A class of iteration methods based on the Moser formula for nonlinear equations in Markov chains (Q1372973)

The author proposes two new methods for numerically determining the minimal nonnegative solutions for a class of nonlinear matrix equations which arises when the stationary probability vector of a transition matrix is computed. The proposed methods are based on the successive substitution iteration and on the Newton iteration, respectively. These new methods avoid the calculating of the inverses of the matrices. The monotone convergence of these methods is also discussed and two numerical examples are performed.