A finitary structure theorem for vertex-transitive graphs of polynomial growth

Matthew Tointon; Romain Tessera

REVIEW 1 major objections 5 minor 26 references

A finitary structure theorem for vertex-transitive graphs of polynomial growth

T0 review · 1 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read For vertex-transitive graphs, one polynomial-growth scale forces a virtually nilpotent Cayley-graph approximation.

desk verdict Finitary Trofimov theorem for arbitrary vertex-transitive graphs is real, well-proved work; the main soft spot is the acknowledged reliance on Carolino's Lie-structure classification, which readers should check before building applications. read the letter →

arxiv 1908.06044 v3 pith:QQZNFGUT submitted 2019-08-16 math.CO math.GNmath.GRmath.PR

Romain Tessera , Matthew Tointon This is my paper · ORCID

classification math.COmath.GNmath.GRmath.PR MSC 05C2520F65

keywords vertex-transitivegraphspolynomialgrowthfinitarystructuretheoremapproximategroupsquasi-isometryCayleymoderaterandomwalks

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 quantitative, finitary version of the classical theorem that a connected, locally finite vertex-transitive graph of polynomial growth is quasi-isometric to a Cayley graph of a virtually nilpotent group. The new result replaces global polynomial growth by a single-scale condition: if $\beta_\Gamma(3n) \le K \beta_\Gamma(n)$ for one sufficiently large $n$, then the whole graph admits a quotient with fibres of diameter $O_K(n)$, on which the action is virtually nilpotent with rank, step, index and vertex stabilisers all bounded by a function of $K$ alone. Because the bound is quantitative, it turns structural statements into usable estimates. The paper draws consequences for finite vertex-transitive graphs of large diameter, showing they have small-fibre virtually abelian quotients and a growth property known to make the mixing time of the lazy random walk quadratic in the diameter. A further corollary describes the growth of such graphs at every larger scale by a piecewise-monomial function. A careful reader will note that the quantitative conclusion inherits two ineffective ingredients from existing classifications of approximate groups, so the constants $n_0$ and the nilpotent index are not explicit.

What carries the argument

The load-bearing object is the ball $S^n$ in the automorphism group, where $S$ is the set of automorphisms moving a fixed vertex by distance at most one. In the pointwise-convergence topology this ball is an open compact generating set, and under the single-scale growth bound its square $S^{2n}$ satisfies the definition of a $K^3$-approximate group: a symmetric identity-containing set whose square is covered by $K^3$ left translates of itself. Two known classification theorems for approximate groups, one for finite approximate groups and one for relatively compact approximate subgroups of locally compact groups, then force a large virtually nilpotent quotient with controlled structure. Around this core, the proof assembles lemmas showing that a normal subgroup with small displacement acts on the graph by a quasi-isometry, and that the graph is naturally quasi-isometric to the Cayley graph of the induced automorphism group. The overall mechanism is therefore: a growth inequality becomes a measure-theoretic product-set estimate, becomes an approximate subgroup, becomes an algebraic structure theorem, becomes a geometric approximation.

What would settle it

Take any sequence of finite connected vertex-transitive graphs satisfying $\operatorname{diam}(\Gamma) \ge (|\Gamma|/\beta_\Gamma(1))^\delta$ with diameters tending to infinity and compute the relaxation or mixing time of the lazy random walk; the paper's Corollary 2.8 predicts these are $\Theta(\operatorname{diam}(\Gamma)^2)$. A single sequence with relaxation time not comparable to $\operatorname{diam}(\Gamma)^2$ would refute the paper's moderate-growth conclusion and hence the main structure theorem as applied. Alternatively, search for a connected locally finite vertex-transitive graph with $\beta_\Gamma(3n) \le K \beta_\Gamma(n)$ at one large $n$ for which every quotient with virtually nilpotent induced action has either a fibre of diameter much larger than $n$ or a nilpotent step growing with $n$; the theorem asserts both are $O_K(n)$ and $O_K(1)$ respectively.

Watch

Extended reading notes

Core claim

The central discovery is that a single scale of polynomial growth, $\beta_\Gamma(3n) \le K \beta_\Gamma(n)$, already forces the intricate algebraic-geometric structure that previously had been established only under polynomial growth at all scales. Concretely, Theorem 2.3 asserts that for every $K$ there is $n_0(K)$ such that for any transitive automorphism subgroup $G$ of such a graph there is a normal subgroup $H$ of $G$ whose fibres have diameter $O_K(n)$, whose induced action has a nilpotent normal subgroup of rank, step and index $O_K(1)$, whose vertex stabilisers have size $O_K(1)$, and which yields a $(1,O_K(n))$-quasi-isometry from $\Gamma$ to a finitely generated Cayley graph of the quotient. Thus the graph is, at scale $n$, indistinguishable from a Cayley graph of a nilpotent group with bounded complexity. The paper further derives that finite vertex-transitive graphs of large diameter have virtually abelian quotients with small fibres and, when the diameter is large in the sense $\operatorname{diam}(\Gamma) \ge (|\Gamma|/\beta_\Gamma(1))^\delta$, that they have $(O(1),O(1))$-moderate growth; by known results this makes random-walk mixing and relaxation times quadratic in the diameter. It also shows that a single-scale bound at $n$ implies growth at all later scales is described by a piecewise-monomial function with $O(1)$ pieces.

Load-bearing premise

The proof depends as a black box on two classification theorems for approximate groups being true with uniform, dimension-free constants; if either classification were false or non-uniform, the quantitative form of the main theorem would lose its content.

Editorial extensions

If this is right

Every connected locally finite vertex-transitive graph satisfying $\beta_\Gamma(3n) \le K \beta_\Gamma(n)$ at one sufficiently large $n$ is $(1,O_K(n))$-quasi-isometric to a locally finite Cayley graph of a virtually nilpotent group with rank, step and index $O_K(1)$.
Finite vertex-transitive graphs with diameter at least $(|\Gamma|/\beta_\Gamma(1))^\delta$ have a quotient with fibres of diameter at most $\operatorname{diam}(\Gamma)^{1/2+\lambda}$ whose induced automorphism group is virtually abelian with bounded index and rank.
For such large-diameter graphs, the lazy random walk has mixing and relaxation times quadratic in the diameter, because they have moderate growth.
A bound $\beta_\Gamma(n) \le n^d \beta_\Gamma(1)$ at one sufficiently large scale determines all growth: $\beta_\Gamma(mn) \asymp_d f(m) \beta_\Gamma(n)$ for a piecewise-monomial $f$ with $O_d(1)$ pieces and non-negative integer degrees.
If $\beta_\Gamma(n) \le n$ at a sufficiently large scale, the graph is already contained in a ball of radius $n$, so it is finite with diameter $O(n)$.

Reading between the lines

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

The factor $3$ in the hypothesis is probably an artifact of the non-unimodular nature of general automorphism groups; a natural test is whether replacing $3n$ by $2n$, as in the Cayley-graph theorem, holds in full generality or requires a bi-invariant Haar measure.
Because the main theorem controls fibres and stabilisers rather than only the quasi-isometry type, it should make Gromov–Hausdorff scaling limits of vertex-transitive graphs with polynomial growth effective, turning the known qualitative tori limits into quantitative ones.
The moderate-growth equivalence suggests a cheap criterion for random-walk mixing on large vertex-transitive graphs: check the single-scale diameter bound, then read off $\Theta(\operatorname{diam}(\Gamma)^2)$ mixing without constructing any quotient.
One could test sharpness by looking for vertex-transitive graphs with $\beta_\Gamma(2n) \le K \beta_\Gamma(n)$ where the optimal fibre diameter in a virtually nilpotent quotient is $\Theta(n)$, or where $K$ must enter the constants in an essential way.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

The one-line take: this is the right paper at the right level of generality, and the proof largely holds up. The stated finitary version of Trofimov's theorem—any vertex-transitive graph with one polynomial-growth ball at scale n is quasi-isometric to a nilpotent Cayley graph with constants O_K(1) and distortion O_K(n)—is the natural quantitative statement, and the applications (large diameter implies virtually abelian quotient, moderate growth implies quadratic mixing time, single-scale growth controls all later growth) are genuinely new for vertex-transitive graphs and not just Cayley graphs. The proof writes out the reduction cleanly: growth gives an approximate subgroup S^{2n} in the closure of G, Carolino gives a Lie quotient of bounded dimension, total disconnectedness forces discreteness, and BGT finishes. I followed the density argument in Theorem 2.3 and the fibre-diameter estimates; no internal gap jumped out.

Credit where due: the paper is self-contained modulo cited external classification theorems, and it says so. The remark after Theorem 1.3 explicitly names n0 and the nilpotent index as the only ineffective quantities and identifies Carolino + BGT as the source. That is honest and it is also the thing to check before relying on the theorem. If Carolino's theorem is non-uniform or its Lie quotient does not have index O_K(1), the bridge at Corollary 6.11 collapses. I did not find evidence of that failure in the text; the deduction of Corollary 6.11 from Theorem 6.10 is coherent, and the constants are tracked in the right places. The reliance is external but it is load-bearing, so a referee should verify Theorem 6.10's stated uniformity rather than take it on faith.

Only real blemishes are minor. Corollary 2.8's converse states δ ≥ 0 where the proof and the rest of the paper treat δ > 0; that is almost certainly a typo. And the wording around Theorem 7.1's conclusion (iv) could be misread—finite vertex stabilisers, not bounded—but the later argument supplies the bound. Neither affects the main theorem.

Who this is for: anyone working on vertex-transitive graphs, random walks on them, or quantitative Gromov-type structure theory. It deserves a careful referee and likely publication. I would recommend accepting after a routine check of the approximate-group dependencies.

Referee Report

1 major / 5 minor

Summary. The paper proves a quantitative finitary structure theorem for connected, locally finite vertex-transitive graphs of polynomial growth. The main result (Theorem 2.3) states that if a graph Γ satisfies a single-scale polynomial growth bound βΓ(3n) ≤ K βΓ(n) for some large n, then for any transitive subgroup G < Aut(Γ) there is a normal subgroup H ⊳ G whose fibres have diameter O_K(n), such that the induced action of G on Γ/H is virtually nilpotent with rank, step and index O_K(1), vertex stabilisers of size O_K(1), and Γ is (1,O_K(n))-quasi-isometric to a locally finite Cayley graph. The proof passes through a topological-group version (Theorem 7.1) which uses the Breuillard–Green–Tao approximate-group classification and Carolino's classification of relatively compact approximate subgroups to produce a virtually nilpotent, discrete quotient. The paper then derives several applications: a finitary version of Trofimov's theorem (Corollary 2.4), structure of large-diameter finite vertex-transitive graphs (Corollaries 2.5 and 2.6), equivalence of large diameter with moderate growth (Corollary 2.8), and a two-sided description of growth at all scales from a single-scale bound (Corollaries 1.4 and 1.5).

Significance. If correct, this is a substantial quantitative extension of Trofimov's classical theorem, bringing vertex-transitive graphs into the framework of finitary approximate-group theory. The paper is careful and self-contained modulo two explicitly acknowledged external classifications: the theorem of Breuillard–Green–Tao and Carolino's theorem, which are the only sources of ineffective constants. The reductions in Sections 7–9 are coherent and detailed: the use of Haar measures to convert the single-scale growth bound into an approximate subgroup, the passage through a totally disconnected Lie quotient, and the subsequent stabiliser bound via Proposition 6.5 are all internally sound. The applications to large-diameter graphs and moderate growth are well motivated and clearly derive from the main theorem. A particular strength is the explicit accounting of which quantities are effective and which are not, which is important for the announced follow-up work.

major comments (1)

[Section 7, Corollary 6.11] The central quantitative conclusion of Theorem 2.3 inherits its uniformity in K entirely from Carolino's classification (Theorem 6.10), quoted from the PhD thesis [5]. I checked the reduction in Corollary 6.11: applying Lemma 6.7 to the cosets of L0 and Lemma 6.8 to pass to a normal subgroup L of index O_K(1), and then taking H = (L ∩ H0)^G with Lemma 6.9, indeed yields L open of index O_K(1), H ⊂ S^{O_K(n)} ∩ L, H compact, and L/H a Lie group of dimension O_K(1). The bridge from the approximate group S^{2n} to the Lie quotient therefore appears internally valid, and the stress-test worry that the constants might not be uniform does not land. I would nevertheless ask the authors to double-check that Theorem 6.10 is stated with exactly the uniformity used here, since any weakening in the statement of that external theorem would propagate directly to Theorem 2.3.

minor comments (5)

[Section 8, proof of Corollary 2.4] The inequality 'βΓ(5m) ≤ 38^{d/λ} βΓ(m)' appears to contain a typographical constant: Lemma 8.1 applied with q = 5, α = λ/4 and β = λ/2 gives K = 5^{8d/λ}. The constant is absorbed into O_{d,λ}(1), so this is harmless, but it should be corrected for accuracy.
[Section 2, after Theorem 2.3] The term 'rank' of a nilpotent group is used repeatedly in Theorems 2.3, 6.4 and elsewhere, but it is never defined. A one-sentence definition (minimal number of generators of the group) would improve self-containedness.
[Section 2, before Corollary 2.6] The definition of the quotient metric space X/H is given in a paragraph between Corollaries 2.5 and 2.6, after the notation is already used informally. Moving the definition up, or adding an explicit pointer, would make the dependence of Corollary 2.6 on this notion clearer.
[Section 5, Lemma 5.2] The proof of Lemma 5.2 refers to the inclusion ψ^{-1}(B_{Γ/H}(ψ(x), d(x,y)-k-1)) ⊂ B_Γ(x,d(x,y)-1); for readability, the justification that this inclusion is 'a consequence of Lemma 3.7' would benefit from expanding the one-line argument, although the implication is correct.
[References] Reference [5] is an unpublished PhD thesis. If a peer-reviewed version of Carolino's result exists or has since appeared, the authors should cite it; otherwise, they may consider adding a note on where the precise statement can be found in the thesis.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the derivation uses external approximate-group classifications and independent Cayley-graph theorems; no conclusion is assumed as an input.

full rationale

The derivation chain is not circular. Theorem 2.3 is proved by converting the single-scale bound beta_Gamma(3n) <= K beta_Gamma(n) into the statement that S^{2n} is a compact open K^3-approximate group (Lemma 4.8 and Proposition 6.1), then applying Carolino's Theorem 6.10 through Corollary 6.11 to obtain an open finite-index subgroup L with compact normal H0 and Lie quotient L/H0, and applying Breuillard-Green-Tao's Theorem 6.4 to the resulting finite generating set to obtain the nilpotent quotient. These are external classification results whose hypotheses do not include the conclusion of Theorem 2.3. The applications (Corollaries 1.4, 1.5, 2.5, 2.6) use the main theorem only to reduce a vertex-transitive graph to an auxiliary Cayley graph, and then invoke Tao's Theorem 1.9 and the authors' earlier theorems [4] and [20] as independent statements about Cayley graphs; the earlier results are applied to the constructed Cayley graph, not to the original graph, so no fitted parameter is renamed as a prediction and no claim is assumed in order to prove itself. The remark after Theorem 1.3 explicitly acknowledges that n0 and the nilpotent index are ineffective because of the two external classifications; this is an honest limitation, not a circular step.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The main theorem is proved from growth assumptions using standard group-theoretic tools and two deep external classification results. No free parameters fitted to data and no invented entities appear; the listed axioms are the unproved external theorems the proof imports.

assumptions (5)

standard math Breuillard-Green-Tao finite approximate group classification (Theorem 6.4)
Cited as [3, Cor 11.2], invoked in Section 7 to pass from a K^3-approximate group S^{2n} to a virtually nilpotent quotient. External black box.
standard math Carolino's structure theorem for locally compact approximate subgroups (Theorem 6.10)
Cited as [5, Theorem 1.9], used via Corollary 6.11 in Section 7 to obtain a Lie group quotient with compact normal subgroup. External black box and source of ineffectiveness.
standard math Tao's inverse theorem for sets and measures of polynomial growth ([19, Theorem 1.9])
Used in the proof of Corollary 1.4 to obtain the piecewise-monomial growth estimate for the approximating Cayley graph. External.
standard math Breuillard-Tointon large-diameter Cayley graph theorem ([4, Theorem 4.1])
Used in proofs of Corollaries 2.5, 2.6 and Theorem 8.2. Self-cited but an independently published theorem; applied after reduction to a Cayley graph, so not circular.
standard math Existence of Haar measure on locally compact Hausdorff topological groups
Used throughout Section 7 for measure inequalities on closed subgroups of Aut(Γ).

how reviews work

0 comments

Cite this review

Pith. "Pith review of A finitary structure theorem for vertex-transitive graphs of polynomial growth." pith.science (2026). https://pith.science/paper/QQZNFGUT

@misc{pith2026190806044,
  author       = {Pith},
  title        = {Pith review of: A finitary structure theorem for vertex-transitive graphs of polynomial growth},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QQZNFGUT}},
  note         = {Machine review of arXiv:1908.06044}
}

read the original abstract

We prove a quantitative, finitary version of Trofimov's result that a connected, locally finite vertex-transitive graph G of polynomial growth admits a quotient with finite fibres on which the action of Aut(G) is virtually nilpotent with finite vertex stabilisers. We also present some applications. We show that a finite, connected vertex-transitive graph G of large diameter admits a quotient with fibres of small diameter on which the action of Aut(G) is virtually abelian with vertex stabilisers of bounded size. We also show that G has moderate growth in the sense of Diaconis and Saloff-Coste, which is known to imply that the mixing and relaxation times of the lazy random walk on G are quadratic in the diameter. These results extend results of Breuillard and the second author for finite Cayley graphs of large diameter. Finally, given a connected, locally finite vertex-transitive graph G exhibiting polynomial growth at a single, sufficiently large scale, we describe its growth at subsequent scales, extending a result of Tao and an earlier result of our own for Cayley graphs. In forthcoming work we will give further applications.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 25 canonical work pages

[5]

P. K. Carolino. The Structure of Locally Compact Approxi mate Groups. PhD thesis: https://escholarship.org/uc/item/8388n9jk

work page
[1]

Benjamini and G

I. Benjamini and G. Kozma. A resistance bound via an isope rimetric inequality, Combinatorica 25(6) (2005), 645–650

work page 2005
[2]

Benjamini, H

I. Benjamini, H. Finucane and R. Tessera. On the scaling l imit of ﬁnite vertex transitive graphs with large diameter. Combinatorica 36 (2016), 1–41

work page 2016
[3]

Breuillard, B

E. Breuillard, B. J. Green and T. C. Tao. The structure of a pproximate groups, Publ. Math. IHES. 116(1) (2012), 115–221

work page 2012
[4]

Breuillard and M

E. Breuillard and M. C. H. Tointon. Nilprogressions and g roups with moderate growth, Adv. Math. 289 (2016), 1008–1055

work page 2016
[6]

Diaconis and L

P. Diaconis and L. Saloﬀ-Coste. Moderate growth and rand om walk on ﬁnite groups, Geom. Funct. Anal. 4(1) (1994), 1–36. 26 ROMAIN TESSERA AND MATTHEW C. H. TOINTON

work page 1994
[7]

Diestel and I

R. Diestel and I. Leader. A conjecture concerning a limit of non-Cayley graphs, J. Algebraic Combin. 14 (2001), 17–25

work page 2001
[8]

Eskin, D

A. Eskin, D. Fisher and K. Whyte. Quasi-isometries and ri gidity of solvable groups, Pure Appl. Math. Q. 3 (2007), Special Issue: In honor of Grigory Margulis. Part 1, 927–947

work page 2007

Show all 26 references

[9]

M. Gromov. Groups of polynomial growth and expanding map s, Publ. Math. IHES 53 (1981), 53–73

1981
[10]

M. Hall. The theory of groups , Amer. Math. Soc./Chelsea, Providence, RI (1999)

1999
[11]

Hermon and R

J. Hermon and R. Pymar. The exclusion process mixes (alm ost) faster than independent particles, arXiv:1808.10846v2

arXiv
[12]

Hewitt and K

E. Hewitt and K. A. Ross. Abstract Harmonic Analysis I (2nd ed.) , Springer-Verlag, Berlin (1979)

1979
[13]

B. Kleiner. A new proof of Gromov’s theorem on groups of p olynomial growth, Jour. of the AMS 23(3) (2010), 815–829

2010
[14]

D. A. Levin and Y. Peres, with contributions by E. L. Wilm er. Markov Chains and Mixing Times (2nd ed.) , American Mathematical Society, Providence, RI (2017)

2017
[15]

V. Losert. On the structure of groups with polynomial gr owth, Math. Z. 195 (1987), 109–117

1987
[16]

N. Ozawa. A functional analysis proof of Gromov’s polyn omial growth theorem, Ann. Sci. ´Ec. Norm. Sup´ er. (4) 51(3) (2018), 549–556

2018
[17]

Shalom and T

Y. Shalom and T. C. Tao. A ﬁnitary version of Gromov’s pol ynomial growth theorem, Geom. Funct. Anal. 20(6) (2010), 1502–1547

2010
[18]

T. C. Tao. Product set estimates for non-commutative gr oups, Combinatorica 28(5) (2008), 547–594

2008
[19]

T. C. Tao. Inverse theorems for sets and measures of poly nomial growth, Q. J. Math. 68(1) (2017), 13–57

2017
[20]

Tessera and M

R. Tessera and M. C. H. Tointon. Properness of nilprogre ssions and the persistence of polynomial growth of given degree, Discrete Anal. 2018:17, 38 pp

2018
[21]

Tessera and M

R. Tessera and M. C. H. Tointon. Sharp relations between volume growth, isoperimetry and resistance in vertex- transitive graphs, arXiv:2001.01467

2001 arXiv
[22]

Tessera and M

R. Tessera and M. C. H. Tointon. Lie group approximation s of balls with polynomial volume in vertex-transitive graphs, in preparation
[23]

M. C. H. Tointon. Freiman’s theorem in an arbitrary nilp otent group, Proc. London Math. Soc. (3) 109 (2014), 318–352

2014
[24]

M. C. H. Tointon. Introduction to approximate groups , London Mathematical Society Student Texts 94, Cam- bridge University Press, Cambridge (2020)

2020
[25]

V. I. Troﬁmov. Graphs with polynomial growth, Math. USSR-Sb. 51 (1985) 405–417

1985
[26]

W. Woess. Topological groups and inﬁnite graphs, Discrete Math. 95 (1991), 373–384. Institut de Math ´ematiques de Jussieu-P aris Rive Gauche, France Email address : tessera@phare.normalesup.org School of Mathematics, University of Bristol, United Kingd om Email address : m.to...

1991

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [5]

P. K. Carolino. The Structure of Locally Compact Approxi mate Groups. PhD thesis: https://escholarship.org/uc/item/8388n9jk

work page

[2] [1]

Benjamini and G

I. Benjamini and G. Kozma. A resistance bound via an isope rimetric inequality, Combinatorica 25(6) (2005), 645–650

work page 2005

[3] [2]

Benjamini, H

I. Benjamini, H. Finucane and R. Tessera. On the scaling l imit of ﬁnite vertex transitive graphs with large diameter. Combinatorica 36 (2016), 1–41

work page 2016

[4] [3]

Breuillard, B

E. Breuillard, B. J. Green and T. C. Tao. The structure of a pproximate groups, Publ. Math. IHES. 116(1) (2012), 115–221

work page 2012

[5] [4]

Breuillard and M

E. Breuillard and M. C. H. Tointon. Nilprogressions and g roups with moderate growth, Adv. Math. 289 (2016), 1008–1055

work page 2016

[6] [6]

Diaconis and L

P. Diaconis and L. Saloﬀ-Coste. Moderate growth and rand om walk on ﬁnite groups, Geom. Funct. Anal. 4(1) (1994), 1–36. 26 ROMAIN TESSERA AND MATTHEW C. H. TOINTON

work page 1994

[7] [7]

Diestel and I

R. Diestel and I. Leader. A conjecture concerning a limit of non-Cayley graphs, J. Algebraic Combin. 14 (2001), 17–25

work page 2001

[8] [8]

Eskin, D

A. Eskin, D. Fisher and K. Whyte. Quasi-isometries and ri gidity of solvable groups, Pure Appl. Math. Q. 3 (2007), Special Issue: In honor of Grigory Margulis. Part 1, 927–947

work page 2007