{"id":"ecef9091-392e-472d-a0cf-3631af6f5635","arxiv_id":"2505.14182","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For any braid, the outermost three T-degree layers of reduced triply-graded HOMFLY homology are determined explicitly, and are almost entirely zero with only a few one-dimensional k-modules.","lead":"Triply-graded link homology is a powerful but difficult knot invariant that refines the HOMFLY polynomial. This paper computes its top and bottom three grading levels for all braid closures and finds that almost everything vanishes except a few uniform pieces.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The reduction from arbitrary positive braids to those containing alternating length-4 subexpressions (Prop 2.4) is not fully proved; if it fails, Theorems C, D, and 1.2 no longer cover all braids.","rationale":"The reader identifies Proposition 2.4 as the weakest assumption, and I agree. It is the linchpin that converts the explicit three-strand and pattern-dependent computations into a result for all positive and negative braids. The proof is brief and leaves the most delicate part of the braid-move argument to intuition, so the central claim is not fully secured. No independent counterexample is known to me, and the rest of the paper is coherent: the Koszul-complex framework in Section 3 is plausible, the three-strand matrix computations in Section 4 are detailed, and the negative-braid results follow formally from Lemma 3.4 once the positive-braid reduction is accepted. I therefore do not move the verdict: the paper should remain CONDITIONAL, with Proposition 2.4 and the cited/asserted base cases needing expansion or independent verification before the unrestricted form of the Main Theorem is accepted.","tokens_in":25597,"tokens_out":15617,"duration_ms":152998,"concrete_test":"Enumerate all positive braid words in B_3 and B_4 up to length 12 (or a large random sample) that, for some adjacent pair i, contain neither σ_iσ_{i+1}σ_iσ_{i+1} nor σ_{i+1}σ_iσ_{i+1}σ_i as a subexpression. For each such word, compute the closure and test whether it is prime and has positive braid index n, using SnapPy/KnotTheory` and Birman-Menasco-style braid-index bounds. A single prime non-split example would disprove Proposition 2.4; if no counterexample appears, the proposition becomes much more credible and the remaining defect is the missing rigorous proof.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The Main Theorem's claim to handle any positive or negative braid depends on Proposition 2.4: a braid β with minimal positive braid index n whose closure is not a connect sum must contain σ_iσ_{i+1}σ_iσ_{i+1} or σ_{i+1}σ_iσ_{i+1}σ_i for every i. Without this, Theorems C and D apply only to braids that already contain the special length-4 subexpressions, so the universe of braids covered is strictly smaller than claimed. The proof of Proposition 2.4 is a sketch: after a sequence of braid moves destroys the property for some k0, it analyzes only the 'final move' of type (A) or (B) and asserts that the surrounding letters contain no s or t. It does not rigorously rule out sequences where the final move involves several adjacent pairs at once, nor does it justify the claim that a violation for k0 can be introduced only by such a move. Lemma 2.3 also checks only the sstt case explicitly and handles stts/tsst by a one-line cyclic-permutation argument. The base-case formulas in Section 2.5 are likewise asserted without proof or citation, but the reduction via Proposition 2.4 is the more load-bearing gap for the 'any braid' statement.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper computes the reduced triply-graded link homology HHH as an R-module in the three top (resp. bottom) T-degrees for closures of arbitrary positive (resp. negative) braids. For positive braids, it claims that T=|β| is concentrated in A=0 degree k(|β|), that T=|β|-1 vanishes, and that T=|β|-2 is k(|β|-4) at A=0, k(|β|) at A=1, and zero for A≥2, under a connectivity condition. For negative braids on n≥3 strands, the two lowest degrees vanish; for 3-strand negative braids there is a nontrivial class at T=-|α|+2, A=2. The proofs use diagrammatic Hochschild cohomology of Soergel bimodules and a reduction (Proposition 2.4) to braids containing alternating length-4 subexpressions.","tokens_in":25804,"tokens_out":5926,"duration_ms":56655,"significance":"If the computations are correct, the paper establishes a striking uniformity in the top T-degrees of triply-graded link homology and gives the first such computations for arbitrary braids, with applications to necessary conditions for positive/negative braid links and to a possible spectral sequence to the (2,3)-torus knot. The explicit kernel/image matrices in Theorem 4.1 and the emphasis on R-module structure (rather than only vector-space grading) are concrete strengths, and the main algebraic results are derived by direct computation rather than by fitting parameters. However, several reduction and base-case proofs are too compressed to verify the full scope of the claims, so the main theorem is plausible but not fully established in the current text.","major_comments":[{"comment":"The proof of the reverse implication is not sufficient to justify that every positive braid of minimal strand count contains σ_iσ_{i+1}σ_iσ_{i+1} or σ_{i+1}σ_iσ_{i+1}σ_i as a subexpression for every i. The argument analyzes only the 'final move' (A)/(B) in a sequence of braid moves and asserts without proof that the surrounding letters contain no s or t; it does not rule out final moves that involve several adjacent pairs at once, nor does it prove that a violation for a fixed k0 can be introduced only by moves (A)/(B). The minimality argument involving Markov move II also needs a more detailed justification. Since Proposition 2.4 is what allows Theorems C, D, and the Main Theorem to apply to all positive/negative braids rather than only to those already containing the special subexpressions, a complete proof is load-bearing.","section":"Proposition 2.4"},{"comment":"The case analysis is incomplete: the proof explicitly treats the sstt subexpression and then asserts that the cases stts and tsst can be reduced to sstt by Markov move I, but a cyclic permutation changes the cyclic word and does not obviously preserve the non-connect-sum hypothesis while converting the subexpression. Since Lemma 2.3 feeds directly into Proposition 2.4, this gap propagates to the claimed scope of the Main Theorem.","section":"Lemma 2.3"},{"comment":"The formulas for HHH(σ_1^m) and HHH(σ_1^{-m}) in Section 2.5 are stated as facts without proof or reference. These formulas are the n=2 base case of the Main Theorem and are also used in the applications in Theorem 1.2. Without a derivation or a citation, the claim that the Main Theorem covers the n=2 case is unsupported.","section":"Section 2.5"},{"comment":"The proof is compressed to the assertions that 'the kernels are fairly easy to work out' and that the only nonzero cohomology is generated by a single displayed element. Since this theorem is one of the main results (the negative 3-strand case), the reader cannot verify the cancellation of all other cohomology groups from the material given. A complete kernel/image computation, or at least a detailed indication of how Lemma 3.4 dualizes the explicit matrices of Theorem 4.1, is needed.","section":"Theorem 4.2"},{"comment":"The proof outlines the structure of the relevant matrices but relies on 'as a result' and 'it is then clear that this is the image of HH^k(d_{|β|-2})^T' after a single illustrative example. The definitions of N, ℓ, and the block decomposition are not fully justified, and the final identification of the kernel of the transposed differential with the image of HH^k(d_{|β|-2})^T is asserted rather than shown. Given that Theorem D is a central component of the Main Theorem for n≥4, this proof needs to be written out in more detail.","section":"Theorem D"}],"minor_comments":[{"comment":"Both statements quantify 'for all 1 ≤ i ≤ n - 1' where σ_iσ_{i+1} appears; this should be '1 ≤ i ≤ n - 2', since σ_{i+1} is undefined for i = n-1.","section":"Theorem C and Theorem D"},{"comment":"The distinction between 'subword' and 'subexpression' is confusing; the example says 'there is only one subword of length 1 of ss while there are 2 possible subexpressions', but the earlier definition of subexpression allows repeated letters. These terms should be defined explicitly and used consistently.","section":"Section 2.3"},{"comment":"The displayed generator for the nonzero cohomology class in Theorem 4.2 is difficult to parse in text form; a diagram or an explicit vector in the basis from Appendix A would improve the presentation.","section":"Section 4.1, Theorem 4.2"},{"comment":"The sentence 'Note such a braid representative also exists for T(2,k) as σ_1^k ∼ σ_1^k σ_2 ∼ ...' is not fully argued; the chain of equivalences is unclear and should be expanded or justified.","section":"Section 1.2.1, proof of Theorem 1.2(2)"},{"comment":"The notation alternates between \\overline{\\mathrm{HHH}} (abstract) and HHH (body). Please standardize the notation.","section":"Abstract and Section 1"}],"recommendation":"major_revision","confidential_remarks":"The paper's main theorems are plausible, and the explicit 3-strand computation in Theorem 4.1 is a solid piece of work. However, the breadth of the claims (all braids) rests on Proposition 2.4 and Section 2.5, which are not fully proved, and the proofs of Theorems 4.2 and D are too sketchy. I would be willing to look at a revised version that supplies complete proofs for these load-bearing reduction and duality steps."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper does something genuinely new: it determines the top three and bottom three T-degrees of reduced triply graded homology for closures of arbitrary positive and negative braids, as R-module isomorphisms, not just ranks. That goes beyond the torus-link results of EH19/HM19 and the 11-crossing algorithm of NS24. The uniformity in the answers is striking and the methods are explicit, not black-box fitting. The negative-braid results via Serre duality are also new.\n\nThe positive 3-strand computation (Thm 4.1) is the heart, and it is written out with matrices and kernel/image analysis. I believe it. The reduction of general n to the 3-strand case in Section 5 is coherent, and Lemmas 5.1 and 5.3 give the right formal framework. The core mathematics has a good chance of being correct.\n\nWhere I would push back: the proof of Proposition 2.4, which is the sole reason the main theorem covers arbitrary braids rather than only braids already containing stst/tsts, is only a sketch. The stress-test note is right that the 'final move' argument does not rigorously rule out sequences where the final braid move affects several adjacent pairs at once, and the handling of the k0=2 and k0=n-1 exceptions is too compressed. If Prop 2.4 fails, Theorems C, D, and 1.2 apply to a strictly smaller class of braids. That does not sink the paper, but the title and abstract promise more than the proof currently delivers.\n\nOther soft spots are in presentation, not substance. Theorem 4.2's kernel computation is dismissed as 'fairly easy' with only a generator noted; Theorem D's proof is a terse block-matrix sketch; and the base-case formulas for HHH(T(2,±k)) in Section 2.5 are asserted without proof or citation. These are fixable by expansion. The black-box use of [Li22] is partially mitigated by the appendix, but the appendix's basis derivation is itself not independently checked.\n\nOverall: a serious computational paper with a plausible central result and real novelty. The main weaknesses are a load-bearing lemma that needs a fuller proof and several compressed arguments. I would send it to a knowledgeable referee, explicitly asking them to verify Prop 2.4 and Theorem 4.2. I would cite it once the proof details are solidified.","headline":"Computes the extreme T-degree part of HHH for all positive/negative braid closures as R-modules; the core computations look plausible, but the 'all braids' claim rests on a sketchy braid-word lemma and a few compressed proofs.","tokens_in":26393,"tokens_out":1911,"would_cite":true,"duration_ms":19970,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":null,"created_at":"2026-08-07T15:39:12.555682+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":null,"supporting_citations":[],"review_version":1}