{"id":"47fb2160-205c-45ed-a1ed-e972ec7b1976","arxiv_id":"2607.24238","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"The axial-gauge partition function of Abelian BF theory recovers the Milnor metric, realizing Fried’s conjecture for Morse–Smale flows via two-step BV pushforward.","lead":"Abelian BF theory in Morse–Smale axial gauge has partition function equal to the Milnor metric on twisted cohomology. This gives a field-theory reading of Fried’s conjecture (already true for these flows) by matching Schwarz’s Ray–Singer result in the metric gauge.","discovery_kind":"extension","skeptic_critique":{"model":"moonshotai/kimi-k3","headline":"The det′(ι_V)=1 step (§8.1) justifies itself via a reparametrization V↦V/‖V‖ that is singular exactly at the fixed points, and invariance of the partition function under that change of variables is asserted, not proved, on anisotropic spaces.","rationale":"The reader correctly located the weakest assumption at det′(ι_V)=1 (Prop. 8.21, Lemmas 8.18–8.19); my stress-test confirms this is the load-bearing point and sharpens the failure mechanism: the normalisation argument is imported from the Anosov setting where V is nonvanishing and the reparametrization is globally smooth, whereas in the Morse–Smale case V̂=V/‖V‖ is singular at fixed points and the needed invariance of the regularised partition function under a singular change of variables is assumed rather than derived. This is an internal-completeness concern about the derivation, not a correctness challenge to the final formula: the target equality (Ray–Singer = Milnor metric for MS flows) is independently a theorem (Shen–Yu 2021), and the metric-gauge half of the paper (§7) is rigorous following Schwarz, so a nontrivial Jacobian would contradict gauge-independence against a solid computation — suggesting the answer is right and the gap is in the proof. The pointwise identity (123) is correct algebra, and a plausible fix exists (define det′ relative to Im(ι_V) with compatible volume elements, or define the axial partition function with the normalised field from the outset and prove equivalence on anisotropic spaces). Given the independent corroboration and the apparently closable nature of the gap, I move the verdict from ACCEPT to CONDITIONAL rather than REJECT: acceptance conditional on a rigorous treatment of the reparametrization Jacobian, verified e.g. by the S¹ computation above. Other aspects of the paper (the two-stage pushforward structure, Theorem 4.10's isometric torsion equality with its self-contained Appendix C proof, the spectral realisation of R_{V,ρ}(0) in Theorem 5.21) are careful and well-referenced and do not present a competing load-bearing concern.","tokens_in":51142,"tokens_out":5746,"duration_ms":121411,"concrete_test":"Take M=S¹ with a C^∞-linearisable Morse–Smale field (two hyperbolic fixed points, no closed orbits) and a generic rank-1 twist so the field is non-aligned (S² with the height function is a fallback). (1) Compute the axial partition function following §8.1 but keeping the Jacobian of η̃=‖V‖_g·η explicit, using a spectral cutoff on the anisotropic space (Fourier modes on S¹): check whether the cutoff-regularised determinant of multiplication by ‖V‖_g on the relevant graded field space tends to 1 as the cutoff is removed. (2) Independently compute the Milnor metric ∥μ_H∥_{M,V} from SY21's definition (Thom–Smale torsion times |R_{V,ρ}(0)|^{−1}, both explicit here). If the Jacobian converges to a value ≠1 and |Z^V_BF|≠∥μ_H∥, the concern lands; if it converges to 1, §8.1 is vindicated and the \"removable singularity\" step is harmless.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Corollary 8.15 (the central claim |Z^V_BF| = ∥μ_H∥_{M,V}) routes through Theorem 8.10, whose proof defers entirely to §8.1 for det′(ι_V)=1. The §8.1 argument has three soft joints. (a) The \"removable singularity\" claim: near a hyperbolic fixed point, V(x)=Ax+O(x²), so V̂=V/‖V‖_g has no continuous limit at the fixed point unless A is scalar — the norm ratio is 1 but the direction is undefined. Hence f=1/‖V‖_g violates the strict-positivity/smoothness hypothesis of Lemma 8.18 globally, precisely at the fixed points, the locus supporting the zero-resonant states and the whole second pushforward. (b) The key inference \"|det′(ι_V)|=|det′(ι_V̂)|\" (Eq. 120) rests on the sentence \"the formal expression of the partition function must remain invariant under this reparametrisation\" — an assumption, not a theorem, in infinite dimensions. The substitution η̃=‖V‖_g·η in the Berezinian integral generates the Jacobian of the multiplication operator by ‖V‖_g, a function vanishing at critical points; flat regularisation does not automatically trivialize it (its formal derivative involves tr♭(log‖V‖·e^{−tL̃}), with log‖V‖ diverging at the fixed points where flat traces localize). (c) Definition 8.20 is only \"formally analogous\" to det(d_∇), and identity (123) is pointwise algebra; concluding det′(ι_V̂)=1 requires a regularised-determinant framework on anisotropic currents where ι_V̂∘(V̂^♭∧)=I holds with unit determinant — not supplied. Notably, in the acyclic Anosov case (HKS20/SS24) the analogous normalisation is innocuous because V never vanishes; the Morse–Smale setting is exactly where the imported argument develops its singularity. If this Jacobian is nontrivial, the axial partition function differs from the Milnor metric by that multiplicative factor, so the gap is load-bearing for the strongest claim.","agreement_with_reader":"agree"},"referee_report":{"model":"moonshotai/kimi-k3","summary":"The paper studies Abelian BF theory on a closed oriented manifold with a flat unitary twist, in the BV formalism, under two gauge fixings. In the Lorenz/metric gauge it recovers (after a first BV pushforward onto harmonic fields) the Ray–Singer torsion, following Schwarz and [HKS20]. In the axial gauge defined by contraction with a C^∞-linearisable, non-aligned Morse–Smale vector field V, it uses the Dang–Rivière anisotropic Sobolev framework: the chain homotopy I = π_0 + d_∇h_V + h_Vd_∇ splits the field space into zero-resonant states plus fluctuations, a first BV pushforward yields R_{V,ρ}(0)^{(−1)^n} times a Jacobian factor |det′(ι_V)|^{−1}, and a second pushforward onto the cohomology of the resonant complex yields the Thom–Smale torsion via an isometric isomorphism Φ between the resonant and Thom–Smale complexes. After arguing in §8.1 that det′(ι_V)=1 by normalising V to V̂ = V/‖V‖, the main claim (Corollary 8.15) is that the absolute value of the axial partition function equals the Milnor metric ∥μ_H∥_{M,V} = |τ(C^•_TS)|·|R_{V,ρ}(0)|^{−1}, thereby reinterpreting the Shen–Yu proof of Fried's conjecture for Morse–Smale flows as gauge-fixing independence of the BF partition function.","tokens_in":51651,"tokens_out":8146,"duration_ms":610540,"significance":"If the argument is completed, the paper gives a conceptually appealing field-theoretic interpretation of the equality of the Ray–Singer and Milnor metrics for Morse–Smale flows (proved analytically by Shen–Yu [SY21]) as gauge-fixing independence of Abelian BF theory, extending the Anosov/Reeb results of [HKS20, SS24] to the non-acyclic setting with zero modes. Strengths worth naming: the derivation is parameter-free (the only non-standard input is the normalisation claim det'(ι_V)=1); the intermediate results — the isometric isomorphism of the resonant and Thom–Smale complexes (Thms 4.9–4.10), the spectral realisation of R_{V,ρ}(0) via flat determinants (Thm 5.21), and the double BV pushforward isolating closed-orbit and fixed-point contributions separately — are explicit and independently checkable; and the central identity |Z^V_BF| = ∥μ_H∥_{M,V} is a concrete, falsifiable statement that can be tested against [SY21]. The paper is careful not to claim a new proof of Fried's conjecture, framing the result as a reinterpretation conditional on gauge-fixing independence in infinite dimensions, which is the appropriate posture.","major_comments":[{"comment":"Section 8.1, Proposition 8.21 and Lemmas 8.18–8.19 (the step det'(i_V)=1, load-bearing for Theorem 8.10 and Corollary 8.15). The argument has three gaps. (a) The 'removable singularity' claim after Eq. (113) conflates the norm with the direction: near a hyperbolic fixed point V(x)=Ax+O(x^2), so \\hat V = V/\\|V\\|_g has no continuous limit at the fixed point unless A is scalar. Consequently f=1/\\|V\\|_g fails the smooth strictly-positive hypothesis of Lemma 8.18 globally, the reparametrisation of integral curves is justified only on M minus the critical set, and the statement in the proof of Proposition 8.21 that g(\\hat V,\\hat V)=1 'everywhere on M' is literally false at the fixed points — precisely the locus supporting the zero-resonant currents U_{a,j} (Remark 3.8). (b) The key identity |det'(i_V)| = |det'(i_{\\hat V})| (Eq. 120) is justified only by the sentence 'the formal expression of t","section":"§8.1, Prop. 8.21, Lemmas 8.18–8.19, Eq. (120)"},{"comment":"There is an unresolved parity/sign discrepancy in the central identification. Theorem 5.21 gives R_{V,rho}(0) = prod_k det^flat(eL^{(k)})^{(-1)^{n+k}}, and Definition 8.6/Corollary 8.8/Theorem 8.10 therefore yield Z ~ R_{V,rho}(0)^{(-1)^n} (Eqs. 101 and 107). But Definition 8.14 defines the Milnor metric with the factor |R_{V,rho}(0)|^{-1}. Corollary 8.15 (|Z^V_BF| = ||mu_H||_{M,V}) thus follows from the manuscript's own equations only when n is odd. No parity hypothesis on dim M appears anywhere in the geometric setup (Section 3 conventions or Appendix A), and even-dimensional manifolds admitting Morse–Smale flows with closed orbits exist, in which case |Z| = |tau(C_TS)|·|R(0)|^{+1} ≠ ||mu_H||_{M,V} generically. The odd-dimensional case subsumes the contact/Anosov setting of [HKS20], which may explain how the sign was inherited unnoticed. The authors must either restrict the main theore","section":"Cor. 8.15 vs. Thm 5.21, Cor. 8.8, Def. 8.14"},{"comment":"The quantity det'(i_V) appears in Definition 8.6 (Eq. 99) and in Theorem 8.10 but is never defined: Definition 8.20 defines only det'(i_{\\hat V}) for the normalised field, and the bridge is the unproved invariance (120). Independently of the regularisation issue in the first comment, the manuscript should give a standalone definition of det'(i_V) (domain, target, regularisation, and the subspace on which the determinant is taken — note i_V is not an endomorphism of a fixed space, so the analogy with det(d_nabla) in Eq. (89) requires care: det(d_nabla) maps between different form degrees but is treated via d^dagger d; an analogous self-map for i_V is i_V o (V^flat wedge), which is multiplication by \\|V\\|^2, not the identity, off the gauge-fixing subspace).","section":"Def. 8.6, Eq. (99), Def. 8.20"}],"minor_comments":[{"comment":"The proof states 'noting that eta^{(k+1)} can be chosen such that d_nabla eta^{(k+1)}=0' in order to replace d_nabla i_V eta by L_{V,nabla} eta. The primitive eta of A = i_V eta is defined only modulo ker(i_V), and it is not obvious a d_nabla-closed representative exists in Im(I - pi_0); alternatively, show directly that the extra term B wedge i_V d_nabla eta integrates to zero for B in the axial Lagrangian. A sentence of justification is needed since this identity produces the quadratic form (98) on which the first pushforward relies.","section":"Prop. 8.5, proof"},{"comment":"The proof invokes 'Equation (90)' to substitute sdet^flat(d_nabla^dagger) = |sdet^flat(d_nabla^dagger d_nabla)|^{1/2}, but Equation (90) is the restricted action functional; the intended reference is presumably Equation (89).","section":"Cor. 7.9, proof"},{"comment":"The commutative diagram following Corollary 8.8 is introduced without labels on the arrows or a caption; the identification of eF^{(1)}_BF with Ker(eDelta) is only explained afterwards. Please label the maps (BV pushforwards vs. isomorphisms) and state in which category the diagram commutes.","section":"Diagram after Cor. 8.8"},{"comment":"The hypotheses 'non-aligned' (Proposition 3.7) and 'C^infty-linearisable' (Definition A.7) are used throughout Sections 4–8 but are not restated in the statements of Theorem 8.10 or Corollary 8.15; the main results should be self-contained about their assumptions. Relatedly, Remark 3.9 covers the non-singular case, but the standing hypotheses when closed orbits are absent (pure gradient-like flows) versus present could be stated once, globally.","section":"Thm 8.10 / Cor. 8.15 hypotheses"},{"comment":"Definition 8.20 should specify the graded subspace on which det' is computed and whether the superdeterminant or the ordinary determinant is intended; as written the notation det' conflicts with the use of det' in Appendix C (product of nonzero eigenvalues of a finite-dimensional Laplacian).","section":"Def. 8.20 notation"},{"comment":"Numerous typographical artifacts appear throughout (likely from text extraction): 'GIOV ANNI MOLINARI AND MICHELE SCHIA VINA' in the running head, 'heuristically though of' (Section 2.1), 'rests at the foudation' (Section 2.3), 'AbelianBF' (abstract and elsewhere), 'e eta' / 'e F' spacing in Section 8.1. Reference [Se26] is dated 2026 and cited for the canonical BV Laplacian; if it is not yet publicly available, a stable alternative citation should be added.","section":"Passim: typography"},{"comment":"The framing around Fried's conjecture (Remark 8.17 and the abstract) is appropriately hedged ('reinterpreted', 'suggests'), but the abstract's phrase 'provides a field-theoretic realisation of Fried's conjecture, which is true for Morse–Smale flows' could be misread as a new proof; consider adding half a sentence noting that gauge-fixing independence of the infinite-dimensional BF partition function is the conjectural ingredient, with the equality of metrics itself already established in [SY21].","section":"Abstract / Remark 8.17"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is an extension of the first author's Master thesis; the companion survey [Mol26] is by the same author, which explains the heavy cross-referencing — I see no problem with this, but the editor may wish to confirm the thesis/survey disclosure. The paper's contribution over [HKS20, SS24] is the non-acyclic Morse–Smale setting with zero modes; it does not claim a new proof of Fried's conjecture, and referees/editors should hold it to that standard. I would recommend soliciting a referee with expertise in the Dang–Rivière microlocal framework specifically for §8.1, since that subsection is where the load-bearing analytic input is asserted rather than developed."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The new piece is the explicit two-step BV pushforward for Abelian BF in the Morse–Smale axial gauge: first integrate fluctuations to get R_{V,ρ}(0) from the restricted Lie derivative on anisotropic spaces, then a second finite-dimensional pushforward that extracts the Thom–Smale torsion. Together they give |Z^V_BF| = ∥μ_H∥_{M,V} (Cor. 8.15). That is a genuine extension of HKS20/SS24 (acyclic Anosov) to the non-acyclic MS setting with fixed points; SY21 already has the geometric equality, but the field-theoretic path and the second pushforward are new.\n\nWhat works: the chain-homotopy decomposition, the isometric identification of the zero-resonant complex with the Thom–Smale complex (Thms 4.9–4.10), the spectral realisation of R(0) (Thm 5.21), and the Lagrangian checks for both gauges are carefully written and properly referenced. Infinite-dimensional BV is handled by the usual flat-superdeterminant + residual half-density recipe; they do not overclaim a new proof of Fried, only a realisation of the already-settled MS case as gauge independence. Citation pattern is clean.\n\nThe soft spot is real and sits exactly where the stress-test puts it. Section 8.1 sets det'(ι_V)=1 by normalising V \to V/∥V∥. Near hyperbolic fixed points the direction is undefined, so the rescaling function is not smooth (or even continuous) at the locus that supports the zero-resonant states and the whole second pushforward. The claim that the partition function is invariant under that change is asserted rather than proved on anisotropic spaces, and the multiplication-by-∥V∥ Jacobian is not controlled by the flat-trace machinery they use elsewhere. In the Anosov case V never vanishes, so the same move is harmless; here it is load-bearing for Cor. 8.15. If that factor is nontrivial the identification with the Milnor metric acquires an extra constant. Everything else looks solid.\n\nThis is for people already working on TQFT/torsion/dynamical zeta. Worth a serious referee; the gap is local and fixable (or at least isolable). I would send it out.","headline":"Solid axial-gauge computation of the Milnor metric for MS flows via two-step BV pushforward; the det'(ι_V)=1 step is the only real soft joint and is load-bearing for the strongest claim.","tokens_in":50371,"tokens_out":619,"would_cite":true,"duration_ms":13969,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["58J52","37C30","81T70","57Q10","37D15"],"pacs":[],"model":"grok-4.5","headline":"The axial-gauge partition function of Abelian BF theory recovers the Milnor metric for Morse–Smale flows.","keywords":["Abelian BF theory","Batalin-Vilkovisky formalism","Milnor metric","Morse-Smale flows","Ruelle zeta function","analytic torsion","Fried conjecture","BV pushforward"],"falsifier":"An independent flat-trace computation of the determinant of the interior product on the fluctuation sector that returns a value other than 1, or a direct evaluation of the axial partition function that differs from the known Milnor metric by a non-trivial multiplicative dynamical factor.","tokens_in":50087,"feed_emoji":"ζ","tokens_out":972,"duration_ms":44976,"temperature":0.7,"pith_summary":"This paper shows that quantizing Abelian BF theory with a Morse–Smale vector field as axial gauge fixing produces, as its partition function, the Milnor metric on the determinant line of the twisted cohomology. The calculation uses the Batalin–Vilkovisky formalism and proceeds by a two-step pushforward to cohomology: the first step integrates out non-zero modes and recovers the Ruelle dynamical zeta function at zero; the second step integrates the residual zero-resonant states and recovers the combinatorial torsion of the Thom–Smale complex. Those two factors together are precisely the Milnor metric, the natural extension of the Ruelle zeta to flows that possess both closed orbits and critical points. Because the same theory in the metric (Lorenz) gauge is already known to yield the Ray–Singer analytic torsion, the equality of the two gauge-fixed partition functions supplies a field-theoretic reading of Fried’s conjecture, which holds for Morse–Smale flows. A reader who cares about the link between dynamics and topology therefore obtains a single quadratic path integral whose gauge choices interchange spectral and dynamical invariants.","feed_headline":"Axial-gauge BF theory yields the Milnor metric","feed_subtitle":"Two BV pushforwards turn a Morse–Smale flow into a field-theoretic Fried conjecture","key_machinery":"The two-step BV pushforward built on the chain-homotopy identity I = π_0 + d_∇ ∘ h_V + h_V ∘ d_∇ for the Lie derivative along the Morse–Smale field. The first pushforward extracts the Ruelle factor as a flat superdeterminant of the restricted Lie derivative; the second, after an isometric identification with the Thom–Smale complex, extracts its combinatorial torsion.","core_discovery":"After a double BV pushforward onto cohomology, the absolute value of the axial-gauge partition function of twisted Abelian BF theory equals the Milnor metric on the determinant line of the twisted de Rham cohomology: |Z^V_BF| = ∥μ_H∥_{M,V}. The dynamical factor is the Ruelle zeta evaluated at zero; the fixed-point factor is the torsion of the Thom–Smale complex.","pith_inferences":["The axial-gauge construction should extend, with suitable anisotropic spaces, to any flow class for which Pollicott–Ruelle resonances are defined.","Triviality of the interior-product Jacobian is the field-theoretic counterpart of the fact that contraction vanishes on the zero-resonant complex.","Both gauges can be evaluated numerically on a low-dimensional manifold carrying an explicit Morse–Smale flow, giving a direct numerical check of the metric equality."],"forward_implications":["Gauge-fixing independence of the Abelian BF partition function is equivalent to equality of the Ray–Singer and Milnor metrics for Morse–Smale flows.","The Milnor metric is the natural generalisation of the Ruelle zeta at zero once fixed points are admitted.","The same double-pushforward strategy realises the Fukaya–Morse A_∞ structure in the non-Abelian theory.","Analytic torsion and dynamical zeta functions become interchangeable outputs of one quadratic field theory under different Lagrangian choices."],"fun_headline_variants":["Axial-gauge BF theory recovers the Milnor metric","Double BV pushforward yields Milnor metric from BF theory","|Z_BF| equals Milnor metric after BV pushforward to cohomology","Morse–Smale axial gauge turns BF theory into Milnor metric","BF partition function realises Fried conjecture via Milnor metric"],"cache_read_input_tokens":32768,"weakest_assumption_plain":"After the Morse–Smale vector field is normalised to unit length, the regularised determinant of the interior-product operator that defines the axial gauge can be set identically to one.","fun_headline_variants_meta":{"raw":{"variants":["Axial-gauge BF theory recovers the Milnor metric","Double BV pushforward yields Milnor metric from BF theory","|Z_BF| equals Milnor metric after BV pushforward to cohomology","Morse–Smale axial gauge turns BF theory into Milnor metric","BF partition function realises Fried conjecture via Milnor metric"]},"model":"grok-4.5","effort":"low","cost_usd":0.00318,"raw_usage":{"total_tokens":1027,"prompt_tokens":691,"num_sources_used":0,"completion_tokens":71,"cost_in_usd_ticks":31804000,"prompt_tokens_details":{"text_tokens":691,"audio_tokens":0,"image_tokens":0,"cached_tokens":128},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":265,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":691,"tokens_out":71,"duration_ms":6055,"temperature":1.0,"reasoning_tokens":265,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-31T20:14:03.270353+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"An independent flat-trace computation of the determinant of the interior product on the fluctuation sector that returns a value other than 1, or a direct evaluation of the axial partition function that differs from the known Milnor metric by a non-trivial multiplicative dynamical factor.","supporting_citations":[],"review_version":1}