{"id":"2f35ecee-d36e-48c0-9ee2-f446f8171416","arxiv_id":"1908.00677","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"On a closed Riemannian manifold, the Euclidean quantitative isoperimetric inequality is false in general, holds for generic metrics with the sharp exponent, and holds for real analytic metrics with a modified sharp exponent.","lead":"This paper asks whether the Euclidean quantitative isoperimetric inequality can hold on curved spaces. It shows the direct analogue fails in general, but a sharp version holds for generic metrics and for real analytic metrics.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.2's proof has a slice-density gap: (g,V) in the projections of the Sard–Smale generic set does not put the pair in the set; the generic claim needs a Kuratowski–Ulam type refinement, while Theorem 1.4's proof appears independent.","rationale":"I read the proof of Theorem 1.4 in good faith: the Lyapunov–Schmidt reduction in Section 3, the finite-dimensional Lojasiewicz inequality, the selection principle, and the global compactness argument are coherent and do not require the questionable projection step of Section 5.2. The reader's weakest assumption identifies a real logical gap in Theorem 1.2, which is one of the paper's advertised central claims. The gap is repairable in principle by a Kuratowski–Ulam argument, but the paper as written does not supply it, and the open-dense conclusion may need to be weakened to residual. The Section 4 counterexamples are sketches and would deserve fuller proofs, but they do not undermine the main analytic theorem. Overall the conditional verdict remains appropriate: the main Theorem 1.4 is plausible and well supported, but the generic Theorem 1.2 is not fully proven as written.","tokens_in":23255,"tokens_out":28810,"duration_ms":301322,"concrete_test":"Independently re-derive the projection step in Section 5.2 without using the sentence \"Now let g∈G and V∈(0,|M|_g)∩U.\" Instead, apply the Kuratowski–Ulam theorem to the residual set ∩_N Reg(Π_N): first confirm that each Reg(Π_N) is a dense Gδ in Γ×(0,|M|_g); then form the set G' of metrics g for which the volume slice {V:(g,V)∈∩_N Reg(Π_N)} is residual in (0,|M|_g); check whether G' is open dense or merely residual. If G' is only residual, Theorem 1.2's \"open and dense\" statement is unsupported and the proof must be revised; if G' is open dense, the gap is purely cosmetic.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing concern is in Section 5.2. After constructing G = ∩_i G_{N_i} as an open dense (or residual) subset of Γ×R, the proof projects G to G⊂Γ and U⊂(0,|M|_g) and then asserts \"Now let g∈G and V∈(0,|M|_g)∩U.\" Membership in the two projections separately does not imply (g,V)∈G; a pair (g,V) could lie in the projections while the pair itself is excluded from G. The proof needs a slice-density statement: for a large set of metrics g, the good volumes {V:(g,V)∈G} should form an open dense (or at least residual) subset of (0,|M|_g). This is not obtained by the projection argument. The standard repair is Kuratowski–Ulam applied to the residual set of regular values of the Fredholm projections Π_N, but that yields a residual, not necessarily open, set of metrics; the paper's open-dense claim requires additional properness/σ-properness information that is not supplied. This gap affects Theorem 1.2, one of the paper's central claims; Theorem 1.4's proof in Section 3 does not depend on this step. Secondary under-derivations include the n≥8 case of Lemma 4.1 and the analytic counterexample in Section 4, which is only sketched.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies quantitative isoperimetric inequalities on closed Riemannian manifolds. It proves three main results. Theorem 1.1 constructs examples showing that the direct Euclidean-type inequality with exponent 2 fails in general, even for real analytic metrics, and that for smooth non-analytic metrics no power-type bound may hold. Theorem 1.2 asserts that, for 2 ≤ n ≤ 7, on an open dense set of C^3 metrics and, for each such metric, on an open dense set of volumes, the quantitative isoperimetric inequality holds with exponent 2. Theorem 1.4 asserts that for real analytic metrics in dimensions 2 ≤ n ≤ 7, for every fixed volume, a quantitative inequality holds with some exponent 2+γ, without any a priori knowledge of the isoperimetric regions. The proof of Theorem 1.4 combines a Lyapunov-Schmidt reduction, a Lojasiewicz-Simon gradient inequality, and a selection principle. The proof of Theorem 1.2 combines Sard-Smale with the strict-stability case of the local inequality. The paper also states a stable-minimal-surface analogue and discusses possible extensions to finite-volume non-compact manifolds.","tokens_in":23502,"tokens_out":3539,"duration_ms":37717,"significance":"If the results are correct, the paper makes a substantial contribution. The main novelty, as stated in the abstract, is that the quantitative inequalities are proved without any classification or structure assumption on the isoperimetric regions; Theorem 1.4 is the first such result for all volumes on real analytic manifolds, and the sharpness examples in Theorem 1.1 clarify the need for the exponent 2+γ and for the analyticity assumption. The proof of Lemma 1.6, combining the Lojasiewicz-Simon inequality with the selection principle, is a promising and exportable technique. The paper is generally well written, and the reliance on external benchmarks (Fusco-Maggi-Pratelli, Bogelein-Duzaar-Scheven, White's bumpy metrics, Tamanini's almost-minimizer regularity) is explicit. However, the proof of Theorem 1.2 contains a genuine logical gap in the passage from a generic subset of the product Γ×R to open dense slices for each metric. That gap is load-bearing because Theorem 1.2 is one of the paper's central claims. The optimality section also leaves some technical points under-derived, especially in dimensions n ≥ 8.","major_comments":[{"comment":"The proof of Theorem 1.2 contains a logical gap in the projection/slice step. After defining G := ⋂_i G_{N_i} as an open dense subset of Γ×R, the text states that its projections G and U are open and dense and then proceeds: 'Now let g ∈ G and V ∈ (0, |M|_g) ∩ U.' Membership of the pair (g,V) in the two projections separately does not imply (g,V) ∈ G. A pair (g,V) can belong to both projections while the pair itself is excluded from G. What is needed is a slice-density statement: for a sufficiently large set of metrics g, the set of volumes V with (g,V) ∈ G should be open and dense in (0,|M|_g). The projection argument does not supply this. A Kuratowski-Ulam type refinement applied to the residual set of regular values would give residual slices, not necessarily open dense slices, and the open-density claim would require additional properness information that is not provided. This gap affects Theorem 1.2 and Corollaries 5.3 and 5.4, since the conclusion that every minimizer for the pair (g,V) is strictly stable is exactly the point at which the argument breaks.","section":"5.2"},{"comment":"The proof of Lemma 4.1 is given only for 3 ≤ n ≤ 7, with the statement that it 'can be easily modified to accommodate for a singular set in higher dimensions.' Since Lemma 4.1 is used in Theorem 1.1, which is asserted for all n ≥ 2, the higher-dimensional case is load-bearing. The modification is not automatic: for n ≥ 8, isoperimetric regions may have a singular set, and the argument that the reduced boundary has exactly one component and that each component has bounded diameter needs to be revisited in the presence of singularities. The delicate limiting argument involving monotonicity and volume comparison should be written out for all dimensions.","section":"4, Lemma 4.1"},{"comment":"The construction of the counterexamples in Theorem 1.1 is only sketched. The calibration argument identifying the unique isoperimetric regions of half volume in the warped product metric, the precise choice of ρ_δ, and the verification that the perturbed regions Γ_δ are valid competitors (with the correct volume and with boundary a graph over the minimizer) are not given in detail. This matters because Theorem 1.1 is used to justify the optimality of Theorems 1.2 and 1.4, including the claim that γ may be arbitrarily large for analytic metrics. The authors should provide a complete argument, or at least indicate exactly which standard calibration and comparison results are being invoked and why they apply uniformly in the parameters.","section":"4, proof of Theorem 1.1"}],"minor_comments":[{"comment":"There are several typographical errors: 'quantiative' in the abstract, 'the the set' in the introduction, and the repeated '/suppress' artifacts before 'Lojasiewicz' throughout the text. These should be corrected.","section":"Abstract and throughout"},{"comment":"In the statement of Lemma 3.3, the text says 'if and only if the function P of Lemma 3.3 is constant,' but P is defined in Lemma 3.1, not Lemma 3.3. The cross-reference should be fixed.","section":"3.1, Lemma 3.3"},{"comment":"The contradiction argument producing α_0 should spell out the relabeling step: after passing to a subsequence E_j → Σ̄ and choosing Σ_j from the finite subcover with ‖χ_{Σ̄} - χ_{Σ_j}‖_{L^1} ≤ δ(Σ_j)/2, the triangle inequality gives ‖χ_{E_N} - χ_{Σ_j}‖_{L^1} ≤ δ(Σ_j) for large N. As written, the line 'we can assume without loss of generality that ‖χ_{E_N} - χ_{Σ_1}‖_{L^1} ≤ δ(Σ_1)' is compressed and could confuse a reader.","section":"3.4, proof of Theorem 1.4"},{"comment":"The discussion of finite-volume non-compact manifolds is intriguing but speculative; it would be helpful to state explicitly whether the authors expect the obstruction to occur for every such manifold or only in a constructed example.","section":"4.1"},{"comment":"The proof of Lemma B.5 is extremely brief and relies on 'standard facts about almost-minimizers.' Since the selection principle is central to the paper, one or two precise references for the Hausdorff convergence and ε-regularity statements would improve readability.","section":"Appendix B, Lemma B.5"}],"recommendation":"major_revision","confidential_remarks":"The slice-density gap in §5.2 is the main obstacle. The authors should be encouraged to repair it, since Theorem 1.2 is a central contribution and the surrounding ideas are sound. The optimality section also needs more detail, but that is a matter of completeness rather than an apparent error in the main mechanism. The paper fits the journal and is likely to be accepted after revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague — this is a serious paper: the first to show the Euclidean quantitative isoperimetric inequality fails in general on closed manifolds, to prove a generic sharp-exponent version, and to get a real-analytic result via Lojasiewicz–Simon without knowing minimizer structure. The main line of thought is coherent, and the local Lojasiewicz-meets-Fuglede estimate is a real advance. The counterexample in Section 4 is instructive; the n=2 case and the analytic perturbation sketch are plausible though compressed.\n\nThe soft spots: the stress-test note is right. In Section 5.2, after Sard–Smale gives open dense G_N in Γ×R, the proof projects to G and U and says 'let g∈G and V∈...∩U'. That does not put (g,V) in the intersection of the G_N. You need a slice-density statement, and mere openness of projections does not provide it. The gap is load-bearing for Theorem 1.2; Theorem 1.4's proof in Section 3 does not rely on this step, and the generic theorem may still be true by a Kuratowski–Ulam refinement, but as written it needs repair.\n\nAlso, the n≥8 case of Lemma 4.1 is only asserted to be 'easily modified' for the singular set; for a paper of this ambition, that deserves at least a reference or a sentence of substance. The analytic counterexample with γ arbitrarily large is only sketched. These are minor-to-moderate.\n\nNo circular reasoning found. The external benchmarks are the right ones: Fusco–Maggi–Pratelli, Bög elein–Duzaar–Scheven, Cicalese–Leonardi, White, Simon. The real-analytic result does depend on the Lojasiewicz–Simon inequality, but that is a legitimate tool, not a missing proof. The citation pattern is honest.\n\nBottom line: Theorem 1.4 and the counterexample are likely correct; Theorem 1.2 has a genuine gap in the proof as written. The paper deserves a serious referee — send it to review, not desk reject — and the authors should be asked to fix the slice argument. I'd bring it to reading group, and I'd cite it even now for the counterexample and the analytic theorem.","headline":"Strong, genuinely novel paper on Riemannian quantitative isoperimetry, but the generic theorem has a real proof gap in the slice argument that needs repair.","tokens_in":24078,"tokens_out":1330,"would_cite":true,"duration_ms":13636,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C21","49Q20"],"pacs":[],"model":"deepseek-v4-flash","headline":"The squared isoperimetric inequality fails on closed Riemannian manifolds; a modified version holds.","keywords":["quantitative isoperimetric inequality","Riemannian manifold","Lojasiewicz-Simon inequality","selection principle","isoperimetric profile","Fraenkel asymmetry","bumpy metrics","stable minimal hypersurfaces"],"falsifier":"To refute the central positive claim, find a closed real-analytic manifold of dimension $2\\le n\\le7$ and a volume $V_0$ with a sequence $E_k$ of competitors of volume $V_0$ such that $\\alpha_g(E_k)\\to0$ but $(P_g(E_k)-I_g(V_0))/\\alpha_g(E_k)^{2+\\gamma}\\to0$ for every $\\gamma>0$; such a sequence would contradict Theorem 1.4. To test sharpness, compute the optimal exponent in the paper's warped-product example, where the warping function vanishes like $x^{2m}$: the predicted growth of the optimal $\\gamma$ with $m$ is a concrete quantitative check of the counterexample mechanism.","tokens_in":23001,"feed_emoji":"📐","tokens_out":11864,"duration_ms":105260,"temperature":0.7,"pith_summary":"The paper asks whether the Euclidean quantitative isoperimetric inequality—perimeter excess controls the squared volume asymmetry to a ball—has a true analogue on a closed Riemannian manifold. It shows the direct analogue with exponent two fails in general, even for real-analytic metrics, and it constructs explicit counterexamples. In the positive direction, it proves that for generic metrics and generic volumes the exponent-two inequality does hold, and that every closed real-analytic manifold of dimension $2\\le n\\le7$ satisfies a modified quantitative inequality, with a possibly larger exponent, for every volume and every competitor. The reason this matters is that the positive results need no classification or shape information about the isoperimetric regions themselves.","feed_headline":"Square isoperimetric law fails on closed curved manifolds","feed_subtitle":"Generic and real-analytic metrics retain quantitative isoperimetric control, with a possibly larger exponent.","key_machinery":"The load-bearing mechanism is an infinite-dimensional Lojasiewicz-Simon inequality for the perimeter functional, stated as Lemma 1.6. It says that near a smooth isoperimetric region $\\Sigma$, the perimeter excess of a volume-preserving competitor controls the distance to the set of nearby minimizers raised to a power $2+\\gamma$, with $\\gamma=0$ when the minimizer is integrable and a stronger quadratic bound when it is strictly stable. The proof combines a Lyapunov-Schmidt reduction to the finite-dimensional kernel of the Jacobi operator, a selection principle that produces penalized minimizers converging to a worst-case minimizer, and a finite-cover compactness argument over the space of all minimizers. For the generic theorem, strict stability is forced by a volume-constrained bumpy-metrics result obtained through an infinite-dimensional Sard-Smale argument.","core_discovery":"On the paper's own terms, the central discovery is that quantitative isoperimetric control on a closed Riemannian manifold $(M^n,g)$ is genuinely different from the Euclidean case. Writing $I_g(V_0)$ for the least perimeter among sets of volume $V_0$ and $\\alpha_g(E)$ for the Fraenkel asymmetry, the least $L^1$ distance from $E$ to any isoperimetric region of volume $V_0$, the paper proves that $P_g(E)-I_g(V_0)\\ge C\\alpha_g(E)^2$ is false in general: there are real-analytic metrics with a uniquely isoperimetric region $\\Omega$ and sets $E_k$ of the same volume such that $|E_k\\,\\Delta\\,\\Omega|_g\\to0$ while $(\\alpha_g(E_k))^2/(P_g(E_k)-I_g(V_0))\\to\\infty$, and smooth metrics where no power of $\\alpha_g(E_k)$ bounds the perimeter deficit. Theorems 1.2 and 1.4 then give the positive statements: for $2\\le n\\le7$, an open dense set of $C^3$ metrics has an open dense set of volumes for which the exponent-two inequality holds, and every real-analytic metric admits constants $C_0>0$ and $\\gamma\\ge0$ such that $P_g(E)-I_g(V_0)\\ge C_0\\,\\alpha_g(E)^{2+\\gamma}$ for every admissible $E$. The paper also shows the exponent $2+\\gamma$ cannot in general be improved to $\\gamma=0$.","pith_inferences":["Editorial inference: the paper's local-to-global structure suggests a general recipe for quantitative stability in analytic variational problems—obtain a local Lojasiewicz-Simon estimate near each smooth minimizer, then run a selection-principle compactness argument—so the absence of a classification of extremizers should not block similar inequalities for other geometric functionals.","Editorial inference: in the warped-product counterexamples with warping function behaving like $x^{2m}$ near the minimizer, the optimal exponent $\\gamma$ should grow with $m$; computing that growth would give a precise test of how the flatness of the isoperimetric profile forces the loss of exponent.","Editorial inference: for finite-volume noncompact manifolds, the paper's closing discussion suggests that the theorem may fail when a minimizer has infinitely many components with unbounded Lojasiewicz-Simon exponents; constructing that example would sharply delimit the noncompact analogue.","Editorial inference: in dimensions $n\\ge8$, a local Lojasiewicz-Simon estimate away from the singular stratum of an isoperimetric boundary might sustain a quantitative inequality with the distance measured to the full minimizer, even though the present proof stops at $n=7$."],"forward_implications":["On every closed real-analytic manifold of dimension $2\\le n\\le7$, every fixed volume admits a quantitative isoperimetric inequality with no structural assumptions on the minimizers; this removes a classification bottleneck present in nearly all previous quantitative results.","The Euclidean exponent $2$ cannot be recovered in general: the paper's counterexamples show the optimal exponent may be strictly greater than $2$, and for smooth non-analytic metrics no power of the asymmetry may control the perimeter deficit.","For a generic $C^3$ metric and generic volumes, the exponent-two inequality holds, and for any prescribed volume one can perturb the metric so that the exponent-two inequality holds at that volume without changing the volume.","The same Lojasiewicz-Simon mechanism produces a quantitative minimality theorem for stable minimal hypersurfaces in real-analytic ambient manifolds, valid in every dimension where the surface is smooth.","The dimension restriction $2\\le n\\le7$ is tied to boundary regularity of isoperimetric regions; extending Theorem 1.4 to $n\\ge8$ would require handling singular minimizers."],"supporting_citations":[{"why":"establishes the sharp Euclidean quantitative isoperimetric inequality that the paper seeks a Riemannian analogue of.","marker":"[24]"},{"why":"introduces the selection principle used to reduce the worst-case competitor to a small graphical perturbation of a minimizer.","marker":"[12]"},{"why":"provides the Fuglede estimate for nearly spherical domains that the Lojasiewicz-Simon inequality replaces in the non-integrable Riemannian setting.","marker":"[23]"},{"why":"introduces the Lojasiewicz-Simon argument for elliptic PDE that the paper adapts to the perimeter functional.","marker":"[39]"},{"why":"supplies the finite-dimensional Lojasiewicz inequality for analytic functions used in the Lyapunov-Schmidt reduction.","marker":"[18]"},{"why":"provides the bumpy-metrics framework that the paper extends to enforce strict stability under a volume constraint.","marker":"[45]"},{"why":"supplies the infinite-dimensional Sard theorem used to pass from bumpiness to open dense genericity.","marker":"[40]"},{"why":"supplies the regularity theory for isoperimetric regions that underlies the smoothness assumptions.","marker":"[31]"},{"why":"supplies the nonexistence of stable constant mean curvature hypersurfaces used in the bounded-mean-curvature lemma.","marker":"[8]"}],"fun_headline_variants":["Curved manifolds break square isoperimetric law","Isoperimetric square law fails, but generically holds","Real-analytic metrics rescue isoperimetric inequality","No sharp square isoperimetric bound on curved spaces","Isoperimetric control survives generically on curved manifolds"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The generic theorem depends on the assumption that each 'good' metric is accompanied by an open dense set of good volumes; the proof only shows that the collection of good metrics and the collection of good volumes are each open and dense, and separate density does not force them to match up.","fun_headline_variants_meta":{"raw":{"variants":["Curved manifolds break square isoperimetric law","Isoperimetric square law fails, but generically holds","Real-analytic metrics rescue isoperimetric inequality","No sharp square isoperimetric bound on curved spaces","Isoperimetric control survives generically on curved manifolds"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000306,"raw_usage":{"total_tokens":1760,"prompt_tokens":958,"completion_tokens":802,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":574,"completion_tokens_details":{"reasoning_tokens":725}},"tokens_in":574,"tokens_out":802,"duration_ms":8226,"temperature":1.0,"reasoning_tokens":725,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:39:33.887293+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"To refute the central positive claim, find a closed real-analytic manifold of dimension $2\\le n\\le7$ and a volume $V_0$ with a sequence $E_k$ of competitors of volume $V_0$ such that $\\alpha_g(E_k)\\to0$ but $(P_g(E_k)-I_g(V_0))/\\alpha_g(E_k)^{2+\\gamma}\\to0$ for every $\\gamma>0$; such a sequence would contradict Theorem 1.4. To test sharpness, compute the optimal exponent in the paper's warped-product example, where the warping function vanishes like $x^{2m}$: the predicted growth of the optimal $\\gamma$ with $m$ is a concrete quantitative check of the counterexample mechanism.","supporting_citations":[{"cited_title":"Fusco, F","cited_arxiv_id":null,"evidence_quote":"establishes the sharp Euclidean quantitative isoperimetric inequality that the paper seeks a Riemannian analogue of."},{"cited_title":"A selection pri nciple for the sharp quanti- tative isoperimetric inequality","cited_arxiv_id":null,"evidence_quote":"introduces the selection principle used to reduce the worst-case competitor to a small graphical perturbation of a minimizer."},{"cited_title":"Stability in the isoperimetric problem f or convex or nearly spherical domains in Rn","cited_arxiv_id":null,"evidence_quote":"provides the Fuglede estimate for nearly spherical domains that the Lojasiewicz-Simon inequality replaces in the non-integrable Riemannian setting."},{"cited_title":"Asymptotics for a class of nonlinear evolut ion equations, with applica- tions to geometric problems","cited_arxiv_id":null,"evidence_quote":"introduces the Lojasiewicz-Simon argument for elliptic PDE that the paper adapts to the perimeter functional."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the finite-dimensional Lojasiewicz inequality for analytic functions used in the Lyapunov-Schmidt reduction."},{"cited_title":"The space of minimal submanifolds for vary ing Riemannian metrics","cited_arxiv_id":null,"evidence_quote":"provides the bumpy-metrics framework that the paper extends to enforce strict stability under a volume constraint."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the infinite-dimensional Sard theorem used to pass from bumpiness to open dense genericity."},{"cited_title":"Sets of finite perimeter and geometric variational problems , volume 135 of Cambridge Studies in Advanced Mathematics","cited_arxiv_id":null,"evidence_quote":"supplies the regularity theory for isoperimetric regions that underlies the smoothness assumptions."},{"cited_title":"A nonexistence theorem for stable cons tant mean curvature hy- persurfaces","cited_arxiv_id":null,"evidence_quote":"supplies the nonexistence of stable constant mean curvature hypersurfaces used in the bounded-mean-curvature lemma."}],"review_version":1}