{"id":"c2b3144a-89d9-4e00-a2ad-b48a0810e465","arxiv_id":"2505.03404","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A unified framework shows that flat-regularized superdeterminants of characteristic operators stay constant along inner variations of generalized codifferentials, covering local constancy of analytic torsion and of Ruelle zeta at zero.","lead":"This paper proves a general theorem: the regularized superdeterminant of certain operators built from differential forms is constant as a family of generalized codifferentials varies. It recovers two known local constancy results, for analytic torsion under metric changes and for the Ruelle zeta function at zero under changes of a contact Anosov flow.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Ruelle local constancy hinges entirely on [DGRS20, Thm 4]; without a self-contained verification of (A.v), Corollary 2 does not independently support the framework.","rationale":"The reader identified Assumption (A.v) as the weakest assumption, and my reading agrees. The proof of Theorem 1 is algebraically sound: Lemma 3.12, Lemma 3.14, and the final evaluation at λ=0,s=0 all work if (A.v) holds. The elliptic case verifies the assumptions self-containedly via the heat kernel construction in Appendix A. The Ruelle case, however, explicitly outsources the decisive analytic continuation to [DGRS20, Theorem 4]. Since DGRS20's Theorem 4 is itself the microlocal engine behind their local-constancy theorem, Corollary 2 is not a new proof of local constancy but a repackaging of a known theorem. This is a missing-support concern, not a fatal error: the general framework and the elliptic corollary remain valuable. The appropriate verdict is therefore CONDITIONAL, as the reader concluded, asking for a self-contained verification of (A.v) or a softened claim for Corollary 2. No change to the reader's verdict is needed.","tokens_in":54658,"tokens_out":13150,"duration_ms":134480,"concrete_test":"Locate the precise statement of [DGRS20, Theorem 4]. If it asserts the analytic continuation of λ ↦ str♭(θτ e^{-t0 L_{Xτ}} R_{Xτ}(λ)) to a neighborhood of λ=0 with local boundedness in τ, then follow the chain: set s=1 in (38) and use Remark 3.7. If the conclusion ζ_{Xτ}(0)=ζ_{X0}(0) follows immediately from that theorem plus the Guillemin trace formula already derived in §5.2, then (A.v) is exactly DGRS20's analytic-continuation result, and Corollary 2 is a consequence of [DGRS20, Thm 4] rather than of the new assumptions (A.i)-(A.iv). If Theorem 4 is strictly weaker and does not by itself imply local constancy, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The weakest load-bearing point is Assumption (A.v), as used after equation (40) in the proof of Theorem 1. In the Ruelle application, the paper does not verify (A.v) from its own microlocal analysis: the proof of Proposition 5.11 ends by stating that the required analytic continuation of G(τ,λ,1) to λ=0 'is precisely [DGRS20, Theorem 4]'. That external theorem is neither stated nor proved in the manuscript, and it is a core analytic ingredient in the DGRS20 proof of the same local-constancy result. Consequently, Corollary 2 is not an independent output of the new framework; it is a reformulation of a result whose key difficulty is imported. This does not invalidate Theorem 1, whose algebraic core appears consistent, but it means the abstract's claim that the general result 'implies' the Ruelle local constancy overstates self-containedness. The paper should either prove (A.v) in the Anosov case or state Corollary 2 as a recovery using DGRS20's theorem.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a general framework for proving local constancy of flat-regularized superdeterminants of characteristic operators Dτ = [δτ, d∇] along smooth families of \"general codifferentials\" δτ. The main theorem, Theorem 1, states that if τ ↦ δτ is a smooth inner variation of a regular general codifferential and five analytic assumptions (A.i)–(A.v) hold, then sdet♭(Dτ|Lτ) is independent of τ. The authors then apply this theorem to two cases: the Hodge codifferential δ_{gτ}, giving a proof of local constancy of Ray–Singer torsion, and the contraction ι_{Xτ} along a family of regular contact Anosov vector fields, giving local constancy of the Ruelle zeta function at zero. The elliptic case is treated in a largely self-contained way, with a detailed appendix constructing the heat kernel for a smooth family of elliptic operators. The Anosov case uses microlocal tools such as wavefront-set control, the Guillemin trace formula, and Pollicott–Ruelle resonances, but a key analytic continuation assumption is imported from [DGRS20].","tokens_in":54851,"tokens_out":20677,"duration_ms":208996,"significance":"If the main theorem is correct, it provides a unified algebraic mechanism behind two previously known local-constancy results: Ray–Singer torsion invariance and the Dang–Guillarmou–Rivière–Shen result for Ruelle zeta functions of regular contact Anosov flows. The proof of Theorem 1 is carefully structured, and the algebraic core in Lemmas 3.10–3.14 is coherent and appears sound. The elliptic verification is self-contained and includes a substantial heat-kernel construction in Appendix A. The paper is also honest about the conditional nature of the main theorem: the analytic assumptions are stated explicitly. However, the paper's strongest advertised consequence, Corollary 2, is not fully independent: the verification of Assumption (A.v) for the Ruelle case is not proved but is asserted to be exactly [DGRS20, Theorem 4]. This means the general result does not by itself supply a new proof of the Ruelle local constancy unless that external theorem is either proved or explicitly imported as a black box. This limits the significance of the Anosov application, though it does not invalidate Theorem 1 as a conditional statement.","major_comments":[{"comment":"The verification of Assumption (A.v) in the Ruelle case is not carried out in the present paper. The proof ends by stating that the analytic continuation of G(τ,λ,1) to λ = 0 'is precisely [DGRS20, Theorem 4]', but that theorem is neither stated nor proved here. This is the same analytic continuation step that is central to the proof of local constancy in [DGRS20], and it is used exactly at the point where the proof of Theorem 1 passes from equation (40) to the evaluation at λ = 0. Consequently, Corollary 2 is not an independent consequence of the new framework; it is a reformulation of a known result whose key analytic input is imported. The authors should either prove (A.v) in the contact Anosov setting from their own microlocal analysis, or state Corollary 2 explicitly as a recovery of [DGRS20, Theorem 2] and soften the corresponding claims in the abstract and introduction.","section":"§5.3, proof of Proposition 5.11, last paragraph (Assumption (A.v))"},{"comment":"Definition 5.4 includes the assumption that the stable and unstable bundles of the Anosov flow are orientable, and this orientability is later used when the paper writes |det(I − Pγ)| = (−1)^{dim Es} det(I − Pγ) and in the proof of Proposition 5.10. However, Corollary 2 and Proposition 5.11 are stated for 'regular contact Anosov vector fields' without repeating the orientability condition. If orientability is not automatic for contact Anosov flows, the statements are wider than the proof. Please add the orientability hypothesis to the statements of Corollary 2 and Proposition 5.11, or explain how the non-orientable case is reduced to the orientable one.","section":"§5.2–5.3, orientability hypothesis"}],"minor_comments":[{"comment":"The proof of Lemma 5.14 shows that the difference quotients of the Schwartz kernel are bounded in the Hörmander seminorms and concludes convergence in the Hörmander topology. This is a valid Montel-space argument, but the authors should state explicitly that D′_Γ is a Montel (or nuclear Fréchet) space so that boundedness plus weak convergence implies convergence; as written, the inference is implicit.","section":"§5.3, proof of Lemma 5.14"},{"comment":"The reduction of Assumption (A.v) to analytic continuation of G(τ,λ,1) in λ relies on the fact that s = 0 lies in the domain of convergence for the Ruelle functions F(k)(λ,s). This is true because tr♭(e^{-tLX}|Ω^k_0) is supported away from t = 0, but the point is only implicit; a short explicit remark would make the application of Remark 3.7 to Proposition 5.11 easier to follow.","section":"§3.2, Remark 3.7 and §5.3"},{"comment":"The general codifferential definition writes the graded symmetry as '±' without specifying the sign in terms of the degrees of homogeneous forms. The later examples fix the sign case by case, but for a paper whose main theorem is algebraic, a precise sign convention in Definition 2.3 would improve clarity.","section":"§2.1, Definition 2.3"},{"comment":"In the convergence estimate for Assumption (A.iv), the sum is over all closed orbits and uses the primitive period Tγ^# in the integral bound while the delta function is supported at the period Tγ. This is correct, but the distinction between Tγ and Tγ^# should be noted explicitly at that point to avoid confusion.","section":"§5.2, equation (72)"}],"recommendation":"major_revision","confidential_remarks":"The paper's framing as a new proof of the Ruelle local constancy theorem is stronger than what is actually shown, because the critical analytic continuation (A.v) is imported from DGRS20. I would ask the authors to either prove that condition in the contact Anosov case or clearly reframe Corollary 2 as a recovery of the DGRS20 result. The algebraic core of Theorem 1 appears sound, so this is a revision issue rather than a rejection issue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: Theorem 1 is a new and nontrivial general statement about local constancy of flat regularized superdeterminants along inner variations of general codifferentials. The proof is detailed, the algebraic machinery in Lemmas 3.10–3.14 hangs together, and the paper is transparent about where the analytic input lives. The two applications are not new results, though: Corollary 1 recovers Ray–Singer, and Corollary 2 recovers the DGRS20 local constancy for regular contact Anosov flows.\n\nWhat the paper does well: it gives a single conceptual framework for two local-constancy theorems that look quite different analytically — elliptic Hodge Laplacians and first-order Lie derivatives along Anosov flows. The notion of a general codifferential, the inner-variation structure, and the reduction of the Ruelle zeta function to a flat superdeterminant are all handled with care. The proof of Theorem 1 is genuinely algebraic in the right places: the Duhamel formula, the wavefront-set bookkeeping, and the trick of eliminating the restriction to im(δ) via cyclicity of the flat trace are all executed honestly. The paper also does not oversell Fried's conjecture; it explicitly notes the dimension-parity mismatch and says the interpolation question remains open.\n\nThe soft spots, in proportion. The load-bearing point is Assumption (A.v), the analytic continuation of the auxiliary function G. In the Ruelle case this is not proved: the proof of Proposition 5.11 ends by saying the required continuation is precisely [DGRS20, Theorem 4]. That external theorem is the core microlocal estimate, so Corollary 2 is not an independent output of the new framework. The abstract's word \"implies\" is a bit strong; \"recovers\" would be more honest. This does not damage Theorem 1, whose proof is self-contained modulo the stated assumptions, but it does mean the paper delivers one new theorem plus a repackaging of known results, not a new proof of the Ruelle local constancy.\n\nAlso: the assumptions (A.i)–(A.v) are heavy and the paper does not claim they are optimal. That is a limitation, not a flaw. The reliance on [SS24] for the flat determinant formalism is legitimate self-citation — that is where the formalism is developed.\n\nWho this is for: spectral geometers and people working on dynamical zeta functions who want a unifying algebraic perspective. It deserves a serious referee. The referee should ask the authors to either prove (A.v) in the Anosov case or explicitly state that Corollary 2 is a recovery of DGRS20 using their theorem. With that change, the paper is publishable.\n\nRecommendation: engage with it, send it to peer review.","headline":"A careful, honest unification of two known local-constancy results, with a genuinely new general theorem whose Ruelle application borrows its hardest analytic step from DGRS20.","tokens_in":55381,"tokens_out":1584,"would_cite":true,"duration_ms":17880,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["58J52","37D40","58J35"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that flat-regularized superdeterminants of commutators $[\\delta_\\tau,d_\\nabla]$ are locally constant along inner variations of regular codifferentials, unifying local constancy of analytic torsion and of the Ruelle zeta…","keywords":["flat superdeterminant","general codifferential","analytic torsion","Ruelle zeta function","Anosov flows","contact geometry","local constancy","inner variation"],"falsifier":"Take any smooth inner variation of regular general codifferentials satisfying Assumptions (A.i)-(A.iv) for which $G(\\tau,\\lambda,s+1)$ has a pole in $\\lambda$ at $\\lambda=0$, so that Assumption (A.v) fails. If $\\lambda G(\\tau,\\lambda,1)$ has a nonzero limit as $\\lambda\\to0$ while $G(\\tau,\\lambda,0)$ stays bounded, then the first term of equation (40) contributes a nonzero $s$-derivative at $(0,0)$ and the flat superdeterminant must change with $\\tau$; the contrast with the constant elliptic and contact-Anosov cases would pinpoint exactly which analytic-continuation property forces the result.","tokens_in":54384,"feed_emoji":"📐","tokens_out":14147,"duration_ms":121520,"temperature":0.7,"pith_summary":"The paper aims to prove that a regularized superdeterminant of the graded commutator $D_\\tau=[\\delta_\\tau,d_\\nabla]$, restricted to the image of a general codifferential $\\delta_\\tau$, is constant along smooth families, provided the family moves by an inner variation and a technical analytic-continuation condition holds. The payoff is a common explanation for two known local-constancy phenomena that look analytically very different: the independence of the analytic torsion from the choice of Riemannian metric, and the local constancy of the value at zero of the Ruelle zeta function along regular contact Anosov flows. The theorem is stated for the twisted de Rham complex, but the proof uses only the graded commutator structure plus flat-trace analysis, so the mechanism is not tied to ellipticity of the operator. The contribution is this unification, with the geometric and analytic assumptions isolated rather than left implicit.","feed_headline":"One mechanism proves both torsion and Ruelle zeta are locally constant","feed_subtitle":"Regularized superdeterminants built from codifferentials do not change along metric or Anosov-flow families.","key_machinery":"$D_\\tau=[\\delta_\\tau,d_\\nabla]$ with $L_\\tau=\\operatorname{im}(\\delta_\\tau)$ is the characteristic operator of the family of general codifferentials $\\delta_\\tau$: degree $-1$ nilpotent differential operators with graded symmetry and an acyclic, $D_\\tau$-invariant complement. The argument is carried by the derivative identity $\\frac{d}{d\\tau}F(\\tau,\\lambda,s)=\\lambda\\Gamma(s)^{-1}G(\\tau,\\lambda,s+1)-s\\Gamma(s)^{-1}G(\\tau,\\lambda,s)$, where $F$ is the Mellin-type function whose $s$-derivative at $(0,0)$ gives the log-determinant and $G$ is built from the flat supertrace $\\operatorname{str}^{\\flat}(\\theta_\\tau e^{-tD_\\tau})$. Assumption (A.v) supplies the analytic continuation of $G$ to $(0,0)$ and $(0,1)$, making the prefactors $\\lambda$ and $s$ kill the derivative at the evaluation point. Lemmas 3.10 and 3.11 remove the restriction to $L_\\tau$ and reduce the $\\tau$-derivative to a commutator using the inner-variation structure $\\dot{\\delta}_\\tau=[\\theta_\\tau,\\delta_\\tau]$.","core_discovery":"Theorem 1 asserts that if $\\tau\\mapsto\\delta_\\tau$ is a smooth family of regular general codifferentials, $\\delta_\\tau$ is an inner variation of $\\delta_0$, and Assumptions (A.i)-(A.v) hold, then $\\operatorname{sdet}^{\\flat}(D_\\tau|_{L_\\tau})=\\operatorname{sdet}^{\\flat}(D_0|_{L_0})$ for all $\\tau\\in(-1,1)$. The proof differentiates the auxiliary function $F(\\tau,\\lambda,s)$ defining the flat superdeterminant, obtains $\\frac{d}{d\\tau}F=\\lambda\\Gamma(s)^{-1}G(\\tau,\\lambda,s+1)-s\\Gamma(s)^{-1}G(\\tau,\\lambda,s)$, and then uses Assumption (A.v) to extend this identity to $\\lambda=s=0$, where both prefactors make the $s$-derivative vanish. The authors then verify the hypotheses in two cases: $\\delta_\\tau=\\delta_{g_\\tau}$ for a smooth family of metrics, where $D_\\tau$ is the twisted Hodge Laplacian and the restricted superdeterminant is the analytic torsion; and $\\delta_\\tau=\\iota_{X_\\tau}$ for a smooth family of regular contact Anosov vector fields, where $D_\\tau$ is the Lie derivative and the same superdeterminant equals the value at zero of the Ruelle zeta function. Corollaries 1 and 2 draw the two local-constancy statements from the single theorem.","pith_inferences":["The mechanism suggests that local constancy is a formal consequence of the graded commutator structure plus one analytic-continuation condition; a natural test is to look for inner variations of regular codifferentials where $G$ develops a singularity at $(0,1)$ and see whether the restricted superdeterminant genuinely moves.","Because Assumption (A.v) in the Ruelle case is cited from [DGRS20, Theorem 4] rather than proved in the flat-determinant language of this paper, Corollary 2 is not yet self-contained; re-deriving that continuation from the Guillemin trace formula would close the gap.","The ratio formulation $Z(\\delta_\\tau)=\\operatorname{sdet}^{\\flat}(D_\\tau|_{L_\\tau})/\\operatorname{sdet}^{\\flat}(\\delta_\\tau)$ is locally constant in the two special cases because the denominator factors are manageable there; a family interpolating between $\\iota_X$ and $\\delta_g$ with a well-defined $\\operatorname{sdet}^{\\flat}(\\delta_\\tau)$ would convert Fried's conjecture into a constancy questi","Non-contact Anosov flows mark the boundary of the method: the required isotropic splitting is only Hölder continuous there, so extending the theorem would most likely need a non-smooth counterpart of the splitting and of the flat-trace argument."],"forward_implications":["Corollary 1: for an acyclic twisted de Rham complex, the analytic torsion is locally constant as a function on the space of Riemannian metrics over $M$.","Corollary 2: the value at zero of the Ruelle zeta function is locally constant along smooth families of regular contact Anosov vector fields.","Both the Hodge and the contact families are integrable, meaning $\\delta_\\tau=\\beta_\\tau\\delta_0\\beta_\\tau^{-1}$, so the inner-variation hypothesis is automatic there; in the metric case $\\beta_\\tau$ comes from the Hodge star operators and in the flow case from a bundle automorphism moving $X_0$ to $X_\\tau$.","The same proof applies to any graded vector bundle with a degree-one differential and a family of degree-minus-one operators forming an inner variation, so the mechanism is not special to the twisted de Rham complex.","The proof displays explicitly, in equation (40), where analytic continuation enters; any extension to a broader class of characteristic operators must reproduce or replace that step."],"supporting_citations":[{"why":"Supplies the original local-constancy theorem for analytic torsion along metric families that Corollary 1 recovers, and fixes the quantity the Hodge case must reproduce.","marker":"[RS71]"},{"why":"Supplies the analytic continuation of $G(\\tau,\\lambda,1)$ to $\\lambda=0$ (their Theorem 4) that provides Assumption (A.v) in the Ruelle case, and the local-constancy result Corollary 2 extends.","marker":"[DGRS20]"},{"why":"Provides the microlocal treatment of Pollicott-Ruelle resonances, the meromorphic resolvent, and the flat-trace identification of $\\operatorname{sdet}^{\\flat}(L_X|_{\\operatorname{im}\\iota_X})$ with $\\zeta_{X,\\rho}(0)$.","marker":"[DZ16]"},{"why":"Defines the flat-regularized superdeterminant and the restriction procedure $\\operatorname{sdet}^{\\flat}(D|_L)$ used throughout Theorem 1.","marker":"[SS24]"},{"why":"Justifies the cyclicity of the flat trace used in Lemmas 3.10 and 3.11 to eliminate the restriction to $L_\\tau$.","marker":"[CD24]"},{"why":"Supplies heat kernel asymptotics and Weyl-law eigenvalue estimates used to verify Assumptions (A.iv) and (A.v) in the elliptic case.","marker":"[Gil95]"},{"why":"Supplies the Duhamel formula and heat kernel construction used for the $\\tau$-derivative of the semigroup in Lemma 3.8 and Appendix A.","marker":"[BGV92]"}],"fun_headline_variants":["Single mechanism: torsion and Ruelle zeta stay put","Constant superdeterminants unify torsion and Ruelle zeta","Local constancy proven via flat-regularized superdeterminants","One theorem covers both Ray-Singer torsion and Ruelle zeta","Superdeterminant invariance with metric and Anosov families"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is Assumption (A.v): an auxiliary function $G(\\tau,\\lambda,s)$ built from the flat trace of $\\theta_\\tau e^{-tD_\\tau}$ must continue analytically to $\\lambda=0,s=0$ and $\\lambda=0,s=1$ and stay locally bounded in $\\tau$; in the Ruelle case this continuation is imported from a cited theorem rather than proved here.","fun_headline_variants_meta":{"raw":{"variants":["Single mechanism: torsion and Ruelle zeta stay put","Constant superdeterminants unify torsion and Ruelle zeta","Local constancy proven via flat-regularized superdeterminants","One theorem covers both Ray-Singer torsion and Ruelle zeta","Superdeterminant invariance with metric and Anosov families"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000767,"raw_usage":{"total_tokens":3454,"prompt_tokens":1052,"completion_tokens":2402,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":668,"completion_tokens_details":{"reasoning_tokens":2324}},"tokens_in":668,"tokens_out":2402,"duration_ms":16286,"temperature":1.0,"reasoning_tokens":2324,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:53:12.897719+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take any smooth inner variation of regular general codifferentials satisfying Assumptions (A.i)-(A.iv) for which $G(\\tau,\\lambda,s+1)$ has a pole in $\\lambda$ at $\\lambda=0$, so that Assumption (A.v) fails. If $\\lambda G(\\tau,\\lambda,1)$ has a nonzero limit as $\\lambda\\to0$ while $G(\\tau,\\lambda,0)$ stays bounded, then the first term of equation (40) contributes a nonzero $s$-derivative at $(0,0)$ and the flat superdeterminant must change with $\\tau$; the contrast with the constant elliptic and contact-Anosov cases would pinpoint exactly which analytic-continuation property forces the result.","supporting_citations":[],"review_version":1}