Combinatorics of affine cactus groups

Hugo Chemin

REVIEW 4 major objections 5 minor 12 references

Combinatorics of affine cactus groups

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read The paper proves that affine cactus groups embed injectively into a semidirect product of a right-angled Coxeter group and a symmetric group, deriving linearity, a solvable word problem, a trivial centre, and a torsion bound of $2^{n-1}$.

desk verdict A promising embedding theorem for affine cactus groups, but the current text has several presentation errors and the main injectivity proof is incomplete. read the letter →

arxiv 2501.16270 v1 pith:3HMBS74O submitted 2025-01-27 math.CO math.GR

Hugo Chemin This is my paper · ORCID

classification math.COmath.GR MSC 20F5520F36

keywords affinecactusgroupsgeneralizedCoxeterright-angledcircularintervalsGaussdiagramtorsionwordproblem

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

Affine cactus groups are algebraic versions of cacti drawn on a cylinder: each generator flips the order of a block of marked points that winds once around the circle. This paper tries to show that these groups are well behaved, in particular that they embed injectively into the semidirect product of an affine Gauss diagram group $AD_n$ and the symmetric group $S_n$. If the embedding is correct, several structural questions are settled at once: the groups are linear, their word problem is decidable, their centre is trivial, and every torsion element has order at most $2^{n-1}$. The paper also identifies affine cactus groups with generalized cactus groups over the affine Coxeter group of type $\tilde{A}_n$, which is what lets Coxeter group theory do the heavy lifting.

What carries the argument

The machinery is the combinatorial algebra of circularity: circular intervals $[i,j]_c$ and circular sets in $\mathbb{Z}/n\mathbb{Z}$, the affine Gauss diagram group $AD_n$ generated by $\tau_I$ with relations $\tau_I^2=1$ and $\tau_I\tau_J=\tau_J\tau_I$ whenever $I\cap J=\varnothing$ or one circular set is cyclically contained in the other, and the action of $S_n$ on these sets that produces the semidirect product $AD_n\rtimes S_n$. The embedding theorem is carried by this action together with Lemma 3.3, which states that a cancellation or commutation visible in the $AD_n$-image of a word corresponds to a cancellation, commutation, or quasi-commutation of the original affine cactus letters. That correspondence is what allows the paper to read the word problem and torsion properties of $AJ_n$ off the right-angled Coxeter structure of $AD_n$.

What would settle it

Enumerate words in the generators $\sigma_{i,j}$ for $n=4$ whose image under $\phi$ is the identity in $AD_4\rtimes S_4$, and test each against the defining relations of $AJ_4$; a single non-trivial word of this kind would disprove Theorem 3.4, while an exhaustive search up to length eight in $AJ_3$ would make the injectivity claim concrete.

Watch

Extended reading notes

Core claim

The central result is Theorem 3.4: for every $n\ge 2$, the map $\phi\colon AJ_n \to AD_n \rtimes S_n$ that sends each generator $\sigma_{i,j}$ to the pair $(\tau_{[i,j]_c}, s_{i,j})$ is injective, where $AD_n$ is the affine Gauss diagram group generated by involutions $\tau_I$ indexed by circular sets over $\mathbb{Z}/n\mathbb{Z}$, and $s_{i,j}$ reverses the circular interval $[i,j]_c$. The proof starts from the identification of $AJ_n$ with the generalized cactus group over the affine Coxeter group of type $\tilde{A}_n$. It then shows that any word in $AJ_n$ whose image in $AD_n \rtimes S_n$ is trivial must already be trivial in $AJ_n$, because cancellations and commutations in the right-angled Coxeter group $AD_n$ lift back to legal moves among the affine cactus generators.

Load-bearing premise

The load-bearing premise is that every reduction of a word in $AD_n$ to the empty word can be mirrored step by step by cancellations and commutations in $AJ_n$, a step the proof asserts by saying 'continuing this process' without formalizing it.

Editorial extensions

If this is right

$AJ_n$ becomes a subgroup of a linear group, so affine cactus groups are linear for all $n\ge2$.
The word problem in $AJ_n$ is solvable: a word is trivial exactly when its image in $AD_n$ reduces to the empty word by the standard right-angled Coxeter algorithm.
The centre of $AJ_n$ is trivial for $n\ge2$, and the centre of the pure affine cactus group $PAJ_n$ is trivial for $n\ge3$.
Torsion is tightly controlled: $PAJ_n$ is torsion-free, every torsion element of $AJ_n$ has even order bounded by $2^{n-1}$, and elements of order $2k$ exist whenever $2k\le n$.
The usual cactus group $J_n$ embeds into $AJ_n$, and the pure affine cactus group $PAJ_n$ is residually nilpotent.

Reading between the lines

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

The gap between the constructed elements of order $2k$ and the bound $2^{n-1}$ leaves the exact maximal torsion order open; computing the maximum for $n=4,5$ would likely show the bound is not sharp.
If the embedding is as strong as stated, the solvability of the word problem should extend to a normal form or rewriting system for $AJ_n$, since right-angled Coxeter groups admit geodesic normal forms.
The same circular-set construction may transfer to affine cactus groups over other affine Coxeter types, where the notion of circular interval would have to respect the different Coxeter diagram.
The injectivity theorems for the subfamilies $AJ^{p,q}_n$ suggest a filtration of $AJ_n$ by the number of strands involved in a crossing, which could support inductive proofs of further structural properties.

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.

Desk Editor's Note

read the letter

Dear X,

The paper has a worthwhile core: it embeds affine cactus groups into a semidirect product of a right-angled Coxeter group and S_n, and the intended consequences (linearity, solvable word problem, trivial centre, torsion bounds, residual nilpotence of the pure subgroup) all follow if that embedding works. That target is genuinely new and worth having.

But the text as it stands has problems that need fixing before the results can be trusted. First, Definition 2.1 of W(~A_n) is inconsistent for n≥3: the commuting rule |i-j|≥2 makes ρ1 and ρn commute, and then the braid relation ρ1ρnρ1=ρnρ1ρn forces ρ1=ρn. This looks like a typo — the commuting condition should be on cyclic distance — but as written it breaks Theorem A. Relatedly, Theorem A as stated cannot be right because the generalized cactus group includes generators for singleton parabolic subgroups, while the map from AJ_n only hits non-singleton ones; for n≥3 the full circular interval [1,n]_c corresponds to an infinite parabolic, so the generator σ_{1,n} has no image. Also, the formula for s_{k,l} in Section 1 is wrong for wrapping intervals (e.g. n=4, [4,2]_c), and the 'otherwise k' clause is clearly a typo for 'otherwise i'; as written the morphism to S_n is not a permutation.

The bigger issue is the proof of Theorem 3.4. The idea is right: reduce a word in AD_n that equals 1 to the empty word, and mirror each cancellation/commutation in AJ_n using Lemma 3.3. But after a quasi-commutation the σ-word is not the same word with two letters swapped; one generator is replaced by its image under a permutation. The proof says 'continuing this process' without giving an induction invariant that keeps the σ-word aligned with the reduced τ-word. Without that, the implication τ=1 ⇒ σ=1 is not formally justified. This is the load-bearing step for the whole paper, so it needs a rigorous treatment, not a hand-wave.

That said, I think the claims are likely true and repairable. The approach is sound, and the local analysis in Lemma 3.3 is mostly correct. A careful revision that fixes the presentation errors and supplies a proper induction for the injectivity proof would make this a solid paper.

The right outcome is to send it to a serious referee, with the expectation of major revision. I wouldn't cite it in its current form, but I'd be interested to see the corrected version.

Referee Report

4 major / 5 minor

Summary. The paper studies affine cactus groups AJ_n, which arise as equivariant fundamental groups of real moduli spaces, from a combinatorial viewpoint. It claims that AJ_n is isomorphic to the generalized cactus group over an affine Coxeter group of type ~A_n (Theorem A / Theorem 2.7), that AJ_n embeds into a semidirect product AD_n \rtimes S_n where AD_n is a newly introduced 'affine Gauss diagram group' (Theorem B / Theorem 3.4), and that this embedding yields linearity, solvability of the word problem, residual nilpotence of pure affine cactus groups, triviality of centres, and torsion bounds. The paper also proves injectivity of natural inclusions of subfamilies AJ_n^{p,q} and derives a semi-direct product decomposition. The arguments are presentation-based and rely on a word-reduction analysis of the embedding.

Significance. If the main embedding were established rigorously, the paper would provide a useful structural framework for affine cactus groups and would unify several known results about cactus groups, with concrete consequences such as linearity, solvable word problem, residually nilpotent pure subgroups, and torsion bounds. The paper also gives explicit conjectures and tools, especially the affine Gauss diagram group, that could be of independent interest. However, the current manuscript contains foundational presentation errors and an incomplete injectivity proof, so the significance of the announced results cannot yet be assessed. No machine-checked proofs or code are included, and the correctness rests entirely on the manuscript's arguments.

major comments (4)

[Definition 2.1] For n≥3 the presentation of W(~A_n) is internally inconsistent. The commuting relation ρ_iρ_j=ρ_jρ_i for |i-j|≥2 includes the pair (1,n), since |1-n|≥2 for n≥3, so ρ_1ρ_n=ρ_nρ_1. Substituting this into the wrap-around braid relation ρ_1ρ_nρ_1=ρ_nρ_1ρ_n gives ρ_1=ρ_n, which collapses the group. Thus W(~A_n) is not the affine Weyl group of type ~A_n as defined, and Theorem A cannot hold as stated. This needs to be repaired with a consistent affine-type presentation, and all later uses of W(~A_n) must be re-examined.
[Theorem 2.7] The proof of the isomorphism AJ_n ≃ C_{W(~A_n)} does not account for singleton parabolic subgroups. The set P^{ir,f}(S) contains {ρ_i} for each i, so the generalized cactus group has generators σ_{ρ_i} that have no counterpart among the generators σ_{i,j} of AJ_n. The step 'we set i=i_1, i+1=i_2, ..., j=i_k' fails for k=1, and the asserted bijection between generators is therefore not a bijection. The author must either prove that all σ_{ρ_i} are trivial or redundant in C_{W(~A_n)}, or explicitly restrict the family of parabolic subsets used to define the generalized cactus group; otherwise the claimed isomorphism is unsupported.
[Section 1, Eq. (1.0.3)] The displayed formula for s_{k,l} is not a permutation for wrapping circular intervals. For example, when n=4 and [k,l]_c=[4,2]_c, the formula gives s_{4,2}(1)=5, which is not an element of {1,2,3,4}; hence the map is not a permutation of [1,n]. Since s_{k,l} is used in the defining relation (1.0.3) and in the morphism π of Proposition 3.1, this invalidates the presentation of AJ_n and the verification that π is well defined. A correct modular cyclic reversal formula must be supplied and the proof of Proposition 3.1 redone.
[Theorem 3.4, proof] The injectivity proof of ϕ is not complete. After observing that τ=1 and choosing a sequence of commutations and cancellations reducing τ to the empty word, Lemma 3.3 only justifies one adjacent move at a time. Following a quasi-commutation, the σ-word is not the original word with two letters interchanged: one generator is replaced by its image under a permutation, so the correspondence between letters of the σ-word and letters of the τ-word is disturbed, and later moves in the chosen τ-reduction need not lift to legal moves in the new σ-word. The sentence 'continuing this process, we deduce that σ=1' does not supply the required induction invariant. A formal induction on the reduction sequence, with an explicit invariant relating the current σ-word to the current τ-word under the S_n-action, is needed. Since Corollary 4.1, Corollary 3.6, and Theorems 4.10–4.16 all rely on this injectivity, this gap is load-bearing.

minor comments (5)

[Abstract] The abstract contains the typo 'world problem'; it should read 'word problem'.
[Figure 3] Figure 3 appears to depict an n-vertex path, not the standard Coxeter diagram of affine type ~A_n, which is a cycle on n+1 vertices. If the author intentionally uses a non-standard presentation, this must be explained.
[Section 1, notation] The notation [i,j]_c uses strict cyclic order, but the membership condition in the formula for s_{k,l} relies on i∈[k,l]_c without clarifying boundary conventions; a precise modular definition would resolve this.
[Definition 4.2] The phrase 'subject to the relations of AJ_n involving only these elements' is ambiguous and should be stated as an explicit presentation, because it is not automatic that an arbitrary subset of relations defines a subgroup.
[Theorem 4.12 proof] In the first sentence of the proof, the phrase 'let σ∈J_n' should presumably be 'let σ∈AJ_n'; otherwise the proof is inconsistent with the statement.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the affine-cactus embedding is constructed from explicit maps and standard external Coxeter facts; the main proof gaps are omissions, not self-referential reductions.

full rationale

The derivation chain is: (1) Theorem A identifies AJ_n with the generalized cactus group C_W(tilde-A_n) by matching generators and relations; (2) the diagram group AD_n and the explicit homomorphism phi are defined; (3) Theorem 3.4 attempts to prove phi injective; (4) linearity, word problem, centre, and torsion bounds are derived. None of these steps is circular: no parameter is fitted, no prediction is renamed, and no conclusion is assumed as a hypothesis. The map phi(sigma_{i,j}) = (tau_{[i,j]_c}, s_{i,j}) is given explicitly, and its well-definedness is checked directly against the defining relations of AJ_n. The cited support for the auxiliary facts is external and independent: Runze-Yu's linearity of generalized cactus groups [12], Bourbaki's Coxeter-group facts [6], and Mostovoy's Gauss-diagram embedding [10]. The only author self-citation, Bellingeri-Chemin-Lebed [2], appears in a comparison sentence before Theorem D ('results are similar to those obtained for Jn by ... [2]') and is not load-bearing. Two rigor gaps should be flagged separately from circularity. First, in the proof of Theorem 3.4, the sentence 'continuing this process, we deduce that sigma = 1' is an unformalized induction: after a quasi-commutation the sigma-word changes one generator, so one needs an invariant showing the correspondence with the reduced tau-word is preserved. This is an omitted proof, not a circular reduction, because the argument does not use the target theorem as an input. Second, the assertion after Theorem 4.13 that every torsion word can be rewritten 'by (quasi-)commuting its letters in such a way that the number of intersecting strands never increases' is used in Lemma 4.14 and Theorems 4.15-4.16 but is not proved where stated. There is also a likely typo in the displayed formula for s_{k,l} for wrapping circular intervals. These are correctness concerns, not circularity. Overall the paper's central claim has independent content and the circularity burden is low.

Assumptions & free parameters 0 free parameters · 2 assumptions · 1 invented entities

No free parameters are fitted. The paper relies on standard facts about Coxeter groups and right-angled Coxeter groups, plus the intended standard presentation of the affine Coxeter group, which the printed text gets wrong.

assumptions (2)

standard math Finite irreducible parabolic subgroups of the affine symmetric group type ~A_n are exactly the connected subdiagrams of the cycle, i.e., consecutive intervals, and their longest elements act by reversing the interval.
Invoked in the proof of Theorem 2.7 to identify generators σ_{i,j} with σ_{ρ_i,...,ρ_j}.
standard math In a right-angled Coxeter group, any word representing the identity can be transformed to the empty word by commuting adjacent letters and cancelling adjacent identical letters.
Used in the proof of Theorem 3.4 to reduce a word τ = 1 in AD_n and then lift the moves to AJ_n.

invented entities (1)

AD_n, the affine Gauss diagram group
purpose: Provides a right-angled Coxeter target for the embedding of AJ_n.
AD_n is a newly defined group in this paper (generated by circular sets with commutation for nested or disjoint sets). It has a formal presentation, but no external evidence beyond the paper's own constructions.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Combinatorics of affine cactus groups." pith.science (2026). https://pith.science/paper/3HMBS74O

@misc{pith2026250116270,
  author       = {Pith},
  title        = {Pith review of: Combinatorics of affine cactus groups},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3HMBS74O}},
  note         = {Machine review of arXiv:2501.16270}
}

read the original abstract

This article deals with the study of affine cactus groups from a combinatorial point of view. Those groups are extensions of cactus groups, which are related to braid and diagram groups and have gained an important place in many mathematics topics. We first show that affine cactus groups may be described as cactus groups on Coxeter groups of type eAn. Then, we prove that these groups embed into a semi-direct product of Coxeter groups, which allows us to obtain a number of combinatorial properties of affine cactus groups, such as the solubility of the world problem or the fact that their centre is trivial.

Figures

Figures reproduced from arXiv: 2501.16270 by the authors.

**Figure 2.** Examples of relations in affine cactus groups The kernel of π is called the pure affine cactus group and is denoted by P AJn. Losev [9] and Bonnaf´e [5] define the notionof generalized cactus group over a Coxeter system (W, S), denoted by CW , as the group generated by {σI , I ∈ Pif f (S)} subject to the following relations: σ 2 I = 1 for I ∈ Pif f (1.0.4) (S), (1.0.5) σIσJ = σJ σI if I ∩ J = ∅, σIσJ = σJ σωJ (I) (1… view at source ↗

**Figure 3.** We will show that the affine cactus group is isomorphic to the generalized cactus group over the affine symmetric group. To do this, we start by recalling some definitions and properties of a Coxeter system (W, S), where S = {s1, . . . , sn} is finite. An interesting fact about Coxeter groups is that they are finite if and only if their Coxeter diagram is of type An, Bn, Dn, E6, E7, E8, F4, G2, H3, H4 or Ip ([6]). A… view at source ↗

Figures from the paper (3 more)

**Figure 3.** Figure 3: Coxeter diagram of type Ae Theorem 2.5 ([6]). Let I = {si1 , . . . , sik } ∈ Pf (S). There exists a unique word of maximal length in WI , denoted by ωI . The element ωI acts on the subsets J of S by conjugacy. On the level of Coxeter diagrams, ωI acts by reversing the …

**Figure 4.** Figure 4: Action of ω{2,3,4,6,7} on the graph of a Coxeter system (W, {1, . . . , 10}) Definition 2.6 ([8],[5]). Let (W, S) be a Coxeter system. The generalized cactus group over W is the group CW with the following presentation : Generators : σI , for I ∈ Pir,f (S). Relations :…

**Figure 5.** Figure 5: Coxeter diagram of (WI , I) and it therefore acts on the Coxeter diagram of (WI , I) in the same way as si,j acts on [i, j]c. Conversely, for a given si,j , we set I = {ρi , . . . , ρj} ∈ Pir,f (S). The action of ωI on the Coxeter diagram of (WI , I) corresponds to the…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

12 extracted references · 11 canonical work pages

[1]

P; Bellingeri and A. Bodin. The braid group of a necklace. Mathematische Zeitschrift , 283:995–1010, 2016

work page 2016
[2]

Bellingeri, H

P. Bellingeri, H. Chemin and V. Lebed. Cactus groups, twi n groups, and right-angled Artin groups. Journal of Algebraic Combinatorics, 59:153–178, 2024

work page 2024
[3]

Right-angled Artin groups are symmetric diagram groups

P. Bellingeri and A. Genevois. Right-angled Artin group s are symmetric diagram groups. arXiv: 2305.11810 , 2023

work page Pith review arXiv 2023
[4]

Trickle groups

P. Bellingeri, E. Godelle and L. Paris. Trickle groups. arXiv:https: 2412.04932 , 2024

work page Pith review arXiv 2024
[5]

Bonnaf´ e

C. Bonnaf´ e. Cells and cacti. International Mathematics Research Notices , 2016:5775–5800, 2016

work page 2016
[6]

Bourbaki

N. Bourbaki. Groupes et alg` ebres de Lie : Chapitres 4 ` a 6 . Springer London NetLibrary, Inc., Guildford, Boulder , OCLC: 467784004, 2007

work page 2007
[7]

D.S. Farley. The planar pure braid group is a diagram grou p. arXiv: 2109.02815 , 2021

work page Pith review arXiv 2021
[8]

A. Ilin, J. Kamnitzer, Y. Li, P. Przytycki and L. Rybnikov . The moduli space of cactus ﬂower curves and the virtual cactus group. arXiv: 2308.06880 , 2023

arXiv 2023

Show all 12 references

[9]

I. Losev. Cacti and cells Journal of the European Mathematical Society , 21:1729–1750, 2019

2019
[10]

Mostovoy

J. Mostovoy. The pure cactus group is residually nilpot ent. Archiv der Mathematik , 113:229–235, 2019

2019
[11]

Mostovoy

J. Mostovoy. Round twin groups on few strands. arXiv: 2303.10737 , 2023

2023
[12]

R. Yu. Linearity of generalized cactus groups. Journal of Algebra , 635:256–270, 2023. Normandie Univ., UNICAEN, CNRS, LMNO, 14000 Caen, France

2023

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

P; Bellingeri and A. Bodin. The braid group of a necklace. Mathematische Zeitschrift , 283:995–1010, 2016

work page 2016

[2] [2]

Bellingeri, H

P. Bellingeri, H. Chemin and V. Lebed. Cactus groups, twi n groups, and right-angled Artin groups. Journal of Algebraic Combinatorics, 59:153–178, 2024

work page 2024

[3] [3]

Right-angled Artin groups are symmetric diagram groups

P. Bellingeri and A. Genevois. Right-angled Artin group s are symmetric diagram groups. arXiv: 2305.11810 , 2023

work page Pith review arXiv 2023

[4] [4]

Trickle groups

P. Bellingeri, E. Godelle and L. Paris. Trickle groups. arXiv:https: 2412.04932 , 2024

work page Pith review arXiv 2024

[5] [5]

Bonnaf´ e

C. Bonnaf´ e. Cells and cacti. International Mathematics Research Notices , 2016:5775–5800, 2016

work page 2016

[6] [6]

Bourbaki

N. Bourbaki. Groupes et alg` ebres de Lie : Chapitres 4 ` a 6 . Springer London NetLibrary, Inc., Guildford, Boulder , OCLC: 467784004, 2007

work page 2007

[7] [7]

D.S. Farley. The planar pure braid group is a diagram grou p. arXiv: 2109.02815 , 2021

work page Pith review arXiv 2021

[8] [8]

A. Ilin, J. Kamnitzer, Y. Li, P. Przytycki and L. Rybnikov . The moduli space of cactus ﬂower curves and the virtual cactus group. arXiv: 2308.06880 , 2023

arXiv 2023