{"id":"f5262775-5404-4846-ad6b-a650e06294a9","arxiv_id":"2507.09486","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper extends the Sobolev-type inequalities of Guo-Phong-Song-Sturm and Guedj-Tô from functions to twisted differential forms using heat kernel estimates.","lead":"Fusheng Deng, Gang Huang, and Xiangsen Qin prove heat kernel, Green form, and Sobolev-type inequalities for differential forms twisted by a holomorphic vector bundle on compact Kähler manifolds. The work extends the recent Sobolev inequalities for functions from Guo-Phong-Song-Sturm and Guedj-Tô to forms, with applications to vanishing theorems and L^{q,p} estimates for the d-bar operator.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The comparison theorem [LX10, Thm 4.3] invoked at (3.12) is the load-bearing step; its hypotheses are not stated and may exceed the paper's assumptions.","rationale":"The reader's weakest assumption—the unverified use of [LX10, Theorem 4.3]—is precisely the step on which the central Gaussian heat kernel bound rests. If that cited domination fails or requires hypotheses not present in the paper, Theorem 1.1(i) and all downstream Sobolev and vanishing results collapse. The paper's proof of the scalar Gaussian bound has an algebra error, but it is repairable and the repaired constant (5/41) is stronger than the stated 1/9, so it does not threaten the main conclusion. The citation issue is the more fundamental concern because it is a black box: the paper does not state the hypotheses of [LX10, Thm 4.3], nor does it verify them. This is not an accusation of error, but a genuine verification gap that a concrete check can settle. Since the paper is otherwise coherent and the remaining arguments follow standard methods, the conditional verdict remains appropriate; no change to the reader's verdict is needed.","tokens_in":24981,"tokens_out":9045,"duration_ms":97155,"concrete_test":"Verify the exact statement of [LX10, Theorem 4.3]. If it applies to E-valued (p,q)-forms on compact Kähler manifolds under only Ric^E_{p,q} ≥ −K, the concern is resolved. If its hypotheses are stronger, check whether those hypotheses are automatically satisfied in the setting of Theorem 1.1, or derive (3.12) directly via Kato's inequality and the Bochner-Weitzenböck formula, confirming both the pointwise domination and the prefactor e^{Kt}.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3, proof of Theorem 1.1(i), line (3.12): the paper asserts |H_{p,q}(t,x,y)| ≤ e^{Kt}H(t,x,y) by [LX10, Theorem 4.3] without stating the theorem's hypotheses. This domination transfers the Gaussian upper bound from the scalar heat kernel to the E-valued (p,q)-form heat kernel, and every later estimate (Green form, Sobolev-type inequalities, Theorems 1.6 and 1.8) depends on it. The paper's assumptions are only the scalar Sobolev inequality (♣) and the Weitzenböck lower bound Ric^E_{p,q} ≥ −K. If [LX10, Thm 4.3] requires more—for instance, a lower Ricci bound on the base manifold, a trivial bundle, or a bound on the full curvature operator of Λ^{p,q}T^*M⊗E—then Theorem 1.1(i) is not proved for general Hermitian holomorphic bundles E, and the advertised generalization to twisted forms does not follow. A secondary algebra issue: the choice b := 10(φ(y)−φ(x))/41 in (3.19) does not yield the stated e^{−d(x,y)^2/(9T)} decay as written, since the quadratic-in-b term carries a factor T; taking b proportional to (φ(y)−φ(x))/T repairs this and gives the slightly stronger constant 5/41, so that part is fixable.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves heat kernel upper bounds for the \\bar\\partial-Laplacian on E-valued (p,q)-forms on a compact K\\\"ahler manifold, assuming a scalar Sobolev inequality and a lower bound on the Weitzenb\\\"ock curvature operator. From these bounds it derives estimates for Green forms, Sobolev-type inequalities for twisted differential forms, a vanishing theorem, L^{k,s}-estimates for \\bar\\partial, and versions for families of K\\\"ahler metrics satisfying entropy bounds, thereby generalizing results of Guo\\-Phong\\-Song\\-Sturm and Guedj\\-T\\^o to the twisted-form setting.","tokens_in":25181,"tokens_out":28550,"duration_ms":307886,"significance":"If the proofs are completed, the paper would give a substantial and useful generalization: a scalar Sobolev inequality is upgraded to a Sobolev-type inequality for forms twisted by a Hermitian holomorphic vector bundle, with explicit dependence of constants on the Sobolev constants, curvature lower bounds, and the spectral gap. The paper also contains a new lower bound for the first nonzero eigenvalue of the twisted \\bar\\partial-Laplacian and a vanishing theorem. The overall strategy is transparent, many steps are standard with tracked constants, and the advertised applications are concrete and falsifiable.","major_comments":[{"comment":"The choice b := 10(φ(y)−φ(x))/41 in (3.19) gives the exponent (41/20)b²T + b(φ(x)−φ(y)) = ((5T−10)/41)(φ(x)−φ(y))², which contains no 1/T factor and cannot yield the claimed Gaussian decay e^{−d(x,y)²/(9T)}; for T→0 the exponent tends to a finite constant rather than to −∞. The argument is repaired by taking b := 10(φ(y)−φ(x))/(41T), which produces e^{−5d(x,y)²/(41T)} and hence implies the stated 1/9 bound, but as written the proof of the central heat kernel estimate contains a concrete algebraic error.","section":"§3, proof of Theorem 1.1(i), Eqs. (3.19)–(3.20)"},{"comment":"The domination |H_{p,q}(t,x,y)| ≤ e^{Kt}H(t,x,y) is invoked from [LX10, Theorem 4.3] without stating the hypotheses of that theorem. This comparison is the only mechanism transferring the scalar Gaussian bound to the E-valued (p,q)-form heat kernel, and every later estimate depends on it. The authors must verify explicitly that [LX10, Theorem 4.3] applies under the paper's assumptions, namely the scalar Sobolev inequality (♣) and Ric^E_{p,q} ≥ −K; if the theorem requires additional hypotheses (for example, a lower bound on the base Ricci curvature or some positivity condition), then Theorem 1.1(i) is not proved at the advertised level of generality.","section":"§3, Eq. (3.12)"},{"comment":"The proof derives the pointwise operator-norm bound |H_{p,q,ω}(t,x,x)| ≤ Vω^{-1}(1 + C e^{−c Iω^{-1}t}) and then states that 'taking traces' yields b_{p,q,ω} + e^{−µ1,ω t} ≤ 1 + C e^{−c Iω^{-1}t}. Since the displayed bound is on the operator norm of the endomorphism-valued heat kernel, the fiber trace is bounded by the fiber dimension of Λ^{p,q}T^*X⊗E times the displayed quantity, not by the same quantity. Moreover, if b_{p,q,ω} > 1, the left-hand side tends to b_{p,q,ω} as t→∞ while the right-hand side tends to 1, so the asserted inequality is impossible in general. This invalidates the derivation of µ1,ω ≥ c/Iω and hence the second part of Theorem 1.8; a different argument is needed.","section":"§6, Proposition 6.6"}],"minor_comments":[{"comment":"The constant in the statement is written as 'C := (k, ℓ, α, β, γ, µ1, K,|M |)' with the 'C' missing after ':='; it should read 'C := C(k, ℓ, α, β, γ, µ1, K,|M |)'.","section":"Corollary 4.3"},{"comment":"The reference [LX10] is cited with page numbers '620-247'; the page range appears to be incomplete (likely 620-647) and should be corrected.","section":"References"},{"comment":"After taking Lipschitz functions to converge to the distance function, the exponent in (3.21) should be d(x,y)²/(9T) rather than (φ(x)−φ(y))²/(9T); the notation currently leaves the distance dependence implicit.","section":"§3, Eq. (3.21)"},{"comment":"The phrase 'Moiser iteration' should read 'Moser iteration'.","section":"Lemma 3.8"},{"comment":"The proof of Theorem 1.8 relies on Lemma 6.1 from the unpublished preprint [GPSS23]; the authors should indicate the current status of [GPSS23] and state the precise hypotheses of the cited result so that the dependence of the main theorem on an external preprint is transparent.","section":"Section 6 / Theorem 1.8"}],"recommendation":"major_revision","confidential_remarks":"The paper has a clear and interesting program, and most of the local estimate chain is standard and likely correct after a small fix in the Gaussian-bound derivation. My main concerns beyond the algebraic slip are the unverified applicability of [LX10, Theorem 4.3] and the trace argument in Proposition 6.6, which currently invalidates the eigenvalue lower bound for families. I recommend major revision rather than rejection, provided the authors supply the missing hypotheses and repair Proposition 6.6."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this paper has a genuinely new idea: instead of trying to define a Monge-Ampère operator for twisted forms, it derives heat kernel estimates for E-valued (p,q)-forms from a Sobolev inequality for scalar functions, then boots those up to Sobolev and Green form estimates. That is a real step beyond GPSS and Guedj-Tô. Second, the proof of the central Gaussian bound, Theorem 1.1(i), has a concrete algebra error in the Davies argument. At (3.19) the authors take b := 10(φ(y)-φ(x))/41 and claim the exponent becomes -d^2/(9T). Direct substitution gives -5(φ(x)-φ(y))^2/41 with no 1/T factor. To get 1/T decay you have to set b proportional to (φ(y)-φ(x))/T, and then the constant is 5/41, not 1/9. The bound may still be correct in substance, but as written it is not proved, and the later Green form and Sobolev estimates inherit the problem.\n\nThe rest of the paper is largely a standard chain: eigenvalue and eigensection estimates via Moser iteration, Green form estimates from the heat kernel, Riesz potential arguments with Marcinkiewicz interpolation, and the family version via the GPSS framework. Those parts look credible, and the paper is honest about what is input and what is derived. The dependence on unpublished preprints ([GPSS23], [GT24], [DHQ25]) is a citation-pattern concern, but not by itself a flaw.\n\nThe other load-bearing point is the comparison theorem (3.12) from [LX10, Thm 4.3], used to transfer the scalar Gaussian bound to the twisted form heat kernel. The paper cites it without stating its hypotheses. If that theorem requires anything beyond Ric^E_{p,q} ≥ -K and the scalar Sobolev inequality, the whole transfer fails. I cannot check the cited theorem from here, so the burden is on the authors to quote it exactly. This is the first thing a referee should ask for.\n\nVerdict: the paper has a real contribution and deserves a serious referee, but it is not ready in its present form. The algebra slip is fixable and probably minor, the comparison theorem should be checkable; the rest of the architecture seems sound. I would send it to review with a request for a corrected Gaussian bound and a full statement of [LX10, Thm 4.3].","headline":"A real extension of the GPSS/Guedj-Tô scalar Sobolev machinery to twisted forms, but the central Gaussian heat kernel bound has a fixable algebra error and the key comparison theorem is invoked without stating its hypotheses.","tokens_in":25840,"tokens_out":3497,"would_cite":false,"duration_ms":34239,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["32W05","58J35","53C55"],"pacs":[],"model":"deepseek-v4-flash","headline":"A scalar Sobolev inequality plus a curvature lower bound forces Gaussian heat kernel bounds for bundle-valued forms on compact Kähler manifolds, and from these bounds come Sobolev inequalities for twisted forms.","keywords":["heat kernel estimates","twisted differential forms","Sobolev inequalities","Kähler manifolds","Weitzenböck curvature","holomorphic vector bundles","Green forms","∂̄-operator"],"falsifier":"On a compact Kähler manifold where the scalar Sobolev inequality holds with explicit constants and the spectrum of $\\square_{p,q}$ is computable (for example $\\mathbb{CP}^1$ with the Fubini-Study metric and a twist line bundle $O(-k)$), evaluate the heat kernel diagonal $H_{p,q}(t,x,x)$ numerically from the eigenfunction expansion and compare it with $C(\\alpha) \\beta^{\\alpha/(\\alpha-1)} e^{Kt+\\gamma t/\\beta} t^{\\alpha/(1-\\alpha)}$; a single point $(t,x)$ where the bound fails would refute Theorem 1.1(i).","tokens_in":24662,"feed_emoji":"🌀","tokens_out":11351,"duration_ms":110724,"temperature":0.7,"pith_summary":"The paper aims to extend Sobolev-type inequalities from plain functions to twisted differential forms—sections of $\\Lambda^{p,q}T^*M \\otimes E$, where $E$ is a Hermitian holomorphic vector bundle—on compact Kähler manifolds. The route is a heat kernel estimate: assuming a scalar Sobolev inequality for functions and a lower bound $\\operatorname{Ric}^E_{p,q} \\ge -K$ on the Weitzenböck curvature operator, the authors prove Gaussian upper bounds for the heat kernel of the $\\bar{\\partial}$-Laplacian on twisted forms, with explicit dependence on the Sobolev constants and $K$. From that bound they derive Sobolev-type inequalities for forms, pointwise estimates of Green forms, a vanishing theorem for Dolbeault cohomology, and $L^{k,s}$-estimates for the $\\bar{\\partial}$-operator. If correct, these results transfer to twisted forms a body of estimates previously known only for functions, with constants explicit enough to be used in Kähler family settings.","feed_headline":"Heat kernel bounds unlock Sobolev inequalities for twisted forms","feed_subtitle":"The ∂̄-heat kernel estimate transfers scalar Sobolev inequalities to bundle-valued forms, with explicit constants.","key_machinery":"The load-bearing object is the heat kernel $H_{p,q}(t,x,y)$ of the $\\bar{\\partial}$-Laplacian $\\square_{p,q}$ on $\\Lambda^{p,q}T^*M \\otimes E$, together with the Weitzenböck curvature operator $\\operatorname{Ric}^E_{p,q} := 2\\square_{p,q} - \\nabla^*\\nabla$, used through the Bochner-Weitzenböck formula. The proof transfers a Gaussian upper bound for the scalar heat kernel—itself obtained from the scalar Sobolev inequality by a standard argument—to the twisted heat kernel through a comparison inequality $|H_{p,q}(t,x,y)| \\le e^{Kt}H(t,x,y)$. A Moser iteration scheme for solutions of the heat equation upgrades the $C^0$ estimate to $C^1$ estimates of $\\bar{\\partial} H_{p,q}$ and $\\bar{\\partial}^* H_{p,q}$. The Sobolev inequalities for forms then come from representing the pseudo-differential operator $\\square_{p,q}^{-1/2}$ as an integral of the heat kernel, proving weak-type bounds for it, and interpolating by Marcinkiewicz to get $L^k$ bounds on $f - Pf$.","core_discovery":"The central claim is Theorem 1.1: under the scalar Sobolev inequality $(\\int_M |f|^{2\\alpha})^{1/\\alpha} \\le \\beta \\int_M |\\nabla f|^2 + \\gamma \\int_M |f|^2$ and the curvature condition $\\operatorname{Ric}^E_{p,q} \\ge -K$, the heat kernel $H_{p,q}(t,x,y)$ of the $\\bar{\\partial}$-Laplacian on $E$-valued $(p,q)$-forms satisfies $|H_{p,q}(t,x,y)| \\le C(\\alpha) \\beta^{\\alpha/(\\alpha-1)} e^{Kt+\\gamma t/\\beta} t^{\\alpha/(1-\\alpha)} e^{-d(x,y)^2/(9t)}$. From this the paper derives, among other things, the Sobolev-type inequality $(\\int_M |f-Pf|^{2\\alpha})^{1/\\alpha} \\le C V^{1/\\alpha-1}(1+\\eta)^2 (\\int_M |\\bar{\\partial} f|^2 + \\int_M |\\bar{\\partial}^* f|^2)$ for forms, with $P$ the Bergman projection, uniform over Kähler families satisfying entropy bounds and $\\operatorname{Ric}^E_{p,q} \\ge -K$. It also proves Green form estimates $|G_{p,q}(x,y)| \\le C d(x,y)^{2/(1-\\alpha)} + C$, a vanishing theorem when $\\operatorname{Ric}^E_{p,q}$ is nonnegative and positive somewhere, and $L^{k,s}$-solutions of $\\bar{\\partial} u = f$ among twisted forms.","pith_inferences":["Beyond the paper: the explicit constant dependence in Theorem 1.1(i) suggests the Gaussian bound survives on complete noncompact Kähler manifolds satisfying the same scalar Sobolev inequality, which the paper notes only in passing.","Beyond the paper: because the argument never defines a Monge-Ampère operator for forms, the same strategy—scalar Sobolev inequality plus a comparison theorem—could produce form-level Sobolev inequalities on spaces where scalar Sobolev bounds are already available, such as singular or non-Kähler settings.","Beyond the paper: the $\\eta$ factor in the family inequality depends on $\\mu_{1,\\omega}$; combining the new eigenvalue lower bound with the main estimate could make the family constants fully explicit and remove the $\\mu_1$ dependence from the Sobolev inequality."],"forward_implications":["Uniform Sobolev inequalities for twisted forms hold on Kähler families satisfying diameter, entropy, and curvature bounds, with constants depending only on the family parameters and the first nonzero eigenvalue.","The Green form of $\\square_{p,q}$ is pointwise controlled by powers of the distance, giving explicit kernel bounds usable in potential theory on forms.","A vanishing theorem: if $\\operatorname{Ric}^E_{p,q} \\ge b \\ge 0$ with $b > 0$ somewhere, then $H^{p,q}(M,E) = 0$, and the $\\bar{\\partial}$-equation is solvable in $L^k$ with $L^s$ data.","The heat kernel estimate yields a lower bound $\\mu_{1,\\omega} \\ge c/I_\\omega$ for the first nonzero eigenvalue of $\\square_{p,q,\\omega}$ in the family setting with $K = 0$."],"supporting_citations":[{"why":"Supplies the comparison inequality $|H_{p,q}(t,x,y)| \\le e^{Kt}H(t,x,y)$ that transfers the scalar Gaussian bound to twisted forms.","marker":"[LX10, Theorem 4.3]"},{"why":"Gives the standard route from a Sobolev inequality to a Gaussian upper bound for the scalar heat kernel, which the proof follows.","marker":"[D90]"},{"why":"Provides the scalar Sobolev inequality and the presentation used to derive the Gaussian bound in the Kähler family setting.","marker":"[GPSS23]"},{"why":"The function-level Sobolev-type inequality for Kähler families that the paper generalizes to twisted forms.","marker":"[GT24]"},{"why":"Source of the C0 and C1 eigensection estimates used to prove the off-diagonal and derivative estimates of the heat kernel.","marker":"[LZZ21]"},{"why":"Supplies the weak-type estimate and Marcinkiewicz interpolation argument that turns heat kernel bounds into Sobolev inequalities via $\\square_{p,q}^{-1/2}$.","marker":"[L12]"}],"fun_headline_variants":["Heat kernel estimates extend Sobolev inequalities to twisted forms","Twisted form heat kernels yield Sobolev and vanishing results","Heat kernel bounds for bundle-valued forms imply Sobolev inequalities","Sobolev inequalities for twisted forms from heat kernel estimates"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof leans on a comparison theorem that bounds the twisted-form heat kernel by the scalar heat kernel times $e^{Kt}$, and the paper uses that theorem without stating its precise hypotheses; if the theorem requires stronger curvature or metric conditions than $\\operatorname{Ric}^E_{p,q} \\ge -K$, the Gaussian bound and every estimate built on it would need reworking.","fun_headline_variants_meta":{"raw":{"variants":["Heat kernel estimates extend Sobolev inequalities to twisted forms","Twisted form heat kernels yield Sobolev and vanishing results","Heat kernel bounds for bundle-valued forms imply Sobolev inequalities","Sobolev inequalities for twisted forms from heat kernel estimates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000562,"raw_usage":{"total_tokens":2694,"prompt_tokens":994,"completion_tokens":1700,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":610,"completion_tokens_details":{"reasoning_tokens":1630}},"tokens_in":610,"tokens_out":1700,"duration_ms":12908,"temperature":1.0,"reasoning_tokens":1630,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:58:18.946103+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On a compact Kähler manifold where the scalar Sobolev inequality holds with explicit constants and the spectrum of $\\square_{p,q}$ is computable (for example $\\mathbb{CP}^1$ with the Fubini-Study metric and a twist line bundle $O(-k)$), evaluate the heat kernel diagonal $H_{p,q}(t,x,x)$ numerically from the eigenfunction expansion and compare it with $C(\\alpha) \\beta^{\\alpha/(\\alpha-1)} e^{Kt+\\gamma t/\\beta} t^{\\alpha/(1-\\alpha)}$; a single point $(t,x)$ where the bound fails would refute Theorem 1.1(i).","supporting_citations":[],"review_version":1}