Geometric property (T) for box spaces and sofic approximations

Basic Information

• Paper ID: 2511.16515
• Title: Geometric property (T) for box spaces and sofic approximations
• Authors: Vadim Alekseev (Technische Universität Dresden), Stefan Drigalla (Universität Leipzig)
• Classification: math.GR (Group Theory), math.MG (Metric Geometry), math.OA (Operator Algebras)
• Publication Date: November 20, 2025 (arXiv preprint)
• Paper Link: https://arxiv.org/abs/2511.16515

Abstract

This paper proves that every sofic approximation of a property (T) group is approximately isomorphic to an approximation with geometric property (T). More generally, graph box spaces with boundary geometric property (T) are approximately isomorphic to box spaces with geometric property (T). The article further proves that sequences of bounded-degree graphs are approximately isomorphic to disjoint unions of expander graphs if and only if the Laplace operator has a spectral gap in the ultraproduct. Finally, in the spirit of Żuk's criterion for property (T) of groups, the paper establishes a local geometric criterion for geometric property (T).

Research Background and Motivation

Problem Background

1. Core Problem: Study the relationship between coarse geometric properties of sofic group approximations and analytic properties of groups. Specifically, for groups with Kazhdan property (T), must their sofic approximations necessarily exhibit corresponding geometric property (T)?
2. Significance:
   • Kazhdan property (T) is a fundamental rigidity property in group theory with wide applications in representation theory, ergodic theory, and geometric group theory
   • Willett and Yu proved in WY14 that geometric property (T) is the correct correspondence of Kazhdan property (T) at the box space level for residually finite groups
   • Understanding geometric properties of sofic approximations is crucial for studying non-residually finite groups
3. Limitations of Existing Methods:
   • Kun Kun19 proved that every sofic approximation of a property (T) group is approximately isomorphic to a sequence of expander graphs, but expansion is strictly weaker than geometric property (T)
   • Earlier work AF19, AB21 resolved the cases of amenable groups and a-T-menable groups, but the property (T) case remained open
   • Existing results primarily address "almost everywhere" properties rather than global geometric properties
4. Research Motivation:
   • Complete the sofic approximation theory framework by incorporating property (T)
   • Establish a bridge between measure property (T) of coarse boundary groupoids and geometric property (T) of box spaces
   • Provide criteria for constructing graph sequences with geometric property (T)

Core Contributions

The main contributions of this paper include:

1. Main Theorem (Theorem A/5.5): Proves that graph box spaces with almost boundary geometric property (T) are approximately isomorphic to box spaces with geometric property (T)
2. Complete Characterization of Sofic Approximations (Theorem B/6.3): For finitely generated sofic groups Γ, the following are equivalent:
   • Γ has property (T)
   • Every sofic approximation of Γ has almost boundary geometric property (T)
   • Γ admits a sofic approximation with geometric property (T)
3. Spectral Gap Characterization (Theorem C/5.4): A box space has a spectral gap in the ultraproduct if and only if it is approximately isomorphic to expander graphs
4. Local Criterion (Theorem D/7.4): Provides a local geometric criterion similar to Żuk's criterion: if the first positive eigenvalue λ₁(Lₓ) ≥ λ > 1/2 of the link graph holds on a subset of density approaching 1, then the box space has almost boundary property (T)
5. Measure Groupoid Property (T) Theory: Introduces property (T) for discrete p.m.p. measure groupoids and proves its equivalence with the classical definition for standard Borel groupoids

Methodology Details

Task Definition

Input:

• Graph box space X = ⊔ᵢXᵢ, where (Xᵢ) is a sequence of finite graphs with uniformly bounded degree, |Xᵢ| → ∞
• For sofic approximations: finitely generated group Γ = ⟨S⟩ and its sofic approximation sequence (Xᵢ)

Output:

• Determine whether X has geometric property (T)
• Construct an approximately isomorphic box space X' with geometric property (T)

Core Concepts:

• Geometric property (T): The Laplace operator Δ has a spectral gap in the maximal uniform Roe algebra C*ᵤ,ₘₐₓ(X)
• Boundary geometric property (T): Δ has a spectral gap in the boundary groupoid C*ₘₐₓ(∂G(X))
• Approximate isomorphism: There exist subgraphs Yᵢ ⊆ Xᵢ and Y'ᵢ ⊆ X'ᵢ that are isomorphic, with |Yᵢ|/|Xᵢ| → 1, |E(Yᵢ)|/|E(Xᵢ)| → 1

Core Architecture

The paper's proof strategy employs a hierarchical structure:

Sofic approximations of property (T) groups
         ↓
Almost boundary geometric property (T) (measure-theoretic property)
         ↓
Spectral gap in ultraproduct
         ↓
Approximately isomorphic to expanders (Theorem 5.3)
         ↓
Geometric property (T) (Proposition 3.4)

Key Technical Components

1. Coarse Boundary Groupoid Framework

Definition: For a box space X, the coarse groupoid is defined as $G(X) = \bigcup_{R>0} \overline{E_R} \subseteq \beta X \times \beta X$ where $E_R = \{(x,y) \in X \times X: d(x,y) \leq R\}$, with closure taken in the Stone-Čech compactification βX.

Core Construction: For Yᵢ ⊆ Xᵢ satisfying |Yᵢ|/|Xᵢ| → 1, define $Z = \text{core}(Y) = (\text{sat}(\partial\beta Y^c))^c$ where sat(A) is the saturation of A. The core Z is a closed, invariant, full-measure subset.

Key Property: $Z = \bigcap_{R>0} \partial\beta(B_R(Y^c))^c$

2. Measure Groupoid Property (T)

Definition 4.2: A discrete p.m.p. measure groupoid (G,μ) has property (T) if there exist c > 0 and finitely many bisections φ₁,...,φₙ generating G such that for every normal representation π:CG → B(H) and ξ ∈ H^π_⊥ (orthogonal to invariant vectors), at least one i satisfies $\|(φ_i - Ψ(φ_i))ξ\| \geq c\|ξ\|$

Key Result (Proposition 4.5): Measure property (T) is equivalent to the Laplace operator having a spectral gap in every normal representation

Connection Theorem (Proposition 4.6): If a compactly generated étale topological groupoid G has topological property (T) and μ is an invariant measure on G⁽⁰⁾, then (G,μ) has measure property (T)

3. Spectral Gap Characterization in Ultraproducts

Markov Operator: Define $M = 1 - \frac{1}{2d}\Delta \in C[X]$

Key Lemma (Lemma 5.8): If X has spectral gap c in the ultraproduct, then for any ε > 0, δ > 0, k ∈ ℕ, there exists I ∈ u such that for all i ∈ I and μᵢ(T) ≥ δ for T ⊆ Xᵢ: $\|M^{k+1}χ_T - M^k χ_T\| \leq ((1-c_M)^k + ε)\|Mχ_T - χ_T\|$ where $c_M = c/2d$.

Decomposition Theorem (Proposition 5.11): There exists a constant C > 0 such that for any α > 0, there exists I ∈ u where each Xᵢ (i ∈ I) can be decomposed as $X_i = \bigsqcup_{\ell=0}^{L_i} P^\ell_i$ satisfying:

• |P⁰ᵢ| < α|Xᵢ| ("bad" part)
• For ℓ ≥ 1: |∂P^ℓᵢ| < α|P^ℓᵢ| (small boundary)
• For ℓ ≥ 1 and T ⊆ P^ℓᵢ with |T| ≤ |P^ℓᵢ|/2: |∂T| ≥ C|T| (interior expansion)

4. Rewiring Technique

Lemma 5.12: If P satisfies interior expansion and |∂P| is sufficiently small, then by adding/deleting at most α|P| edges and removing at most (α/C)|P| vertices, P can be made into a connected graph with Cheeger constant at least C/6.

Core Idea:

1. Choose a sparse point set F ⊆ E(P,P) at distance at least 2r = 8/C
2. Rewire edges in ∂P to endpoints of F
3. Remove newly created small connected components

Technical Innovations

1. Lifting from Measure to Geometric Property:
   • The key innovation is handling expansion of "small measure" subsets
   • Kun's method relied on applying spectral gap to arbitrary small subsets, while this paper can only apply it to "large measure" subsets
   • This difficulty is resolved through a dichotomy of "good"/"bad" subsets and refined measure estimates
2. Precise Control of Core Construction:
   • The core Z = core(Y) construction ensures compatibility of geometric and measure properties
   • Lemma 2.24 proves that approximate isomorphism preserves core structure: ∂G(X)|W ≅ ∂G(X')|{W'}
3. Decomposition-Reconstruction Paradigm:
   • First step: decompose graphs into almost-expander parts via Proposition 5.11
   • Second step: correct to true expanders via Lemma 5.12's rewiring technique
   • Key: prove that reconstructed space maintains approximate isomorphism
4. Local-to-Global Lifting Mechanism:
   • Theorem 7.4 derives global geometric property (T) from local spectral information of link graphs Lₓ
   • Uses sum-of-squares representation: proves Δ²_τ - cΔ_τ is a sum of squares in CX
   • Introduction of weighted Laplace operator Δ_τ is a key technical innovation

Experimental Setup

Note: This is a pure theoretical mathematics paper with no experimental component. All results are rigorous mathematical theorems and their proofs.

Example Constructions

Example 2.18:

• Take box space X = ⊔ᵢXᵢ of a property (T) group Γ (such as SL(n,ℤ), n ≥ 3)
• Construct X'ᵢ by adding an edge between two copies of Xᵢ
• X' is not an expander graph (since connected) but has almost boundary property (T)
• X' is approximately isomorphic to a (non-connected) expander graph with geometric property (T)

Example 6.4: Constructs a complex sofic approximation example

• Expander sofic approximation X'ᵢ of Γ = SL(n,ℤ)
• Expander box space (Yᵢ) of Λ = free group with |Yᵢ|/|X'ᵢ| → 0
• Define Xᵢ = X'ᵢ ⊔ Yᵢ with appropriate connections
• Prove Xᵢ is an expander sequence with geometric property (T)
• But there exist invariant subsets "approximating from Λ", demonstrating structural complexity

Experimental Results

Main Theoretical Results

Theorem 5.5 (Main Result): A box space with almost boundary property (T) is approximately isomorphic to a box space with geometric property (T).

Proof Path:

1. By Corollary 5.4: X is approximately isomorphic to expanders X''
2. By Proposition 2.25: X'' has almost boundary property (T)
3. By Theorem 3.8: X'' is approximately isomorphic to X' with geometric property (T)

Theorem 6.3 (Characterization of Sofic Groups): For finitely generated sofic group Γ, the following are equivalent:

• (i) Γ has property (T)
• (ii) Every sofic approximation of Γ has almost boundary geometric property (T)
• (iii) Γ admits a sofic approximation with geometric property (T)

Quantitative Spectral Gap Estimates (Remark 3.5): If the expansion constant is C, then the spectral gap of geometric property (T) is at least $\frac{c^2}{144d^2}$

Applications of Criteria

Specific Form of Theorem 7.4: If link graphs Lₓ satisfy λ₁(Lₓ) ≥ λ > 1/2 on a subset of density → 1, then:

• Spectral gap is at least c = 2 - 1/λ
• Corresponding Kazhdan constant ε = √(2c)

Correspondence with Żuk's Criterion: For groups, λ₁ > 1/2 of link L(Γ,S) implies property (T); for box spaces, the "almost everywhere" version is needed.

Quantitative Results of Key Lemmas

Proposition 5.10:

• There exists constant C = c²_M/72
• For any α > 0, there exists K such that "bad" subset T can be replaced by "good" subset U
• |U△T| < |T|/4, |∂U| < α|U|

Lemma 5.9 (Level Set Estimate): For 0 < a < b < 1 and f ∈ ℓ²G, there exists t ∈ (a,b) such that U = {f > t} satisfies $|∂U|^2 \leq \frac{4d^2}{a^2(b-a)^2}\|Mf-f\|\|f\|^3$

Related Work

Connections between Coarse Geometry and Group Properties

1. Willett-Yu WY14:
   • First established geometric property (T) as the correspondence of Kazhdan property (T) at box space level
   • Proved that residually finite group Γ has property (T) ⟺ its box space has geometric property (T)
2. Alekseev-Finn-Sell AF19:
   • Introduced coarse boundary groupoid framework for studying sofic approximations
   • Proposed the concept of "almost boundary property"
   • Identified the problem of coarse geometric correspondences of amenability, a-T-menability, and property (T) in sofic setting
3. Alekseev-Biz AB21:
   • Resolved the a-T-menable case
   • Recognized that results naturally apply to general bounded-degree graph sequences
   • Established correspondence between measure amenability and property A

Sofic Groups and Expander Graphs

4. Kun Kun19:
   • Proved that sofic approximations of property (T) groups are approximately isomorphic to expander graph sequences
   • Introduced key technical methods for handling sofic approximations
   • This paper is an important generalization and deepening of his methods
5. Kun-Thom KT19, Kun Kun21:
   • Constructed non-amenable groups not admitting connected expander sofic approximations
   • Revealed the essential distinction between diffuse and atomic decompositions
   • Proposition 8.5 in this paper generalizes their key steps

Criteria for Property (T)

6. Żuk Ż02, Bekka-de la Harpe-Valette BdV08:
   • Established criteria for determining group property (T) through spectral information of link graphs
   • Theorem 7.4 in this paper is a generalization to the box space level
7. Ozawa Oza16:
   • Used noncommutative real algebraic geometry methods to study property (T)
   • This paper employs his sum-of-squares representation technique

Property (T) for Groupoids

8. Anantharaman-Delaroche AD05:
   • Defined property (T) for standard Borel groupoids
   • This paper generalizes to non-standard measure groupoid case
9. Dell'Aiera-Willett DW22:
   • Studied property (T) for topological groupoids
   • This paper establishes connections between topological and measure property (T)

Unique Contributions of This Paper

Compared to existing work, this paper:

• Completes the theory for property (T): Resolves the open problem posed in AF19
• Establishes complete equivalence chains: Provides correspondences at all levels from group property (T) to geometric property (T)
• Provides operational criteria: Theorem 7.4 gives practical methods for verifying geometric property (T)
• Deepens technical methods: Generalizes Kun's method to cases with only ultraproduct spectral gap

Conclusions and Discussion

Main Conclusions

1. Complete Characterization Theorem: For property (T) groups, "geometrization" of sofic approximations is possible—every sofic approximation is approximately isomorphic to an approximation with geometric property (T)
2. Necessity and Sufficiency of Spectral Gap: Spectral gap in the ultraproduct precisely characterizes the property of being approximately isomorphic to expanders
3. Local-to-Global Principle: Global geometric property (T) can be derived from spectral information of local link graphs
4. Measure-Geometry Bridge: Establishes systematic connections between measure groupoid property (T) and geometric property (T)

Limitations

1. Connectivity Assumption:
   • Proposition 3.4 requires the assumption that Xᵢ is connected
   • Example 3.7 demonstrates the necessity of this assumption
   • However, Theorem 3.8 circumvents this restriction through decomposition techniques
2. Quantitative Precision of Spectral Gap:
   • The lower bound $\frac{c^2}{144d^2}$ in Remark 3.5 may not be optimal
   • The relationship with original spectral gap c involves large constant factors
3. Sufficiency of Criteria:
   • Theorem 7.4 only provides sufficient conditions
   • Question 8.2 asks whether a characterization as necessary and sufficient conditions is possible
4. Diffuse vs. Atomic Cases:
   • The paper primarily focuses on diffuse cases, with atomic cases being relatively simple
   • Example 6.4 demonstrates the existence of complex structures

Future Directions

The paper proposes several important open problems in Section 8:

Question 8.1 (Random Construction): Does there exist a randomized construction of geometric property (T) box spaces analogous to Żuk's group construction?

Question 8.2 (Sum-of-Squares Characterization): Is geometric property (T) equivalent to Δ² - cΔ being a sum of squares in CX?

Question 8.3 (Independence of Boundary Property (T)): Does there exist a box space with boundary geometric property (T) that is not an expander graph?

• This would distinguish boundary property (T) from full geometric property (T)

Question 8.4 (Sufficiency of Measure Property (T)): If the coarse boundary groupoid has measure property (T) for every invariant measure, must it have geometric property (T)?

Question 8.6 (Diffuseness of Subgroup Restrictions): For Λ ≤ Γ, under what conditions do restrictions of sofic approximations of Γ to Λ yield diffuse sofic approximations?

• Related to Kun-Thom's non-LEF constructions

Question 8.7 (Superlinear Case): Does spectral gap of the Laplace operator in the ultraproduct imply uniform spectral gap (superlinear analogue)?

Theoretical Significance

1. Unified Framework: This paper completes the research program initiated in AF19, providing a unified framework for coarse geometric theory of sofic approximations
2. Methodological Contributions:
   • Decomposition-reconstruction paradigm
   • Delicate balance between measure theory and geometry
   • Local-to-global lifting mechanism
3. Deep Connections: Reveals profound connections between group theory, coarse geometry, operator algebras, and ergodic theory

In-Depth Evaluation

Strengths

1. Theoretical Completeness

• Strong Systematicity: Completely resolves the property (T) case posed in AF19, forming a complete system with results on amenability Kai19 and a-T-menability AB21
• Multi-level Characterization: Complete chain from group property (T) → measure property (T) → almost boundary property (T) → geometric property (T)
• Equivalence Theorem: Theorem 6.3 provides multiple equivalent conditions, enhancing theoretical robustness

2. Technical Innovation

• Method Generalization: Successfully generalizes Kun's method from "spectral gap for arbitrary small subsets" to "spectral gap only in ultraproduct", a substantial technical breakthrough
• Core Construction: Precise definition of core(Y) and its properties (Lemma 2.20, 2.24) are key tools for handling "almost everywhere" properties
• Rewiring Technique: Lemma 5.12's construction maintains expansion while controlling degree, with elegant design

3. Depth of Results

• Spectral Gap Characterization: Theorem 5.3/Corollary 5.4 establishes equivalence between ultraproduct spectral gap and approximate isomorphism to expanders, which is non-trivial
• Practical Criterion: Theorem 7.4 provides verifiable sufficient conditions with practical value
• Measure Groupoid Theory: Generalization to non-standard cases (Definition 4.2, Proposition 4.3) enriches groupoid property (T) theory

4. Writing Quality

• Clear Structure: Logical progression from preliminaries → boundary property (T) and expansion → measure property (T) → spectral gap characterization → sofic applications → criteria, with rigorous logic
• Well-Motivated: Introduction excellently explains problem background and technical challenges
• Rich Examples: Examples 2.18, 3.7, 6.4 enhance understanding, with Example 6.4 particularly showcasing theoretical subtleties

Weaknesses

1. Technical Complexity

• Lengthy Proofs: Core theorems (e.g., Proposition 5.11) involve multi-layered estimates with tedious technical details
• Implicit Constants: Many places involve "there exists constant C" without explicit expressions, affecting computability of results
• Assumption Propagation: Some results (e.g., Proposition 3.4) require connectivity assumptions; while Theorem 3.8 circumvents this, it adds theoretical complexity

2. Precision of Quantitative Estimates

• Spectral Gap Lower Bound: The $\frac{c^2}{144d^2}$ in Remark 3.5 involves significant loss; optimality is unknown
• Approximate Isomorphism Control: While existence of approximate isomorphism is proven, quantitative control of "loss" lacks refinement

3. Remaining Open Problems

• Question 8.2: Sum-of-squares characterization's necessity remains unresolved, limiting Theorem 7.4's applicability
• Question 8.3: Distinction between boundary property (T) and geometric property (T) unresolved; all current examples satisfy or fail both simultaneously
• Diffuse Theory: While Proposition 8.5 generalizes Kun-Thom's result, general theory of diffuse sofic approximations remains incomplete

4. Limited Application Scope

• Primarily for Property (T) Groups: Though framework applies to general bounded-degree graph sequences, main results and examples concentrate on property (T) groups
• Lack of Constructiveness: Theory is existential; no algorithms for explicitly constructing geometric property (T) box spaces

Impact Assessment

Contributions to the Field

1. Completes Theoretical Puzzle: Resolves key missing piece in coarse geometric theory of sofic approximations
2. Methodological Inspiration: Decomposition-reconstruction paradigm and measure-geometry lifting techniques may apply to other problems
3. Opens New Directions: Proposed open problems (especially Questions 8.6, 8.7) point toward important research directions

Practical Value

• Operational Criterion: Theorem 7.4 provides practical tools for verifying geometric property (T)
• Approximate Isomorphism Theory: Provides theoretical foundation for constructing and analyzing sofic approximations
• Groupoid Framework: Measure groupoid property (T) theory may have applications in ergodic theory

Reproducibility

• Pure Theoretical Results: All theorems have complete proofs, verifiable in principle
• Technical Details: Estimates and construction steps in proofs are clearly described
• Examples: Provided examples aid understanding of theoretical applications

Potential Impact

• Short-term: Will advance sofic group theory and coarse geometry
• Medium-term: May inspire new property (T) criteria and expander graph construction techniques
• Long-term: May influence group classification theory and rigidity theory in operator algebras

Applicable Scenarios

1. Theoretical Research:
   • Study geometric properties of sofic groups
   • Analyze coarse geometric invariants of graph sequences
   • Develop property (T) theory for groupoids
2. Concrete Applications:
   • Verify whether specific group sofic approximations have geometric property (T)
   • Construct graph sequences with desired properties
   • Study relationships between residually finite and sofic groups
3. Technical Methods:
   • Applications of ultraproduct techniques in coarse geometry
   • Combination of measure theory and geometric properties
   • Local-to-global lifting principles

Overall Evaluation

This is a high-quality theoretical mathematics paper achieving important progress in coarse geometric theory of sofic approximations. Main strengths are:

1. Resolves important open problem in the field (property (T) case)
2. Significant technical innovations (handling ultraproduct spectral gap)
3. Complete theoretical system (multiple equivalent characterizations)
4. Provides operational criteria

Main limitations are:

1. High technical complexity; some quantitative estimates lack refinement
2. Some natural questions remain unresolved (Questions 8.2, 8.3)
3. Applications primarily limited to property (T) groups

Overall, this is an important contribution that will become foundational literature in cross-disciplinary research between sofic groups and coarse geometry. The paper's technical methods and theoretical framework have significant reference value for subsequent research.

Recommendation Index: ⭐⭐⭐⭐⭐ (5/5)

• Theoretical Significance: ⭐⭐⭐⭐⭐
• Technical Innovation: ⭐⭐⭐⭐⭐
• Completeness: ⭐⭐⭐⭐⭐
• Readability: ⭐⭐⭐⭐
• Practical Value: ⭐⭐⭐⭐

References

The paper cites 42 references, with key works including:

1. WY14 Willett-Yu: Geometric property (T) - Foundational work
2. AF19 Alekseev-Finn-Sell: Sofic boundaries - Direct predecessor of this work
3. AB21 Alekseev-Biz: a-T-menability - Sister result
4. Kun19 Kun: Sofic approximations of Property (T) groups - Core technical source
5. DW22 Dell'Aiera-Willett: Topological property (T) for groupoids - Groupoid theory foundation
6. BdV08 Bekka-de la Harpe-Valette: Kazhdan's Property (T) - Classical reference
7. AD05 Anantharaman-Delaroche: Cohomology of property T groupoids - Measure groupoid property (T)
8. KT19, Kun21 Kun-Thom: Non-LEF group constructions - Motivation for diffuseness theory