Schur parameters, factorization and dilation problems (Q1917142)

The book under review is devoted to the study and applications of Schur parameters. Schur parameters were introduced by I. Schur in 1917 to solve interpolation problems for functions analytic in the unit disk. Since that time Schur parameters were used in many different problems of function theory and operator theory, and in applications. From the author's introduction: ``In Chapter 1 we describe the structure of positive definite kernels on the set of integers in terms of Schur parameters. This construction explores the interplay between some dilation theoretic ideas (Kolmogorov decomposition) and (Cholesky) factorization. A multiplicative structure of the Kolmogorov decompositions is also emphasized as a key technical tool. Chapter 2 explores the role of the Schur parameters in the structure of the triangular contractions and in the realization theory of unitary systems. Then, for another perspective on these results, the models for families of contractions given by Sz.-Nagy and Foias and, respectively, de Branges and Rovnyak are presented in some details. Chapter 3 deals with interpolation and moment problems. A method employed by M. A. Naimark and M. G. Krein is further developed, allowing us to view all these problems in a general framework of extending certain ``structured matrices''. The positivity of these matrices turns into the solvability criterion and the Schur parameters can be easily used to provide a description of all the solutions. The commutant lifting theorem is also analyzed in this chapter. The term of ``structured matrix'' is explained in Chapter 4 by the introduction of the concept of matrix with displacement structure. Here, the main achievement is a generalized Schur algorithm that encompasses all the algorithms mentioned above, starting with the algorithms of Schur and Nevanlinna and ending with least squares algorithms in adaptive filtering. Sometimes, it is possible to associate to this generalized Schur algorithm a generalized transmission-line model. The consideration of so-called transmission zeros of the line leads to the connection with the interpolation problems analyzed in Chapter 3. We also show that this connection can be explained as a consequence of the method of the commutant lifting. In Chapter 5 we unify the various factorization results used in the previous chapters by proving the existence of a spectral factor of a positive definite kernel. Chapter 6 contains an application to the study of nonstationary processes. Schur parameters are used to compute various angles between parts of past and future, as well as to discuss the Szegö limit theorems. In the last two chapters we pass from structured matrices to arbitrary patterns. It is shown that the Schur parameters still play a role if we stick to patterns associated to chordal graphs. In Chapter 7 we discuss completion problems, while in Chapter 8 we focus on determinantal formulae. We prove a general determinantal formula using the Schur parameters, we discuss a maximum determinant principle which is the right analogue of a maximum entropy principle analyzed in Chapter 5 and we show an inheritance property of the maximum determinant solution. And, of course, we give some hints about the limitations of the use of the Schur parameters''.