{"id":"ffa07ff8-4c3c-4676-bd80-7919a7220a56","arxiv_id":"2412.02614","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"In type D, the cactus group action on the crystal B(nω1) is generated by length-one and length-two subdiagram elements and factors through the toggle group.","lead":"This paper proves that a group action from Lie theory, the cactus group, can be described using simple toggle moves on stacks of labelled beads called reverse plane partitions. It resolves a conjecture in type D and reduces the generating set for the action to just one-node and two-node pieces.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The paper's central computation rests on the unproved toggle action asserted in §5; this description is load-bearing, while the braid relation flagged by the reader is a standard normal-crystal fact. The proof of Prop 6.1 should be conditional on a verified toggle action.","rationale":"The reader's CONDITIONAL verdict is appropriate, but the primary missing support is the §5 toggle description, not the braid relation. The braid relation is a known property of the Weyl-group action on normal crystals; the paper's reliance on it is under-cited but not a correctness risk. The toggle description, however, is asserted in a garbled sentence and is used directly in every computation. The r_k construction would be unnecessary if t_i were merely c_i, confirming that the toggle action is genuinely subtle. I found no contradiction with the claimed results, and the case analyses are plausible, but the proof is not self-contained. A direct small-case check would settle the matter, so the conditional verdict should stand pending that check.","tokens_in":14122,"tokens_out":37589,"duration_ms":349380,"concrete_test":"Compute, by hand or with a computer, the toggle t_1 defined in Definition 5.5 on the reverse plane partition of shape H(w_J^0) (type D_4, n=2) that corresponds under Theorem 4.5 to the SSYT 1 1̄. Translate the resulting RPP back to an SSYT and compare with the §5 predicted image 2 2̄. Any disagreement invalidates the §5 description and forces a recheck of Proposition 6.1's base cases and inductive step.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 5 asserts, without proof or citation, that the toggle t_i on B(nϖ1) (type D_m) acts on SSYT by swapping the counts of i and i+1 (and of the barred entries), with a special rule for the spin nodes. This is the computational input to Proposition 6.1: the base cases for c_m and c_{m−1} and the entire eight-case induction for c_k ∼ r_k are written in terms of these count-tuples. The claim is not routine. In type D, simply exchanging counts cannot, by itself, equal the single-node cactus generator c_i, because the toggle may move an element out of its i-string component. For example, in B(2ϖ1) of type D_4, the tableau 1 1̄ lies on an sl2 string of length three, whereas the count-swapped tableau 2 2̄ is isolated under e_1 and f_1. The definition of r_k as a product of toggles is precisely designed to repair this, so the entire paper depends on knowing the actual toggle action. The text's wording is also garbled ('swaps the number of i fillings and i+1 fillings and swaps the number of i fillings and i+1 fillings'), and no derivation from the RPP/heap bijection is supplied. If this description is wrong, the tuple formulas in Prop 6.1 and the subsequent Theorems 6.5/6.6 have no foundation. The braid relation used in Theorem 6.6, by contrast, is standard for normal crystals (c_i equals the Weyl-group simple reflection), though it should be cited.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","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].","tokens_in":14434,"tokens_out":1883,"duration_ms":19629,"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":[{"comment":"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.","section":"Section 5 (paragraph after Proposition 5.7)"},{"comment":"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.","section":"Theorem 6.6, proof, first paragraph"}],"minor_comments":[{"comment":"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.","section":"Section 5, toggle description"},{"comment":"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...'","section":"Proposition 6.1 proof"},{"comment":"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'.","section":"Theorem 6.6 statement"}],"recommendation":"major_revision","confidential_remarks":"The two major concerns are both fixable in principle: the toggle action can be proved from the RPP bijection (or supplied with a reference), and the braid relation is standard but must be explicitly cited. If those are addressed, the paper would be a solid contribution. One additional editorial point: the proof of Proposition 6.1 relies on many 'analogous' cases; the authors might consider including a verification script or a more complete case analysis to increase confidence, though this is not a mathematical objection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","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.\n\nThe 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.\n\nThe 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.\n\nI 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.\n\nSo: 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.","headline":"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.","tokens_in":14972,"tokens_out":9535,"would_cite":true,"duration_ms":89998,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05E10","17B37","20F55"],"pacs":[],"model":"deepseek-v4-flash","headline":"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…","keywords":["cactus group","crystal bases","reverse plane partitions","toggles","type D","minuscule representations","Schützenberger involution","heaps"],"falsifier":"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.","tokens_in":13909,"feed_emoji":"💎","tokens_out":17320,"duration_ms":142758,"temperature":0.7,"pith_summary":"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.","feed_headline":"Small diagrams generate the full cactus action on type-D crystals","feed_subtitle":"Every cactus move on B(nϖ1) in type D becomes a simple toggle on reverse plane partitions.","key_machinery":"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.","core_discovery":"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.","pith_inferences":["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 $n=2$."],"forward_implications":["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."],"supporting_citations":[{"why":"Supplies the reverse plane partition model for minuscule crystals, the toggle action, and the conjecture this paper proves for $B(n\\varpi_1)$.","marker":"[Dra+22]"},{"why":"Establishes the cactus group action on crystals via partial Schützenberger involutions, the action under study.","marker":"[HKRW20]"},{"why":"Gives the heap model for fully commutative Weyl-group elements and the order ideal isomorphism underlying RPPs.","marker":"[Ste96]"},{"why":"Characterizes the local structure of heaps of minuscule elements, including the $D_m$ interval shape used in the theorem statements.","marker":"[Ste01]"},{"why":"Provides the sign-pattern tensor product rule used throughout the case computations comparing $c_k$ with $r_k$.","marker":"[Tin08]"},{"why":"Introduces the cactus group action on crystals and the coboundary framework this paper builds on.","marker":"[HeKa06]"}],"fun_headline_variants":["Tiny toggles reveal full cactus action on type-D crystals","Length-1 and -2 subdiagrams generate type-D cactus action","Small subdiagrams generate cactus action in type D","Toggles show small subdiagrams run cactus action"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"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.","fun_headline_variants_meta":{"raw":{"variants":["Tiny toggles reveal full cactus action on type-D crystals","Length-1 and -2 subdiagrams generate type-D cactus action","Small subdiagrams generate cactus action in type D","Toggles show small subdiagrams run cactus action"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000785,"raw_usage":{"total_tokens":3479,"prompt_tokens":978,"completion_tokens":2501,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":594,"completion_tokens_details":{"reasoning_tokens":2443}},"tokens_in":594,"tokens_out":2501,"duration_ms":16817,"temperature":1.0,"reasoning_tokens":2443,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T23:16:21.833540+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"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.","supporting_citations":[],"review_version":1}