{"id":"c3b7ceb8-33ae-4801-b62e-48bf69c10e54","arxiv_id":"2608.08057","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A spherical finite tensor category with a modified trace gives a finite-dimensional open-closed 3d TFT via a new handlebody invariant and the universal construction.","lead":"The paper constructs a three-dimensional topological field theory for non-semisimple tensor categories, extending Turaev-Viro type state-sum TFTs. It builds a new 3-manifold invariant using spanning trees to avoid vanishing red loops, then applies the universal construction to obtain a finite-dimensional open-closed TFT.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the central TFT construction is coherent; the only delicate point is the lengthy proof of Proposition 5.8, which I could not falsify.","rationale":"The reader's verdict of ACCEPT with medium correctness risk is reasonable. The paper's central construction is long but internally coherent: Theorem 4.8 gives topological invariance of the invariant, and Theorem 5.10 follows formally from the universal construction once Proposition 5.8 is available. I agree with the reader that Proposition 5.8 is the most delicate assumption: it is the point where the abstract universal construction is shown to produce concrete, finite-dimensional state spaces via admissible skein modules. However, I did not find a specific flaw in its proof. The construction of the collar element u is plausible: red-to-blue modification introduces projectively coloured edges, and the cutting move propagates projective edges across all cuts, so admissibility should be preserved. The reduction to connected M via Lemma 5.6 is legitimate because radical membership of a disjoint union tensor state is controlled by the same lemma. The remaining risk is that a topological subtlety in the lengthy PLCW argument was missed; this is a verification risk, not an identified mathematical error. No formal verification exists, and the proof is intricate enough that independent checking of at least one nontrivial example is warranted. I therefore recommend leaving the verdict unchanged.","tokens_in":40786,"tokens_out":22333,"duration_ms":274173,"concrete_test":"Re-run the Proposition 5.8 construction in the minimal nontrivial case: let the target (Sigma,L) have two connected components, and let M be obtained from the collar Sigma x [0,1] by attaching a single 1-handle, with the free-boundary graph Gamma a single P-coloured strand (P projective). Explicitly compute the resulting element u after the red-to-blue modification and cutting along the meridian of the 1-handle, then check (i) that u is a genuine collar element, i.e. its induced graph on each component of Sigma contains a projectively coloured edge, and (ii) that for N the disk bordism (Sigma,L) -> empty, the equality tau_C(N union M, Gamma) = tau_C(N union u) holds. If either check fails, Proposition 5.8 has a real gap; if both hold, the most delicate step of the paper is supported in the simplest nontrivial configuration.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I read the central claim as requiring three ingredients: (1) the invariant tau_C is well-defined and diffeomorphism invariant, (2) the universal construction produces a symmetric monoidal functor, and (3) Proposition 5.8, which says every state is equivalent to a collar element, whence the state spaces are finite-dimensional via admissible skein modules. I found the arguments for (1) and (2) internally consistent, and the algebraic input from chromatic maps and the multi-handlebody invariant is used in a way that is compatible with the cited results. The weakest point is indeed Proposition 5.8: its proof is long, uses a custom PLCW decomposition, contraction of 1-cells, red-to-blue modification, and repeated cutting moves. The potential failure mode would be that the constructed collar element u is not actually an admissible C-graph on the target, or that the equality of pairings with an arbitrary left bordism N fails in some corner case. I could not locate such a corner case in the text. The red-to-blue step produces projectively coloured G-edges, and the cutting move propagates projective edges to both sides of each meridian disk, so the resulting graph on each target component should be admissible. The sliding and red-capping moves used are backed by Lemmas 3.14-3.18. Thus the concern is one of proof complexity and absence of independent verification, not a detected error.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper defines an invariant τC of compact oriented 3-manifolds with boundary decorated by a bulk-admissible C-coloured graph, where C is a spherical finite tensor category with a two-sided modified trace. The construction combines the multi-handlebody invariant F of CGPT18 with the chromatic maps of CGPV23, and introduces a new spanning-tree rule for adding red loops when passing from a PLCW decomposition to a handlebody. The authors prove that τC is independent of the PLCW decomposition, the spanning tree, and the auxiliary interior points (Theorem 4.8), that it agrees with the closed-manifold invariant KC of CGPT18 after removing a ball (Proposition 4.11), and that it is invariant under cutting moves (Proposition 4.17). Via the universal construction, τC is then extended to a symmetric monoidal functor V: Bord^nc_oc(C) -> Vect_k (Theorem 5.10). The key technical step is Proposition 5.8, which asserts that every state of a ∂-marked surface is a linear combination of collar elements; this yields a surjection from the finite-dimensional admissible skein module to the state space (Proposition 5.9).","tokens_in":41051,"tokens_out":13858,"duration_ms":167626,"significance":"If correct, the paper provides a genuinely non-semisimple open-closed 3d TFT with finite-dimensional state spaces and an explicit skein-theoretic description, going beyond the semisimple alterfold and Turaev-Viro frameworks. The main strengths are the detailed classical proof of the new invariant, the novel use of spanning trees to avoid red loops that would evaluate to zero, and the clear delineation of the algebraic input from earlier work. The paper also gives a transparent comparison with the closed-manifold invariant of CGPT18 and indicates expected but not fully proved relations with the alterfold theory and with the noncompact TFT of CGPV23. The proofs are lengthy, especially the proof of Proposition 5.8, but I did not find a concrete error; the main risk is the complexity of the PLCW and handlebody arguments, which are not independently verified here.","major_comments":[],"minor_comments":[{"comment":"The sentence 'By Lemma 5.6, we may assume that M is connected' is not immediate from the statement of Lemma 5.6, which only proves injectivity of the map Ψ. The reduction can likely be justified by applying the connected statement to each connected component of M and using the containment R1⊗~V2 + ~V1⊗R2 ⊂ R1⊔2 shown in Lemma 5.6, but this should be spelled out for the reader.","section":"§5.3, beginning of the proof of Proposition 5.8"},{"comment":"The bullet point 'Γ intersects δ^1(Σ) only on the edges of Γ' should read 'only at points on the edges of Γ'; as written it suggests that the intersection is contained in the edges rather than consisting of transverse intersection points on those edges.","section":"§4.2, Definition 4.5"},{"comment":"In the chain of isomorphisms (5.12), the equality dim C(P,1) = dim Sadm(D,P) is asserted without proof. A one-line justification, for instance by exhibiting a basis of Sadm(D,P) represented by the single P-labelled strand with a morphism P -> 1, would make the dimension comparison transparent.","section":"§5.2, Example 5.11"},{"comment":"The passage from cutting along the meridian disks D(M1 \\ Σ) to a collar element is very compressed: it is stated that evaluating the resulting 3-balls with F yields a linear combination of admissible graphs on Σ, but the role of the newly introduced dual-basis coupons in making the ball graphs admissible is not explicitly justified. A short explanatory sentence would help.","section":"§5.3, after Eq. (5.17)"},{"comment":"The comparisons with the alterfold invariant are explicitly labelled as remarks and several verifications are left to the reader. Since these comparisons are not needed for the main theorem, this is acceptable, but the sentence 'We skip the details' in Remark 5.13 should perhaps be accompanied by a precise statement of which of the listed assertions are proven and which are expected equivalences.","section":"Remarks 4.12 and 5.13"}],"recommendation":"minor_revision","confidential_remarks":"I read the paper in good faith and paid particular attention to Lemma 4.16 and Proposition 5.8, the two most intricate parts. I did not find a load-bearing error. The main potential weakness is the length and complexity of the proof of Proposition 5.8, and the reduction to the connected case is underexplained; however, the argument appears to be repairable with clarification. The paper is a solid contribution to the non-semisimple TFT literature and is within the scope of math.QA."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThis paper gives the first construction I've seen of a non-semisimple open-closed 3d TFT with finite-dimensional state spaces, starting from a spherical finite tensor category with a two-sided modified trace. The invariant tau_C extends the multi-handlebody invariant F of CGPT18 to arbitrary 3-manifolds with boundary and bulk-admissible graphs, and the universal construction then yields the TFT V. The genuinely new technical device is the spanning-tree choice for which 2-cell belts get red loops. That avoids red loops bounding disks that would evaluate to zero, which would have killed the invariant in the presence of boundary. That is a real obstruction, and the fix is neat.\n\nThe paper does a lot well. The proof of Theorem 4.8 (independence of PLCW decomposition, spanning tree, and interior points) is detailed and structured by elementary subdivisions. Proposition 4.11 matching tau on M minus a ball with the CGPT18 closed invariant is a useful sanity check. The comparison remarks to alterfold invariants are explicitly flagged as sketches, so they don't overclaim.\n\nThe soft spot is exactly where the reader put it: Proposition 5.8, the spanning of state spaces by collar elements. The proof is long, uses a custom PLCW decomposition, contraction of 1-cells, red-to-blue modification, and repeated cutting moves. I could not find a gap, but it is the kind of argument that deserves a careful referee's eye, especially the step where the red-to-blue modification produces projectively coloured G-edges and the cutting move propagates them correctly. If that proposition fails, finite-dimensionality of state spaces and monoidality would both be in question. The stress-test note's assessment matches mine: no detected error, but proof complexity is high.\n\nAlso minor: Remarks 4.12 and 5.13 state expected equivalences with alterfold invariants without proofs. That's fine for the main result, but a referee should ask whether those claims can be supported or should be rephrased as conjectures.\n\nOverall, this deserves a serious referee. I'd recommend sending it to peer review. It is a real step forward for non-semisimple TFTs, and the main risk is contained in one long proof rather than in the overall strategy.","headline":"A genuine non-semisimple open-closed 3d TFT with finite-dimensional state spaces; the spanning-tree red-loop device is the real new idea, and the main risk is the long proof of Prop. 5.8.","tokens_in":41577,"tokens_out":1802,"would_cite":true,"duration_ms":19114,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper constructs an open-closed three-dimensional TFT from any spherical finite tensor category with a two-sided modified trace, with finite-dimensional state spaces described by admissible skein modules.","keywords":["spherical finite tensor category","modified trace","open-closed topological field theory","admissible skein module","bulk-admissible graph","chromatic maps","universal construction","non-semisimple 3d TFT"],"falsifier":"Carry out the universal-construction pairing on a once-punctured torus for a non-semisimple spherical category in which red loops bounding disks evaluate to zero, for instance representations of a finite-dimensional unimodular unibalanced Hopf algebra over a field where the dimension of the algebra is zero. The surjection from the admissible skein module predicts a specific finite dimension for $V$; a smaller dimension, or any nonzero class of the skein module lying in the radical of the pairing, would contradict Proposition 5.8 and with it Theorem 5.10.","tokens_in":40589,"feed_emoji":"🧶","tokens_out":12675,"duration_ms":128464,"temperature":0.7,"pith_summary":"State-sum constructions of three-dimensional topological field theories normally require semisimple input categories. This paper removes that restriction: starting from a spherical finite tensor category, possibly non-semisimple, equipped with a two-sided modified trace, it constructs an open-closed three-dimensional topological field theory with values in vector spaces. The construction first defines a new invariant of compact oriented 3-manifolds whose boundary carries an admissible $\\mathcal{C}$-coloured graph, then applies the universal construction. The payoff is that every such category now yields a concrete 3d TFT whose state spaces are finite-dimensional and are quotients of admissible skein modules, so skein theory gives an explicit description of the quantum state spaces.","feed_headline":"Every spherical tensor category yields a finite-dimensional 3d TFT","feed_subtitle":"A spanning-tree trick controls red loops, and skein modules bound the TFT's state spaces.","key_machinery":"The load-bearing machinery is the $(2,3)$-graph of a PLCW decomposition of a 3-manifold: one vertex for each 3-cell plus a boundary vertex, and one edge for each 2-cell. Given a spanning tree of this graph, the construction adds red loops only along the belts of 2-cells outside the tree; those red loops are then converted into blue projective strands by chromatic maps, the red-to-blue procedure, turning the boundary into an admissible bichrome graph on a handlebody that is evaluated by the multi-handlebody invariant $F$. The spanning-tree choice is what avoids red loops that bound disks, which would evaluate to zero in many non-semisimple categories. Invariance under changing the decomposition, tree, or interior points is transported by sliding moves, cutting moves, and red capping and digging moves, and the universal construction [BHMV95] converts the resulting invariant into the TFT.","core_discovery":"The paper's central claim is Theorem 5.10: for any spherical finite tensor category $\\mathcal{C}$ with a chosen two-sided modified trace, the invariant $\\tau_{\\mathcal{C}}$ extends through the universal construction to a symmetric monoidal functor $V:\\mathrm{Bord}^{\\mathrm{nc}}_{\\mathrm{oc}}(\\mathcal{C})\\to\\mathrm{Vect}_k$, an open-closed 3-dimensional TFT. The invariant $\\tau_{\\mathcal{C}}(M,\\Gamma)$ is defined for a compact oriented 3-manifold $M$ with boundary and a bulk-admissible $\\mathcal{C}$-coloured graph $\\Gamma$ in $\\partial M$, meaning each connected component of each boundary component carries a projectively coloured edge. Theorem 4.8 states that $\\tau_{\\mathcal{C}}$ depends only on the diffeomorphism class of $(M,\\Gamma)$, not on the auxiliary PLCW decomposition, spanning tree, or interior points. The main structural step is Proposition 5.8, which says every vector in every state space is a linear combination of collar elements; from this, Proposition 5.9 gives a surjection $E_{(\\Sigma,L)}:S_{\\mathrm{adm}}(\\Sigma,L)\\to V(\\Sigma,L)$ from the admissible skein module to the state space, and finite-dimensionality follows.","pith_inferences":["The spanning-tree trick is transferable: any state-sum or surgery construction that decorates 2-skeleton belts with red loops can prune loops using a spanning tree, which may remove spurious zero evaluations in categories whose global dimension or cointegral evaluation vanishes.","If the expected agreement with the admissible-skein TFT on closed surfaces holds, then the surjection $E_{(\\Sigma,L)}$ should actually be an isomorphism on closed surfaces, giving a skein-theoretic presentation of the TFT rather than a radical quotient; this is testable surface-by-surface.","Following the usual state-sum/surgery correspondence, $V$ should coincide with the surgery-based TFT associated to the Drinfeld center of $\\mathcal{C}$ from [DGG+19]; a proof would extend that correspondence to open-closed non-semisimple bordisms.","Explicit low-genus computations for categories like representations of small quantum groups at roots of unity would give concrete dimensions of the new state spaces and test the collar-element spanning statement against known admissible skein module dimensions."],"forward_implications":["Every spherical finite tensor category with a two-sided modified trace, semisimple or not, determines an open-closed 3d TFT with finite-dimensional state spaces.","For any marked surface $(\\Sigma,L)$, the state space $V(\\Sigma,L)$ is a quotient of the admissible skein module $S_{\\mathrm{adm}}(\\Sigma,L)$, so skein-theoretic computations give concrete upper bounds on state-space dimensions.","The new invariant $\\tau_{\\mathcal{C}}$ reproduces the closed 3-manifold invariant of [CGPT18] when a ball is removed from a closed manifold, so the known non-semisimple closed invariants fit inside the open-closed theory.","When $\\mathcal{C}$ is semisimple, $\\tau_{\\mathcal{C}}$ is a non-zero scalar multiple of the alterfold invariant, and the open-closed state spaces agree with alterfold state spaces up to isomorphism (Remark 5.13).","Even though the bordism category is not rigid, the universal construction here runs through finite-dimensional vector spaces, so the TFT is genuinely finite-dimensional on every object."],"supporting_citations":[{"why":"Supplies the multi-handlebody invariant $F$ and its red capping, digging, and cutting moves, which the new invariant $\\tau_{\\mathcal{C}}$ extends and, for closed manifolds, reproduces.","marker":"[CGPT18]"},{"why":"Supplies chromatic maps, admissible skein modules, the red-to-blue map, and the sliding moves; without these the decorated handlebody cannot be evaluated.","marker":"[CGPV23]"},{"why":"Gives the universal construction that turns the invariant $\\tau_{\\mathcal{C}}$ into the TFT $V$.","marker":"[BHMV95]"},{"why":"Establishes two-sided modified traces on projective ideals and their evaluation as linear forms on admissible skein modules, the algebraic input for the whole construction.","marker":"[GPV11]"},{"why":"Provides PLCW decompositions and elementary subdivisions, used to prove $\\tau_{\\mathcal{C}}$ is independent of the chosen decomposition.","marker":"[Kir10]"},{"why":"Proves boundary triangulations extend to the whole manifold, an input to the collar-element proof of Proposition 5.8.","marker":"[Arm67]"},{"why":"Proves admissible skein modules are finite-dimensional, which is what makes the state spaces of $V$ finite-dimensional via the surjection.","marker":"[RST24]"}],"fun_headline_variants":["Non-semisimple tensor categories build full 3d TFTs","Modified trace yields open-closed 3d TFT for spherical categories","From spherical tensor categories to finite 3d TFTs","Open-closed 3d TFT from non-semisimple categories"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on Proposition 5.8, that every vector in every state space is a linear combination of cylinder-shaped collar elements; if that spanning statement fails, the state spaces need not be finite-dimensional and monoidality of $V$ is not established.","fun_headline_variants_meta":{"raw":{"variants":["Non-semisimple tensor categories build full 3d TFTs","Modified trace yields open-closed 3d TFT for spherical categories","From spherical tensor categories to finite 3d TFTs","Open-closed 3d TFT from non-semisimple categories"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000228,"raw_usage":{"total_tokens":1473,"prompt_tokens":941,"completion_tokens":532,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":557,"completion_tokens_details":{"reasoning_tokens":455}},"tokens_in":557,"tokens_out":532,"duration_ms":5845,"temperature":1.0,"reasoning_tokens":455,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T00:29:01.448120+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Carry out the universal-construction pairing on a once-punctured torus for a non-semisimple spherical category in which red loops bounding disks evaluate to zero, for instance representations of a finite-dimensional unimodular unibalanced Hopf algebra over a field where the dimension of the algebra is zero. The surjection from the admissible skein module predicts a specific finite dimension for $V$; a smaller dimension, or any nonzero class of the skein module lying in the radical of the pairing, would contradict Proposition 5.8 and with it Theorem 5.10.","supporting_citations":[],"review_version":1}