A note on the spectral distribution of non-Hermitian block matrices with Toeplitz blocks
Andrea Adriani (1) and Giacomo Tento (2) · 2026
Read A note on the spectral distribution of non-Hermitian block matrices with Toeplitz blocks by Andrea Adriani (1) and Giacomo Tento (2) (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-03823.txt.
Introduction
Andrea Adriani and Giacomo Tento studied how eigenvalues spread for block matrices that are not Hermitian and whose blocks have Toeplitz form. In plain terms, they asked where the eigenvalues pile up as the matrices grow large. The point is to predict the pattern before computing it. Using geometric means of matrices and the theory of Generalized Locally Toeplitz sequences, they derived the limiting spread. Numerical tests and pictures backed the derivations.
Catalog record
| CREATOR | Andrea Adriani (1) and Giacomo Tento (2) |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51216 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.03823
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A note on the spectral distribution of non-Hermitian block matrices with Toeplitz blocks Andrea Adriani (1) and Giacomo Tento (2) Abstract Abstract In the present paper, we are concerned with the study of the spectral distribution of matrix-sequences showing a non-Hermitian block structure with Toeplitz blocks. We use the notion of geometric mean of matrices and the theory of Generalized Locally Toeplitz (GLT) sequences to perform our analysis and produce some numerical tests and visualizations to confirm our theoretical derivations. (1) Vanguard Center, University Mohammed VI Polytechnic, Morocco ([email protected], ORCID ID: 0000-0003-3390-7891); (2) University of Insubria, Via Valleggio 11, 22100, Como, Italy ([email protected], ORCID ID:0009-0000-5773-7829) Keywords Block matrices, Toeplitz matrices, Spectral distribution 1 Introduction A matrix-sequence \left\{A_{n}\right\}_{n} is an ordered collection of matrices A_{n}\in\mathbb{C}^{d_{n}\times d_{n}} , with d_{n} strictly increasing and n\in\mathbb{N} . Such sequences naturally arise from the numerical discretizaton of integral, ordinary and partial differential equations; most often they possess a Generalized Locally Toeplitz (GLT) symbol (see, for instance, [ 11 ] ), that is, a measurable function f:\Omega\subset\mathbb{R}\rightarrow\mathbb{C}^{s\times s} that, under additional circumstances, is able to provide a description of the asymptotic distribution of the eigenvalues of the sequence. Namely, we say that \{A_{n}\}_{n} is distributed in the eigenvalue sense as f over \Omega and we write \{A_{n}\}_{n}\sim_{\lambda}(f,\Omega) , if \lim_{n\to\infty}\sum_{i=1}^{d_{n}}F\left(\lambda_{i}\left(A_{n}\right)\right)=\frac{1}{\mathcal{L}\left(\Omega\right)}\int_{\Omega}\frac{\sum_{j=1}^{s}F\left(\lambda_{j}(f(x))\right)}{s}\mathrm{d}x (1) for every continuous function F:\mathbb{C}\rightarrow\mathbb{C} with compact support. For our purposes, we are interested in uni-variate matrix-valued symbols, but the notion of spectral distribution can been extended also to multilevel matrix-sequences and multilevel block matrix-sequences with multi-variate matrix-valued symbols (see [ 13 , 14 ] ). The theory of GLT sequences stems from previous inquiries on Locally Toeplitz (LT) sequences ( [ 22 ] ), and underwent a systematic study in [ 12 ] . Notable applications of this theoretical apparatus comprise finite differences methods (see, e.g., [ 21 ] ), finite elements methods ( [ 12 , Section 10.6] ), finite volumes ( [ 10 ] ) and fractional derivatives ( [ 16 ] ). Moreover, information provided by means of GLT sequences are not only of interest from a theoretical standpoint, but often carries significant practical purposes. For instance, it is known that the convergence rate of Krylov methods strongly depends on the spectral features of the matrices to which they are applied; thus, the spectral distribution notion can be employed to predict and improve their performances via preconditioning techniques (see [ 12 , Section 10] ). In this paper, we consider sequences of block matrices k\times k , k\geq 2 , with the following structure: \mathcal{A}_{n}=\begin{bmatrix}T_{n}(f_{1,1})&T_{n}(f_{1,2})&&\\ T_{n}(f_{2,1})&T_{n}(f_{2,2})&\ddots&\\ &\ddots&\ddots&T_{n}(f_{k-1,k})\\ &&T_{n}(f_{k,k-1})&T_{n}(f_{k,k})\end{bmatrix}, (2) where each f_{i,j} is a function in L^{\infty}([-\pi,\pi]) and T_{n}(f) denotes the Toeplitz matrix of dimension n generated by f . In the case where the sequence is Hermitian, it is possible to use the theory developed in [ 1 , 2 ] (see also [ 6 ] ) to perform a complete spectral analysis of the sequence, both from the singular value and the eigenvalue perspective. Moreover, even when the sequence is not Hermitian, and the ratio of the dimensions of the various blocks with the global dimension tends (as n\to\infty ) to any rational or irrational number in (0,1) , the tools in [ 1 , 2 ] allow us to find the singular value distribution of the sequence. However, in the case where the sequence does not present a Hermitian structure, the theory does not explicitly give the eigenvalue distribution, and a new approach is needed to tackle the problem. In this paper, we use the notion of geometric means of matrices to solve this issue for a class of matrices as in ( 2 ). Note that sequences of matrices presenting a similar non-Hermitian structure are already known in applications (see, for example, [ 8 ] , where the authors study electro-diffusion equations). One of the first formulations of geometric means of matrices can be found in [ 17 ] , where many properties are discussed. We make use of their definition, that is, for two Hermitian positive definite (HPD) matrices A and B , we let G(A,B):=A^{1/2}\left(A^{-1/2}BA^{-1/2}\right)^{1/2}A^{1/2}=G(B,A) (3) be their geometric mean. For a recent account on how the Generalized Locally Toeplitz (GLT) theory is related to this topic, we refer the reader to [ 15 ] . The main contribution of the present work is the proof of the following theorem. Theorem 1.1 . Let \mathcal{A}_{n} be as in Equation ( 2 ). Assume that f_{i,j}\in L^{\infty}([-\pi,\pi]) and real-valued almost everywhere for every i,j . Moreover, assume that \mathop{\mathrm{ess\,inf}}f_{i,i+1},\mathop{\mathrm{ess\,inf}}f_{i+1,i}>0 for every i=1,\dots,k-1 . Then \{\mathcal{A}_{n}\}_{n}\sim_{\lambda}F(\theta), with F(\theta)=\begin{bmatrix}f_{1,1}&f_{1,2}&&\\ f_{2,1}&f_{2,2}&\ddots&\\ &\ddots&\ddots&f_{k-1,k}\\ &&f_{k,k-1}&f_{k,k}\end{bmatrix}. Equivalently, \{\mathcal{A}_{n}\}_{n}\sim_{\lambda}\tilde{F}(\theta), with \tilde{F}(\theta)=\begin{bmatrix}f_{1,1}&f_{1,2}^{1/2}f_{2,1}^{1/2}&&\\ f_{1,2}^{1/2}f_{2,1}^{1/2}&f_{2,2}&\ddots&\\ &\ddots&\ddots&f_{k-1,k}^{1/2}f_{k,k-1}^{1/2}\\ &&f_{k,k-1}^{1/2}f_{k-1,k}^{1/2}&f_{k,k}\end{bmatrix}. In order to prove our main theorem, we will make use of a fine approximation of each Toeplitz block with a circulant matrix depending on the generating function (see [ 23 ] and references therein) and of [ 7 , Theorem 1] . The paper is organized as follows. In Section 2 we recall the concepts of spectral distribution, Toeplitz sequences, GLT sequences (and related properties). There, we also recall the definition of geometric mean of two matrices that we use throughout the paper and discuss the correlation between GLT theory and geometric means. Section 3 is the core of the present note and is devoted to the proof of our main result. In Section 4 we perform some numerical tests that corroborate the theory developed and, finally, Section 5 is devoted to draw conclusions, provide a discussion on possible improvements, and state some open problems for future research. 2 Preliminary tools: Spectral distribution, Toeplitz matrices, GLT sequences We start this section with the definition of spectral distribution, which is the main objective of investigation in this work. After this, we recall the definition of approximating class of sequences (a.c.s.) and list some classes of meaningful sequences, ending with the class of GLT sequences. Definition 2.1 . Let \{A_{n}\}_{n} be a matrix-sequence, with A_{n} of size d_{n} increasing, and let f:\Omega\subset\mathbb{R}\to\mathbb{C}^{s\times s} be a measurable function defined on a set \Omega with Lebesgue measure 0<\mathcal{L}(\Omega)<\infty . • We say that \{A_{n}\}_{n} has a (asymptotic) singular value distribution described by f on \Omega , and we write \{A_{n}\}_{n}\sim_{\sigma}(f,\Omega) , if \lim_{n\to\infty}\frac{1}{d_{n}}\sum_{i=1}^{d_{n}}F(\sigma_{i}(A_{n}))=\frac{1}{\mathcal{L}(\Omega)}\int_{\Omega}\frac{\sum_{j=1}^{s}F(\sigma_{j}(f(x)))}{s}{\rm d}x,\qquad\forall\,F\in C_{c}(\mathbb{R}). (4) • We say that \{A_{n}\}_{n} has a (asymptotic) spectral (or eigenvalue) distribution described by f on \Omega , and we write \{A_{n}\}_{n}\sim_{\lambda}(f,\Omega) , if \lim_{n\to\infty}\frac{1}{d_{n}}\sum_{i=1}^{d_{n}}F(\lambda_{i}(A_{n}))=\frac{1}{\mathcal{L}(\Omega)}\int_{\Omega}\frac{\sum_{j=1}^{s}F(\lambda_{j}(f(x)))}{s}{\rm d}x,\qquad\forall\,F\in C_{c}(\mathbb{C}), (5) where C_{c}(\mathbb{R}) and C_{c}(\mathbb{C}) represent the sets of continuous and compactly supported functions on \mathbb{R} and \mathbb{C} , respectively. If \{A_{n}\}_{n} has both a singular value and an eigenvalue distribution described by f , we write \{A_{n}\}_{n}\sim_{\sigma,\lambda}\left(f,\Omega\right) . Definition 2.2 . Let \{A_{n}\}_{n} be a square matrix-sequence of size d_{n} , such that d_{n}\nearrow\infty , and let \left\{\{B_{n,t}\}_{n}\right\}_{t} be a sequence of matrix-sequences of the same size d_{n} . We say that \left\{\{B_{n,t}\}_{n}\right\}_{t} is an approximating class of sequences (a.c.s.) for \{A_{n}\}_{n} if the following condition is met: for every t there exists n_{t} such that, for n>n_{t} , A_{n}=B_{n,t}+R_{n,t}+N_{n,t}, and \textnormal{rank}\left(R_{n,t}\right)\leq c(t)d_{n}, \left\|N_{n,t}\right\|\leq\omega(t), where \|\cdot\| denotes the spectral norm of a matrix, n_{t},c(t),\omega(t) depend only on t , and \lim_{t\to\infty}c(t)=\lim_{t\to\infty}\omega(t)=0. The following theorem constitutes a link between the notions of a.c.s. and of spectral distribution. Theorem 2.3