A note on the quasi-local algebra of expander graphs
Bruno M. Braga, Ján Špakula, Alessandro Vignati · 2025
Read A note on the quasi-local algebra of expander graphs by Bruno M. Braga, Ján Špakula, Alessandro Vignati (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-2508-07327.txt.
Introduction
Bruno Braga, Jan Spakula, and Alessandro Vignati prove a point about the math of expander graphs, in work from twenty twenty-five. Expander graphs are networks that stay well linked even though each point has few links. The authors study an algebra built from such graphs, called the quasi-local algebra.
They show that the algebra holds no special sub part of a named simple form. A recent result by Ozawa had already shown these algebras differ from a close cousin built from the same graphs. The new proof adds one more clear gap between the two. The note is short and aimed at readers who work on graph algebras and large scale shape.
Catalog record
| CREATOR | Bruno M. Braga, Ján Špakula, Alessandro Vignati |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 36 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-07327
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 quasi-local algebra of expander graphs Bruno M. Braga, Ján Špakula, Alessandro Vignati Abstract . We show that the quasi-local algebra of a coarse disjoint union of expander graphs does not contain a Cartan subalgebra isomorphic to ℓ ∞ \ell_{\infty} . N. Ozawa has recently shown that these algebras are distinct from the uniform Roe algebras of expander graphs, and our result describes a further difference. 1. Introduction This paper focuses on the study of Roe-like algebras of metric spaces arising from expander graphs and their properties. We will in particular focus on quasi-local algebras associated to such spaces and on their Cartan algebras. In a recent breakthrough, N. Ozawa showed that the quasi-local algebra of a coarse disjoint union of expander graphs is strictly larger than its uniform Roe algebra ( [ Oza25 , Corollary C] ). Up until this result, the question of whether these algebras coincided was one of the important open problems in the field (see [ ŠT19 , ŠZ20 , LNvZ21 , BBF+24b ] ) which went back to J. Roe (see [ Roe96 , Page 20] ). In his proof, Ozawa showed that uniform Roe algebras can never contain the product of matrix algebras \prod_{n\in\mathbb{N}}\mathrm{M}_{n}(\mathbb{C}) while the quasi-local algebra of expander graphs always contains this product. This result suggests a new question: how distinct can the quasi-local and the uniform Roe algebra be? The goal of the present note is to show that the quasi-local algebra of a coarse disjoint union of expander graphs does not have a Cartan subalgebra isomorphic to \ell_{\infty} (see Theorem 1.1 ), a property shared by all uniform Roe algebras. The proof of this result relies both on earlier work on Cartan subalgebras of Willett and White ( [ WW20 ] ) as well as on Ozawa’s ( [ Oza25 ] ); as Ozawa, we use the concentration of measure phenomenon in order to rule out the existence of such subalgebras. Let us carefully introduce our objects. In coarse geometry, there are several algebras of bounded operators which serve as models to code in \mathrm{C}^{*} -algebraic terms the large scale geometry of metric spaces; these are known as Roe-like algebras. Important for us are two of these, the uniform Roe and the quasi-local algebra , which we shall now define. Firstly, as coarse geometry is the study of metric spaces from afar, local properties of metric spaces are irrelevant in this context and one can restrict themselves to discrete spaces. In fact, throughout these notes, we will only deal with uniformly locally finite (abbreviated as u.l.f. ) metric spaces, i.e., spaces where balls of a given radius are finite and uniformly bounded in cardinality. Given a u.l.f. metric space (X,d) , \ell_{2}(X) denotes the Hilbert space of square-summable functions X\to\mathbb{C} and \mathcal{B}(\ell_{2}(X)) the algebra of bounded linear operators on \ell_{2}(X) . Given an operator a\in\mathcal{B}(\ell_{2}(X)) , its propagation is given by \propg(a)=\sup\{d(x,y)\mid\langle a\delta_{x},\delta_{y}\rangle\neq 0\}, where (\delta_{x})_{x\in X} denotes the canonical orthonormal basis of \ell_{2}(X) ; so, \propg(a) is in [0,\infty] . The uniform Roe algebra of X , denoted by \mathrm{C}^{*}_{u}(X) , is the norm closure of all operators in \mathcal{B}(\ell_{2}(X)) whose propagation is finite. Identifying \ell_{\infty}(X) , the \mathrm{C}^{*} -algebra of bounded functions X\to\mathbb{C} , with the (multiplication) operators on \ell_{2}(X) with propagation zero, i.e. the operators which are diagonal with respect to the basis (\delta_{x})_{x\in X} , it is clear that \ell_{\infty}(X)\subseteq\mathrm{C}^{*}_{u}(X) . In fact, \ell_{\infty}(X) is a Cartan subalgebra of \mathrm{C}^{*}_{u}(X) , in the sense that \ell_{\infty}(X) is a maximal abelian subalgebra (abbreviated as masa ), the normalizer of \ell_{\infty}(X) in \mathrm{C}^{*}_{u}(X) generates the whole \mathrm{C}^{*}_{u}(X) as a \mathrm{C}^{*} -algebra, and there is a faithful conditional expectation \mathrm{C}^{*}_{u}(X)\to\ell_{\infty}(X) (see [ WW20 , Proposition 4.1] ). It is straightforward that, for any operator in \mathrm{C}^{*}_{u}(X) , its cutoffs by projections in \ell_{\infty}(X) given by subsets of X which are far enough apart have small norm. More precisely, given A\subseteq X , let \chi_{A} denote the orthogonal projection of \ell_{2}(X) onto \ell_{2}(A) . Then, for any a\in\mathrm{C}^{*}_{u}(X) , the following holds: (1.1) \forall\varepsilon>0\ \exists r>0\ \forall A,B\subseteq X\ \text{if }d(A,B)>r\text{ then }\|\chi_{A}a\chi_{B}\|\leq\varepsilon. Indeed, if b is a finite propagation operator with \|a-b\|\leq\varepsilon , then r=\propg(b) has the desired property, since \|\chi_{A}b\chi_{B}\|=0 for all A,B\subseteq X with d(A,B)>r . It is readily seen that the subset of all operators in \mathcal{B}(\ell_{2}(X)) satisfying ( 1.1 ) forms a \mathrm{C}^{*} -subalgebra of \mathcal{B}(\ell_{2}(X)) . We call this \mathrm{C}^{*} -algebra the quasi-local algebra of X and denote it by \mathrm{C}^{*}_{\!\mathit{ql}}(X) . By the discussion above, we have that \mathrm{C}^{*}_{u}(X)\subseteq\mathrm{C}^{*}_{\!\mathit{ql}}(X) and, as mentioned in the first paragraph of this introduction, it was a long standing open problem whether these algebras were actually the same, regardless of X . This is an important problem: in fact, in case quasi-local operators correspond to the operators that can be approximated by finite propagation operators, one has a method for checking whether an operator belongs to a uniform Roe algebra merely by estimating norms of off-diagonal block restrictions of the operator, without having to explicitly produce finite propagation approximants. This is crucial, for instance, in the work of Engel (see [ Eng14 , Eng18 ] ) on the index theory of pseudo-differential operators. In general, checking whether an operator is quasi-local is much easier than checking whether it belongs to \mathrm{C}^{*}_{u}(X) . As of now, the largest class of u.l.f. metric spaces for which it is known that \mathrm{C}^{*}_{u}(X)=\mathrm{C}^{*}_{\!\mathit{ql}}(X) is the class of spaces satisfying Yu’s property A (see [ ŠZ20 , Theorem 3.3] , and [ Oza25 , Theorem 5] for a much shorter proof). On the other hand, as proved by Ozawa, these algebras are never the same if X is the coarse disjoint union of expander graphs ( [ Oza25 , Corollary C] ). Recall that for k\in\mathbb{N} and h>0 , a finite (undirected) graph G=(V,E) is a (k,h) -expander if any vertex is incident to at most k edges, and for all A\subseteq V we have that A\leq|V|/2\text{ implies }|\partial A|\geq h|A|, where \partial A=\{v\in V\setminus A\mid\exists u\in A,\ (v,u)\in E\} . This condition readily implies that G is connected, and thus we consider V as a metric space endowed with the shortest path distance. Let now (X_{n},d_{n})_{n\in\mathbb{N}} be a sequence of finite metric spaces, each being a vertex set of some (k,h) -expander graph, for k\in\mathbb{N} and h>0 fixed and independent of n\in\mathbb{N} , and such that |X_{n}|\to\infty as n\to\infty . We endow X=\bigsqcup_{n\in\mathbb{N}}X_{n} with a metric d such that 1. d\restriction X_{n}\times X_{n}=d_{n} for all n\in\mathbb{N} , and 2. d(X_{n},X_{m})\to\infty as n,m\to\infty with n\neq m . We refer to such a disjoint union X as a coarse disjoint union of expander graphs . Note that while the metric d above is by no means unique, properties ( 1 ) and ( 2 ) above characterize d coarsely. For details on expander graphs, we refer to [ Lub10 ] . We want to understand the possible differences between uniform Roe and quasi-local algebras in the setting of coarse disjoint union of expander graphs, with a specific focus on Cartan C*-subalgebras. While \ell_{\infty}(X) is always a Cartan subalgebra of \mathrm{C}^{*}_{u}(X) , in the case of expander graphs, not only \ell_{\infty}(X)