Iterations for solving a rational Riccati equation arising in stochastic control (Q2458700)

Two types of stochastic matrix Riccati equations are studied. Different iterative methods for computing approximations of Hermitian or maximal Hermitian solutions are proposed. The corresponding recurrence equations are based on the solution of matrix equations of Lyapunov's type. The proposed approach is less expensive compared with algorithms based on the classical Newton's method but requires stronger assumptions. Some convergence properties are also established.