Pith. sign in

REVIEW 3 major objections 3 minor 16 references

Centralizers of sofic approximations of Kazhdan groups

T0 review · 3 major / 3 minor · reviewed 2026-08-08 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper proves that a Kazhdan group with a sofic embedding whose centralizer acts ergodically on the Loeb space must be locally embeddable into finite groups, and residually finite if finitely presented.

desk verdict Theorem A is new and the proof is carefully built, but the argument depends on a uniform Becker–Chapman stability variant that the paper states without proof or precise citation. read the letter →

arxiv 2608.05362 v1 pith:H2XEBN6W submitted 2026-08-05 math.GR

Vadim Alekseev , Andreas Thom This is my paper · ORCID
classification math.GR MSC 20F6520F6920F0537A1546L10
keywords soficgroupsKazhdanproperty(T)centralizersmetricultraproductsLoebspacesLEFresidualfinitenessstabilityofapproximateactions
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proves a conditional rigidity statement about sofic approximations of groups with Kazhdan's property (T). If a Kazhdan group G admits a sofic embedding into a universal sofic group whose centralizer acts ergodically on the associated Loeb probability space, then G must be locally embeddable into finite groups (LEF); when G is finitely presented, this forces G to be residually finite. The point is that the additional ergodicity demand on the centralizer, which a 2024 conjecture predicts every sofic group should satisfy, is extremely restrictive for Kazhdan groups. Its technical engine is a description of such a centralizer as essentially a metric ultraproduct of finite permutation groups (Theorem 3.1).

What carries the argument

The central object is the cluster groupoid $\mathcal{C}_n$ built from partial bijections between the high-expansion components of a sofic approximation. Its arrows are clusters of almost-equivariant partial bijections that are close in Hamming distance; composition is defined only after repairing the raw composition via a two-component repair lemma. The finite full group $[[\mathcal{C}_n]]$ of total bisections of this groupoid is shown, by a two-sided majority argument, to exhaust the centralizer up to asymptotic error, and a stability theorem for near-actions of finite groups then corrects the approximate action to a genuine homomorphism $\rho_n: [[\mathcal{C}_n]] \to \mathrm{Sym}(Y_n)$. The ultraproduct of the groups $\rho_n([[\mathcal{C}_n]])$ is exactly the centralizer, yielding the metric-ultraproduct-of-permutation-groups description.

What would settle it

Find a sequence of approximate actions of finite groups on finite sets whose defect tends to zero but where every genuine homomorphism on a slightly enlarged set remains Hamming distance at least a fixed c>0 away; that would refute the uniform stability statement used in Section 5 and break Proposition 5.1, the step that makes Theorem 3.1 and Theorem A depend on an unproved premise.

Watch

Extended reading notes

Core claim

The central claim is the paper's Theorem A: a Kazhdan group that has a sofic embedding into a universal sofic group with the property that the centralizer acts ergodically on the Loeb probability space is LEF, and if finitely presented, residually finite. The proof establishes the stronger structural statement Theorem 3.1: after replacing the finite models by essentially equivalent ones, the centralizer of the sofic embedding is exactly a metric ultraproduct of finite permutation groups $A_n \leq \mathrm{Sym}(Y_n)$. This rigidity of the centralizer, combined with a discreteness argument for right-translation actions of normalizer quotients, converts the embedding into a genuine embedding into an algebraic ultraproduct of finite groups, which is precisely an LEF certificate.

Load-bearing premise

The proof uses a uniform stability theorem: for every epsilon there is a delta such that any map from a finite group to permutations with defect below delta is epsilon-close to a genuine homomorphism after adding a small number of new points; if this uniform statement is not a proved consequence of the cited stability results, the centralizer equality that the whole argument rests on does not follow.

Editorial extensions

If this is right

  • Every finitely presented Kazhdan group that admits a sofic embedding with ergodic centralizer is residually finite; hence no such group can serve as a finitely presented, non-residually finite counterexample to the soficity conjecture.
  • Combining Theorem A with the conjecture that every sofic group admits an embedding with ergodic centralizer would make every sofic Kazhdan group LEF and every finitely presented sofic Kazhdan group residually finite, sharply constraining the geometry of sofic approximations of Kazhdan groups.
  • The centralizer of any sofic embedding of a Kazhdan group is, up to essentially equivalent finite models, a metric ultraproduct of finite permutation groups; this structural rigidity holds for all sofic approximations of Kazhdan groups, not just those with ergodic centralizers.
  • Known Kazhdan groups that are LEF but not residually finite must fail the ergodic-centralizer condition in every sofic embedding they admit.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The theorem suggests that the ergodicity of the centralizer is a 'finite-nearness' condition: it forces the sofic model to admit a co-large subsystem on which the model is genuinely finite. One could test whether weaker invariants, such as the non-existence of invariant Loeb-measurable subsets with intermediate measure, already imply LEF.
  • If the recently announced proof of a non-sofic finitely presented Kazhdan group is correct, it does not contradict Theorem A; that group would simply fail the hypothesis of admitting any sofic embedding at all. It would instead shift the question to whether the ergodic-centralizer conjecture is refuted for non-sofic groups.
  • The two-component repair and cluster-groupoid construction is plausibly adaptable to tracial ultraproducts of matrix algebras, an extension the paper itself poses as an open problem; the permutation-group centralizer rigidity proven here is a natural template for a von Neumann algebraic centralizer rigidity statement.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 3 minor

Summary. The paper proves Theorem A: if a Kazhdan group G admits a sofic embedding into a metric ultraproduct of symmetric groups whose centralizer acts ergodically on the associated Loeb probability space, then G is locally embeddable into finite groups (LEF), and if G is finitely presented then G is residually finite. The main technical result, Theorem 3.1, asserts that after an essentially equivalent modification of the finite models, the centralizer of such a sofic embedding is a metric ultraproduct of finite permutation groups. The proof proceeds through Kun's expander decomposition, a two-component repair argument for approximate intertwiners, a cluster-groupoid construction, and Becker–Chapman flexible stability. The paper also formulates two open problems in Section 6.

Significance. If the proof can be made fully self-contained, the result is significant: it gives a structural obstruction to ergodic centralizers of sofic approximations of Kazhdan groups and connects the Hayes–Kunnawalkam Elayavalli conjecture with residual finiteness. The internal developments in Sections 3 and 4 are carefully structured, and the reduction in Section 2 from rigid centralizers to LEF is elegant and clearly explained. The paper is also honest about its reliance on external results, but this honesty exposes the main risk: the crucial uniform stability input in Section 5 is only asserted, not proved or precisely cited.

major comments (3)
  1. [Section 5, preamble to Proposition 5.1] The stated 'uniform variant' of Becker–Chapman flexible stability is load-bearing: Proposition 5.1 applies the stability theorem to the sequence of finite groups F_n = [[C_n]] with |F_n| tending to infinity, so the same delta must work uniformly for all finite groups. If the stability modulus in [3] depends on |Gamma|, the proof cannot pass to the ultrafilter, and the equality C(pi'(G)) = prod_U rho_n(F_n) in Proposition 5.1, hence Theorem 3.1 and Theorem A, would not follow. The manuscript neither proves the uniform statement nor cites a theorem number in [3]; please supply a proof or an exact reference, including the precise hypotheses on Gamma, V, and the enlargement bound.
  2. [Section 3, Proposition 3.3] Proposition 3.3 is the technical engine of the cluster groupoid construction, but its proof invokes '[12, Proposition 3.3]' without restating that result. Since [12] is a preprint and the cited proposition is described only in prose, the manuscript should state it precisely, or prove it, so that the constants C_1 and rho and the hypotheses on the regular S-labelled graphs are verifiable.
  3. [Section 2, Theorem 2.5] Theorem 2.5, Kun's expander decomposition, is also load-bearing and is cited to the unpublished preprint [11]. The manuscript should either include a proof, state the theorem in full with its hypotheses, or cite a published version; otherwise the expander-component reduction used at the start of Theorem 3.1 rests on an external statement whose status the reader cannot verify from the present paper.
minor comments (3)
  1. [Section 5, Proposition 5.1, converse direction] The modification of sigma_n on (Y_n \ X_n) cup sigma_n^{-1}(Y_n \ X_n) to make it preserve X_n is only sketched; since the argument is needed in order to apply Proposition 4.5(c), please spell out the construction of the modified permutation and verify that its commutation defects with alpha_n(s) still tend to zero along U.
  2. [Section 1, first footnote] The footnote about the OpenAI announcement is extraneous to the mathematical content of the paper and should be removed or moved to a clearly separated editorial note.
  3. [Section 6, Open problem 6.2] The notation pi: G -> U(prod_U M_{d_n}(C)) is terse; identifying the target as the unitary group of the tracial ultraproduct, or writing the map coordinatewise, would improve readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the centralizer rigidity theorem is proved by explicit construction, and the cited prior work is independent support rather than a restatement of the conclusion.

full rationale

The paper's derivation chain is not circular. Theorem A is obtained from Theorem 3.1 and Proposition 2.9. Theorem 3.1 is proved by a concrete construction: Kun's expander decomposition (Theorem 2.5), the two-component repair Proposition 3.3 (using [12, Proposition 3.3]), the cluster groupoid (Definition 4.1 and Lemma 4.2), and Proposition 4.5, which shows that the finite full groups of the cluster groupoid exhaust the centralizer. Proposition 5.1 then applies Becker–Chapman stability as a black-box external theorem. Whether the stated uniform variant of Becker–Chapman actually follows from [3] is a correctness or dependency concern, not a circularity: the statement concerns arbitrary finite approximate actions and does not assume the paper's conclusion. The self-citations, notably [12] co-authored by the second author, are load-bearing but are cited as prior mathematical theorems with proofs; they are not definitions of the target result, nor do they forbid alternatives by fiat. Proposition 2.9 converts centralizer rigidity into an LEF conclusion using the general fact that subgroups of ultraproducts of finite groups are LEF; no fitted parameter or predicted quantity is reused as an input. The manuscript therefore contains no step in which a claimed derivation reduces, by its own equations or by a self-citation, to its own assumptions.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The proof rests on a small stack of external theorems: Kun's expander decomposition, a proposition from [12], and a strong uniform flexible stability theorem attributed to Becker-Chapman. No free parameters are fitted to data, and no new physical or formal entities are postulated. The cluster groupoid is a constructed tool, not an assumed object.

assumptions (6)
  • standard math Kun's expander decomposition theorem (Theorem 2.5)
    Stated as Theorem 2.5 and used to decompose sofic approximations of Kazhdan groups into expander components; cited to [11], a preprint.
  • standard math [12, Proposition 3.3] on approximating subsets with small boundary in Kazhdan Schreier graphs
    Used in the proof of Proposition 3.3 to find a set U close to the graph of a partial bijection with small boundary; cited to Kun-Thom [12].
  • standard math Becker-Chapman uniform flexible stability theorem
    Stated in Section 5 and used in Proposition 5.1 to correct approximate actions of finite full groups [[C_n]] to genuine actions; cited to [3].
  • standard math Properties of metric ultraproducts of finite permutation groups and Loeb probability spaces
    Used throughout, especially in Lemma 2.4 and Proposition 2.7, to pass from finite approximations to limit objects.
  • standard math Kazhdan groups are finitely generated and admit a Kazhdan constant
    Used via Kun's theorem and the expander constant h; a standard consequence of property (T).
  • standard math Existence of a non-principal ultrafilter U on the natural numbers
    Required for the definition of metric ultraproducts; standard set-theoretic background in the subject.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Centralizers of sofic approximations of Kazhdan groups." pith.science (2026). https://pith.science/paper/H2XEBN6W

@misc{pith2026260805362,
  author       = {Pith},
  title        = {Pith review of: Centralizers of sofic approximations of Kazhdan groups},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H2XEBN6W}},
  note         = {Machine review of arXiv:2608.05362}
}
read the original abstract

We prove that a Kazhdan group admitting a sofic embedding into a metric ultraproduct of symmetric groups with a centralizer that acts ergodically on the associated Loeb probability space is locally embeddable in finite groups (LEF). In particular, every finitely presented Kazhdan group admitting such an embedding is residually finite. The main technical theorem says that the centralizer of a sofic embedding of a Kazhdan group is itself a metric ultraproduct of permutation groups.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 11 canonical work pages

  1. [3]

    Becker and M

    O. Becker and M. Chapman,Stability of approximate group actions: uniform and probabilistic, J. Eur. Math. Soc.25(2023), no. 9, 3599–3632. 1, 5

    work page 2023

  2. [12]

    Kun and A

    G. Kun and A. Thom,Inapproximability of actions and Kazhdan’s property (T), preprint arXiv:1901.03963. 1, 1, 2.1, 3.1, 3.1, 3.1, 4.1

    arXiv 1901

  3. [11]

    Kun,On sofic approximations of property T groups, preprint arXiv:1606.04471

    G. Kun,On sofic approximations of property T groups, preprint arXiv:1606.04471. 1, 2.1

    arXiv

  4. [1]

    Alekseev and A

    V. Alekseev and A. Thom,On non-isomorphic universal sofic groups, preprint arXiv:2406.06741. 1

    arXiv

  5. [2]

    ,Remarks on approximability and stability for groups, preprint arXiv:2512.15494. 1

    arXiv

  6. [4]

    de Cornulier,Finitely presentable, non-Hopfian groups with Kazhdan’s property (T) and infinite outer automorphism group, Proc

    Y. de Cornulier,Finitely presentable, non-Hopfian groups with Kazhdan’s property (T) and infinite outer automorphism group, Proc. Amer. Math. Soc.135(2007), no. 4, 951–959. 5

    work page 2007

  7. [5]

    Elek and E

    G. Elek and E. Szabó,Hyperlinearity, essentially free actions andL2-invariants. The sofic prop- erty, Mathematische Annalen332(2005), no. 2, 421–441. 1

    work page 2005

  8. [6]

    Gohla and A

    L. Gohla and A. Thom,High-dimensional expansion and soficity of groups, preprint arXiv:2403.09582, to appear in Israel J. Math. 1

    arXiv

Show all 16 references
  1. [7]

    Gromov, Endomorphisms of symbolic algebraic varieties, J

    M. Gromov, Endomorphisms of symbolic algebraic varieties, J. Eur. Math. Soc.1(1999), no. 2, 109–197. 1

    1999

  2. [8]

    Hayes and S

    B. Hayes and S. Kunnawalkam Elayavalli,On sofic approximations of non amenable groups, Math. Z.307(2024), Paper No. 38. 1, 1

    2024

  3. [9]

    Kar and N

    A. Kar and N. Nikolov,A non-LEA sofic group, Proc. Math. Sci.127(2017), no. 2, 289–293. 5

    2017

  4. [10]

    Kerr and H

    D. Kerr and H. Li,Combinatorial independence and sofic entropy, Commun. Math. Stat.1(2013), no. 2, 213–257. 1 CENTRALIZERS OF SOFIC APPROXIMATIONS 19

    2013

  5. [13]

    Păunescu,Sofic actions and equivalence relations, J

    L. Păunescu,Sofic actions and equivalence relations, J. Funct. Anal.261(2011), no. 9, 2461–2485. 1

    2011

  6. [14]

    Systems34(2014), no

    ,A convex structure on sofic embeddings, Ergodic Theory Dynam. Systems34(2014), no. 4, 1343–1352. 1

    2014

  7. [15]

    Thom,Examples of hyperlinear groups without factorization property, Groups Geom

    A. Thom,Examples of hyperlinear groups without factorization property, Groups Geom. Dyn.4 (2010), no. 1, 195–208. 5

    2010

  8. [16]

    Weiss, Sofic groups and dynamical systems, Sankhy¯ a Ser

    B. Weiss, Sofic groups and dynamical systems, Sankhy¯ a Ser. A62(2000), no. 3, 350–359. 1 V adim Alekseev, TU Dresden, 01062 Dresden, Germany Email address:vadim.alekseev@tu-dresden.de Andreas Thom, TU Dresden, 01062 Dresden, Germany Email address:andreas.thom@tu-dresden.de

    2000

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.