Topics in multidimensional linear systems theory (Q1586215)

The behavioural approach to system theory, introduced in the 1980s by J. C. Willems to provide a framework in which all signals connected with a system could be treated equally, without necessarily introducing a notion of causality, proved to be particularly fruitful for systems depending on several independent variables. The book provides a systematic and up to date presentation of the behavioural approach to the theory of multidimensional linear systems and contains a rich and detailed exposition of the algebraic framework for the study of multidimensional behaviours, in particular the theory of co-prime factorisations of multivariate rational matrices and a number of important applications. Chapter 1 is devoted to the fundamental notion of controllability, which in the behavioural approach is an intrinsic property of the system depending neither on the distinction between inputs and outputs nor on causality. The primeness of multivariate polynomial matrices is the key point in the definition of a notion of controllability for multidimensional systems. Various characterizations of primeness for polynomial matrices are considered and are shown to be mutually equivalent in the one-dimensional case, while in the multivariate case they lead to different concepts. The notion that can be interpreted in terms of concatenability of trajectories is chosen as more appropriate for the multidimensional case. Chapter 2 is devoted to the problem of left (or right) co-prime factorisation of rational matrices and to the associated kernel or image representations of multidimensional systems. The generalized notion of factor primeness is presented, and a procedure to obtain the coprime factors of a multivariate rational matrix is given. The properties of such a decomposition are discussed also in relation with the factorisation ideals. The case of 2D systems presenting distinctive features is considered in the last section. Chapter 3 is devoted to the notion of stability and stabilizability by means of dynamic output feedback in terms of coprime factorisations over the ring of stable transfer functions. In Chapter 4 a class of admissible systems, for which a transfer function is well defined, is studied, and a notion of system equivalence is introduced. Finally, general, non causal, first-order representations for multidimensional systems are studied in Chapter 5. Generalized state space behaviours, involving manifest and latent system variables, that give rise to output nulling representations can be derived from kernel representations by suitable linear fractional transformation. Chapter 6 contains a number of results concerning linear fractional transformations that are prominent in \(H_\infty\) and robust control, in particular a constructive procedure that can be seen as a realization algorithm for multidimensional finite-impulse-response filters. For robust stability issues, in fact, it is often desirable to represent uncertain transfer matrices in terms of linear fractional transformations, and several results previously presented for multivariate polynomial matrices therefore apply. In the final Chapter 7 a number of known facts on the mathematical theory of networks and circuits are recalled.