On the spectral norm of a random Toeplitz matrix (Q2461038)

Let \(X_0, X_1, \dots\) be a family of independent random variables. For \(n \geq 2\), let \(T_n\) denote the \(n \times n\) random symmetric Toeplitz matrix \(T_n = [X_{| j-k| }]_{1 \leq j,k \leq n}\). \textit{Z. D. Bai} [Stat. Sin. 9, No.~3, 611--677 (1999; Zbl 0949.60077)] asked whether the spectral measure of \(n^{-1/2}T_n\) approaches a deterministic limit measure \(\mu\) as \(n \to \infty\). \textit{W. Bryc, A. Dembo} and \textit{T. Jiang} [Ann. Probab. 34, No.~1, 1--38 (2006; Zbl 1094.15009)] and \textit{C. Hammond} and \textit{S. J. Miller} [J. Theor. Probab. 18, No.~3, 537--566 (2005; Zbl 1086.15024)] independently proved that this is so when the \(X_j\) are identically distributed with variance \(1\), and that with these assumptions \(\mu\) does not depend on the distribution of the \(X_j\). In this paper the author shows that the spectral norm \(\| T_n\| \) is of the order \(\sqrt{n\log n}\), in the case that the \(X_j\)'s need not be identically distributed, but satisfy stronger moment or tail conditions than in the above cited papers. It is shown that \(\| T_n\| \) is also of the same order for random Hankel matrices as well as other variants of random Toeplitz matrices.