Multidimensional circuit synthesis and multivariable dilation theory (Q632161)

A map \(S\) from the unit disk \(\mathbb{D}\) on the complex plane into the space of bounded linear operators between two Hilbert spaces is said to be in the \textit{Schur class} if \(S\) is holomorphic and \(S(z)\) is a contractive operator for all \(z\) in \(\mathbb{D}\). These functions arose in operator theory as characteristic functions for completely non-unitary contractions on Hilbert spaces in the Sz.-Nagy-Foiaş model theory. Such a function is called inner if its values on the unit circle are unitary operators. Inner functions, in addition to their appearance in operator model theory, also came up in the Lax-Phillips scattering theory as the scattering matrix of a Lax-Phillips scattering system. One of the main results in the field is the state-space realization theorem, which says that a function \(S\) belongs to the Schur class if and only if there exists a unitary colligation matrix \(U=\left[\begin{smallmatrix} A & B\\ C & D\end{smallmatrix}\right]\) on a Hilbert state-space \(\mathcal{X}\) so that \(S(z)\) has the form \(S(z)=D+zC(I-zA)^{-1}B\). The function \(S\) is inner if and only if the operator \(A\) above is stable, that is, \(A^n\to 0\) as \(n\to\infty\) in the strong operator topology. In the context of circuit theory, engineers have become interested in square matrix- or operator-valued rational functions \(W\) defined on the right half plane \(\mathbb{C}_{+}\) with values \(W(s)\) having positive-semidefinite real part for all \(s\) in \(\mathbb{C}_{+}\). They are called positive-real functions. The analogue of inner functions in this setting consists of positive-real functions \(W\) with the additional condition that \(W(i\omega)\) has zero real part at all regular points \(i\omega\) on the imaginary line. Such functions are referred to as positive-real lossless (or PRL) functions in the paper. The result for PRL functions corresponding to the above state-space realization theorem asserts that a rational square matrix-valued function \(W\) is PRL if and only if \(W\) can be represented as \(W(s)=Es+D+B^{*}(sI-A)^{-1}B\), where \(A\) and \(D\) are skew Hermitian and \(E\) is positive semi-definite. The paper under review concerns extensions of the above results to multivariate settings, where the underlying domain is the polydisk \(\mathbb{D}^d\) or the poly-half-plane \((\mathbb{C}_{+})^{d}\). Approaches and recent progress obtained in both the mathematics and engineering communities are discussed. In both communities, the case of one or two variables has been well established, but the case of more than two variables remains incomplete. This paper will be useful not only for operator theorists, but also for engineers working on the subject.