Szegő's theorem and its probabilistic descendants (Q1950169)

The paper deals with the theory of orthogonal polynomials on the unit circle (OPUC). Its aim is to give some probabilistically motivated results, which complement the recent work by \textit{B. Simon} presented in his books [Orthogonal polynomials on the unit circle. Part 1: Classical theory. Providence, RI: American Mathematical Society (2005; Zbl 1082.42020); Orthogonal polynomials on the unit circle. Part 2: Spectral theory. Providence, RI: American Mathematical Society (2005; Zbl 1082.42021); Szegő's theorem and its descendants. Spectral theory for \(L^2\) perturbations of orthogonal polynomials. Princeton, NJ: Princeton University Press (2011; Zbl 1230.33001)], and in his survey article [Bull. Am. Math. Soc., New Ser. 42, No. 4, 431--460 (2005; Zbl 1108.42005)]. The subject of OPUC dates from two papers by \textit{G. Szegő} [Math. Z. 6, 167--202 (1920); ibid. 9, 167--190 (1921; JFM 48.0376.03)], and the key result, the three-term recurrence relation, appeared first in 1939 in his book [Orthogonal polynomials. New York: American Mathematical Society (AMS)(1939; Zbl 0023.21505)]. The present paper starts with the Kolmogorov isomorphism theorem, relating a complex-valued zero-mean covariance stationary stochastic process \( X = (X_n : n\in{\mathbb Z})\) to a process \(Y\) on the unit circle \(\mathbb T\) with orthogonal increments, and a probability measure \(\mu\) on \(\mathbb T\). After that, Verblunsky's theorem, stating the existence of a bijection between the sequences \(\alpha = ({\alpha}_n)^{\infty}_{n=1}\), \({\alpha}_n\in{\mathbb C}\), \(|\alpha|<1\), and nontrivial probability measures \(\mu\) on \(\mathbb T\), is recalled. The Verblunsky coefficients \({\alpha}_n\) are identified as the partial correlation functions as well as the parameters occurring in the Szegő recursion. The rest of the paper deals with conditions involving the Verblunsky coefficients and the spectral density (with respect to the normalized Lebesgue measure) of the measure \(\mu\), which were considered in the cited survey article by Simon. In the present paper, some of them are viewed as the conditions for the process \(X\) to be non-deterministic (Szegő's theorem), and are supplemented by some new time-series motivated conditions. For instance, following \textit{Ł. Dȩbowski} [Stat. Probab. Lett. 77, No. 7, 752--759 (2007; Zbl 1116.62091)] and \textit{A. Inoue} [Probab. Theory Relat. Fields 140, No. 3--4, 523--551 (2008; Zbl 1138.62049)], the author advocates a new definition of long-range dependence in the form of the condition \(\alpha \notin l_{1}\) (the negation of Baxter's condition). The article includes an extensive list of references on the subject.