{"id":"da6794bc-55a0-4d5c-a4a3-03de2e4f5692","arxiv_id":"2412.12714","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The squared Lorentzian Dirac operator on small Minkowski perturbations has real spectrum plus isolated resonances, and a zeta-function residue equals the scalar curvature plus twisting curvature.","lead":"This paper proves that the square of the Dirac operator on small perturbations of Minkowski space has real spectrum apart from possible isolated resonances, and that the poles of its complex powers encode the spacetime's scalar curvature. It gives the first rigorous Lorentzian-signature analogue of Connes' spectral action principle for Dirac operators, opening noncommutative geometry to curved spacetimes without Euclideanization.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 5.7's displayed residue formula differs by a factor of -2 from the derivation via Eq. (5.19) and u1(0)=R_g/12+F^E; the scalar-curvature coefficient in the central claim is unsupported as stated.","rationale":"The paper develops a serious and technically sophisticated microlocal framework to define complex powers of the Lorentzian Dirac square and extract local invariants. The reader's conditional verdict is reasonable: the engineering of the resolved scattering calculus and the semiclassical estimates is substantial, and the non-trapping smallness hypothesis, while only asserted qualitatively, is a standard type of open condition that could plausibly be established with a more quantitative argument. However, the single most load-bearing concern is the internal contradiction in the headline residue formula. The proof's own equations (5.19) and the computed transport coefficient u1(0)=R_g/12+F^E lead to a residue that differs by a factor of −2 in both the scalar curvature and twisting-curvature terms from the formula displayed in Theorem 5.7 and in the abstract. This is not an outside-consensus disagreement; it is a direct algebraic inconsistency within the manuscript. The displayed formula is the central advertised result, so as written the theorem is not proven by the paper's own derivation. Since the error appears to be a localized coefficient/sign error rather than a failure of the whole microlocal strategy, the appropriate verdict remains conditional: the paper should not be accepted without the authors reconciling Eq. (5.19) with Theorem 5.7 and Remark 5.8, and correcting whichever expression is wrong. I therefore keep the reader's CONDITIONAL verdict unchanged, while flagging that the residue-coefficient issue is more immediately damaging to the stated central claim than the non-trapping smallness gap, with which I only partially agree as the weakest point.","tokens_in":43989,"tokens_out":9412,"duration_ms":80578,"concrete_test":"Recompute the residue in two independent ways: (a) evaluate Eq. (5.19) at k=1 with u1(0)=R_g/12 1_E + F^E and trace; (b) derive the h^{−n+2} coefficient in Remark 5.8 from the spectral action formula and compare with the residue using the Mellin relation. If the two agree with each other but not with Theorem 5.7, the theorem's displayed constants are erroneous and must be replaced by the derived values (or the derivation, if wrong, must be corrected with a precise sign/factor explanation).","verdict_should_be":"UNCHANGED","load_bearing_attack":"Substituting k=1 into Eq. (5.19) gives res_{α=n/2−1}(P−iε)^{−α}(x,x) = i u1(x,x)/(2^n π^{n/2} Γ(n/2−1)). With u1(0) = (R_g/12)1_E + F^E from the proof of Theorem 5.7, the traced residue equals i(rk(E)R_g/12 + tr_E(F^E))/((4π)^{n/2}Γ(n/2−1)), because 2^n π^{n/2} = (4π)^{n/2}. The displayed Theorem 5.7 asserts rk(E)R_g/(i6(4π)^{n/2}Γ) + 2tr_E(F^E)/(i(4π)^{n/2}Γ) = −i rk(E)R_g/(6(4π)^{n/2}Γ) − 2i tr_E(F^E)/((4π)^{n/2}Γ), which is exactly −2 times the derived value for each geometric term. Thus the central formula does not follow from the paper's own parametrix calculation: both the coefficient (R_g/12 vs R_g/6) and the sign are off. Remark 5.8's small-h expansion also lists rk(E)R_g/12 + tr(F^E) as the h^{−n+2} geometric coefficient, corroborating the derivation rather than the displayed theorem. This is a load-bearing inconsistency because the advertised Lorentzian spectral action coefficient changes.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies P = -/D^2, the square of a Lorentzian Dirac operator on a small perturbation of Minkowski space, viewed as a non-self-adjoint scattering pseudodifferential operator with a small or decaying imaginary part relative to an auxiliary positive Hermitian form. The main results are: (i) Theorem 1.1 (cf. Theorems 3.3 and 3.9), asserting that the closure of P has spectrum consisting of the real line plus isolated resonances in a horizontal strip, with smooth resonant states; and (ii) Theorem 1.2 (cf. Theorem 5.7), asserting that the zeta-density trace tr_E((P - i epsilon)^{-alpha})(x,x) extends meromorphically with poles at n/2, n/2 - 1, ..., 1 and that the residue at alpha = n/2 - 1, in the limit epsilon -> 0+, is a local expression in the Lorentzian scalar curvature R_g and the twisting curvature F^E, intended as a Lorentzian spectral action coefficient. The proof introduces a resolved scattering calculus Psi^{m,k,ell}_{sc,qsc} and a further resolved classical-semiclassical calculus Psi^{m,k,ell,p,q,r}_{qsc,sc,hbar^2,hbar,cl} to prove radial, propagation, and large-parameter elliptic estimates, and then applies a vector-bundle Hadamard parametrix following the authors' earlier work [14].","tokens_in":44280,"tokens_out":15413,"duration_ms":125717,"significance":"The strengths are substantial if the central formula holds: the paper extends the Lorentzian spectral zeta program from the scalar wave operator to Dirac-type operators, overcoming the indefinite-form obstruction via the auxiliary positive scalar product (5.14); the two new calculi are original tools with likely wider applicability; the resolvent estimates are made uniform with O(<lambda>^{-1}) decay along the integration contour, allowing the zeta continuation without a global hyperbolicity assumption; and Lemma 5.6 quantifies the contour ambiguity as finite-rank smoothing, so the residues are unambiguous. The residue formula is a concrete, parameter-free, falsifiable prediction, and the vector-bundle Bochner-Lichnerowicz computation is explicit (u_0 and u_1 transport equations). These merits are real, but they are presently undercut by the internal inconsistency described in Major Comment 1: the displayed Theorem 5.7 does not follow from the paper's own Eq. (5.19) and u_1(0) computation, and differs from it by a factor of -2 on both geometric terms.","major_comments":[{"comment":"The displayed residue formula of Theorem 5.7 does not follow from the paper's own computation. Substituting k = 1 into Eq. (5.19) gives res_{alpha = n/2 - 1} (P - i epsilon)^{-alpha}(x,x) = i u_1(x,x)/(2^n pi^{n/2} Gamma(n/2 - 1)) = i u_1(x,x)/((4 pi)^{n/2} Gamma(n/2 - 1)). With u_1(0) = (R_g/12) 1_E + F^E as computed in Step 3, the traced residue equals i(rk(E) R_g/12 + tr_E(F^E))/((4 pi)^{n/2} Gamma(n/2 - 1)). The theorem instead states rk(E) R_g/(i 6 (4 pi)^{n/2} Gamma) + 2 tr_E(F^E)/(i (4 pi)^{n/2} Gamma) = -i rk(E) R_g/(6 (4 pi)^{n/2} Gamma) - 2i tr_E(F^E)/((4 pi)^{n/2} Gamma), which is exactly -2 times the derived value for each geometric term. Remark 5.8's small-h expansion lists rk(E) R_g/12 + tr(F^E) as the h^{-n+2} geometric coefficient, agreeing with the derivation and not with the displayed theorem. Thus the central advertised formula is internally inconsistent as written; the authors must determine which expression is correct, correct the other, and re-verify the consequences for the Lorentzian spectral action claim.","section":null},{"comment":"The assertion that 'small perturbations of Minkowski space' satisfy the non-trapping hypothesis is not proved. The manuscripts states that version (2) of Definition 5.3 is verified for sufficiently small perturbations of the Minkowski metric in the sense of scattering metrics, but no smallness bound is given and no argument is supplied that the radial source/sink structure L_-/L_+ of Definition 2.9 persists under such perturbations, or that no trapped bicharacteristics appear (which would invalidate Propositions 2.14-2.15 and hence Theorems 3.3 and 5.7). Since the abstract and Theorems 1.1-1.2 are stated for 'small perturbations of Minkowski space', this is a load-bearing gap: the spectral and residue results are proved only for the abstract class of Definition 5.3, and the advertised application requires a quantitative stability lemma (e.g., explicit Psi^{2,0} seminorm bounds under which the radial points remain sources/sinks and all bicharacteristics escape). Lemma 5.4 (fast-decaying perturbations) proves only the decay of P - P*, not non-trapping, so it does not fill the gap.","section":null},{"comment":"The large-|Im lambda| elliptic and propagation estimates are load-bearing for Theorem 3.3 and for the O(<lambda>^{-1}) resolvent decay used in Step 1 of the proof of Theorem 5.7, but they are presented as sketches. The proof of Proposition 2.12 explicitly defers the needed elliptic estimate to Section 4, and Theorem 4.2 is stated after a discussion in which microlocalizers and error terms are repeatedly suppressed and the passage from the resolved semiclassical algebra to classical spaces is described as 'obtained similarly as in [51]' with only a reference. For the framework to be verifiable, the composition rules, the normal operator and full ellipticity condition of Definition 4.1, and the derivation of (4.12) from the resolved calculus should be stated as lemmas with proofs, or at least with precise statements of the symbol classes and mapping properties involved.","section":null}],"minor_comments":[{"comment":"The phrase 'similarly as in as in [14]' contains a duplicated 'as in' and should be corrected.","section":null},{"comment":"The sentence 'which is in dot C' uses an undefined symbol; presumably a fast-decay space is meant, but this should be spelled out.","section":null},{"comment":"The endomorphism space is written End(S_x) although the setup of Theorem 5.7 uses a Clifford module E; the formula should read End(E_x), or the trace should be taken in E.","section":null},{"comment":"The threshold is written 'k - s + beta_tilde_- < -1/2 on L_+' with a minus subscript on beta_tilde at the sink L_+; the subscript should be matched to L_+/L_- as in Eq. (2.16) to avoid confusion about which subprincipal correction is used.","section":null},{"comment":"The caption spans nearly a full page and is very difficult to follow; moving the technical details of the alternative blow-up order into the text and shortening the caption would improve readability.","section":null}],"recommendation":"major_revision","confidential_remarks":"The most likely fix is the displayed Theorem 5.7 (and the corresponding Theorem 1.2 and introduction formulas), since Eq. (5.19), the u_1(0) computation, and Remark 5.8 are mutually consistent; nevertheless the authors should re-check the normalization of (5.19), including the factor 2^n and the sign of the contour integral, against [14], since the orientation of Z_epsilon and the branch of (z - i epsilon)^{-alpha} affect exactly such factors. No citation-pattern issues: the reliance on [14] and [48] is transparent and appropriate. If the authors correct the residue formula and supply the non-trapping persistence argument, the paper should be publishable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Thanks for the report. I read the paper and I think you're right about the main issue, and I'd phrase it even more strongly. The central residue formula in Theorem 5.7 does not follow from the proof. From (5.19) with k=1, res = i u1 / (2^n π^{n/2} Γ(n/2-1)). With u1(0) = Rg/12 + F^E, the traced residue is i (rk Rg/12 + tr F^E) / ((4π)^{n/2} Γ). The theorem gives rk Rg/(i6(4π)^{n/2}Γ) + 2 tr F/(i(4π)^{n/2}Γ), which is exactly -2 times that. Remark 5.8's small-h coefficient (Rg/12 + trF) backs the derivation, not the theorem. So the advertised Lorentzian spectral action coefficient is wrong as stated. This is not a philosophical quibble; the whole point of the paper is that residue. It might be a typo in the theorem and the correct formula is probably the one from the computation, but the authors need to reconcile it. The spectrum theorem (Theorem 1.1) is the other main result and it looks credible. The novelty is real: the non-self-adjoint Dirac square is not just a routine extension of [14], and the resolved scattering / semiclassical calculi in Sections 2.4 and 4 are serious new tools. The non-trapping smallness in Definition 5.3 is treated qualitatively, which is a bit unsatisfying but common in this literature. Several proofs are sketchy (Prop 2.12 and the semiclassical section), so checking this paper is hard work. Still, the overall architecture is coherent and the authors engage honestly with prior work; this is not a paper with a hidden fitted parameter. It deserves a proper referee. I'd send it out, with a specific request to audit the residue computation and reconcile Theorem 5.7 with (5.19), u1(0), and Remark 5.8. My own verdict is conditional: I would not cite this version, but I expect the fix to be straightforward and then the paper becomes a solid contribution. For a reading group, it could be useful to go through the discrepancy, but the length and technicality make it a maybe.","headline":"Technically serious paper whose central residue formula appears to be off by a factor -2 and a sign from its own derivation.","tokens_in":44853,"tokens_out":3905,"would_cite":false,"duration_ms":33686,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35L05","58J40","58J50","53C27"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that on small perturbations of Minkowski space, the squared Lorentzian Dirac operator has only real spectrum plus isolated resonances, and the residue of its spectral zeta density at $\\alpha=n/2-1$ is a local invariant…","keywords":["Lorentzian Dirac operator","spectral zeta function density","scattering calculus","radial estimates","resonances","non-trapping spacetimes","local invariants","Bochner–Lichnerowicz formula"],"falsifier":"Take Minkowski space in dimension 4 and add a small compactly supported bump metric engineered to create a stable, trapped null geodesic (a photon sphere). Compute or numerically estimate whether the closure of $P=-\\slash{D}^2$ still has spectrum only on $\\mathbb{R}$ plus isolated resonances in a strip, and whether the residue at $\\alpha=n/2-1$ still matches the scalar-curvature formula; a trapped geodesic producing resonances accumulating on $\\mathbb{R}$ or a residue one full derivative different from the formula would falsify the small-perturbation claim as stated.","tokens_in":117,"feed_emoji":"⚛️","tokens_out":9335,"duration_ms":142393,"temperature":0.7,"pith_summary":"The paper's goal is to give the Lorentzian Dirac operator the same spectral-zeta treatment that works for wave operators on asymptotically Minkowski spaces. For $P = -\\slash{D}^2$ on a small perturbation of Minkowski space, it claims the spectrum consists of the real axis plus isolated resonances in a strip, with smooth resonant states. Its sharper claim is that the meromorphic residue of the spectral zeta density $\\operatorname{tr}_E (P-i\\varepsilon)^{-\\alpha}(x,x)$ at $\\alpha = n/2-1$, taken in the limit $\\varepsilon\\to 0^+$, is a local invariant built from the Lorentzian scalar curvature and the twisting curvature. If true, the spectral action principle of noncommutative geometry has a well-defined Lorentzian version, despite the Dirac operator's indefinite Hermitian form. The route is microlocal: radial estimates in a resolved scattering calculus control the resolvent uniformly along a complex contour, and a Hadamard parametrix converts the contour integral into curvature data.","feed_headline":"Dirac zeta residues recover Lorentzian scalar curvature","feed_subtitle":"A microlocal proof makes the spectral action well-defined on small Minkowski perturbations.","key_machinery":"The central technical object is the resolved scattering phase space $\\operatorname{sc,res}T^*M = [\\operatorname{sc}T^*M; \\operatorname{sc}S^*_{\\partial M}M]$, obtained by blowing up the corner at fiber infinity and base infinity, together with the further blow-up at $h=0$ in fiber infinity that connects the semiclassical and classical pseudodifferential algebras. In the resolved calculus $\\Psi^{m,k,\\ell}_{sc,qsc}$ the relative order $k-s$ replaces the uncomfortable threshold $-\\tfrac12$ in the radial estimates, so propagation and radial estimates give Fredholm estimates with adjustable trade-offs between regularity and decay. A second ingredient, the Bochner–Lichnerowicz formula $-\\slash{D}^2 = \\nabla^{E*}\\nabla^E + F^E + \\tfrac14 R_g\\,\\mathrm{Id}_E$, identifies the first transport coefficient in the Hadamard parametrix with the scalar and twisting curvature, turning the contour-integral residue into the displayed local invariant.","core_discovery":"The paper establishes that, although $-\\slash{D}^2$ is only formally self-adjoint for a non-positive Hermitian form, its spectral theory is controllable after endowing the spinor bundle with the auxiliary positive scalar product $\\langle u,v\\rangle = \\langle u,\\gamma(e)v\\rangle_S$. The main theorem states that for every $\\varepsilon>0$ the diagonal restriction $(P-i\\varepsilon)^{-\\alpha}(x,x)$ is meromorphic in $\\alpha$ with poles at $n/2, n/2-1, \\dots, 1$, and\n$$\\lim_{\\varepsilon\\to0^+}\\operatorname{res}_{\\$\\alpha$=n/2-1} \\operatorname{tr}_E (P-i\\varepsilon)^{-\\$\\alpha$}(x,x) = \\frac{\\operatorname{rk}(E)R_g(x)}{i6(4\\pi)^{n/2}\\Gamma(n/2-1)} + \\frac{2\\operatorname{tr}_E(F^E)(x)}{i(4\\pi)^{n/2}\\Gamma(n/2-1)},$$\nwhere $R_g$ is the scalar curvature and $F^E$ the twisting curvature of the Clifford module $E$. In the same package, the resolvent is shown to be meromorphic with finite-multiplicity poles whose resonant states are smooth, and complex powers are defined up to finite-rank smoothing ambiguities that do not affect the residue. This is the Lorentzian analogue of the heat-kernel coefficient computation that underlies the spectral action.","pith_inferences":["Beyond the paper, one testable extension is to compute the same residue for a metric whose null geodesic flow is non-trapping but has a normally hyperbolic trapped set; the framework suggests the local invariant would persist, but this is not claimed in the paper.","The formula pins the $h^{-n+2}$ coefficient of the Lorentzian spectral action to $R_g/12+\\operatorname{tr}(F^E)$, which in a physical model would tie the gravitational and matter actions to the same constant; this numerical identification is an editorial extrapolation of the displayed residue.","A quantitative smallness bound for 'small perturbation' would let the result be verified by explicit examples, e.g. perturbing $\\mathbb{R}^{1,3}$ by a bump metric and checking the resonance-strip width; the paper leaves the size implicit."],"forward_implications":["The spectral action principle can be formulated for Lorentzian perturbations of Minkowski: $f(h(P+i\\varepsilon))$ has a small-$h$ asymptotic expansion whose leading $h^{-n+2}$ coefficient contains $R_g/12 + \\operatorname{tr}(F^E)$, the Lorentzian analogue of the Riemannian heat-kernel coefficient.","The resonance structure is qualitatively the same as for wave operators: off the real axis the spectrum is a discrete set of poles of finite multiplicity, and every generalized resonant state is smooth in the interior; this supports studying Dirac-type fields with the same Feynman-resolvent tools used for $\\square_g$.","The definition of $(P-i\\varepsilon)^{-\\alpha}$ depends on the contour only by a finite-rank smoothing operator, so all diagonal residues and the associated local invariants are unambiguous.","Because the resolvent estimates replace the global hyperbolicity assumption used in earlier parametrix constructions with non-trapping dynamics, the same heat-kernel coefficients should be computable without assuming global hyperbolicity."],"supporting_citations":[{"why":"the scalar-case Hadamard parametrix and residue computation whose bundle version is proved here","marker":"[14]"},{"why":"the scattering-calculus radial estimates on Lorentzian scattering spaces that the resolved estimates extend","marker":"[48]"},{"why":"the blow-up at $h=0$ linking semiclassical and classical pseudodifferential algebras used for high-energy estimates","marker":"[51]"},{"why":"the quadratic scattering calculus identified with the resolved calculus at the front face","marker":"[55]"},{"why":"the Clifford-module Bochner–Lichnerowicz formula used to identify the coefficient with curvature","marker":"[36]"},{"why":"the propagation and radial-estimate framework the paper refines","marker":"[46]"},{"why":"the model semiclassical resolvent estimates for asymptotically Euclidean scattering","marker":"[53]"}],"fun_headline_variants":["Dirac zeta poles on Minkowski perturbations give scalar curvature","Microlocal calculus shows Dirac zeta residues encode Lorentzian curvature","Small Minkowski perturbations: Dirac zeta poles tie to local invariants","Resonant states of Dirac operator on Minkowski-like spaces traced","From Dirac zeta poles to Lorentzian scalar curvature via microlocal analysis"],"cache_read_input_tokens":46848,"weakest_assumption_plain":"The load-bearing premise is that every sufficiently small perturbation of Minkowski space is non-trapping, meaning no lightlike geodesic is trapped: each one flows from a radial source at past infinity to a radial sink at future infinity, and no explicit bound or persistence proof is given for this property.","fun_headline_variants_meta":{"raw":{"variants":["Dirac zeta poles on Minkowski perturbations give scalar curvature","Microlocal calculus shows Dirac zeta residues encode Lorentzian curvature","Small Minkowski perturbations: Dirac zeta poles tie to local invariants","Resonant states of Dirac operator on Minkowski-like spaces traced","From Dirac zeta poles to Lorentzian scalar curvature via microlocal analysis"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000222,"raw_usage":{"total_tokens":1444,"prompt_tokens":925,"completion_tokens":519,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":541,"completion_tokens_details":{"reasoning_tokens":423}},"tokens_in":541,"tokens_out":519,"duration_ms":4892,"temperature":1.0,"reasoning_tokens":423,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T13:49:25.900217+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take Minkowski space in dimension 4 and add a small compactly supported bump metric engineered to create a stable, trapped null geodesic (a photon sphere). Compute or numerically estimate whether the closure of $P=-\\slash{D}^2$ still has spectrum only on $\\mathbb{R}$ plus isolated resonances in a strip, and whether the residue at $\\alpha=n/2-1$ still matches the scalar-curvature formula; a trapped geodesic producing resonances accumulating on $\\mathbb{R}$ or a residue one full derivative different from the formula would falsify the small-perturbation claim as stated.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"the scalar-case Hadamard parametrix and residue computation whose bundle version is proved here"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"the scattering-calculus radial estimates on Lorentzian scattering spaces that the resolved estimates extend"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"the blow-up at $h=0$ linking semiclassical and classical pseudodifferential algebras used for high-energy estimates"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"the quadratic scattering calculus identified with the resolved calculus at the front face"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"the Clifford-module Bochner–Lichnerowicz formula used to identify the coefficient with curvature"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"the propagation and radial-estimate framework the paper refines"},{"cited_title":"Vasy and M","cited_arxiv_id":null,"evidence_quote":"the model semiclassical resolvent estimates for asymptotically Euclidean scattering"}],"review_version":1}