Szegö difference equations, transfer matrices and orthogonal polynomials on the unit circle (Q5953700)

The sequence of orthonormal polynomials \(\{ \varphi_n\}\) with respect to a probability measure \(\mu\) on the unit circle \(\mathbb T\) satisfies the Szegő recurrences, which written in the matrix form read as \[ X_n(z)=T_n(z) X_{n-1}, \quad T_n(z)={1\over \rho_n} \begin{pmatrix} z & a_n \\ \overline a_n z & 1 \end{pmatrix}, \quad X_0=\begin{pmatrix} 1 \\ 1 \end{pmatrix}, \quad X_n=\begin{pmatrix} \varphi_n(z) \\ \varphi_n^*(z) \end{pmatrix}, \tag{*} \] where \(\varphi_n^*(z)\) are the reversed polynomials, \(a_n\) are the reflection coefficients, and \(\rho_n=(1-|a_n|^2)^{1/2}>0\). This is the starting point of the paper. Using this ``matrix point of view'', the authors elegantly derive some well-known algebraic relations for \(\varphi_n\), and extend the Weyl's approach to second order linear differential equations on the half-line, showing how \(\mu\) arises from the unique \(l^2\) solution of (*). The main tools used are transfer matrices \(\mathcal T_n=T_n \dots T_1\) and the notion of subordinacy from the Gilbert-Pearson theory, applied to Szegő equation. Another remarkable aspect is the study of the Alexandrov measures \(\mu_\lambda \), corresponding as measure of orthogonality to the sequence of reflection coefficients \(\{ \lambda a_n\}\), \(|\lambda |=1\). The authors produce some formulas for orthogonal polynomials with respect to \(\mu _\lambda \) and prove a version of the Aronszajn-Donoghue theorem. Finally, it is shown that the behavior of norms \(\|\mathcal T_n \|\) is closely related to the absolute continuity of the orthogonality measure, and these two notions are connected by the property of subordinacy. This approach is illustrated by the study of Szegő recurrences with reflection coefficients having bounded variation. This is a nice and important contribution to the theory of orthogonal polynomials on the unit circle.