{"id":"cc8bfc79-a064-4830-87c5-ff5d8a3889bd","arxiv_id":"2506.03579","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Polystable saturated reflexive parabolic sheaves admit admissible Hermitian-Einstein metrics compatible with the parabolic structure, and semistable ones admit approximate such metrics; a Bogomolov-Gieseker inequality for nef and big classes follows.","lead":"This paper claims a Kobayashi-Hitchin correspondence for saturated reflexive parabolic sheaves on compact Kähler manifolds: stable sheaves are exactly those carrying special Hermitian metrics. The result would give analytic tools for moduli spaces and Chern number inequalities in a broader setting.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 4.16 is asserted without proof and is load-bearing: analytic stability (Prop 5.1) and compatibility of the limiting metric (Prop 6.10) both rely on bH being adapted to every parabolic subsheaf in codimension 1.","rationale":"The paper's central claim is the Kobayashi–Hitchin correspondence for saturated reflexive parabolic sheaves. The reader's weakest-assumption analysis identifies Prop 4.16 as the most load-bearing missing derivation, and my reading agrees. The construction of bH in §4.2 is explicit, and Prop 4.10 establishes the adaptedness identity for F_* itself. But the step to arbitrary subsheaves is not a formal consequence: the smooth decomposition used to define H0 is compatible with the filtration of F_*, while a proper parabolic subsheaf S_* need not align with that decomposition. Thus the identity ch1(S_*) = (√−1/2π) Tr(F_{bH|S}) requires a separate argument, involving cancellation of the second fundamental form and of boundary terms along D. The paper's phrase 'as it is tautological' covers precisely this missing argument, so the concern is internal rather than a disagreement with a consensus. The most direct way to settle it is to test the identity in the simplest non-trivial local model where the decomposition is not holomorphic: a non-split extension of parabolic line bundles. If the identity fails there, the equivalence of parabolic and analytic stability (Prop 5.1) and the compatibility of the limiting metric (Prop 6.10) would both lose their foundation, undermining Theorem 6.14. If it holds, the omitted proof is probably routine enough to fill, and the reader's CONDITIONAL verdict can be upgraded. Other gaps noted by the reader, such as Lemma 3.2 and Prop 6.12, are also real but are more localized; Prop 4.16 is the single point where both directions of the main theorem converge.","tokens_in":31454,"tokens_out":6369,"duration_ms":75070,"concrete_test":"Take a non-split extension 0→L(a)_*→F_*→L(b)_*→0 of parabolic line bundles over (Δ^n,D) with a≠b, and construct bH following §4.2 with the smooth decomposition of Prop 4.7. For S_*=L(a)_*, compute ch1(S_*)·[η] and (√-1/2π)∫_{Δ^n∖D} Tr(F_{bH|S})∧η for a compactly supported (n−1,n−1)-form η. If the two numbers differ for some η, Prop 4.16 is false and Theorem 6.14 collapses; if they agree for all such extensions, the omitted proof is likely fillable.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Prop 4.16 states that the metric bH constructed in §4.2 is adapted in codimension 1 to every parabolic subsheaf S_* of F_*, and the proof is omitted as 'tautological'. This is the hinge that converts parabolic stability into analytic stability (Prop 5.1) and that lets Prop 6.10 pass from ch1(F_*) to the inequality for all S_*; both directions of Theorem 6.14 inherit from it. The assertion is not tautological: bH is built from a C∞ decomposition (§4.2.1) that is only adapted to the parabolic filtration of F_* itself, not to an arbitrary holomorphic subsheaf S_*. For a general S_*, the Chern–Weil comparison involves the second fundamental form of S_* in F_*, and on X∘∖D it also involves boundary terms along D. Equality between ch1(S_*)·[η] and (√-1/2π)∫ Tr(F_{bH|S})∧η requires these contributions to vanish, which is a genuine analytic statement about the weights and the choice of H1. The paper gives no derivation, and no nearby lemma supplies it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims a Kobayashi–Hitchin correspondence for saturated reflexive parabolic sheaves over compact Kähler manifolds with a simple normal crossing divisor: a saturated reflexive parabolic sheaf F_* is μ_ω-polystable if and only if there exists an admissible Hermitian–Einstein metric with respect to ω on the regular part F|_{X°\\D} that is compatible with the parabolic structure (Theorem 6.14), with an analogous statement for semistability and approximate Hermitian–Einstein metrics (Theorem 6.15). The proof constructs a parabolic metric bH adapted in codimension 2 via resolution of singularities and a smooth decomposition of the pulled-back locally abelian parabolic bundle, then runs the Hermitian–Yang–Mills flow on the regular part, using analytic stability as an intermediate equivalence. The paper also derives a Bogomolov–Gieseker inequality for semistable parabolic sheaves with respect to nef and big classes (Theorem 7.1).","tokens_in":31689,"tokens_out":14321,"duration_ms":142355,"significance":"If the main theorem is correct, it extends the classical Kobayashi–Hitchin correspondence from vector bundles and reflexive sheaves to the parabolic setting, replacing the conical-metric Hermitian–Einstein metrics of [Li] by metrics adapted to the fixed Kähler form ω, and it provides a parabolic Bogomolov–Gieseker inequality even for nef and big classes. The paper contains substantial original constructions: the resolution procedure reducing a saturated reflexive parabolic sheaf to a locally abelian parabolic bundle (Section 3), the explicit adapted metric with Chern–Weil representatives for the first and second parabolic Chern characters (Propositions 4.10, 4.14, 4.15), and uniform Sobolev and heat-kernel estimates for conical Kähler metrics (Section 4.1). The main gap is a missing proof of the load-bearing Proposition 4.16, together with several steps delegated to previous work of the authors; these need to be supplied before the theorems can be considered established.","major_comments":[{"comment":"Proposition 4.16 asserts that the metric bH constructed in §4.2 is adapted in codimension 1 to every parabolic subsheaf S∗ of F∗, with the proof omitted as 'tautological'. This assertion is load-bearing: it is used in Proposition 5.1 to equate analytic stability with parabolic stability, and in Proposition 6.10 to show that the limiting Hermitian–Yang–Mills metrics are compatible with all parabolic subsheaves. The claim is not tautological: bH is built from a C∞ decomposition (Proposition 4.9) adapted only to the parabolic filtration of F∗ itself, whereas for a general subsheaf S∗ the Chern–Weil comparison between ch1(S∗) and (√−1/2π)Tr(F_{bH|S}) involves the second fundamental form of S∗ in F∗ and boundary terms along D. A complete proof, or a precise reference to one, is required.","section":"§4.2.3, Proposition 4.16"},{"comment":"The proof that the HYM flow converges to an approximate (or exact) Hermitian–Einstein metric depends on [Li-Zh-Zh, Proposition 4.1] for the semistable case and on 'the same trick used in [Li-Zh-Zh, Proposition 4.1]' for the stable case. The text does not state the hypotheses of that proposition or verify them in the present setting (noncompact base X∘∖D, reflexive parabolic sheaf with singular locus of codimension at least 3, conical Kähler metrics ω_{ϵδ}). Because this convergence is one half of Theorem 6.14, the authors should either provide the proof or carefully transcribe the relevant estimates with the precise conditions under which the cited result applies.","section":"§6.2, convergence of the HYM flow"},{"comment":"Proposition 6.12 establishes the L∞ and L^2_1 regularity of sections of an admissible Hermitian–Einstein metric compatible with a saturated reflexive parabolic sheaf on a polydisk with a simple normal crossing divisor. This proposition is needed in the converse direction of Theorem 6.14 to obtain the holomorphic splitting when equality holds in the slope inequality. Its proof appeals to [Ba-Si, Section 1] and to Biquard's extension theorem [Bq2, Theorem 2.1], which is stated for a smooth divisor; the extension to the simple normal crossing case is relegated to Remark 6.13 with a one-sentence assertion about solving a ∂-problem on Δ∗ × Δ∗. This extension is nontrivial and is load-bearing, so the details of the ∂-problem and the slice argument should be supplied.","section":"§6.2, Proposition 6.12 and Remark 6.13"},{"comment":"The parabolic Chern character is defined by formula (1), which the authors acknowledge is 'by no means standard' and for which there is no hint for general parabolic sheaves. Since the main theorems concern stability and Bogomolov–Gieseker inequalities with respect to this Chern character, the paper should clarify the scope: the results are for this specific definition. The agreement with the classical formulas (2)–(3) for locally abelian parabolic bundles in codimension 2 (cf. Lemma 2.4) is reassuring, but the proof of Lemma 2.8 and the additivity invoked in Section 7 should be expanded to make clear that they use only properties of the formula (1).","section":"§2.3, Definition 2.11"}],"minor_comments":[{"comment":"The expressions '∆(F0_∗) · [η^{n−1}_ϵ] ≥ 0' and the following lines should use [η^{n−2}_ϵ] to match the statement of Theorem 7.1; also 'F0_∗' is not defined and appears to denote a graded piece of the Jordan–Hölder filtration, which should be introduced.","section":"§7, proof of Theorem 7.1"},{"comment":"In the proof of Proposition 5.1 the reference 'Lemma 4.16' should be 'Proposition 4.16'.","section":"§5, proof of Proposition 5.1"},{"comment":"The heat kernel estimate in Proposition 4.6 writes the exponential factor as 'exp(− d_{ϵδ}(x,y)(4+τ)t)', which should presumably be 'exp(−d_{ϵδ}(x,y)^2/((4+τ)t))' to be dimensionally correct.","section":"§4.1, Proposition 4.6"},{"comment":"There are numerous typographical errors (e.g., 'comtatible', 'respcet', 'adpated', 'Metha' for 'Mehta') and a few notation ambiguities: X∘ is used for X\\W while X\\D also appears, and the notation D∗ in Section 6 is introduced only implicitly. A careful proofreading pass is recommended.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper's main theorem extends the second author's earlier result [Li] from parabolic bundles to saturated reflexive parabolic sheaves, while also replacing the conical-metric Hermitian–Einstein condition by compatibility with the fixed Kähler form ω. The heavy reliance on [Li] and [Li-Zh-Zh] as black boxes, and the missing proof of Proposition 4.16, make the novelty and correctness difficult to assess. The journal's scope is appropriate for this material. I recommend requesting a major revision with the missing details."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a serious extension of the parabolic Kobayashi-Hitchin program, and the main theorem is plausible, but the manuscript as written has several load-bearing steps that are asserted rather than proved. It deserves a serious referee, but the referee should demand real derivations, not just 'tautological'.\n\nWhat's genuinely new: the correspondence for saturated reflexive parabolic sheaves, which does not reduce to the parabolic-bundle case; the upgrade from conical-metric Hermitian-Einstein to approximate Hermitian-Einstein for the original Kähler class; and the nef-and-big Bogomolov-Gieseker inequality. The metric construction in §4 is substantial: conical Kähler metrics, uniform Sobolev via Nash-Yau entropy, heat kernel bounds, and the curvature computations for adaptedness in codimension 2. That is real analytic work, and the paper is clearly organized.\n\nThe soft spot is the adaptedness of bH to every parabolic subsheaf. Proposition 4.16 says it is 'tautological' and omits the proof. It is not tautological. bH is constructed to respect the parabolic filtration of F_* itself; for an arbitrary holomorphic subsheaf S_*, the Chern-Weil comparison involves the second fundamental form of S_* and, along D, boundary terms. Equality ch1(S_*)·[η] = (√-1/2π)∫ Tr(F_{bH|S})∧η is exactly what is needed to make analytic stability match parabolic stability (Prop 5.1) and to get compatibility of the limiting metric (Prop 6.10). Both directions of Theorem 6.14 inherit from it. The stress-test note is right on this point.\n\nTwo smaller items: Lemma 3.2 is a one-line consequence of the nonstandard Chern character definition and deserves a few more lines; and Prop 6.12's snc-divisor adaptation of Biquard is waved off as 'easily adapted' in a remark, which is an unproved claim. The stable-convergence step is delegated to [Li-Zh-Zh]; that one is less alarming if the cited result is genuinely the same trick, but it is still a black box.\n\nOverall: the central argument is credible and the contributions are real. The paper is not ready as is; the gaps are structural but probably fillable. I would send it to a capable referee and expect at least one revision. It belongs in a reading group.","headline":"A serious extension of the parabolic Kobayashi-Hitchin program to saturated reflexive sheaves, worth a careful referee, but with a few load-bearing steps asserted rather than proved.","tokens_in":32218,"tokens_out":2911,"would_cite":true,"duration_ms":32063,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C07","14J60","32L05","53C55"],"pacs":[],"model":"deepseek-v4-flash","headline":"A saturated reflexive parabolic sheaf is polystable if and only if it admits an admissible Hermitian-Einstein metric compatible with its parabolic structure.","keywords":["Kobayashi-Hitchin correspondence","parabolic sheaves","reflexive sheaves","Hermitian-Yang-Mills flow","Hermitian-Einstein metrics","Bogomolov-Gieseker inequality","Kähler manifolds","simple normal crossing divisor"],"falsifier":"On a Kähler surface with D = {xy = 0}, take a rank-2 saturated reflexive parabolic sheaf whose filtration jumps at the crossing, construct bH as in Section 4, and compare the trace of the restricted curvature on a parabolic subsheaf with its first parabolic Chern class as currents: any mismatch away from a codimension-2 set would refute Proposition 4.16 and undermine Theorem 6.14. Independently, a μω-polystable F_* with no admissible Hermitian-Einstein metric compatible with its parabolic structure, or an admissible H-E metric compatible with a non-polystable F_*, would refute the correspondence itself.","tokens_in":31180,"feed_emoji":"📐","tokens_out":10496,"duration_ms":109315,"temperature":0.7,"pith_summary":"This paper aims to establish a Kobayashi-Hitchin correspondence for saturated reflexive parabolic sheaves on compact Kähler manifolds with a simple normal crossing divisor. The main theorem, Theorem 6.14, states that such a parabolic sheaf is μω-polystable if and only if it admits an admissible Hermitian-Einstein metric with respect to the original Kähler form ω, compatible with the parabolic structure. The semistable analogue, Theorem 6.15, replaces the single metric by a family of approximate Hermitian-Einstein metrics. From this the paper derives the parabolic Bogomolov-Gieseker inequality and extends it to semistability with respect to a nef and big class. The consequence is that an algebraic stability condition becomes a curvature condition, bringing flow and metric techniques to bear on these sheaves.","feed_headline":"Parabolic polystability equals a Hermitian-Einstein metric","feed_subtitle":"The Kobayashi-Hitchin correspondence now covers reflexive parabolic sheaves on Kähler manifolds.","key_machinery":"The mechanism is the Hermitian-Yang-Mills flow with fixed determinant, starting from a specially constructed metric bH on the regular part of F_*. The paper first resolves the sheaf's singularities by blowing up, embeds F into a locally free sheaf E, and twists by an effective exceptional divisor so that the pullback becomes a locally abelian parabolic bundle E'_* = E ⊗ [-P_0]. A metric H_0 adapted to this parabolic bundle is built from local smooth decompositions compatible with the parabolic filtrations; its curvature reproduces the first and second parabolic Chern characters. Tensoring with a line bundle metric gives bH, adapted to F_* in codimension 2 and with polynomial curvature growth near the divisor. The flow then deforms bH; under polystability it converges to an admissible Hermitian-Einstein metric, and under semistability to approximate Hermitian-Einstein metrics. The transition from parabolic to analytic stability uses Proposition 4.16, which asserts that bH is adapted to every parabolic subsheaf in codimension 1.","core_discovery":"The central claim is Theorem 6.14: for a saturated reflexive parabolic sheaf F_* over (X, ω, D), μω-polystability is equivalent to the existence of an admissible Hermitian-Einstein metric with respect to ω on F|_{X^∘∖D} that is compatible with F_*. Compatibility means that the metric's trace curvature represents ch_1(F_*) and that, for every proper parabolic subsheaf S_*, the induced metric's first Chern class is at least ch_1(S_*) in the sense of currents. Admissibility requires the curvature to be in $L^{2}$ and its trace contraction with ω to be bounded. The paper also proves Theorem 6.15, where semistability is equivalent to a family of approximate Hermitian-Einstein metrics all compatible with the parabolic structure, and Corollary 6.16, the parabolic Bogomolov-Gieseker inequality. Theorem 7.1 extends that inequality to semistability with respect to a nef and big class.","pith_inferences":["A natural next step, not taken in the paper, is to feed the same adapted metric and flow into the parabolic Higgs bundle setting; the convergence and extension arguments appear to carry over once the Higgs field has tame singularities along the divisor.","If Proposition 4.16 can be supplied with a direct proof, the correspondence gives a practical criterion: checking parabolic stability could be reduced to finding a single admissible Hermitian-Einstein metric, avoiding a search over all parabolic subsheaves.","The compatibility inequality in Definition 1.1, rather than equality, is likely the correct notion for singular sheaves; it degenerates to the usual equality exactly when the underlying sheaf is locally free.","The nef-and-big Bogomolov-Gieseker inequality suggests that the discriminant bounds the geometry of semistable parabolic sheaves outside the Kähler cone, which may matter for compactness or moduli questions not addressed in the paper."],"forward_implications":["Polystability of F_* is equivalent to the existence of an admissible Hermitian-Einstein metric compatible with the parabolic structure, giving a curvature-theoretic criterion for the algebraic condition.","Semistability is characterized by a family of approximate Hermitian-Einstein metrics all compatible with the parabolic structure, extending the correspondence beyond stable objects.","Every semistable saturated reflexive parabolic sheaf satisfies the Bogomolov-Gieseker inequality Δ(F_*) · [ω]^{n-2} ≥ 0.","The same inequality holds when semistability is taken with respect to a nef and big class [η], not only a Kähler class.","Equality in the inequality occurs exactly when F|_{X\\D} is a vector bundle admitting a projectively flat Hermitian-Einstein connection compatible with the parabolic structure."],"supporting_citations":[{"why":"Supplies the notion of admissible Hermitian-Einstein metrics on reflexive sheaves and the regularity and removable-singularity results used in the converse direction.","marker":"[Ba-Si]"},{"why":"Proves the earlier parabolic bundle case with H-E metrics for conical Kähler metrics and gives curvature formulas for the second parabolic Chern character; this paper extends it to sheaves and to the fixed Kähler form.","marker":"[Li]"},{"why":"Provides the Hermitian-Yang-Mills flow, analytic stability, approximate H-E metrics, and the convergence framework on noncompact Kähler manifolds.","marker":"[Si1]"},{"why":"Supplies the semistability flow estimate lim ||Φ(t)||_{L^2} = 0 and its use in producing approximate Hermitian-Einstein metrics.","marker":"[Li-Zh-Zh]"},{"why":"Establishes the definitions and basic properties of parabolic sheaves, including reflexive saturated filtrations and the metric constructions used in Section 4.","marker":"[Mo1]"},{"why":"Provides the desingularization theorem used to blow up the singular locus and make the sheaf embed into a locally free bundle with parabolic structure.","marker":"[Hr]"},{"why":"Gives the parabolic Chern character for locally abelian parabolic bundles and the explicit formula the paper uses as definition.","marker":"[Iy-Si]"},{"why":"Proves the extension result for admissible H-E metrics on parabolic bundles over surfaces, used in the polystable converse to split F into parabolic subsheaves.","marker":"[Bq2]"},{"why":"Verifies that the Chern character formulas for locally abelian parabolic bundles agree with the explicit first and second Chern character expressions used throughout.","marker":"[Ta]"}],"fun_headline_variants":["Parabolic polystability = Hermitian-Einstein metric","Hermitian-Einstein metrics decode parabolic sheaf stability","Kobayashi-Hitchin for reflexive parabolic sheaves on Kähler","Parabolic sheaves: Hermitian-Einstein metric = polystability"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the unproved assertion, Proposition 4.16, that the constructed metric bH computes the first Chern class of every parabolic subsheaf S_* away from a set of codimension 2; if this assertion fails, the bridge from parabolic stability to the analytic stability used by the flow argument collapses.","fun_headline_variants_meta":{"raw":{"variants":["Parabolic polystability = Hermitian-Einstein metric","Hermitian-Einstein metrics decode parabolic sheaf stability","Kobayashi-Hitchin for reflexive parabolic sheaves on Kähler","Parabolic sheaves: Hermitian-Einstein metric = polystability"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000706,"raw_usage":{"total_tokens":3099,"prompt_tokens":780,"completion_tokens":2319,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":396,"completion_tokens_details":{"reasoning_tokens":2243}},"tokens_in":396,"tokens_out":2319,"duration_ms":21072,"temperature":1.0,"reasoning_tokens":2243,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T10:59:23.506776+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On a Kähler surface with D = {xy = 0}, take a rank-2 saturated reflexive parabolic sheaf whose filtration jumps at the crossing, construct bH as in Section 4, and compare the trace of the restricted curvature on a parabolic subsheaf with its first parabolic Chern class as currents: any mismatch away from a codimension-2 set would refute Proposition 4.16 and undermine Theorem 6.14. Independently, a μω-polystable F_* with no admissible Hermitian-Einstein metric compatible with its parabolic structure, or an admissible H-E metric compatible with a non-polystable F_*, would refute the correspondence itself.","supporting_citations":[],"review_version":1}