Cacti, Toggles, and Reverse Plane Partitions

Balazs Elek; Devin Brown; Iva Halacheva

REVIEW 2 major objections 3 minor 18 references

Cacti, Toggles, and Reverse Plane Partitions

T0 review · 2 major / 3 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read This paper proves that on the crystal $B(n\varpi_1)$ of type $D_m$, the cactus group action is generated by elements attached to subdiagrams of length one and two, and identifies each single-node cactus generator with an explicit product…

desk verdict Proves a real conjecture, but the toggle description in §5 is unproved and the base-case argument for Prop 6.1 uses a false uniqueness claim; the braid issue is minor by comparison. read the letter →

arxiv 2412.02614 v1 pith:NUDLITJS submitted 2024-12-03 math.CO math.RT

Devin Brown , Balazs Elek , Iva Halacheva This is my paper · ORCID

classification math.COmath.RT MSC 05E1017B3720F55

keywords cactusgroupcrystalbasesreverseplanepartitionstogglestypeDminusculerepresentationsSchützenbergerinvolutionheaps

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 cactus group is a group of involutions acting on crystals, combinatorial models of representations of Lie algebras. In type $A$, this action is known to be governed by Bender-Knuth involutions. This paper extends that description to the family of crystals $B(n\varpi_1)$ in type $D_m$: the cactus action is shown to coincide with the action of certain toggles on reverse plane partitions of height $n$. The main result is that every cactus generator attached to a subdiagram acts as an explicit product of toggles, and that the whole cactus action is generated by elements coming from one- and two-node subdiagrams only. This confirms the conjecture stated in the paper's introduction for this crystal family.

What carries the argument

The central objects are the toggle operators $t_i$ acting on the order ideals of the heap $H(w)$, extended to reverse plane partitions by applying $t_i$ independently to each layer of the chain of order ideals; a heap is the poset attached to a reduced word in the Weyl group. The load-bearing identity is the recursive toggle $r_k$, defined by $r_m=t_m$, $r_{m-1}=t_{m-1}$, and $r_k=r_{k+1}t_k r_{k+1}t_k r_{k+1}$ for $1\le k\le m-2$, together with the statement that the single-node cactus generator $c_k$ acts identically to $r_k$ on $B(n\varpi_1)$. This reduces the cactus action to toggle computations on reverse plane partitions. Theorems 6.5 and 6.6 then express the interval and spin subdiagram generators as products of these single-node actions, which makes the generation by length-one and length-two diagrams follow.

What would settle it

Compute, on every element of $B(2\varpi_1)$ in type $D_4$, the two sides of the braid relation $c_1 c_2 c_1 = c_2 c_1 c_2$; a single mismatch would break the step in Theorem 6.6 and refute the claimed generation result.

Watch

Extended reading notes

Core claim

For the minuscule weight $\varpi_1$ in type $D_m$, the crystal $B(n\varpi_1)$ is isomorphic to the set of reverse plane partitions of shape $H(w_0^J)$ and height $n$, where $w_0^J$ is the minimal-length representative of the longest element of the parabolic subgroup stabilizing $\varpi_1$. The paper proves that each cactus generator $c_J$ acts on this crystal as an explicit product of toggles and single-node cactus generators. For a single node $k$, $c_k$ acts as the toggle $r_k$ defined by $r_m=t_m$, $r_{m-1}=t_{m-1}$, and $r_k=r_{k+1}t_k r_{k+1}t_k r_{k+1}$ for $1\le k\le m-2$; for an interval $J=[i,j]$, $c_J$ acts as $(c_j t_{j-1}\cdots t_i)(c_j t_{j-1}\cdots t_{i+1})\cdots(c_j t_{j-1})c_j$; and for a spin subdiagram $J=\{j,\ldots,m-1,m\}$, $c_J$ acts as a product of single-node cactus generators with toggles inserted, a product that is independent of the chosen reduced word for the longest element of $W_J$. From these identities the authors conclude that the cactus action on $B(n\varpi_1)$ in type $D_m$ is generated by elements corresponding to length 1 and 2 subdiagrams.

Load-bearing premise

The argument depends on the fact that neighboring one-node cactus moves satisfy the same braid relation as the Weyl group simple reflections they mimic; the paper uses this without proof, and if it failed, the claimed independence of the spin subdiagram element from the chosen reduced word would collapse.

Editorial extensions

If this is right

If the paper's central claim is correct, the full cactus action on $B(n\varpi_1)$ in type $D_m$ can be computed by toggles alone, without constructing the ambient tensor product crystal.
The action is generated by the order-two involutions attached to one- and two-node subdiagrams, so any computation that respects the cactus relations can be carried out with a much smaller generating set.
The explicit toggle description of $c_k$ gives a direct way to compute the affine crystal operators $e_0$ and $f_0$ on the Kirillov-Reshetikhin crystal $B^{1,n}$ of affine type $D_m^{(1)}$, a connection the paper notes.
For the $\varpi_1$ family in type $D$, the conjecture stated in the introduction is fully settled; the remaining minuscule cases are the two spin nodes of type $D$ and the two minuscule weights of $E_6$ and one of $E_7$.
Because each toggle acts locally on the entries of a generalized tableau, the cactus involution on a tableau becomes an entry-wise operation that can be applied directly to the labels.

Reading between the lines

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

Inference: the same recursive toggle construction may transfer directly to the two spin-node crystals $B(n\varpi_{m-1})$ and $B(n\varpi_m)$, since the local heap shape around the spin nodes is the same $D_m$ interval used in this paper; the paper leaves these cases open.
Inference: if the braid relation for adjacent cactus generators, which the proof of Theorem 6.6 assumes without proof, is verified directly from the RPP model, the argument becomes self-contained and the generation statement would follow for the spin cases by the same product arguments.
Inference: the toggle realization embeds the cactus action into the toggle group on RPPs, so one could study how the cactus group sits inside the full toggle group and whether the length-two generators are in fact necessary or could be replaced by length-one generators alone.
Inference: the explicit formulas make the remaining minuscule cases in $E_6$ and $E_7$ computationally testable: enumerate the corresponding RPPs and check the same toggle identities on a small example, such as height $n=2$.

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

Let me give you the short version. The paper proves a real conjecture from Dranowski–Elek–Kamnitzer–Morton-Ferguson: for B(nϖ1) in type D, the cactus action factors through toggles and is generated by length-one and length-two subdiagram elements. The explicit toggle formula for the single-node cactus generator (Prop 6.1) and the product formulas for intervals and spin subdiagrams (Theorems 6.5, 6.6) are genuinely new and look correct in spirit. The RPP/heap framework is the right tool, and the paper is honest about what it borrows from [Dra+22]. No circularity.

The problem is that the toggle action on tableaux, stated in Section 5, is load-bearing and unproved. The description is garbled — the sentence for t_i literally repeats itself — and no derivation from the RPP bijection is given. Everything downstream, meaning the base cases of Prop 6.1, the eight-case induction, Lemmas 6.2–6.4, and the theorems, rests on that description. Worse, the base case proof of Prop 6.1 contains a false uniqueness claim: it says weight plus N(v), the total number of spin entries, determines all entry counts. That is not true. In D4, [1,\bar{1}] and [2,\bar{2}] have the same weight (0) and the same N (0), but different counts. So the base case as written is not justified.

The braid relation invoked in Theorem 6.6 is a different matter. For normal crystals, the single-node cactus generator acts as the Weyl simple reflection, and adjacent ones braid. That is standard, and a citation would fix it. The stress-test note is right to distinguish the two.

I also want to say the case checks are plausible. I did not find a contradiction, and the "remaining cases follow analogously" language is common, though for a proof this computational a Sage check or a table would make it much easier to trust.

So: this is a serious paper with a real result, but it is not yet referee-ready. The toggle action needs to be proved or cited, the base case of Prop 6.1 needs a correct argument, and the typos in Section 5 should be cleaned up. I would send it to a good combinatorics journal after major revision. For the reading group, I'd bring it—the gap is instructive.

Referee Report

2 major / 3 minor

Summary. The paper studies the action of the cactus group on the type D crystal B(nϖ1) via reverse plane partitions and toggles. The authors introduce toggles r_k defined inductively, prove that a single-node cactus generator c_k acts as r_k (Proposition 6.1), derive commutation and intertwining relations (Lemmas 6.2, 6.3, 6.4), and use these to show that interval subdiagram cactus generators act by explicit toggle products (Theorem 6.5) and that spin-containing subdiagram generators act by products of single-node generators (Theorem 6.6). As a corollary they conclude that the cactus action on B(nϖ1) is generated by elements corresponding to subdiagrams of length 1 and 2, confirming part of Conjecture 1.1 from [Dra+22].

Significance. If the main results hold, the paper provides a concrete and computable description of the cactus action on a non-type-A family of crystals, reducing the generating set to length-one and length-two subdiagrams. This is a substantive step beyond the type A case and would validate a conjecture from [Dra+22]. The paper's strengths include a clear structural strategy, explicit combinatorial computations in representative cases, and the use of the established RPP/crystal isomorphism. However, two load-bearing assertions are currently unsupported: the explicit description of the toggle action on tableaux in Section 5, and the braid relation for adjacent single-node cactus generators used in Theorem 6.6. Both need to be proved or precisely cited before the main claims are fully justified.

major comments (2)

[Section 5 (paragraph after Proposition 5.7)] The description of the toggle action on tableaux is stated without proof or citation, and this description is the computational basis for Proposition 6.1. For example, the claim that t_i merely swaps the counts of i and i+1 entries (and the barred analogues) is not routine in type D: in B(2ϖ1) of type D4, the tableau 1 1̄ lies on an sl2-string of length three, while the count-swapped tableau 2 2̄ is isolated under e_1 and f_1. Thus the toggle action must be derived from the RPP/heap bijection or supplied with a precise reference; otherwise the eight-case induction in Proposition 6.1 has no foundation.
[Theorem 6.6, proof, first paragraph] The proof asserts 'use the fact that adjacent single node cactus generators braid' to conclude that c_J ∼ c_{i1} ... c_{il} is independent of the reduced word for the longest element of W_J. This braid relation is standard for normal crystals when c_i is realized as the Weyl group simple reflection on the crystal, but the paper gives neither a proof nor a citation. If this relation failed, the exact form of Theorem 6.6 would not follow; the authors should either prove the braid relation in the crystal B(nϖ1) or cite a reference that establishes it in this context.

minor comments (3)

[Section 5, toggle description] There is a repeated fragment: 'the toggle t_i swaps the number of i fillings and i+1 fillings and swaps the number of i fillings and i+1 fillings'; the second clause should presumably read 'swaps the number of i fillings and i+1 fillings' only once, or should refer to barred entries consistently.
[Proposition 6.1 proof] The sentence 'So, we need to consider We use the previous cases...' is grammatically incomplete; it should be rephrased, e.g., 'So we need to consider eight cases, and we use the previous cases to compute...'
[Theorem 6.6 statement] The phrase 'then c_J ∼ c_{i1} · · · c_{il} in for the crystal B(nϖ1)' contains a typo: 'in for' should be 'in' or 'for'.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the cactus-to-toggle theorems are proved by explicit case computations, and the overlapping-author citations are to established prior theorems, not to the conjecture being proved.

full rationale

The derivation chain is self-contained in the relevant sense: Conjecture 1.1 is stated as an open conjecture from [Dra+22] that this paper proves rather than assumes, and the central comparisons c_k ∼ r_k, Lemma 6.2, Lemma 6.3, Lemma 6.4, Theorems 6.5, and Theorem 6.6 are established by explicit case-by-case computations on the same tableau count tuples (a,b,c,c̄,b̄,ā), not by invoking the target statement. The citations to [Dra+22] (Theorem 4.5 and Proposition 5.7) and to [HKRW20] (Theorem 5.4) are to independent, previously established results—the RPP model for B(nλ) and the existence and uniqueness characterization of the cactus action—and are not the conjecture being proved. Two support gaps are present but are not circularity: Section 5's description of the toggles on tableaux is stated without a proof or citation, and Theorem 6.6 uses without proof or citation the fact that adjacent single-node cactus generators braid; these are omitted justifications for the computation, not reductions of the conclusion to its inputs. Hence the circularity score is 0.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No free parameters are fitted and no new mathematical entities are introduced. The argument relies on standard crystal theory, the prior RPP/heap framework from [Dra+22], and two unproved-but-standard supporting facts flagged in red flags.

assumptions (3)

domain assumption The isomorphism B(nλ) ≅ RPP(w_J^0, n) compatible with tensor product inclusion.
Invoked in Section 5.7 and throughout Section 6; taken from [Dra+22, Theorem 2.29/4.5].
standard math The unique characterization of the cactus action by weight and by intertwining with e_i and f_i, and the braid relations of single-node cactus generators on normal crystals.
Used in Theorem 6.5 and Theorem 6.6; c_i is the Schützenberger involution on the sl2 subcrystal, which coincides with the Weyl group simple reflection on a normal crystal.
domain assumption The action of toggles on tableaux is the explicit swap-and-replacement rule stated at the end of Section 5.
Stated without proof; all case computations in Section 6 rely on this description of t_i.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Cacti, Toggles, and Reverse Plane Partitions." pith.science (2026). https://pith.science/paper/NUDLITJS

@misc{pith2026241202614,
  author       = {Pith},
  title        = {Pith review of: Cacti, Toggles, and Reverse Plane Partitions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NUDLITJS}},
  note         = {Machine review of arXiv:2412.02614}
}

abstract

The cactus group acts combinatorially on crystals via partial Sch\"utzenberger involutions. This action has been studied extensively in type $A$ and described via Bender-Knuth involutions. We prove an analogous result for the family of crystals $B(n\varpi_1)$ in type $D$. Our main tools are combinatorial toggles acting on reverse plane partitions of height $n$. As a corollary, we show that the length one and two subdiagram elements generate the full cactus action, addressing conjectures of Dranowski, the second author, Kamnitzer, and Morton-Ferguson.

Figures

Figures reproduced from arXiv: 2412.02614 by the authors.

**Figure 1.** The crystal B(̟1) in type Dm. Crystals have a particularly nice tensor product rule, which we recall next. Note that we use a different convention from Kashiwara’s convention (as in [HoKa02]), instead following [BS17, Section 2.3] and the tensor product rule implemented in SageMath [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗

**Figure 2.** H(w) and H(w ′ ) in type A3. Definition 3.3. An element w ∈ W is called fully commutative if any two reduced words for w can be obtained from each other only by commuting elements, e.g. s2s1s3s2 = s2s3s1s2 in A3. If w is fully commutative, then any two reduced words for w give rise to isomorphic posets, so we may refer to the heap H(w). We next recall a generalization of such heaps that captures the structure of cry… view at source ↗

**Figure 3.** An element Φ of RPP(w, 3) and the corresponding chain of order ideals. is of minimal length such that w J 0 λ = w0λ, and WJ = {v ∈ W : v ≤L w J 0 } where v ≤L w is the weak left order (i.e. there exist reduced expressions such that v appears as a terminal subword of w). Theorem 4.1 ([Ste96], Lemma 3.1, Theorem 6.1, Theorem 7.1). (1) If w ∈ W is fully-commutative, then {v ∈ W : v ≤L w} =∼ J(H(w)) as posets. (2) For λ… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

18 extracted references · 17 canonical work pages

[1]

E. A. Bender, and D. E. Knuth, Enumeration of plane partitions, J. Combinatorial Theory Ser. A 13 (1972), no. 1, 40--54

work page 1972
[2]

Berenstein, and A

A. Berenstein, and A. Kirillov, Groups generated by involutions, Gelfand-Tsetlin patterns, and combinatorics of Young tableaux, Algebra i Analiz 7 (1995), no. 1, 92--152

work page 1995
[3]

Bump, and A

D. Bump, and A. Schilling, Crystal Bases: Representations and Combinatorics, World Scientific Publishing, Hackensack, NJ. (2017)

work page 2017
[4]

Chmutov, M

M. Chmutov, M. Glick, and P. Pylyavskyy, The B erenstein- K irillov group and cactus groups , J. Comb. Algebra 4 (2020), no. 2, 111--140

work page 2020
[5]

Dranowski, B

A. Dranowski, B. Elek, J. Kamnitzer, and C. Morton-Ferguson, Heaps, crystals, and preprojective algebra modules, Sel. Math. New Ser. 30 (2024), no. 4, 94

work page 2024
[6]

Halacheva, Skew Howe duality for crystals and the cactus group

I. Halacheva, Skew Howe duality for crystals and the cactus group. (2020). https://arxiv:2001.02262

arXiv 2020
[7]

Halacheva, J

I. Halacheva, J. Kamnitzer, L. Rybnikov, and A. Weekes, Crystals and monodromy of Bethe vectors, Duke Math. J. 169 (2020), no. 12, 2337--2419

work page 2020
[8]

Henriques, and J

A. Henriques, and J. Kamnitzer, Crystals and coboundary categories, Duke Math. J. 132 (2006), no. 2, 191--216

work page 2006

Show all 18 references

[9]

Hong, and S.-J

J. Hong, and S.-J. Kang, Introduction to Quantum Groups and Crystal Bases, Graduate Studies in Mathematics, AMS, Providence, RI. 42 (2002)

2002
[10]

J. E. Humphreys, Introduction to Lie Algebras and Representation Theory, Graduate Texts in Mathematics, Springer Verlag, New York, NY. 9 (1972)

1972
[11]

Kashiwara, Crystalizing the q-analogue of universal enveloping algebras, Commun

M. Kashiwara, Crystalizing the q-analogue of universal enveloping algebras, Commun. Math. Phys. 133 (1990), no. 2, 249--260

1990
[12]

Kashiwara, On crystal bases of the q-analogue of universal enveloping algebras, Duke Math

M. Kashiwara, On crystal bases of the q-analogue of universal enveloping algebras, Duke Math. J. 63 (1991), no. 2, 465--516

1991
[13]

Kleschev, and A

A. Kleschev, and A. Ram, Homogeneous representations of Khovanov-Lauda algebras, J. Eur. Math. Soc. (JEMS) 12 (2010), no. 2, 1293-1306

2010
[14]

R. A. Proctor, Minuscule Elements of Weyl groups, the numbers game, and d-complete posets, J. Algebra 213 (1999), no. 1, 272-303

1999
[15]

J. R. Stembridge, On the fully commutative elements of Coxeter groups, J. Algebraic Combin. 5 (1996), no. 4, 353-385

1996
[16]

J. R. Stembridge, Minuscule elements of Weyl groups, J. Algebra 235 (2001), no. 2, 722-743

2001
[17]

J. R. Stembridge, A concise proof of the L ittlewood-- R ichardson rule , Electron. J. Combin 9 (2002), no. 1, Note 5

2002
[18]

Tingley, Three combinatorial models for sl _n crystals, with applications to cylindric plane partitions , Int

P. Tingley, Three combinatorial models for sl _n crystals, with applications to cylindric plane partitions , Int. Math. Res. Not. IMRN 2008 (2008), 2374–2375

2008

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [1]

E. A. Bender, and D. E. Knuth, Enumeration of plane partitions, J. Combinatorial Theory Ser. A 13 (1972), no. 1, 40--54

work page 1972

[2] [2]

Berenstein, and A

A. Berenstein, and A. Kirillov, Groups generated by involutions, Gelfand-Tsetlin patterns, and combinatorics of Young tableaux, Algebra i Analiz 7 (1995), no. 1, 92--152

work page 1995

[3] [3]

Bump, and A

D. Bump, and A. Schilling, Crystal Bases: Representations and Combinatorics, World Scientific Publishing, Hackensack, NJ. (2017)

work page 2017

[4] [4]

Chmutov, M

M. Chmutov, M. Glick, and P. Pylyavskyy, The B erenstein- K irillov group and cactus groups , J. Comb. Algebra 4 (2020), no. 2, 111--140

work page 2020

[5] [5]

Dranowski, B

A. Dranowski, B. Elek, J. Kamnitzer, and C. Morton-Ferguson, Heaps, crystals, and preprojective algebra modules, Sel. Math. New Ser. 30 (2024), no. 4, 94

work page 2024

[6] [6]

Halacheva, Skew Howe duality for crystals and the cactus group

I. Halacheva, Skew Howe duality for crystals and the cactus group. (2020). https://arxiv:2001.02262

arXiv 2020

[7] [7]

Halacheva, J

I. Halacheva, J. Kamnitzer, L. Rybnikov, and A. Weekes, Crystals and monodromy of Bethe vectors, Duke Math. J. 169 (2020), no. 12, 2337--2419

work page 2020

[8] [8]

Henriques, and J

A. Henriques, and J. Kamnitzer, Crystals and coboundary categories, Duke Math. J. 132 (2006), no. 2, 191--216

work page 2006