Indefinite stochastic linear quadratic control with Markovian jumps in infinite time horizon (Q1407245)

The authors study a class of stochastic linear quadratic (LQ) control problems in the infinite time horizon with Markovian jumps in the parameters [for the problems, cf. \textit{M. Ait Rami} and \textit{L. El Ghaoui}, IEEE Trans. Autom. Control 41, 1666-1671 (1996; Zbl 0863.93087)]. In contrast to the cases studied in the literature, here the cost weighting matrices of the state and control are allowed to be indefinite. The problem is now equivalent to the solvability of a system of coupled generalized algebraic Riccati equations (CGAREs) under some equality and inequality constraints. Under the assumption of the nonsingularity of certain matrices, a linear matrix inequalities algorithm is proposed here to solve the associated CGAREs. Numerical examples are also given.