Generalized linear-quadratic stochastic control problem with incomplete information (Q1281101)

The classical results on linear-quadratic control problems are generalized to the case where the observable and nonobservable processes are described by stochastic differential equations with measure and the perturbing processes are not Gaussian. So one considers a wide class of processes which are non-Gaussian semimartingales whose evolution is described by the Itô stochastic differential equations. The admissible controls of a nonobservable process are confined to the linear class, which enables one to solve the problem analytically and prove results analogous to the well-known separation theorems, getting the answer in a closed form.