{"id":"e9d22f28-1010-4b20-8f92-18c65f7bdac2","arxiv_id":"2505.04428","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Configuration space integrals produce characteristic classes of framed fibre bundles with closed manifold fibres, generalizing Kontsevich's graph complex classes to arbitrary closed fibres.","lead":"This paper constructs new characteristic classes for bundles whose fibres are any closed manifold and whose vertical tangent bundle is trivialized, extending Kontsevich's original construction for homology sphere fibres. It links the resulting graph complex integrals to recent algebraic models of automorphism spaces of configuration spaces, giving a unified picture of these invariants.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 5.2 depends on Proposition 4.4, whose proof hides the key computation behind 'straightforward yet lengthy' and defines a fibre integral over a non-submersion by naturality; this is the least secure step in the central claim.","rationale":"The paper's central theorem is a family version of Kontsevich's configuration-space integrals. The architecture is standard: choose a differential Ω(B)-module model for Ω(E), choose a compatible propagator, define Z_E by fibre integrals, and prove it is a chain map by Stokes' theorem and Kontsevich's vanishing lemma. The theorem is plausible and much of the proof is detailed. The place where the proof stops being a proof is Proposition 4.4. The compatible propagator is the input to the whole integral construction; its existence and uniqueness up to coboundary are needed for the independence-of-choice statement in Theorem 5.2, and Proposition 4.4 is also invoked in the proof of Theorem 6.1. The construction of λ12 requires fibre integrals that are not literally defined, as the paper acknowledges, and the verification of its crucial properties is omitted. This is not a disagreement with consensus; it is an internal gap in a load-bearing lemma. The triviality assumption on the π1(B)-action is a genuine limitation, but it is part of the hypothesis rather than an unproved step in the argument. The dependence on [Wil23] mostly affects Theorem B, not Theorem A. The reader's conditional verdict is exactly right: the result should be accepted only after Proposition 4.4 is fully proved. I therefore recommend keeping the CONDITIONAL verdict, hence UNCHANGED relative to the reader's assessment.","tokens_in":34463,"tokens_out":11992,"duration_ms":124574,"concrete_test":"Fully compute dλ12 and the boundary restriction of λ12 using the fibrewise Stokes formula (Lemma B.1) for the three integrals defining λ12, including all signs and the contributions of the strata of ∂FMfw_E[3] and ∂FMfw_E[4]. In particular, verify the claimed identity ∫_2(φ12−λ12)∧π_2^*α=0 for α running over a basis {x_i, x_i^#} of H(M), and check explicitly that λ12 restricts to zero on the boundary face E×S^{d−1} of FMfw_E[2]. If any sign or boundary term fails, the compatible propagator of Proposition 4.4 does not exist as stated and Theorem 5.2 needs revision; if the computation closes, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central construction Z_E in Theorem 5.2 is only well-defined and choice-independent if the compatible propagator φ12 of Proposition 4.4 exists and is unique up to coboundary. The proof of Proposition 4.4 is not self-contained at exactly that point. The form λ12 is defined using the fibre integrals ∫_3 φ13∧Δ23, ∫_3 φ23∧Δ13, and ∫_{3,4} φ34∧Δ13∧Δ24. Here π12: FMfw_E[3]→FMfw_E[2] is not a proper submersion: over the interior of FMfw_E[2] its fibre is M minus two points, which is non-compact, and the paper itself notes that ∫_3 is not directly defined. The proposed fix is a naturality argument through the submersion Id×π2 with fibre M\\D^d, but diagram (4.4) only embeds FMfw_E[3] as an open dense subset of E×_B FMfw_E[{2,3}]; no support or vanishing condition is proved that would make the non-proper fibre integral well-defined. The key identities — that λ12 is closed, (−1)^d-symmetric, satisfies ∫_2 λ12∧π_2^*α = ∫_2 φ12∧π_2^*α for all α∈H(M), and vanishes on the boundary of FMfw_E[2] — are relegated to 'a straightforward yet lengthy calculation' with no signs or boundary terms shown. These identities are precisely what makes φ12−λ12 satisfy properties (i)–(iv) and (4.3), and the uniqueness statement is used in Lemma 5.4 to prove independence of Z_E from the choices. Thus a gap here directly undermines Theorem 5.2, and Proposition 4.4 is also invoked in the proof of Theorem 6.1. This is a correctness risk internal to the proof, not a scope limitation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper generalizes Kontsevich's configuration-space-integral construction of characteristic classes for framed fibre bundles with homology-sphere fibres to arbitrary closed-manifold fibres. The main result, Theorem A (Theorem 5.2), constructs a fibrewise partition function Z_E from the Chevalley-Eilenberg complex of the complete dg Lie algebra osp_{H(M)}^{<0} ⋉ GC_{H(M)} to the de Rham algebra of the base, for framed smooth submersions with closed fibre M, dim M > 2, and trivial π_1(B)-action on H(M). The map is claimed to be a cdga map, independent of choices, and natural under pullback. Theorem B (Theorem 6.1) refines this to a map from C^*_CE(g_M), where g_M is the dg Lie algebra appearing in Willwacher's model for automorphisms of configuration-space modules, thus connecting the construction to rational homotopy-theoretic classifying spaces. The proof combines a relative Sullivan model for the fibration, a construction of compatible propagators on fibrewise Fulton-MacPherson compactifications, and the graph-complex formalism of Campos--Willwacher.","tokens_in":34786,"tokens_out":5089,"duration_ms":53303,"significance":"If the central construction is fully justified, this is a substantial contribution: it provides an explicit geometric source of characteristic classes for all framed bundles with closed fibres, extends Kontsevich's original framework beyond homology spheres, and makes a concrete connection to Willwacher's algebraic models for configuration-space modules and their automorphisms. The manuscript is transparent about its reliance on prior results, especially [CW23], [CM10], and [Wil23], and it gives a detailed account of the differential, boundary terms, and orientation conventions. The weakest point is Proposition 4.4, whose proof hides the key compatibility calculation and defines fibre integrals over non-proper maps; this is load-bearing for both main theorems. The paper does not appear circular: it imports established algebraic models as inputs rather than assuming the target statement.","major_comments":[{"comment":"The definition of λ12 via the fibre integrals ∫_3 φ13∧Δ23, ∫_3 φ23∧Δ13, and ∫_{3,4} φ34∧Δ13∧Δ24 is not justified as written. The map π12: FMfw_E[3]→FMfw_E[2] is not a proper submersion: over the interior of FMfw_E[2] its fibre is M minus two points, which is non-compact, and the paper itself notes that ∫_3 is not directly defined. The naturality argument through diagram (4.4) only embeds FMfw_E[3] as an open dense subset of E×_B FMfw_E[{2,3}]; no support, vanishing, or compactly-supported representative condition is proved that would make the non-proper fibre integral well-defined. This is not a cosmetic gap: the form λ12 is the correction making φ12−λ12 satisfy (4.3), and condition (4.3) is used in Lemma 5.4 and in the proof of Theorem 6.1.","section":"§4, Proposition 4.4 and diagram (4.4)"},{"comment":"The key identities in Proposition 4.4 are asserted without proof. That λ12 is closed, is (−1)^d-symmetric, satisfies ∫_2 λ12∧π_2^*α = ∫_2 φ12∧π_2^*α for all α ∈ H(M), and vanishes on the boundary of FMfw_E[2] are exactly the properties needed to conclude that φ12−λ12 is a compatible propagator satisfying (i)–(iv) and (4.3). The subsequent uniqueness statement for compatible propagators is also used in Lemma 5.4 to prove independence of Z_E from the choice of propagator, and Proposition 4.4 is invoked again in the proof of Theorem 6.1. Since no signs or boundary terms are shown, the central choice-independence claim rests on an omitted calculation rather than on a proof contained in the paper.","section":"§4, Proposition 4.4, 'straightforward yet lengthy calculation'"},{"comment":"The proof that Z_E is compatible with multiplication uses the equality Z_E(Γ1⊔Γ2)=Z_E(Γ1)Z_E(Γ2), justified by naturality of fibre integration along the map pr1×pr2: FMfw_E[V(Γ1)⊔V(Γ2)]→FMfw_E[V(Γ1)]×_B FMfw_E[V(Γ2)]. As in the case of π12 in Proposition 4.4, this map is not a proper submersion; the forms are pulled back from the two factors, and on interiors the map is an open dense embedding. No support or completion argument is provided to justify the interchange of fibre integration with pullback along this non-proper map. This gap affects the claim that Z_E is a map of cdga's, not merely a chain map.","section":"§5, proof of Theorem 5.2, multiplicativity step"}],"minor_comments":[{"comment":"The statement of Lemma 4.3 says 'smooth submersion of closed manifolds', but only the fibre is assumed closed; the base is not assumed closed in the paper. Please rephrase to 'with closed fibre'.","section":"§4, Lemma 4.3"},{"comment":"In the proof of Theorem 5.2, the reference to 'property (iii) of Proposition 4.3' should be to Lemma 4.3, since that is where the propagator properties are stated.","section":"§5, proof of Theorem 5.2"},{"comment":"Remark 5.1 leaves unresolved whether the sign convention in the definition of Z_E agrees with the conventions in [CW23] and [Idr19]. Since Theorem B identifies the constructed map with cohomology classes coming from Willwacher's model, a sign mismatch would affect that comparison; please either fix the convention or state precisely how the signs differ.","section":"§1, Remark 1.2 and §5, Remark 5.1"},{"comment":"In the definition of z0, the text says 'xi∈H_*(M;R) denotes a basis of the homology of M', but the surrounding notation uses cohomology classes and the Poincaré pairing. This should be clarified to avoid confusion.","section":"§6, proof of Theorem 6.1"},{"comment":"There is a typo: 'we see that see that Z_E is a chain map' should read 'we see that Z_E is a chain map'.","section":"§5, proof of Theorem 5.2, final paragraph"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope and the central idea is attractive, but the referee report focuses on Proposition 4.4 because that proposition is load-bearing for both main theorems and its proof is not currently self-contained at the crucial point. If the author can supply the missing calculation and a correct treatment of the non-proper fibre integrals, I would be willing to see a revised version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the paper really does something new: Kontsevich's configuration space integral construction is extended from homology sphere fibres to arbitrary closed framed fibres, with a non-trivial base. That requires the osp-action and a family propagator, so this is not a cosmetic change. Second, the main theorem is not fully proven as written. Proposition 4.4 is the load-bearing step, and it contains the weakest link: the compatible propagator is constructed using a fibre integral over a non-submersion, justified by a naturality diagram, and then the key identities—closedness, symmetry, equation (4.3)—are relegated to 'a straightforward yet lengthy calculation'. Since uniqueness of that propagator is used in Lemma 5.4 to show Z_E is choice-independent, Theorem 5.2 depends on this gap. This is exactly the kind of omission that can hide sign errors or a missing support condition.\n\nThe rest of the paper is in much better shape. Section 3, on models for fibrations with trivial monodromy, is careful and largely self-contained; the orthogonal basis lemma (3.8) is fully proved. The boundary analysis in Theorem 5.2 follows the standard Kontsevich vanishing argument and is written out in detail. The author also flags his own limitations in Remarks 1.1, 5.1, 5.3, which I appreciate. The reliance on [Wil23] and forthcoming work of Berglund is a dependency but not a flaw.\n\nThe soft spots, in proportion: the gap in Prop 4.4 is real and central. The sign convention mismatch in Remark 5.1 is minor. And the paper computes no examples, so the significance is potential rather than demonstrated. If the propagator lemma is fixable, this is a strong paper.\n\nI'd send it to a serious referee, with the specific instruction to pin down Prop 4.4. The intended reader is someone who works on graph complexes and characteristic classes; they will get value from the conceptual unification with Willwacher's model. I wouldn't cite it yet until the gap is closed.","headline":"A genuine extension of Kontsevich's construction to all closed framed fibres, but the key propagator step is not fully proven; worth a serious referee.","tokens_in":35352,"tokens_out":2512,"would_cite":false,"duration_ms":26285,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["55R40","57R20","55R10","55R80"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that every framed fibre bundle with closed fibre and trivial base action on fibre cohomology carries characteristic classes built from configuration space integrals over graphs decorated by the fibre's cohomology.","keywords":["characteristic classes","framed fibre bundles","configuration space integrals","graph complexes","Fulton-MacPherson compactification","ortho-symplectic Lie algebra","Maurer-Cartan elements","Chevalley-Eilenberg cochain complex"],"falsifier":"Take the pullback of a framed $S^3$-bundle over $S^4$ along a degree-2 self-map of $S^4$. The theorem predicts $f^*[Z_E] = [Z_{f^*E}]$ in $H^*(S^4)$ for every graph; exhibiting one graph for which these cohomology classes differ would falsify the naturality claim. Even more directly, a single graph whose integral is not closed, or whose integral changes when the model or propagator is replaced, would disprove the chain-map statement.","tokens_in":34187,"feed_emoji":"🕸️","tokens_out":11637,"duration_ms":111887,"temperature":0.7,"pith_summary":"The paper proves a sweeping extension of a classical graph-integral construction: from framed bundles with odd-dimensional homology-sphere fibres to all framed smooth fibre bundles with closed fibre $M$, connected base and total space, dimension at least $3$, and trivial action of the fundamental group of the base on the cohomology of $M$. For every such bundle, a configuration space integral defines a chain map from a Chevalley-Eilenberg graph complex to the de Rham complex of the base. The map is independent of all choices up to homotopy and is natural under pullback, so it produces characteristic classes of the framed bundle itself. If correct, the construction makes the whole machine of graph cohomology available for studying families of arbitrary closed manifolds.","feed_headline":"Graph integrals produce characteristic classes for framed bundles","feed_subtitle":"Graphs decorated by fibre cohomology become de Rham classes on the base via configuration-space integrals.","key_machinery":"The mechanism is the fibrewise configuration space integral over the Fulton-MacPherson compactification of configurations of points in the fibres. To a connected graph whose vertices carry classes $\\alpha_i \\in H^*(M)$ and whose edges are dressed with a propagator form $\\varphi_{12}$ satisfying $d\\varphi_{12} = \\Delta^!(1)$, the Poincaré dual of the diagonal of the fibre, one associates a differential form on the compactified fibre-wise configuration space and integrates along the fibres to a form on the base. The key identities are Stokes' theorem, the boundary behaviour of the propagator (which is controlled by the vertical framing), and the fact that the differential of the model of the total space is a Maurer-Cartan element in $osp^{<0}_{H(M)}$. These identities make the graph complex differential dual to the exterior derivative on forms, so that graph cocycles evaluate to closed forms on $B$.","core_discovery":"The central claim is that fibre-wise configuration space integrals assemble into a fibrewise partition function $$Z_E: C^*_{\\mathrm{CE}}($osp^{{<0}}$_{H(M)} \\ltimes GC_{H(M)}) \\to \\Omega_{dR}(B),$$ which is a chain map of commutative differential graded algebras, independent of the choice of module model and propagator up to homotopy, and natural with respect to pullbacks. Here $GC_{H(M)}$ is the graph complex whose vertices are decorated by classes in the reduced cohomology of the fibre, $osp^{<0}_{H(M)}$ is the Lie algebra of degree-negative endomorphisms of $H(M)$ preserving the Poincaré pairing, and $\\ltimes$ denotes the semi-direct product. The theorem further refines $Z_E$ to a map from the Chevalley-Eilenberg complex of the sub-Lie algebra $g_M$ generated by trivalent graphs and the Maurer-Cartan element encoding the rational homotopy type of the fibre's Fulton-MacPherson compactification. A consequence drawn in the paper is that the graph integral provides a geometric construction of pullbacks of real cohomology classes of the classifying space of automorphisms of configuration-space modules.","pith_inferences":["The triviality assumption is likely not essential; a nilpotent action should be handled by twisted coefficients and a larger dg Lie algebra, and the paper says the same construction should extend in that direction.","For fibres with rich cohomology, such as products of spheres, the graph complex has many additional decorations, so the construction should produce classes not visible to classical Chern-Weil invariants; computing the Chevalley-Eilenberg cohomology of $osp^{<0}_{H(M)} \\ltimes GC_{H(M)}$ would locate them.","If the algebraic model of configuration-space modules is a complete invariant, then the partition function gives a practical way to distinguish framed bundles by evaluating a finite set of graph integrals, a testable numerical scheme in low dimensions.","The framing assumption enters only through the boundary term of the propagator; one consequence is that changing the framing should change $Z_E$ by exact forms, so the induced cohomology classes are candidates for invariants of the underlying unframed bundle when the framing ambiguity is finite."],"forward_implications":["For every such framed bundle, every cocycle in the graph complex evaluates to a closed differential form on the base, so combinatorial graph cohomology produces explicit characteristic classes.","The invariants are natural under pullback, so they define cohomology classes on the classifying space of framed $M$-bundles and, via the refined theorem, on the classifying space of configuration-space module automorphisms.","When $B$ is a point, the fibrewise partition function reduces to the previously known partition function of a framed manifold, making the new construction a family version of that invariant.","When the fibre is an odd-dimensional homology sphere, the construction recovers the original integral classes, giving a common framework for the old and new invariants.","The refined map means the real homotopy type of the family of fibre-wise configuration spaces is enough to organise all these classes."],"supporting_citations":[{"why":"The classical construction for homology sphere fibres that this paper generalizes to arbitrary closed fibres.","marker":"[Kon94]"},{"why":"Supplies the graph complex $GC_{H(M)}$, the partition function for framed manifolds, and the quasi-isomorphism model of configuration spaces that the fibrewise construction extends.","marker":"[CW23]"},{"why":"Provides the minimal relative Sullivan models used to build the $\\Omega_{dR}(B)$-module model of the total space with differential $D$.","marker":"[Hal83]"},{"why":"Gives the Fulton-MacPherson compactification and the right module structure that underlie fibrewise configuration spaces.","marker":"[Sin04]"},{"why":"Earlier version of the configuration-space model whose propagator construction is adapted in the fibre bundle setting.","marker":"[CW16]"},{"why":"Algebraic model of the classifying space of automorphisms of configuration-space modules; the refined theorem maps from its Chevalley-Eilenberg complex.","marker":"[Wil23]"}],"fun_headline_variants":["Configuration-space integrals define characteristic classes","New characteristic classes from framed fibre bundle integrals","Framed bundles get characteristic classes via graph integrals","Generalizing Kontsevich: classes for framed fibre bundles","Characteristic classes via fibre-wise graph integrals"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction assumes that every loop in the base leaves the cohomology of the fibre unchanged; under that assumption the flat connection term in the model can be removed and the differential is controlled by the ortho-symplectic Lie algebra. If the action were non-trivial, the fibre integrals would need twisted coefficients and the whole Lie-algebra setup would have to be replaced.","fun_headline_variants_meta":{"raw":{"variants":["Configuration-space integrals define characteristic classes","New characteristic classes from framed fibre bundle integrals","Framed bundles get characteristic classes via graph integrals","Generalizing Kontsevich: classes for framed fibre bundles","Characteristic classes via fibre-wise graph integrals"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000647,"raw_usage":{"total_tokens":2898,"prompt_tokens":797,"completion_tokens":2101,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":413,"completion_tokens_details":{"reasoning_tokens":2032}},"tokens_in":413,"tokens_out":2101,"duration_ms":14984,"temperature":1.0,"reasoning_tokens":2032,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:28:35.499418+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the pullback of a framed $S^3$-bundle over $S^4$ along a degree-2 self-map of $S^4$. The theorem predicts $f^*[Z_E] = [Z_{f^*E}]$ in $H^*(S^4)$ for every graph; exhibiting one graph for which these cohomology classes differ would falsify the naturality claim. Even more directly, a single graph whose integral is not closed, or whose integral changes when the model or propagator is replaced, would disprove the chain-map statement.","supporting_citations":[],"review_version":1}