On coordination strategies for the interaction prediction principle using gradient and multiplier methods

Abstract
This paper deals with supremal level coordination strategies for the interaction (input) prediction principle. The decision-making problem of thoe supremal level (the coordinator) is to find a saddle point of a dual function of the coordination variables. The differentiability properties of the dual function are studied and explicit formulae for the gradients are derived. Application of the multiplier method of Hestenes results in a heuristical coordination strategy. The proposed algorithm is tested by two numerical examples.
Keywords