Bounded geometry version of property A
V. Manuilov · 2025
Read Bounded geometry version of property A by V. Manuilov (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-2506-00890.txt.
Introduction
Manuilov proposes, in a 2025 math paper, a new version of property A for spaces that lack bounded geometry. Property A is a large-scale trait that helps prove a space can be embedded, in a coarse sense, into a Hilbert space. His fix builds the trait from smoother metrics laid over the given one. The new version still forces the embedding result, and some spaces that fail the classic trait pass it.
Catalog record
| CREATOR | V. Manuilov |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 33 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-00890
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Bounded geometry version of property A V. Manuilov Abstract . For uniformly dicrete metric spaces without bounded geometry we suggest a modified version of property A based on metrics of bounded geometry greater than the given metric. We show that this version still implies coarse embeddability in Hilbert spaces, and that some examples of non-property A spaces of unbounded geometry satisfy this version. We also relate this version of property A to our version of uniform Roe algebras for spaces without bounded geometry and introduce an appropriate equivalence relation. Introduction Bounded geometry for discrete metric spaces is a useful property in coarse geometry. For Rips complexes of metric spaces it means their finite dimension. But many natural examples of metric spaces do not have bounded geometry, e.g. Cayley graphs of infinitely generated groups or some box spaces. Making in such spaces the distances between points much greater, we may obtain metrics of bounded geometry, and we suggest to use these metrics of bounded geometry to get information about the original metric. For Rips complexes this approach reduces to study of finitedimensional approximations instead of the complex itself. In [ 3 ] we have used this approach to define a bounded geometry version for the uniform Roe algebras. Here we use it to expose what we think should be a more suitable version of property A of Guoliang Yu for spaces without bounded geometry, which we denote by \mathcal{B}\mathcal{G} -A and call the bounded geometry version of property A. P. Nowak in [ 4 ] has shown that the unbounded geometry space \sqcup_{i\in\mathbb{N}}(\mathbb{Z}/2\mathbb{Z})^{i} does not have property A, although should be viewed as ‘amenable’ in any reasonable sense. In particular, it admits a coarse embedding in a Hilbert space. Property \mathcal{B}\mathcal{G} -A coincides with property A on bounded geometry spaces and includes more spaces without bounded geometry, in particular, the Nowak’s example. Also property \mathcal{B}\mathcal{G} -A implies coarse embeddability in a Hilbert space by the standard argument with negative type kernels, thus justifying the approach based on bounded geometry metrics. Unlike spaces with bounded geometry, general metric spaces may be coarsely equivalent, but still very different. For example, an infinite space of finite diameter is coarsely equivalent to a single point, but the uniform Roe algebra for such space is even not an exact C^{*} -algebra. Considering on a metric space (X,d) all metrics of bounded geometry greater than d gives a more subtle structure on (X,d) and allows to distinguish such spaces using \mathcal{B}\mathcal{G} -coarse equivalence relation. In particular, we show that if (X,d_{X}) and (Y,d_{Y}) are \mathcal{B}\mathcal{G} -coarsely equivalent and one of them has property \mathcal{B}\mathcal{G} -A then the other one has this property too. We also establish relations between this structure and our version of the uniform Roe algebras for uniformly discrete metric spaces. 1. Notation and property A Let X=(X,d) be a discrete metric space. We write B^{d}_{S}(x)=\{y\in X:d(x,y)\leq S\} for the ball of radius R with respect to the metric d centered at x (as we will consider several metrics on X , we specify the metric as a superscript), and, for A\subset X , we denote by |A| the number of points in A . Recall that (X,d) has bounded geometry if for any R>0 there exists N\in\mathbb{N} such that |B^{d}_{R}(x)|\leq N for any x\in X . A metric space is called uniformly discrete if there is \alpha>0 such that d(x,y)\geq\alpha for any x\neq y\in X . Let l_{2}(X) denote the Hilbert space of square-summable complex-valued functions on X . A map \xi:X\to l_{2}(X) , x\mapsto\xi_{x} , is said to have (R,\varepsilon) variation if d(x,y)\leq R implies \|\xi_{x}-\xi_{y}\|<\varepsilon . It is known that for metric spaces of bounded geometry the following conditions are equivalent (cf. Theorem 1.2.4 of [ 8 ] ): (i) (X,d) has property A; (ii) for all R>0 , \varepsilon>0 , there exists S=S(d,R,\varepsilon)>0 and a map \xi:X\to l_{2}(X) such that (1) \xi has (R,\varepsilon) variation, and \|\xi_{x}\|=1 for any x\in X ; (2) \operatorname{supp}\xi_{x}\subset B_{S}(x) for any x\in X . For general discrete metric spaces, not necessarily of bounded geometry, property A is defined as follows: (X,d) has property A if for any R>0 and any \varepsilon>0 there exists S>0 and a family of finite subsets \{A_{x}\}_{x\in X} in X\times\mathbb{N} such that (1) d(x,y)<R implies \frac{|A_{x}\Delta A_{y}|}{|A_{x}\cap A_{y}|}<\varepsilon ; (2) A_{x}\subset B_{S}(x)\times\mathbb{N} for any x\in X . Note that it was shown in [ 9 ] that, for bounded geometry spaces, one can get rid of the seemingly redundant factor \mathbb{N} and to use subsets A_{x}\subset X , x\in X , in the above definition. 2. Metrics of bounded geometry Given two metrics, d_{1} and d_{2} , on a set X , we say that d_{2} dominates d_{1} and write d_{1}\preceq d_{2} if d_{1}(x,y)\geq d_{2}(x,y) for any x,y\in X . We say that d_{2} coarsely dominates d_{1} and write d_{1}\preceq_{c}d_{2} if there exists a homeomorphism \psi of (0,\infty] such that \psi(d_{1}(x,y))\geq d_{2}(x,y) for any x,y\in X . Both \preceq and \preceq_{c} give a partial order on the set of all metrics on X . Given a metric space X=(X,d) with a uniformly discrete metric d , consider the set \mathcal{B}\mathcal{G}(X,d) of all metrics \rho on X such that (m1) \rho\preceq d ; (m2) \rho is of bounded geometry. The following result was shown in [ 3 ] . As the set \mathcal{B}\mathcal{G}(X,d) was defined in [ 3 ] a bit differently, we provide a proof here. Lemma 2.1 . The set \mathcal{B}\mathcal{G}(X,d) is directed, i.e. for any \rho_{1},\rho_{2}\in\mathcal{B}\mathcal{G}(X,d) there exists \rho\in\mathcal{B}\mathcal{G}(X,d) such that \rho_{i}\preceq\rho , i=1,2 . Proof. Let \rho_{1},\rho_{2}\in\mathcal{B}\mathcal{G}(X,d) . As (X,d) is uniformly discrete, we may assume without loss of generality that d(x,y)\geq 1 for any x\neq y\in X . Turn X into a complete graph G by connecting each two points, x,y\in X , by an edge, and assign length l(x,y)=\min(\rho_{1}(x,y),\rho_{2}(x,y)) to the edge that connects x with y . As \rho_{1},\rho_{2}\preceq d , l(x,y)\geq 1 for any x\neq y\in X . If \gamma is a path in G that connects two vertices, define its length l(\gamma) as the sum of lengths of the edges contained in \gamma . Set \rho(x,y)=\inf_{\gamma}l(\gamma) , where the infimum is taken over all paths that connect x with y . Clearly, \rho is a metric. Given x,y\in X and \varepsilon>0 , there is a path \gamma connecting x with y such that \rho(x,y)>l(\gamma)-\varepsilon . Let x_{0}=x,x_{1},x_{2},\ldots,x_{n}=y be the vertices on \gamma , i.e. \gamma=(x=x_{0},x_{1},\ldots,x_{n}=y) . Then l(\gamma)=\sum_{i=1}^{n}l(x_{i-1},x_{i}) . Therefore, d(x,y)\leq\sum_{i=1}^{n}d(x_{i-1},x_{i})\leq\sum_{i=1}^{n}l(x_{i-1},x_{i})<\rho(x,y)-\varepsilon. As \varepsilon is arbitrary, we conclude that d(x,y)\leq\rho(x,y) for any x,y . Note that if we assign length l_{1}(x,y)=\rho_{1}(x,y) to each edge of G and define a metric in the same way as the infimum of lengths of paths connecting two vertices then this metric would coincide with \rho_{1} . Therefore, \rho(x,y)\leq\rho_{1}(x,y) for any x,y\in X . Similarly, \rho(x,y)\leq\rho_{2}(x,y) for any x,y\in X . It remains to show that \rho has bounded geometry. Fix x\in X , R\in\mathbb{N} , and let \rho(x,y)<R . Let \gamma be a path connecting x with y with length l(\gamma)<R . As \rho(u,v)\geq 1 for any u,v\in X , the path \gamma contains not more than R edges, and the length of each edge does not exceed