Anderson localization in high-contrast media with random spherical inclusions

Matteo Capoferri, Matthias Täufer · 2025

Read Anderson localization in high-contrast media with random spherical inclusions by Matteo Capoferri, Matthias Täufer (2025) 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-2512-02568.txt.

Introduction

Matteo Capoferri and Matthias Täufer study waves in materials made of stiff matter dotted with soft balls. The soft spherical inclusions have random sizes, scattered through the stiff matrix. Operators of this kind model composite materials with strong contrast between their parts.

The pair examines the spectrum of these operators, the set of allowed vibration patterns and energies. Random high contrast media also arise in stochastic homogenization, which replaces fine detail with average behavior. Understanding when waves stay trapped, called Anderson localization, helps predict how real composites carry sound, heat, or light. The same math can guide the design of foams and gels built to block sound or shake in a planned way.

Catalog record

CREATORMatteo Capoferri, Matthias Täufer
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE71 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-02568

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Anderson localization in high-contrast media with random spherical inclusions Matteo Capoferri, Matthias Täufer Abstract We study spectral properties of partial differential operators modelling composite materials with highly contrasting constituents, comprised of soft spherical inclusions with random radii dispersed in a stiff matrix. Such operators have recently attracted significant interest from the research community, including in the context of stochastic homogenization. In particular, it has been proved that the spectrum of these operators may feature a band-gap structure in the regime where heterogeneities take place on a sufficiently small scale. However, the nature of the limiting (as the small scale tends to zero) spectrum in the above setting is non-classical and not completely understood. In this paper we prove for the first time that Anderson localization occurs near band edges, thus shedding light on the limiting spectral behaviour. Our results rely on recent nontrivial advancements in quantitative unique continuation for PDEs, in combination with assumptions on the model that are standard in the Anderson localization literature, and which we plan to relax in future works. Keywords: Anderson localization, high-contrast composites, random media, stochastic homogenisation, unique continuation. 2020 MSC classes: Primary 47B80, 74A40. Secondary 35B27, 35R60, 74S25, 82B44. 1 Introduction The goal of this paper is to make a first step towards a rigorous mathematical understanding of Anderson localization in the context of high-contrast random media. Over the years, high-contrast media, that is, composite materials whose constituents possess very different (or “highly contrasting”) physical properties, have attracted the attention of both mathematicians and applied scientists, because, unlike more traditional composites, they possess special features allowing for effective manipulation of propagating waves. Indeed, in this class of materials, often modelled as a collection of “soft” inclusions dispersed in a “stiff” matrix, the interplay between the characteristic size \varepsilon of heterogeneities and the ratio \varepsilon^{2} between the diffusion coefficients of inclusions and matrix brings about resonances at the microscopic scale which yield a spectrum with band-gap structure. This particular relative scaling is often referred to as double-porosity resonant regime. From a mathematical perspective, the traditional approach to study composite materials with small heterogeneities is via homogenisation theory. An immediate mathematical challange one encounters when trying to do so for high-contrast media is that the family of operators \mathcal{A}^{\varepsilon}:=-\operatorname{div}\,a^{\varepsilon}\,\nabla ( a^{\varepsilon} being the diffusion coefficients) describing them is not uniformly elliptic as \varepsilon\to 0 , ultimately leading to a loss of compactness in the limit. Hence, classical homogenisation does not work. In the high-contrast periodic setting — that is, when inclusions are arranged periodically in space — pioneering works by Allaire first [ All92 ] and by Zhikov shortly thereafter [ Zhi00 , Zhi04 ] demonstrated that one can still define a limiting “homogenised” operator in the sense of the so-called two-scale convergence . This limiting operator possesses a two-scale nature, in that it accounts for both macroscopic and microscopic properties of the material, coupled in a particular way. Remarkably, Zhikov showed that the two-scale limit yields convergence of spectra in the sense of Hausdorff, thus making this technique well-suited for the spectral analysis of high-contrast media. From the point of view of applications, requiring perfect periodicity is often too strong a constraint, and one would like to allow for some randomness in the coefficients, the geometry of the inclusions, or both. Whilst there exists extensive literature on high-contrast periodic media, see, e.g., [ CSZ06 , CC15 , CC18 , CKVv25 , CKS23 , CK+25 ] , a systematic study of the random setting was initiated only recently, so that the literature is limited and much less is known. It should be mentioned that this is not accidental: moving from periodic to random operators involves a significant step change in both conceptual and technical challenges. Building upon earlier work by Zhikov and Piatnitski [ ZP06 ] , Cherdantsev, Cherednichenko and Velčić in [ CCV18 ] proved that on bounded domains one has stochastic two-scale resolvent convergence as well as Hausdorff spectral convergence for high-contrast operators with random coefficients. In the whole space \mathbb{R}^{d} the picture is rather different: indeed, a few years later the same set of authors showed in [ CCV21 ] that, although one still has two-scale convergence, in general the spectrum of the limiting operator is a proper subset of the limiting spectrum \lim_{\varepsilon\to 0}\sigma(\mathcal{A}^{\varepsilon}) . That is, the operators \mathcal{A}^{\varepsilon} shortly before the limit possess spectrum that ‘is not seen’ by the limiting operator. Nevertheless, by examining a range of physically meaningful examples, they showed that the limiting spectrum (and, thus, the spectrum of the homogenised operator) can exhibit infinitely many gaps. So as not to overload the paper and disperse the focus, we refrain here from discussing subsequent literature on high-contrast stochastic homogenisation (mostly qualitative so far, with a couple of very recent quantitative results), limiting ourselves to listing those references closest to the setting of our paper, for the benefit of the reader: [ CCV23 , CCV25 , BC+25 , BDG25 , APZ23 ] . Informal, non-rigorous examination of specific examples where Hausdorff convergence of spectra fails in the stochastic setting seems to suggest that this type of materials may exhibit dense pure point spectrum with well localized eigenfunctions near the edges of the spectral bands — a phenomenon known under the name of Anderson localization . The theory of Anderson localization goes back to Anderson’s pioneering work [ And58 ] who identified it as absence of diffusion in certain random lattice operators. Since then, the mathematical treatment of Anderson localization has had a long history with numerous contributions. We highlight [ GMP77 ] as the first mathematically rigorous work proving localization in continuum models in one dimension, and [ HM84 ] for the first proof of Anderson localization in continuum models in \mathbb{R}^{d} at the bottom of the spectrum. Furthermore, we should like to mention [ FS84 ] where the most common technique to prove localization, that is, multiscale analysis , has been developed (albeit for the special case of discrete Laplacians), and refer the reader to the monographs [ CL90 , Sto01 , Ves08 , AW15 ] for more details and references. The bulk of works on Anderson localization for operators in L^{2}(\mathbb{R}^{d}) treat the alloy-type or continuum Anderson model H_{\omega}=-\Delta+V_{\omega}\quad\text{where}\quad V_{\omega}(x)=\sum_{j\in\mathbb{Z}^{d}}\omega_{j}\,,u(x-j) (1.1) where \omega=(\omega_{j})_{j\in\mathbb{Z}^{d}} is a family of (usually bounded) independent and identically distributed random variables and u is a (usually bounded, nonnegative, and compactly supported) profile function, modeling the influence of a single atom. The advantage of this model is that it gives rise to linear perturbation theory: changes to the elementary random variables correspond to linear perturbations of the operator H_{\omega} . There are some exceptions to this in which localization has been proved, the most noteworthy being the Poisson model [ GHK07 ] V_{\omega}(x)=\sum_{y\in\mathcal{P}(\omega)}u(x-y)\,, where \mathcal{P}\subset\mathbb{R}^{d} is a Poisson point process, the random displacement model [ KLNS12 ] V_{\omega}(x)=\sum_{j\in\mathbb{Z}^{d}}u(x-j-\eta_{j})\,, where the family (\eta_{j})_{j\in\mathbb{R}^{d}} is a family of independent and identically distributed, \mathbb{R}^{d} -valued random displacements , the random breather model [ TV15 , NTTV18 ] V_{\omega}(x)=\sum_{j\in\mathbb{Z}^{d}}\mathbf{1}_{B_{\omega_{j}}(j)}\,, where B_{\omega_{j}}(j) are balls, centered at j with independent and identically distributed random radii (\omega_{j})_{j\in\mathbb{Z}^{d}} — amongst others, including Gaussian random potentials [ FLM00 ] . A reasonably systematic understanding of random operators where the randomness does

Read the complete text