Conservative realizations of Herglotz-Nevanlinna functions (Q547468): Difference between revisions

Many natural, physical and engineered processes can be described by open systems which were introduced in early 1960s by \textit{M. S. Livšic} [Operators, oscillations, waves (open systems). Providence: AMS (1973; Zbl 0254.47001)]. An open system \(\Theta\) is described by the equations \[ \begin{cases} i d\chi/dt + T\chi(t) = KJ\psi_-(t),\\ \chi(0) = x \in\mathcal{H},\\ \psi_+(t) =\psi_-(t)-2iK^*\chi(t),\end{cases}\tag{*} \] where \(T\) is a bounded linear operator from a Hilbert space \(\mathcal{H}\) into itself, \(K\) is a linear operator from a Hilbert space \(\mathcal{E}\) (\(\dim \mathcal{E} < \infty\)) into \(\mathcal{H}\), \(J=J^*=J^{-1}\) maps \(\mathcal{E}\) onto itself, and \(\mathrm{Im}\,T=KJK^*\). If \(\psi_-\) is a continuous function on \([0,\tau_0]\) and we have that \(\chi(t)\in\mathcal{H}\) and \(\psi_+\in L^2_{[0,\tau_0]}(E) \) satisfies (*), then the following metric conservation law holds: \[ 2\|\chi(\tau)\|^2-2\|\chi(0)\|^2=\int_0^\tau[(J\psi_-(t),\psi_-(t))_{\mathcal{E}}-(J\psi_+(t),\psi_+(t))_{\mathcal{E}}]\,dt,\quad \tau\in[0,\tau_0]. \] Open systems are often characterized in terms of their transfer functions, which have the form \[ W_\Theta(z)=I-2iK^*(T-zI)^{-1}KJ. \] Such a transfer function satisfies, for \(z\in\rho(T)\), the properties: \[ W_\Theta(z)^*JW_\Theta(z)\begin{cases}\geq J\text{ for }\mathrm{Im}\,z>0,\\ = J\text{ for }\mathrm{Im}\,z=0,\\\leq J\text{ for }\mathrm{Im}\,z<0.\end{cases} \] Closely related to \(W_\Theta\) is the so-called impedance function of the system given by \[ V_\Theta(z)=K^*(\mathrm{Re}\,T-zI)^{-1}K=i[W_\Theta(z)+I]^{-1}[W_\Theta(z)-I]J, \] which is a Herglotz--Nevanlinna function (in the authors' terminology), i.e., analytic and with positive imaginary parts of its matrix values in the upper half-plane. In addition, \(V_\Theta(z)\) is self-adjoint when \(z\in\mathbb{R}\). When a physical system has distributed parameters, the state-space operator \(T\) of the system \(\Theta\) becomes unbounded and \(\mathrm{Im}\,T\) does not make sense. However, in some cases, the system can be considered in a generalized sense, where the construction of rigged Hilbert spaces \(\mathcal{H}_+\subset\mathcal{H}\subset\mathcal{H}_-\) associated with \(T\) is used. The importance of a study of open systems with unbounded operators was emphasized already by Livšic [op.\,cit.], and a series of works (Arov--Nudelman, Arov--Dym, Staffans, Ball--Staffans, and others) developed different approaches in this direction. The present authors develop their own approach, where the main emphasis is put on self-adjoint extensions of (unbounded) symmetric operators and the main object of study is a Livšic rigged canonical system (L-system). The applications include: the description of L-systems with Schrödinger operator, Nevanlinna--Pick interpolation problems for the class of scalar Herglotz--Nevanlinna functions, inverse spectral problems for finite dissipative Jacobi matrices. The book chapters are as follows: Chapter 1. Extensions of symmetric operators; Chapter 2. Geometry of rigged Hilbert spaces; Chapter 3. Bi-extensions of closed symmetric operators; Chapter 4. Quasi-self-adjoint extensions; Chapter 5. The Livšic canonical systems with bounded operators; Chapter 6. The Herglotz--Nevanlinna functions and rigged canonical systems; Chapter 7. Classes of realizable Herglotz--Nevanlinna functions; Chapter 8. Normalized L-systems; Chapter 9. Canonical L-systems with contractive and accretive operators; Chapter 10. L-systems with Schrödinger operator; Chapter 11. Non-self-adjoint Jacobi matrices and system interpolation; Chapter 12. Non-canonical systems. The book demonstrates the deep and powerful interconnection between operator theory, systems theory, and the theory of analytic functions of one variable (which was observed and explored in the pioneering works of M. Livšic). It will be of interest for researchers in each of these fields.