{"id":"c04da2d3-4265-487b-8122-ec128d16e09d","arxiv_id":"2412.12707","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"If a complete noncompact n-manifold with at least two ends satisfies lambda1(-gamma Delta + Ric) >= 0 for some gamma < 4/(n-1), then it splits isometrically as R x N with compact N and Ric_N >= 0; the constant is sharp.","lead":"A new theorem sharpens the classical Cheeger-Gromoll splitting result to a spectral Ricci condition. It shows that complete spaces with at least two ends and a spectral lower bound with constant gamma below 4/(n-1) must split as a line times a compact nonnegatively curved piece.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"High-dimensional stability estimate is omitted and deferred to an unpublished appendix; the n≥8 case is not self-contained.","rationale":"I read the proof in good faith. The core mechanism is coherent: the μ-bubble minimizers exist, the first and second variation computations on the regular part are correct, the discriminant estimate (3.16) gives the required negative-definite quadratic form for γ < 4/(n−1), and the contradiction argument using the uniform density estimates is logically sound. The FS80 step invoked at (3.1) is a standard theorem for Schrödinger operators with locally bounded potentials; the locally Lipschitz function Ric is acceptable, and the scaling from γ to 1 is harmless, so I do not regard that as the most load-bearing weakness. The most concrete gap is the omitted singular-set computation for n≥8: the footnote explicitly defers the essential stability estimate to an unpublished appendix of a different preprint. This does not make me doubt the theorem, but it does mean the submitted proof is not fully self-contained for the full claimed range, matching the reader's CONDITIONAL verdict. The further point about ends and compactness of N is a standard topological fact and does not need separate remedy.","tokens_in":13187,"tokens_out":48588,"duration_ms":424401,"concrete_test":"Retrieve the omitted Appendix A of [AX24] and verify the cutoff argument for the functional Pε. Concretely, for n≥8, take a sequence of cutoffs η_j on ∂Ωε that equal 1 away from an r_j-neighborhood of the singular set and vanish on it. Check that ∫|∇Σ(η_j u^{-γ/2})|^2 and all error terms in the second variation converge to the corresponding integrals for u^{-γ/2} as r_j→0, using the uniform density estimates of Lemma 2.3(2). If the appendix does not contain this computation for the functional with the bulk term hεu^γ, supply it; if the approximation fails, the contradiction leading to (3.20) is unjustified for n≥8.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The theorem asserts all n≥2, but for n≥8 the proof of the decisive stability inequality (3.19) is not contained in the paper. Minimizers Ωε have a singular set of Hausdorff dimension at most n−8 (Lemma 2.3(1)). The footnote after (3.11) states that to obtain (3.19) one multiplies test functions by a cutoff vanishing on the singular set and carries out computations 'as in [AX24, Appendix A]', which are omitted and located in an unpublished self-cited preprint. The stability inequality is what forces the contradiction c0η ≤ ... = 0; without a complete proof that the cutoff approximation passes to the limit for the weighted perimeter functional with the bulk term hεu^γ, the argument that |∇u0|(x)=0 is not established in dimensions n≥8. Because the singular set has codimension at least 7 in a hypersurface of dimension n−1, its H^1 capacity is zero, so the result is very likely repairable; but as submitted, the main theorem is not self-contained for the full stated range.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a sharp spectral generalization of the Cheeger–Gromoll splitting theorem. The main result, Theorem 1.1, states that a complete noncompact Riemannian n-manifold (n≥2) with at least two ends satisfying λ1(-γΔ+Ric)≥0 for some γ<4/(n-1) must have Ric≥0 and split isometrically as R×N with N compact and Ric_N≥0. The proof uses an existence result of an eigenfunction u0 for the Schrödinger operator, constructs weighted µ-bubbles, and obtains a stability inequality that forces |∇u0|=0 at every point, making u0 constant. The paper also includes remarks showing that the constant 4/(n-1) is sharp and that the multiple-end assumption is necessary for γ>0, and derives a corollary on stable minimal hypersurfaces in manifolds with biRic≥0.","tokens_in":13419,"tokens_out":17732,"duration_ms":149224,"significance":"If the proof is completed as claimed, the result is a significant and attractive contribution: it extends the classical Cheeger–Gromoll splitting theorem to a spectral Ricci lower bound, gives a sharp threshold 4/(n-1), and introduces a µ-bubble/surface-capturing approach that differs from the original Busemann-function proof. The discriminant computation leading to (3.16) is clean, and the sharpness examples in Section 4 are explicit and checkable. However, the proof as submitted has a substantial gap for dimensions n≥8, where the key stability inequality is deferred to an unpublished self-cited appendix; this prevents the main theorem from being fully established in the stated range.","major_comments":[{"comment":"For n≥8 the decisive stability inequality (3.19) is not proved in the paper. The footnote states that one multiplies test functions by a cutoff vanishing on the singular set ∂Ωε\\∂*Ωε and carries out computations 'as in [AX24, Appendix A]', which are omitted and appear only in a self-cited arXiv preprint. Equation (3.19) is the estimate used to force the contradiction at the end of the proof; without a complete argument that the cutoff approximation passes to the limit for the weighted perimeter functional with the bulk term hεu^γ, the claim |∇u0|(x)=0 is not established in dimensions n≥8. The gap is likely repairable (the singular set has codimension at least 7, hence capacity zero), but as submitted the proof of Theorem 1.1 is not self-contained for the full stated range n≥2.","section":"§3, footnote to (3.11) and equation (3.19)"},{"comment":"The proof starts from the existence of a positive C^{2,α} function u0 satisfying -γΔu0+Ric·u0=0, which is deduced from the quadratic-form condition via [FS80, Theorem 1]. Since Ric is only locally Lipschitz and the manifold is noncompact, the hypotheses under which [FS80, Theorem 1] applies should be stated precisely, or a short justification should be given. This is load-bearing because every subsequent µ-bubble construction depends on u0; if the equivalence of (i) and (ii) fails for this class of potentials, the argument cannot start.","section":"§1, definition of λ1 and equation (3.1)"},{"comment":"The abstract states that the multiple-end assumption is necessary for any γ>0, but the counterexample in Remark 4.2 is explicitly constructed only for n≥3. For n=2 the necessity of the multiple-end assumption is not demonstrated. Please qualify the abstract statement to match the proven range, or supply a two-dimensional example.","section":"Abstract and Remark 4.2"}],"minor_comments":[{"comment":"In the Gauss equation display, the sign of the second fundamental form term in '0 = Ric_Σ(∂t, ∂t) = -II^2(∂t, ∂t) + Ric_M(∂t, ∂t) - Sect(∂t∧ν_Σ)' appears to be opposite to the standard minimal-hypersurface Gauss equation; please check the sign convention and notation (what II^2(∂t,∂t) denotes).","section":"§4, proof of Corollary 1.2"},{"comment":"The statement of Lemma 2.2(1) gives -2u0≤w<0, but the proof only establishes -u0≤w<0 outside U and says the lemma follows 'by extending w inside U keeping -2u0≤w<0'. It would help to indicate explicitly how the extension across ∂U is made while preserving C^{2,α} regularity.","section":"§2, Lemma 2.2"},{"comment":"In the paragraph after (3.19), the argument that ∂*Ωε∩φ^{-1}([-1,1])≠∅ implicitly uses that Ωε is admissible (so ∂*Ωε is nonempty); it may be worth stating this explicitly.","section":"§3, proof of Theorem 1.1"},{"comment":"There are minor typos in the abstract/OCR rendering, e.g., 'Cheeger–Gro moll' and 'greaterorequalslant'; these should be corrected in the final version.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper makes heavy use of two self-cited preprints, [AX24] and [Xu24], for technical tools (the n≥8 variation formulas and the spectral inequality on stable minimal hypersurfaces). The editor may wish to check the publication status of these references; in any case, the omitted computation for (3.19) should be moved into the paper or replaced by a complete argument. The main theorem is attractive and, to my reading, very likely correct once the n≥8 gap is filled; the central computation in Section 3 is sound and the sharpness examples are convincing."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"It's a good paper. The main theorem is real: if a complete noncompact n-manifold with at least two ends satisfies λ1(−γΔ+Ric) ≥ 0 for γ < 4/(n−1), then Ric ≥ 0 and the manifold splits as R × N with compact N. The constant is sharp, and the two counterexamples—the hyperbolic quotient at γ = 4/(n−1), and the compact perturbation of R^n showing two ends can't be weakened to just having a line—are genuinely new. The proof strategy is also new, at least to me: use the [FS80] positive solution u0 to convert the spectral bound into a pointwise equation, run µ-bubbles with a surface-capturing perturbation, and get |∇u0| = 0 at an arbitrary point. The algebra in Section 3 is clean; the discriminant computation (3.16) and the resulting bound (3.17) are the heart, and the contradiction works.\n\nThe real soft spot is the n ≥ 8 range. The decisive stability estimate (3.19) requires a second variation computation on the singular set of the µ-bubble minimizer. The authors defer it to [AX24, Appendix A], an unpublished self-cited preprint. That is a genuine self-containedness gap for the stated range of Theorem 1.1. It is probably repairable—the singular set has codimension at least 8, and a cutoff vanishing on it should pass to the limit—but as written the theorem is not proved for n ≥ 8.\n\nTwo smaller notes. The equivalence (i)⇔(ii) is cited tersely to [FS80]; since Ric is only locally Lipschitz, I'd like the precise statement and hypotheses spelled out. And after u is constant, the compactness of N in the splitting is invoked without comment; it's standard for a two-ended Ric ≥ 0 manifold, but worth a line.\n\nThe citation pattern is fine. [AX24] and [Xu24] provide auxiliary lemmas, not the target result. The addendum about the independent overlap with Catino–Mari–Mastrolia–Roncoroni is honest.\n\nBottom line: send it to peer review. The core mathematics is solid. Ask the authors to either include the n ≥ 8 computation or restrict the theorem to n ≤ 7. Once that is fixed, this is a strong paper.","headline":"Sharp spectral splitting theorem with a clean core computation; the n≥8 case relies on an unpublished appendix, which is fixable before acceptance.","tokens_in":13919,"tokens_out":3174,"would_cite":true,"duration_ms":27897,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C21","53C42"],"pacs":[],"model":"deepseek-v4-flash","headline":"If a complete two-ended manifold satisfies λ1(−γΔ + Ric) ≥ 0 for γ < 4/(n−1), then it splits as R × N with N compact and Ric ≥ 0.","keywords":["spectral splitting theorem","Cheeger–Gromoll splitting theorem","Ricci curvature","µ-bubble","spectral lower bound","stable minimal hypersurfaces","biRicci curvature","sharp constant"],"falsifier":"For the warped product in Remark 4.1 (M = R × $T^{{n−1}}$, g = dt² + $e^{{2t}}$g), compute the bottom of the spectrum of −γΔ + Ric at a value γ slightly below 4/(n−1). The theorem predicts λ1 < 0, so the manifold would not satisfy the hypothesis; a numerical or analytic demonstration that λ1 ≥ 0 for such γ would create a counterexample to the theorem.","tokens_in":12993,"feed_emoji":"📐","tokens_out":15307,"duration_ms":121786,"temperature":0.7,"pith_summary":"The paper proves a sharp spectral version of the Cheeger–Gromoll splitting theorem: a complete noncompact Riemannian manifold of dimension at least 2 with at least two ends, whose Ricci curvature is nonnegative only in the spectral sense that the operator −γΔ + Ric has nonnegative first eigenvalue for some γ < 4/(n−1), must in fact have genuinely nonnegative Ricci curvature everywhere and split isometrically as a line times a compact nonnegatively curved manifold. Spectral curvature bounds are much weaker than pointwise ones, so the result shows that a large family of almost-rigid geometries is actually rigid. The threshold 4/(n−1) is optimal, and the multiple-end assumption is necessary for every γ > 0. The proof replaces the classical geodesic-line argument with a µ-bubble technique (minimizing a carefully weighted surface area), yielding an alternative route to splitting rigidity.","feed_headline":"Spectral Ricci bound forces two-ended manifolds to split","feed_subtitle":"Below the threshold 4/(n−1), spectral nonnegative Ricci still forces a line product with a compact factor.","key_machinery":"The central object is the operator L = −γΔ + Ric, a weighted Laplacian with the pointwise minimum Ricci curvature as potential. The spectral hypothesis λ1(L) ≥ 0 is converted, via a theorem of Fischer-Colbrie and Schoen, into a positive solution u0 of L u0 = 0, which acts as a weight for the perimeter functional. The proof then runs a µ-bubble scheme: it minimizes Pε(Ω) = ∫_{∂*Ω} u^γ − ∫ (χΩ − χΩ0) hε u^γ, with hε a rapidly diverging function of a proper 1-Lipschitz coordinate φ whose level sets divide the ends. The second variation of Pε yields a stability inequality whose quadratic part has discriminant γ(γ − 4/(n−1)); this discriminant is negative exactly in the theorem's range, giving a nonzero penalty term. A surface-capturing perturbation of u0 (Lemma 2.2) forces the minimizers to pass through any prescribed point, and the density estimates on the bubbles let the inequality pass to the limit to show |∇u0| = 0 everywhere.","core_discovery":"Theorem 1.1 states that for a complete noncompact manifold (M^n, g) with n ≥ 2, at least two ends, and λ1(−γΔ + Ric) ≥ 0 for some γ < 4/(n−1), the Ricci curvature satisfies Ric ≥ 0 pointwise and M is isometric to R × N, where N is compact and has Ric_N ≥ 0. The constant 4/(n−1) is sharp: for every γ ≥ 4/(n−1) the warped product (R × $T^{{n−1}}$, dt² + $e^{{2t}}$g) satisfies the spectral condition while having two ends but not splitting, and the paper details this in Remark 4.1. The multiple-end assumption is also necessary, as any sufficiently small compact perturbation of flat R^n satisfies the spectral condition for every γ > 0 while containing lines but not splitting (Remark 4.2). As a corollary, in a manifold with biRic ≥ 0 of dimension at most 5, any noncompact, two-sided stable minimal hypersurface either has one end or splits as R × Σ′ with Σ′ compact, is totally geodesic, and has vanishing normal Ricci curvature.","pith_inferences":["For dimensions n ≥ 5, the theorem does not cover γ = 1 (the standard operator −Δ + Ric); since the sharpness examples are perturbations of the threshold, one might suspect the splitting conclusion could fail for γ = 1 in high dimensions, and checking this case directly would be a natural test of the result's boundary.","The proof actually shows the stronger local fact that |∇u0| = 0 at each point x (by constructing a bubble through x), so a variation with a local spectral bound, e.g., λ1 ≥ 0 on a single end, might yield a partial splitting or an end-stability statement.","The discriminant γ(γ − 4/(n−1)) suggests that the constant 4/(n−1) is the sharp threshold for positivity of the stability form; the same algebraic structure may govern other weighted-area problems, for instance with the Bakry–Émery operator, and could be probed with the same warped-product examples."],"forward_implications":["Any complete two-ended manifold with λ1(−γΔ+Ric) ≥ 0 for γ < 4/(n−1) automatically has Ric ≥ 0, so the entire Cheeger–Gromoll rigidity apparatus applies: the universal cover splits, and the fundamental group is isomorphic to the fundamental group of the compact factor N (since R is simply connected).","Corollary 1.2 extends the one-end rigidity of stable minimal hypersurfaces from ambient nonnegative sectional curvature to ambient biRic ≥ 0 in dimension n ≤ 5, without requiring properness of the immersion.","The sharpness examples at γ = 4/(n−1) and for compact perturbations of R^n delineate exactly where the theorem stops: the two-end hypothesis and the range of γ cannot be relaxed.","The proof supplies a new proof of the classical Cheeger–Gromoll splitting theorem under the assumptions Ric ≥ 0, at least two ends, and inf_{x} |B(x,1)| > 0, using µ-bubbles instead of geodesic lines."],"supporting_citations":[{"why":"Bridges the spectral inequality to the existence of a positive C^{2,α} solution u0 of −γΔu0 + Ric · u0 = 0, the starting point of the weighted bubble construction.","marker":"[FS80, Theorem 1]"},{"why":"The classical splitting theorem that the paper generalizes and invokes once Ric ≥ 0 is established.","marker":"[CG71]"},{"why":"Supplies the model function hε and the µ-bubble approximation scheme used to define the weighted perimeter and produce the approximating minimizers.","marker":"[Zhu23, Lemma 2.3]"},{"why":"Origin of the surface-capturing technique: perturbing the potential so that a weighted area minimizer cannot avoid a prescribed ball; the proof adapts this to force the bubbles through x.","marker":"[CCE16]"},{"why":"Extends surface-capturing to splitting-type rigidity for scalar curvature; the argument here transfers that idea to the spectral Ricci setting.","marker":"[CEM19]"},{"why":"Provides existence of a minimizer for the weighted perimeter functional Pε within the admissible class, a step used in the proof of Theorem 1.1.","marker":"[CL24, Proposition 12]"},{"why":"Regularity theory for minimizers guaranteeing the reduced boundary is a C^{2,α} hypersurface away from a small singular set, needed for the stability computation.","marker":"[Mag12, Theorem 27.5]"},{"why":"Density estimates for quasiminimizers, used to keep the perimeter of the bubbles controlled locally near the point x.","marker":"[Kin+13, Lemma 5.1]"}],"fun_headline_variants":["Sharp spectral splitting: two-ended manifolds split below threshold","Spectral Ricci bound forces two-ended splitting at sharp constant","Two-ended manifolds: spectral positivity below 4/(n−1) forces splitting","Sharp constant 4/(n−1) for spectral splitting theorem","Spectral condition below sharp threshold implies isometric product"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the bridge from the spectral inequality λ1(−γΔ+Ric) ≥ 0 to a global positive solution u0 of −γΔu0+Ric·u0 = 0, a conversion that must hold for the locally Lipschitz potential Ric on the whole noncompact manifold for the proof to start.","fun_headline_variants_meta":{"raw":{"variants":["Sharp spectral splitting: two-ended manifolds split below threshold","Spectral Ricci bound forces two-ended splitting at sharp constant","Two-ended manifolds: spectral positivity below 4/(n−1) forces splitting","Sharp constant 4/(n−1) for spectral splitting theorem","Spectral condition below sharp threshold implies isometric product"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000339,"raw_usage":{"total_tokens":1848,"prompt_tokens":901,"completion_tokens":947,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":517,"completion_tokens_details":{"reasoning_tokens":859}},"tokens_in":517,"tokens_out":947,"duration_ms":8692,"temperature":1.0,"reasoning_tokens":859,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T13:50:20.340326+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For the warped product in Remark 4.1 (M = R × $T^{{n−1}}$, g = dt² + $e^{{2t}}$g), compute the bottom of the spectrum of −γΔ + Ric at a value γ slightly below 4/(n−1). The theorem predicts λ1 < 0, so the manifold would not satisfy the hypothesis; a numerical or analytic demonstration that λ1 ≥ 0 for such γ would create a counterexample to the theorem.","supporting_citations":[],"review_version":1}