{"id":"5fbb3c46-fedf-4f0e-a76d-9e2b80e5103e","arxiv_id":"1908.06878","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Links in genus-g handlebodies are assigned a triply-graded homology built from singular Soergel bimodules and Hochschild cohomology, generalizing colored HOMFLYPT homology.","lead":"This paper constructs a triply-graded invariant of links inside a genus-g handlebody, a 3D space with g holes, extending the colored HOMFLYPT homology of links in ordinary 3-space. It matters because it is the first such invariant for handlebodies and it can distinguish handlebody links that standard 3-space invariants conflate.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The handlebody-sensitivity claim (Prop. 4.8 / Cor. 4.13) rests on an unstated HOMFLYPT computation; if that polynomial comparison is wrong, the claimed non-invariance under classical conjugation collapses.","rationale":"I read the construction in good faith. The main invariance theorem is built on standard machinery and, aside from compressed quasi-isomorphism arguments, I found no internal inconsistency in the stabilization proof. The reader's weakest_assumption focuses on degree shifts in Lemma 4.12 and the normalization (4-45); that is a legitimate place to worry, but I did not find a concrete error there. The more vulnerable spot is Proposition 4.8, which the paper itself flags only as 'a computation shows'. That computation is the sole evidence for the claim that the invariant is not just the classical colored HOMFLYPT of the S^3 closure of the non-core strands. If the polynomial difference is wrong, the invariant could still be well-defined, but the paper's advertised handlebody sensitivity would fail. The reader notes this as a caveat but does not elevate it to the load-bearing concern; I do, because the central novelty explicitly relies on it. A conditional acceptance requiring the displayed computation, or an independent check, is therefore appropriate.","tokens_in":24523,"tokens_out":26148,"duration_ms":299414,"concrete_test":"Draw the two closures in (4-12) for M=1, g=2, n=1 and compute their reduced HOMFLYPT polynomials using the skein relation obtained by decategorifying (3-21), e.g. with a computer algebra system. Verify that the difference is exactly (a-a^{-1})^2-(q-q^{-1})^2; if it is not, Proposition 4.8 and the second assertion of Corollary 4.13 are unsupported. As a second check, evaluate the Euler characteristics of HH_Hg(t2 t1) and HH_Hg(t1 t2) at M=1 directly from (4-39)-(4-45) and compare them.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Proposition 4.8 is the only evidence that HHH_Hg is genuinely sensitive to the handlebody, rather than being a pullback of the classical S^3 invariant. Its proof asserts that HH_Hg(t2 t1) and HH_Hg(t1 t2) cannot be homotopy equivalent because their Euler characteristics would then force equality of the HOMFLYPT polynomials of the two closures shown in (4-12), and 'a computation shows' the difference is (a-a^{-1})^2-(q-q^{-1})^2. No computation is displayed, and the step from Euler characteristics of HH_Hg to HOMFLYPT polynomials of those closures relies on an unspoken identification of the trace of the trivalent web with the closure invariant (via the decategorification of (3-21)). A wrong sign, a wrong color normalization, or equality of the two polynomials would leave the handlebody-sensitivity claim unsupported; the invariant could then coincide, up to normalization, with colored HOMFLYPT of the S^3 closure of the non-core strands, which is exactly what the paper claims to improve on. This does not by itself invalidate the invariance half of Cor. 4.13, but it is load-bearing for the paper's central novelty.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs a triply-graded homology invariant for links in a genus g handlebody, generalizing the colored HOMFLYPT (Khovanov–Rozansky) homology of links in the 3-sphere. The construction embeds the handlebody braid group Br(g,n) into the classical braid group Br(g+n), assigns complexes of singular Soergel bimodules via Rouquier complexes, and then applies partial Hochschild trace functors to achieve invariance under the handlebody Markov moves. The main results are Theorem 4.7, stating invariance of HH_Hg under handlebody conjugation and stabilization (up to grading normalization) and non-invariance under classical conjugation, and Corollary 4.13, which defines the normalized invariant HHH_Hg(b,M) of handlebody links and asserts that it is not in general an invariant of the S^3 closure of the non-core strands. The key example for handlebody sensitivity is Proposition 4.8, where the braids t2t1 and t1t2 are conjugate in Br(g+n) but are claimed to have non-homotopy-equivalent invariants, with the verification reduced to an unstated HOMFLYPT polynomial computation.","tokens_in":24700,"tokens_out":4106,"duration_ms":44528,"significance":"If correct, this is a substantive advance: it provides the first triply-graded homological link invariant for handlebodies of arbitrary genus, going beyond the doubly-graded annular theories in the literature. The paper uses established published machinery (singular Soergel bimodules, Rouquier complexes, partial Hochschild traces) rather than introducing ad hoc axioms, and it gives explicit statements of the new topological sensitivity. The main construction is clearly laid out, and the proof of stabilization invariance is largely explicit. However, the claimed novelty of handlebody sensitivity hinges on a single omitted computation in Proposition 4.8, and the proof of Lemma 4.12, which underpins stabilization invariance, is somewhat compressed. These points are fixable within the scope of the manuscript, so the appropriate outcome is major revision rather than rejection.","major_comments":[{"comment":"The proof that HH_Hg(t2t1) and HH_Hg(t1t2) are not homotopy equivalent is the only evidence for the handlebody-sensitivity claim in Corollary 4.13, but the key computation is omitted: the text says 'a computation shows' that the difference of the reduced HOMFLYPT polynomials of the closures in (4-12) is (a-a^{-1})^2-(q-q^{-1})^2. Please display this computation, including the closures and the coloring conventions, and explain the reduction from Euler characteristics of HH_Hg to HOMFLYPT polynomials, especially how the decategorification of Example 3.10 and the Jones–Ocneanu trace identification are applied. Without this, the second sentence of Corollary 4.13 and the claimed improvement over S^3 closure invariants rely on an unverifiable calculation.","section":"§4, Proposition 4.8"},{"comment":"Stabilization invariance of the normalized invariant depends on the exact degree shifts in (4-39), and the proof of Lemma 4.12 uses an induction in which the base case cites [Hog18, Proposition 3.10] to identify differentials and then performs 'Gaussian elimination' of terms with coinciding aq-degrees. Please spell out the cancellation leading to (4-41) and justify the use of the Krull–Schmidt property of the relevant derived category, or give a precise statement of [Wu14, Lemma 4.20] and explain why it applies to the bounded derived category of finitely generated graded singular Soergel bimodules over a polynomial ring. A small error in these shifts would invalidate the normalization factor x(b,M) in (4-45).","section":"§4, Lemma 4.12 and equations (4-40)–(4-42)"}],"minor_comments":[{"comment":"In the relation 'bjbi = bibj if |i-j] > 2', the bracket should be '|i-j| > 2'.","section":"§2, equation (2-2)"},{"comment":"The passage to 'half-integral values of the at-gradings' is not fully explained; please clarify when the exponents in x(b,M) are half-integers and how the resulting homology is an object of KVec_{atq}.","section":"§4, equation (4-45)"},{"comment":"The first expression in the chain of isomorphisms is ambiguous: '(q^k-q^{-k})/(q-q^{-1}) k' should be written as [k]_q times the diagram, or parenthesized, to avoid confusion with the diagram itself.","section":"§4, equation (4-42)"},{"comment":"In the stabilization move (2-10), the braids b and c are drawn in a way that is hard to disambiguate at a glance; labeling the strands or enlarging the figure would improve readability, especially because the move is used later in the paper.","section":"§2, Definition 2.7 and Figure (2-10)"},{"comment":"The statement that integral versions are not currently available for Section 4 is useful for honesty, but it would be helpful to indicate whether the missing input is the singular Soergel diagrammatics over Z or something else, so that readers can assess the obstruction.","section":"§1, Convention 1.1"}],"recommendation":"major_revision","confidential_remarks":"The omitted HOMFLYPT computation in Proposition 4.8 is central to the paper's novelty claim, but it is likely easy to supply. I recommend asking the authors to include it explicitly and to clarify the decategorification step. The rest of the invariance argument appears sound, and I do not see grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a genuine new invariant — triply-graded HOMFLYPT homology for links in a genus g handlebody — and the construction is clever. The key move is to embed the handlebody braid group into a classical braid group, then attach an extra merge/split singular Soergel bimodule. That breaks classical conjugation while preserving the handlebody Markov moves. It's the right fix, and it fills a real gap: prior annular and sutured constructions only gave doubly-graded invariants.\n\nThe invariance half reads well. The use of partial Hochschild trace and singular Soergel bimodules is standard machinery, the normalization x(b,M) is explicit, and the characteristic-0 restriction is honestly stated. I don't see circularity: the decategorification comparison to HOMFLYPT is a consistency check, not a parameter fit. The citation pattern is fine; the self-citations are technical inputs.\n\nThe soft spots are real but concentrated. Proposition 4.8 is the only evidence that the invariant is genuinely sensitive to the handlebody rather than a pullback of the S^3 invariant. The proof says \"a computation shows\" the HOMFLYPT difference is (a-a^{-1})^2 - (q-q^{-1})^2, with no computation displayed. That is load-bearing: if that comparison is wrong, the non-invariance claim collapses. The step from Euler characteristics to HOMFLYPT also relies on an unstated identification of the trace of the trivalent web with the closure invariant. A referee should ask for the calculation. Lemma 4.12, which carries stabilization invariance, is also sketched rather than fully written; the degree shifts are stated, but the chain of quasi-isomorphisms is not fully displayed. This is less worrying, since it follows known results, but it's still where an error would hide.\n\nWho is this for? People in categorified link invariants, Soergel bimodules, and handlebody topology. It's specialized but it fills a clear gap.\n\nSend it to a serious referee. In revision I'd demand an expanded proof of Prop 4.8 and more detail in Lemma 4.12. If those check out, it's a clean accept.","headline":"A real new invariant, but the key sensitivity example is asserted rather than shown; deserves a serious referee.","tokens_in":25292,"tokens_out":4126,"would_cite":true,"duration_ms":37035,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["57K18","20F36","16E40"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper constructs a triply-graded homology invariant for links in any genus-g handlebody, extending colored HOMFLYPT homology from the 3-sphere.","keywords":["HOMFLYPT homology","handlebody","Soergel bimodules","triply-graded homology","braid groups","Hochschild cohomology","Markov moves","categorification"],"falsifier":"Compute the normalized invariant HHH_Hg for the genus-two braids t2t1 and t1t2 with one strand color 1; if the two triply-graded vector spaces coincide up to degree shift, Corollary 4.13's non-invariance statement fails. Alternatively, verify Lemma 4.12's degree shifts by a direct k=2 calculation and see whether the stabilization isomorphism preserves the claimed shifted gradings.","tokens_in":24264,"feed_emoji":"🧶","tokens_out":8633,"duration_ms":84671,"temperature":0.7,"pith_summary":"The paper constructs a triply-graded homology theory for links in a genus-g handlebody, the compact three-manifold obtained by attaching g handles to a ball. For a braid in the handlebody braid group with a balanced coloring, the assignment produces a triply-graded vector space that is invariant under the handlebody analogues of Markov moves, and this invariant is sensitive to the handlebody structure: it can distinguish links whose closures inside the handlebody are not isotopic even though their ordinary closures in the three-sphere would agree. This extends the colored HOMFLYPT homology of links in the three-sphere, recovered at genus zero, and gives the first triply-graded link homology in a three-manifold other than the three-sphere. The construction matters because existing doubly-graded annular invariants did not lift to the triply-graded setting, and the handlebody braid group is not generally an Artin-Tits group, so a new categorical mechanism is needed.","feed_headline":"A finer link invariant for handlebodies","feed_subtitle":"The triply-graded homology detects handlebody topology that ordinary braid closure misses.","key_machinery":"The load-bearing object is the 2-category of singular Soergel bimodules, a categorification of the Hecke/Schur algebroid built from bimodules over rings of symmetric polynomials. Braid complexes of these bimodules give a categorical action of the classical braid group, and hence of its handlebody subgroup; the handlebody braid is first sent to a complex of bimodules by forgetting the special role of core strands. To make the invariant invariant under handlebody stabilization but not under full classical conjugation, the authors glue merge and split bimodules onto the complex, then apply the partial Hochschild trace functors I_M and T_M, which close one strand at a time. Lemma 4.12 supplies the exact degree shifts for the merge-split stabilization isomorphisms, and the normalization factor x(b,M) in equation (4-45) cancels these shifts using half-integer powers of a and t and integer powers of q. That combination of partial trace and normalization is what converts a homotopy-equivalence-up-to-shift into a genuine invariant of handlebody links.","core_discovery":"The central claim is that for any balanced, colored handlebody braid (b,M) in Br(g,n), the triply-graded vector space HHH_Hg(b,M) is an invariant of the handlebody link obtained by closing b inside H_g, and in general it is not invariant under the classical closure of the non-core strands in $S^{3}$. Invariance is established under the conjugation and stabilization moves of the handlebody Markov theorem (Theorem 4.7), and the non-invariance under ordinary braid conjugation is demonstrated explicitly in Proposition 4.8 by braids t2t1 and t1t2 in genus two, whose classical HOMFLYPT polynomials differ. The authors achieve this by embedding the handlebody braid group into the classical braid group, realizing braids by complexes of singular Soergel bimodules, then modifying the closure procedure with merge and split bimodules before applying Hochschild cohomology; the key technical step is a stabilization lemma with explicit q-, t-, and a-degree shifts, and a normalization factor x(b,M) that compensates those shifts. The upshot is a well-defined triply-graded invariant that is finer than the invariant obtained by ignoring the handlebody embedding.","pith_inferences":["The half-integer shifts in the normalization x(b,M) suggest that the invariant may admit a natural square-root grading or a relation to spin structures on the handlebody, a direction the paper does not pursue.","The partial-trace mechanism could be iterated to compute the invariant recursively strand by strand, in the style of annular evaluation, potentially making explicit computations practical for small braid index.","If the conjectured connection between genus-two braid groups and affine type C structures is realized, the invariant may match a yet-to-be-built type C homology, allowing a cross-check on examples.","One can test sensitivity by computing the invariant on the explicit pair in Proposition 4.8 with larger colors; if the difference persists for all colorings, the invariant sees the full handlebody Markov quotient."],"forward_implications":["At genus zero the construction recovers the classical colored HOMFLYPT homology of links in S^3, so the known structural properties of that invariant restrict to the new invariant.","At genus one it yields a triply-graded invariant of annular links, lifting the doubly-graded annular theories and carrying information about winding around the core strand.","Because the invariant is not invariant under ordinary conjugation, it distinguishes handlebody links whose classical closures are isotopic, giving a genuinely stronger invariant in the handlebody category.","The invariant is unchanged under exactly the handlebody Markov moves, so it factors through the Markov quotient of the handlebody braid group.","The construction works uniformly for all genera, including g at least 2, where the handlebody braid group is not known to be an Artin-Tits group."],"supporting_citations":[{"why":"Supplies Alexander's and Markov's theorems for links in handlebodies, the topological foundation of the invariant.","marker":"[HOL02]"},{"why":"Gives the presentation of the handlebody braid group used throughout the paper.","marker":"[Ver98]"},{"why":"Provides the alternate braid structure and discusses the related Hecke-like algebras in handlebodies.","marker":"[Lam00]"},{"why":"Establishes that singular Soergel bimodules categorify the Hecke/Schur algebroid, the setting for the categorical action.","marker":"[Wil11]"},{"why":"Introduces the braid complexes of bimodules that realize braid generators as chain complexes.","marker":"[Rou06]"},{"why":"Supplies the partial Hochschild trace functors and the uncolored stabilization base cases adapted here.","marker":"[Hog18]"},{"why":"Provides the geometric construction of colored HOMFLYPT homology whose stabilization behavior Lemma 4.12 imports.","marker":"[WW17]"},{"why":"Gives the colored stabilization statements used to fix the degree-shift conventions.","marker":"[Cau17]"},{"why":"Shows that the braid action on complexes of singular Soergel bimodules is independent of the chosen braid expression.","marker":"[QRS18]"}],"fun_headline_variants":["Handlebody link homology that beats braid closure","Triply-graded invariant for handlebody links, finer","New HOMFLYPT-type homology for handlebody links","Handlebody invariant that detects what closure misses","A more powerful link invariant for handlebodies"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The invariant is well-defined only if the chain of quasi-isomorphisms in Lemma 4.12 carries exactly the stated q-, t- and a-degree shifts; if any of those shifts is off, the normalization factor x(b,M) cannot repair the stabilization invariance and the construction collapses.","fun_headline_variants_meta":{"raw":{"variants":["Handlebody link homology that beats braid closure","Triply-graded invariant for handlebody links, finer","New HOMFLYPT-type homology for handlebody links","Handlebody invariant that detects what closure misses","A more powerful link invariant for handlebodies"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000148,"raw_usage":{"total_tokens":1139,"prompt_tokens":845,"completion_tokens":294,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":461,"completion_tokens_details":{"reasoning_tokens":219}},"tokens_in":461,"tokens_out":294,"duration_ms":3808,"temperature":1.0,"reasoning_tokens":219,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:34:38.328956+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the normalized invariant HHH_Hg for the genus-two braids t2t1 and t1t2 with one strand color 1; if the two triply-graded vector spaces coincide up to degree shift, Corollary 4.13's non-invariance statement fails. Alternatively, verify Lemma 4.12's degree shifts by a direct k=2 calculation and see whether the stabilization isomorphism preserves the claimed shifted gradings.","supporting_citations":[],"review_version":1}