modeled by Markov chains. The main properties of the Möbius mapping are discussed to relate different resolution algorithms having quadratic convergence. These algorithms include logarithmic reduction, cyclic reduction (they extend Graeffe's iteration to matrix polynomials), and the invariant subspace approach, which comprises \textit{J. P. Cardinal's} algorithm [Lect. Appl. Math. 32, 165-188 (1996; Zbl 0893.65031)]. A new algorithm based on Chebyshev iteration is developed. Numerical experiments are presented.