Sketch of a theory of nonlinear partial information Min-Max control
Résumé
We sketch a dynamic programing type of theory for both continuous time and discrete time non linear partial information min-max control, using the "cost to come" function as the informational state. We use this theory to derive conditions under which a certainty equivalence principle holds. The condition derived is less powerful than what was known from direct investigation in the continuous time case, but more powerful in the discrete.