Limiting spectral distributions of some band matrices

DOI10.1007/S10998-011-7113-5MaRDI QIDQ663104zbMATH OpenOpenAlexFDO

Publication date 13 February 2012

Published in Periodica Mathematica Hungarica (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1003.5254

band matrix circulant matrix empirical spectral distribution Hankel matrix large dimensional random matrix limiting spectral distribution moment method reverse circulant matrix Toeplitz matrix

Mathematics Subject Classification ID

Random matrices (algebraic aspects) (15B52) Random matrices (probabilistic aspects) (60B20) Central limit and other weak theorems (60F05) Strong limit theorems (60F15)

Abstract: Except the Toeplitz and Hankel matrices, the common patterned matrices for which the limiting spectral distribution (LSD) are known to exist, share a common property--the number of times each random variable appears in the matrix is (more or less) same across the variables. Thus it seems natural to ask what happens to the spectrum of the Toeplitz and Hankel matrices when each entry is scaled by the square root of the number of times that entry appears in the matrix instead of the uniform scaling by

n^{- 1 / 2}

. We show that the LSD of these balanced matrices exist and derive integral formulae for the moments of the limit distribution. Curiously, it is not clear if these moments define a unique distribution.

Review text

The authors establish the limiting spectral distribution (LSD) of appropriately scaled large-dimensional random symmetric circulant, reverse circulant, Toeplitz and Hankel matrices which have suitable band structures, by the method of moments. The existence of an LSD means that the limit of the empirical spectral distributions in probability exists. A matrix has band structure if the top right corner and the bottom left corner elements are zeros, with larger dimension, the band of non-zero elements around the main diagonal may be larger. The authors define Type I and Type II bands in the introduction. \textit{V.~Kargin} [Electron. Commun. Probab. 14, 412--423 (2009; Zbl 1188.15036)] and \textit{D.-Z.~Liu} and {Z.-D.~Wang} [J. Theor. Probab. 24, No. 4, 988--1001 (2011; Zbl 1237.15034)] have obtained similar results as in the paper under review via different methods. Kargin dealt with Toeplitz matrices with Type I band structures, and Liu and Wang dealt with self-adjoint Toeplitz matrices of Type I band structure and real symmetric Hankel matrices of Type II band structure. Under certain assumptions, the authors prove that the limiting results hold for various matrices (Theorem 1). The main idea is to use the first moment condition (M1), second moment condition (M2), the Carleman condition (C) and the fourth moment condition (M4). The authors introduce the bounded Lipschitz metric in Section 3.1 and reduce the main result under the assumption that the input sequence is uniformly bounded in Section 3.2, by using the trace formula they verify (M1), with the asymptotically negligible (M1) condition can be verified for each matrices in Section 3.5. Note that the verification of (M1) essentially yields the corresponding LSD. Both conditions (C) and (M4) are verified in Section 3.6 and 3.7 for all matrices. Section 3 present the complete proof of the main result Theorem 1 in each subcase. The authors show some simulations for the cases in which the LSD is not known explicitly. Examples on the LSD clearly suggest a dependency on the parameter in Section 4, and suggest some further properties of the LSD.

Recommendations

Cites work

Cited in

(20)

This page was built for publication: Limiting spectral distributions of some band matrices

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q663104)