State representations of linear systems with output constraints (Q1264117)

This paper investigates the state space models of linear systems whose outputs are subjected to constant linear constraints and shows how, in the case of general linear systems, such representations are essentially nonunique and that their indeterminacy can be described by the freedom left in specifying a complement to a given subspace in the input space. It is then proved that, in the more specific case of linear Hamiltonian and gradient systems a notion of energy, expressed by means of special vector space structures, can be used to remove indeterminancies and to obtain results that are well known in physical modeling. Such results establish a link between the theory of transformations of linear systems and the methods of physical modeling. The paper is completed by two examples.