An overview of matrix factorization theory and operator applications.

Authors I. Gohberg, Marinus Kaashoek, Ilya M. Spitkovsky

Publication date 2003

factorization Fredholm theory indices invertibility matrix function ordinary differential equations with operator coefficients realization Riemann-Hilbert problem state method systems of singular integral equations systems theory Wiener-Hopf and Toeplitz operators

Mathematics Subject Classification ID

Boundary value problems in the complex plane (30E25) Riemann-Hilbert problems in context of PDEs (35Q15) Systems of singular linear integral equations (45F15) Integro-ordinary differential equations (45J05) Introductory exposition (textbooks, tutorial papers, etc.) pertaining to operator theory (47-01) Numerical range, numerical radius (47A12) Operator colligations (= nodes), vessels, linear systems, characteristic functions, realizations, etc. (47A48) (Semi-) Fredholm operators; index theories (47A53) Factorization theory (including Wiener-Hopf and spectral factorizations) of linear operators (47A68) Toeplitz operators, Hankel operators, Wiener-Hopf operators (47B35)

Review text

The present lecture notes present an extensive review of the factorization theory of matrix functions relative to a curve with emphasis on the developments of the last 20-25 years. The material consists of four chapters. The first chapter contains the notion of factorization with continuous and bounded measurable factors. Given a matrix function \(G\) on a bounded Jordan curve \(\Gamma\), its left factorization is a representation \[ G(t)=G_+(t)\Lambda(t)G_-(t),\quad t\in\Gamma, \] where \(\Lambda(t)=\text{diag}\, [t^{\kappa_1},\ldots,t^{\kappa_n}]\) with integers \(\kappa_1,\ldots,\kappa_n\) called left factorization (partial) indices, and the \(n\times n\) matrix functions \((G_+)^{\pm 1}\) and \((G_-)^{\pm 1}\) are analytic, respectively, in the inner and outer domains determined by \(\Gamma\), and their non-tangential limits on \(\Gamma\) are continuous or bounded measurable. A right factorization \(G=F_-MF_+\), where \(M(t)=\text{diag}\, [t^{\mu_1},\ldots,t^{\mu_n}]\) and \(\mu_1,\ldots,\mu_n\in\mathbb{Z}\), is defined analogously. The existence of a factorization is established for rational matrix functions and for positive definite matrix functions defined on the unit circle. The uniqueness of the factorization indices and their behavior under small perturbations, relations between the factorization and numerical range of the matrix, between the indices of left and right factorizations, and the connection with the Birkhoff factorization are considered. Applications to the Riemann-Hilbert boundary value problem and to the Fredholm theory of singular integral operators based on the factorization concept are also considered. The second chapter deals with the theory of factorization in abstract decomposing Banach algebras and its applications to concrete decomposing algebras of continuous matrix functions on a curve. Such algebras include, for example, Wiener type algebras or algebras of Hölder continuous matrix functions. Applications to the invertibility and the Fredholm theory of block Toeplitz operators and Wiener-Hopf integral operators are described as well. In the third chapter, the notion of generalized \(L_p\) factorization for a matrix function \(G\in L_\infty^{n\times n}\) is introduced and properties of this factorization are considered. Unlike the factorization with bounded factors, the generalized \(L_p\) factorization of a discontinuous matrix function \(G\) can explicitly depend on the exponent \(p\in(1,\infty)\) of the Lebesgue space \(L_p^n(\Gamma)\) and is equivalent to the Fredholmness on \(L_p^n(\Gamma)\) of the singular integral operator \(P_++GP_-\) related to the Riemann-Hilbert boundary value problem. Factorization of different classes of discontinuous matrix functions (piecewise continuous, matrices with entries in Douglas algebras, sectorial) is studied. Furthermore, the so-called \(APW\) and \(AP\) factorizations of almost periodic matrix functions \(F\in (APW)^{n\times n}\) on the real line are introduced and the equivalence of these factorizations and the invertibility of the singular integral operators \(P_++FP_-\) on the space \(L_p^n(\mathbb{R})\) is stated. Several sufficient conditions for the existence of the \(AP\) factorization are cited. In contrast to rational matrix functions, there are matrix \(AP\) polynomials that do not have such a factorization. A Fredholm criterion for the operators \(P_++FP_-\) with semi almost periodic matrix functions \(F\) on the space \(L_p^n(\mathbb{R})\) is presented. Finally, new phenomena in the relations between left and right generalized factorizations are discussed: not only the sets of left and right factorization indices might be different, but their sums as well. Moreover, the left and right factorizations do not always exist simultaneously. In the fourth chapter, the authors describe the state space approach to the factorization of rational matrix functions and its generalizations to non-rational matrix or operator-valued functions. Such an approach originated from the theory of linear input-output systems and is based on a representation of any rational \(n\times n\) matrix function \(F\) in the form \[ F(\lambda)=I+C(\lambda G-A)^{-1}B \] where \(I\) is the \(n\times n\) identity matrix, \(A\) and \(G\) are \(m\times m\) matrices, and the matrices \(C\) and \(B\) are of sizes \(n\times m\) and \(m\times n\), respectively. Such a representation, called a realization of \(F\), allows one to construct a factorization for \(F\) explicitly in terms of the four matrices \(A\), \(G\), \(B\), and \(C\). Thus, the factorization problem for \(F\) is reduced to finding a realization for \(F\). The authors review applications of the state space method to the inversion and the Fredholm theory of Toeplitz operators, Wiener-Hopf operators, and singular integral operators. They also consider its applications to spectral factorization, to linear integro-differential equations arising in the linear transport theory, and their generalizations with operator coefficients. Each chapter contains notes describing the history of the subject and corresponding references. For the entire collection see [Zbl 1012.00042].

Recommendations

Cited in

(86)

This page was built for publication: An overview of matrix factorization theory and operator applications.

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4425382)