Optimal state feedback input-output stabilization of infinite-dimensional discrete time-invariant linear systems (Q841664)

The paper deals with the optimal input-output stabilization of discrete time-invariant linear systems of \[ x_{n+1}= Ax_n+ Bu_n,\quad y_n= Cx_n+ Du_n,\quad x_0= z, \] in Hilbert spaces, described by bounded linear operator \(A, B, C, D\). In the optimal version, one does not only require this input-output map to be bounded, but to have the smallest possible norm. Special solutions to this optimal control problem are known if all the spaces are finite-dimensional, but the problem is much more difficult in infinite-dimensional spaces. A necessary and sufficient condition for the problem formulated above is that the transfer function has a right factorization over \(H^\infty\). A function with \(H^\infty\)-factorization in the unit disc satisfies the finite cost condition. In the paper under review, a condition is introduced which is called the finite future incremental cost condition, which is weaker than the finite cost condition. It is proved that every realization of a function with \(H^\infty\)-factorization satisfies the new condition and conversely, from this new condition follows the \(H^\infty\)-factorization. An equivalent condition is derived, which is connected with the solution of the control Riccati equation. The optimal feedback operator can be unbounded! The optimal state feedback input-output stabilization problem can be solved explicitly in terms of a solution of this control Riccati equation. There are new treatments of the control Riccati equation. An appendix presents several alternative forms of the control Riccati equation. A second one gives results on extensions and compression of systems.