{"id":"39be2d1f-969e-4d92-b9cf-4aa0814f42e1","arxiv_id":"2501.01252","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Scalar curvature on any closed even-dimensional manifold with nonzero Euler characteristic can be increased by at most an explicit constant, the gap, which is a function of the minimal eigenvalue of the curvature operator of the starting metric.","lead":"On any even-dimensional closed space with nonzero Euler characteristic, every metric is epsilon-gap extremal: no larger metric can increase the scalar curvature by more than a number determined by the most negative curvature of the original metric. A companion result for spaces with boundary bounds how much boundary mean curvature can exceed that of a Euclidean domain, addressing a question of Gromov about extending boundary metrics inward.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 6's boundary condition is the weak point: the paper asserts, without proof, that the Lott/Wang-Xie-Yu boundary condition gives a nonnegative boundary Dirac integral and a nonzero index once the nonnegativity assumptions are removed.","rationale":"The reader's weakest assumption and my stress-test converge on the same point: Theorem 6 is the least secure link. I examined the closed-manifold proof before accepting this. Inequality (9) is justified because R−Rmin is nonnegative, so the minimum of the Clifford operator −1/2∑ a_m(ω_m⊗ω_m) is −1/2∑ a_m, yielding exactly the claimed lower bound; inequality (8) similarly follows from λ_iλ_j≤1. The index formula for closed spin maps is standard. Thus Theorem 3 and Theorem 2 do not appear to have a hidden flaw. The boundary theorem, however, contains an unsupported assertion: the Lott/Wang-Xie-Yu boundary condition with nonnegative boundary Dirac integral and nonzero index is cited, not proved, after the nonnegativity assumptions have been dropped. If that condition fails, the boundary results collapse, but the central gap phenomenon survives. The appropriate verdict is CONDITIONAL, unchanged: the fix is to supply the missing boundary-condition argument or to restrict the statement.","tokens_in":12239,"tokens_out":37741,"duration_ms":342666,"concrete_test":"Check Lott [19, §2.2] and Wang-Xie-Yu [26, §3.1] for the exact hypotheses of the boundary condition: (i) does ∫∂N⟨φ,D∂Nφ⟩≥0 hold for all φ in the chosen boundary space without sign conditions on scalar curvature or mean curvature; and (ii) is Ind(D_E) proved to equal deg(f)χ(M), or does the argument require a vanishing eta invariant or nonnegativity of the curvature operator or second fundamental form? If either source uses such an assumption, Theorem 6 lacks the proof needed for the stated generality.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The closed-manifold estimate (Theorem 3) and its application (Theorem 2) appear sound: the Lichnerowicz-Weitzenböck-Bochner calculation follows the Goette-Semmelmann trick after splitting off Rmin, and the index identity Ind(D_E)=deg(f)χ(M) is standard for spin maps. The load-bearing weak point is the boundary theorem (Theorem 6, Section 3). After equation (11), the proof invokes 'the boundary condition as in section 2.2 in [19] or section 3.1 in [26]' to get ∫∂N⟨φ,D∂Nφ⟩≥0 and Ind(D_E)≠0. The paper has just removed all nonnegativity assumptions on the curvature operator and the second fundamental form, but it does not show that the cited boundary condition still has these properties. For a manifold with boundary, the Atiyah-Patodi-Singer index formula contains boundary eta terms, so Ind(D_E) is not automatically deg(f)χ(M). If the cited sources establish the nonvanishing index only under nonnegativity (where a Lichnerowicz-type estimate controls the boundary term), then the disjunction (4) or (5), Corollary 7, and the discussion of Gromov's Question 8 do not follow as stated. This is a missing proof, not a contradiction, so the closed-manifold gap theorem is unaffected.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper derives scalar curvature comparison estimates for spin manifolds without imposing nonnegativity of the curvature operator. For closed manifolds, Theorem 3 gives an upper bound for inf_x(Sc_g(x)-Sc_{g0}(f(x))) in terms of the minimum eigenvalue Rmin of the curvature operator of the target metric, assuming an area-nonincreasing map of nonzero degree to a manifold with nonzero Euler characteristic. Theorem 2 converts this into an epsilon-gap distance extremality statement for any metric on a closed even-dimensional manifold with nonzero Euler characteristic. For compact manifolds with boundary, Theorem 6 states a similar interior-or-boundary alternative involving scalar curvature and mean curvature, and Corollary 7 applies this to Euclidean domains to address a question of Gromov. The proof of the closed-manifold theorem follows the Goette-Semmelmann Lichnerowicz-Weitzenbock-Bochner argument with a splitting of the curvature operator into Rmin plus a nonnegative part. The boundary theorem follows Lott and Wang-Xie-Yu but relies on a boundary condition whose availability is not proved in the present generality.","tokens_in":12453,"tokens_out":8606,"duration_ms":84795,"significance":"If the closed-manifold theorem is correct, it is a useful extension of Llarull and Goette-Semmelmann results: it removes the nonnegativity assumption on the target curvature operator and produces an explicit, parameter-free constant in the scalar curvature estimate. The application to epsilon-gap distance extremality in Theorem 2 is a clean and interesting consequence. The proof of Theorem 3 is essentially a direct Bochner-type calculation, and the index-theoretic input is standard. The boundary results, if made rigorous, would address a concrete aspect of Gromov's extension question. However, the boundary part currently contains unproved assumptions, and the Corollary 7 application does not verify the hypotheses of Theorem 6 as stated.","major_comments":[{"comment":"The proof of Theorem 6 invokes, without proof, a boundary condition from section 2.2 of [19] or section 3.1 of [26] that is required to satisfy both ∫∂N⟨φ,D∂Nφ⟩≥0 and Ind(DE)≠0. The present setting has removed the nonnegativity assumptions on the curvature operator and the second fundamental form, while the cited constructions may depend on such assumptions to control the boundary Dirac term and to identify the index. Since the Atiyah-Patodi-Singer formula contains boundary eta contributions, Ind(DE) is not automatically equal to deg(f)χ(M) once the boundary condition is changed. This is a load-bearing point: the disjunction (4) or (5), Corollary 7, and the discussion of Question 8 all depend on it. The authors should state the boundary condition as a lemma and prove that it exists in the present generality, or cite precisely the theorem that supplies it under the stated hypotheses.","section":"Section 3, paragraph after Eq. (11)"},{"comment":"Corollary 7 is stated as an application of Theorem 6 with f the identity map on a Euclidean domain, but the hypotheses of Theorem 6 are not verified. For f:(Ω,g)→(Ω,g_E), the area-nonincreasing condition requires |v∧w|_{g_E}≤|v∧w|_g for all tangent bivectors; this does not follow from Sc_g≥0 and is not implied by the definition of ĉ, which only controls tangent vectors on the boundary. The appearance of ĉ in the boundary inequality suggests that a variant of Theorem 6 with a conformally scaled target metric or a rescaled map is being used, but such an argument is not supplied. As written, Corollary 7 does not follow from Theorem 6 and needs either additional hypotheses or a separate proof.","section":"Corollary 7"},{"comment":"In the rigidity part of Theorem 6, the conclusion '∂f:∂N→∂M is local isometric if AM_min≠0' is asserted after saying all inequalities become equalities. The tracking of equalities is only sketched, and it is not shown that the equalities force the pointwise eigenvalue condition that makes ∂f a local isometry. Since this rigidity statement is not used in the application to Corollary 7, the authors could either supply the missing details or clearly mark this part as a secondary claim.","section":"Theorem 6, equality case"}],"minor_comments":[{"comment":"There is a typo: 'Stierel-Whitney c1ass' should be 'Stiefel-Whitney class'.","section":"Remark 4"},{"comment":"For manifolds with boundary, the degree of f is not explicitly defined; the authors should specify that f is orientation-preserving with respect to the boundary orientations and that deg(f) is the relative degree in H_{2n}(N,∂N).","section":"Theorem 6 statement"},{"comment":"The notation AM_min(g0)αi is used without defining the subscript notation; it would be clearer to write the components of the operator AM - AM_min g0 explicitly.","section":"Equation (14) and surrounding text"},{"comment":"The symbol f^*HM appears where (∂f)^*HM would be more precise, since HM is a function on ∂M and the pullback is along ∂f.","section":"Section 3, proof of inequality (13)"},{"comment":"The statement says 'a complete Riemannian metric g0 on M' although M is closed; completeness is automatic, so the wording is redundant but not harmful.","section":"Theorem 2"}],"recommendation":"major_revision","confidential_remarks":"The closed-manifold part of the paper is sound and likely publishable after routine corrections. The main risk is the boundary theorem: the missing justification of the Lott/Wang-Xie-Yu boundary condition is a genuine gap, and Corollary 7 seems not to follow from the stated Theorem 6 without additional argument. These issues are fixable within the scope of the paper, so I do not recommend rejection, but the revision needs to address them carefully."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nThe thing to know about this paper: the closed-manifold core is a genuine and likely correct advance, while the boundary theorem is asserted on top of an unverified borrowed boundary condition. If you read only one thing, read Section 2.\n\nWhat's new: Sun and Wang remove the nonnegative curvature operator assumption that was essential in Goette-Semmelmann and Lott. The trick is to split the target curvature operator as R = (R - Rmin) + Rmin and estimate both parts separately, using area nonincreasingness to control the Rmin term. The resulting inequality (Theorem 3) gives an explicit constant depending on Rmin. This is not a repackaging; I don't see it in earlier literature. The application to epsilon-gap extremality for every even closed manifold with nonzero Euler characteristic is a natural and attractive consequence. The index identity Ind(D_E) = deg(f)chi(M) is standard.\n\nI checked the algebra in (8) and (9) and it seems right. The proofs follow Llarull and Goette-Semmelmann closely; the square trick for the nonnegative operator R - Rmin is standard and correctly applied. The count of no free parameters is accurate.\n\nSoft spots, in order of importance:\n\n1. Theorem 6 (boundary case). After equation (11), the proof says 'we impose the boundary condition as in section 2.2 in [19] or section 3.1 in [26] such that int <phi, D_partial phi> >= 0 and Ind(D_E) != 0.' That is doing a lot of work. The paper has just discarded Lott's nonnegativity assumptions, but gives no argument that the boundary condition still delivers those two properties. This is not a minor gap: Theorem 6, Corollary 7, and the remarks on Gromov's Question 8 all rest on it. It may be fixable by importing Lott's actual boundary analysis, but as written it's an assertion, not a proof.\n\n2. The equality-case rigidity in Theorem 3 is sketched. In dimension 2, lambda_i lambda_j = 1 for i != j only gives lambda_1 lambda_2 = 1, not that both equal 1. So the claim that f is an isometry in the equality case may need a separate argument for surfaces. This doesn't affect the main estimate, but it should be fixed.\n\n3. Example 5 claims the estimate is optimal, but the calculation only shows both sides tend to 0 as c -> 0. That does not establish optimality of the coefficient. The example is harmless, but the claim is stronger than the evidence.\n\nBottom line: the closed-manifold gap theorem is a solid, useful result and deserves a serious referee. The boundary part needs real work before it can be taken as stated. I'd send it to review with a request that the referee focus on Theorem 6's boundary condition and the dimension-2 rigidity detail.\n\nBest,\n[You]","headline":"The closed-manifold scalar curvature estimate is a genuine and likely correct advance; the boundary theorem relies on an unverified borrowed boundary condition that needs real work.","tokens_in":13019,"tokens_out":10336,"would_cite":true,"duration_ms":90543,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C21","53C27","58J20"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves a scalar-curvature gap estimate with no sign restriction on the curvature operator, and uses it to show every metric on a closed even-dimensional manifold with nonzero Euler characteristic is ε-gap distance extremal.","keywords":["scalar curvature","epsilon-gap extremality","Dirac operator","curvature operator","spin manifold","Euler characteristic","mean curvature","fill-in problem"],"falsifier":"A counterexample would be a pair of closed spin manifolds $(M^{2n},g_0)$ and $(N^{2n},g)$ with $\\chi(M)\\neq 0$ and an area-nonincreasing map $f:N\\to M$ of nonzero degree for which $\\inf_x [\\mathrm{Sc}_g(x)-\\mathrm{Sc}_{g_0}(f(x))]$ is strictly greater than $-2n(2n-1)R_{\\min}+2n(2n-1)|R_{\\min}|$; such a pair would refute Theorem 3. The product $S^2\\times \\Sigma$ with varying hyperbolic surface factors is a concrete family in which the bound can be checked numerically, and the paper's Example 5 shows equality can be approached by taking the surface curvature to zero.","tokens_in":11977,"feed_emoji":"📐","tokens_out":10231,"duration_ms":87946,"temperature":0.7,"pith_summary":"This paper establishes a quantitative upper bound on how much scalar curvature can rise when a metric is enlarged on an even-dimensional closed manifold with nonzero Euler characteristic. The main estimate says that, for closed spin manifolds $(M^{2n}, g_0)$ and $(N^{2n}, g)$ linked by an area-nonincreasing map $f: N \\to M$ of nonzero degree, the infimum of $\\mathrm{Sc}_g - \\mathrm{Sc}_{g_0}(f)$ is at most $-2n(2n-1)R_{\\min} + 2n(2n-1)|R_{\\min}|$, where $R_{\\min}$ is the smallest eigenvalue of the curvature operator of $g_0$. Previous estimates of this type required the curvature operator to be nonnegative; the new argument removes that assumption. Taking $f$ to be the identity shows that every complete metric on such a manifold is $\\epsilon$-gap distance extremal for $\\epsilon = 2n(2n-1)(|R_{\\min}|-R_{\\min})$. A boundary version compares scalar curvature and mean curvature, and yields an obstruction to extending boundary metrics with prescribed lower bounds into Euclidean domains.","feed_headline":"Every metric is gap-extremal on even closed manifolds with χ≠0","feed_subtitle":"A Dirac-operator estimate with no sign condition on curvature caps how much scalar curvature can rise when a metric is enlarged.","key_machinery":"The twisted Dirac operator $D_E$ on the bundle $S(N) \\otimes f^*S(M)$, together with the Lichnerowicz-Weitzenböck-Bochner formula, is the load-bearing object. The new step is to write the curvature operator $R$ of the target as $(R - R_{\\min}) + R_{\\min}$, with $R_{\\min}$ the minimum eigenvalue, so that $R - R_{\\min}$ is nonnegative and admits a square root $L$; Clifford algebra estimates on $L$ then yield a lower bound on the curvature term without any sign assumption on $R$. In the boundary theorem the same split is applied to the second fundamental form $A^M$, with its minimum eigenvalue $A^M_{\\min}$, and a boundary condition borrowed from the cited literature makes the boundary spinor term nonnegative.","core_discovery":"The central claim is Theorem 3: under the hypotheses above, $\\inf_x [\\mathrm{Sc}_g(x) - \\mathrm{Sc}_{g_0}(f(x))] \\leq -2n(2n-1)R_{\\min} + 2n(2n-1)|R_{\\min}|$, and equality forces rigidity, namely $R_{\\min} \\geq 0$ and $\\mathrm{Sc}_g(x) = \\mathrm{Sc}_{g_0}(f(x))$. The proof runs through a twisted Dirac operator on $N$ with coefficients in the pullback spinor bundle, and the new ingredient is an estimate of the curvature term in the Lichnerowicz-Weitzenböck-Bochner formula that subtracts the minimum eigenvalue $R_{\\min}$ and controls the remainder with Clifford algebra inequalities. Because the index of the twisted Dirac operator is nonzero, being equal to $\\deg(f)\\chi(M)$, a harmonic spinor must exist, forcing the scalar-curvature gap. The identity-map case gives Theorem 2, the $\\epsilon$-gap distance extremality of every metric; the boundary case, Theorem 6, gives a similar alternative between a scalar-curvature gap in the interior and a mean-curvature gap on the boundary, with an application to Euclidean domains in Corollary 7.","pith_inferences":["A natural next step is to run the same minimum-eigenvalue subtraction through other Dirac-type operators; the signature operator, for instance, could yield gap bounds for metric variations of the curvature tensor, which the paper does not discuss.","Because the gap $\\epsilon$ grows with $|R_{\\min}|$, the extremality statement is strongest when the base metric has nonnegative curvature operator; this suggests that the new theorem is best viewed as the correct replacement for the old zero-gap result in the presence of negative curvature, not as a sharp rigidity statement.","The boundary theorem could be tested by checking whether the cited boundary condition can be constructed for arbitrary second fundamental forms; if it cannot, Corollary 7 would require an extra hypothesis on the boundary and the non-existence statement would be narrower than stated."],"forward_implications":["On any closed even-dimensional manifold with nonzero Euler characteristic, every complete metric is $\\epsilon$-gap distance extremal for $\\epsilon = 2n(2n-1)(|R_{\\min}|-R_{\\min})$, so no larger metric can raise the scalar curvature by more than $\\epsilon$.","If the scalar-curvature gap bound is attained with equality, the metric comparison is rigid: the map is an isometry, $R_{\\min} \\geq 0$, and the scalar curvatures agree, so the base metric is distance rigid in that case.","The estimate extends sphere-target and nonnegative-curvature-operator results to arbitrary curvature operators, interpolating between the nonnegative and nonpositive cases listed in the paper's table.","For a Euclidean domain with nonzero Euler characteristic, the boundary version rules out any metric with nonnegative scalar curvature that agrees with the Euclidean metric on the boundary while making the boundary mean curvature exceed a computable threshold.","The same theorem covers maps between different manifolds, so it constrains scalar-curvature gaps not only for identity maps but for all area-nonincreasing maps of nonzero degree onto such targets."],"supporting_citations":[{"why":"Supplies the curvature-term decomposition and square-root trick that the paper adapts to arbitrary curvature operator.","marker":"[9]"},{"why":"Provides the Clifford algebra inequalities and the index argument used in the equality case.","marker":"[18]"},{"why":"Supplies the boundary Dirac operator framework and the boundary condition used in Theorem 6.","marker":"[19]"},{"why":"Supplies the boundary condition ensuring a nonnegative boundary Dirac term and nonzero index.","marker":"[26]"},{"why":"Provides the original index-theoretic method and the nonpositive-curvature-operator case listed in the table.","marker":"[11]"},{"why":"Defines $\\epsilon$-gap distance extremality, the target concept of Theorem 2.","marker":"[12]"},{"why":"Poses the boundary metric extension question addressed by Corollary 7.","marker":"[13]"},{"why":"Gives the convex-boundary nonexistence result that Corollary 7 extends.","marker":"[4]"}],"fun_headline_variants":["Scalar curvature gap forced by nonzero Euler characteristic","Every metric gap-extremal on even closed χ≠0 manifolds","No curvature sign needed: scalar gap on even manifolds","Dirac operator caps scalar curvature on even closed manifolds","Gap phenomenon: χ≠0 makes all metrics gap-extremal"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The boundary theorem depends on the existence of a boundary condition, taken from the cited literature, that makes a boundary spinor integral nonnegative and the twisted Dirac operator have nonzero index, and the paper does not prove that this condition remains available after dropping the nonnegativity assumptions on the curvature operator and the second fundamental form.","fun_headline_variants_meta":{"raw":{"variants":["Scalar curvature gap forced by nonzero Euler characteristic","Every metric gap-extremal on even closed χ≠0 manifolds","No curvature sign needed: scalar gap on even manifolds","Dirac operator caps scalar curvature on even closed manifolds","Gap phenomenon: χ≠0 makes all metrics gap-extremal"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000869,"raw_usage":{"total_tokens":3757,"prompt_tokens":930,"completion_tokens":2827,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":546,"completion_tokens_details":{"reasoning_tokens":2752}},"tokens_in":546,"tokens_out":2827,"duration_ms":20902,"temperature":1.0,"reasoning_tokens":2752,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T22:33:35.887467+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A counterexample would be a pair of closed spin manifolds $(M^{2n},g_0)$ and $(N^{2n},g)$ with $\\chi(M)\\neq 0$ and an area-nonincreasing map $f:N\\to M$ of nonzero degree for which $\\inf_x [\\mathrm{Sc}_g(x)-\\mathrm{Sc}_{g_0}(f(x))]$ is strictly greater than $-2n(2n-1)R_{\\min}+2n(2n-1)|R_{\\min}|$; such a pair would refute Theorem 3. The product $S^2\\times \\Sigma$ with varying hyperbolic surface factors is a concrete family in which the bound can be checked numerically, and the paper's Example 5 shows equality can be approached by taking the surface curvature to zero.","supporting_citations":[{"cited_title":"Goette and U","cited_arxiv_id":null,"evidence_quote":"Supplies the curvature-term decomposition and square-root trick that the paper adapts to arbitrary curvature operator."},{"cited_title":"Sharp estimates and the Dirac operato r","cited_arxiv_id":null,"evidence_quote":"Provides the Clifford algebra inequalities and the index argument used in the equality case."},{"cited_title":"Index theory for scalar curvature on manifol ds with boundary","cited_arxiv_id":null,"evidence_quote":"Supplies the boundary Dirac operator framework and the boundary condition used in Theorem 6."},{"cited_title":"Blaine Lawson, Jr","cited_arxiv_id":null,"evidence_quote":"Provides the original index-theoretic method and the nonpositive-curvature-operator case listed in the table."},{"cited_title":"A dozen problems, questions and conjectu res about positive scalar curva- ture","cited_arxiv_id":null,"evidence_quote":"Defines $\\epsilon$-gap distance extremality, the target concept of Theorem 2."},{"cited_title":"Four lectures on scalar curvature","cited_arxiv_id":null,"evidence_quote":"Poses the boundary metric extension question addressed by Corollary 7."},{"cited_title":"Rigid ity of spin fill-ins with non-negative scalar curvature","cited_arxiv_id":null,"evidence_quote":"Gives the convex-boundary nonexistence result that Corollary 7 extends."}],"review_version":1}