Abstract
The quadratic optimal control problems in linear continuous time stochastic systems with a nonclassical information structure are studied. To specify the information structure the systems are described by the integral equations which describe the causal relation between control variables and measurement variables. Under the assumption of Gaussian version of random variables, sufficient conditions for information structures which guarantee the linear optimal solutions are derived. Sufficient conditions under which the optimal solutions are the linear functions of state estimates are also obtained. As a special case of nonclassical information structures the problems of delayed information structures are discussed and the features of classical information structures are clarified.
