{"id":"c7fdd592-ee53-4ad4-b0b0-14ec60ee9ae6","arxiv_id":"2506.07059","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For semistable fibrations with general type fibers, a Kodaira-Spencer class supported on a suitable non-movable, foliation-invariant divisor forces the general fibers to be birational.","lead":"This paper studies families of shapes built over a curve and asks when all the shapes are the same up to mild modifications. It shows that if the infinitesimal change in the shapes is concentrated on a special lower-dimensional piece, and that piece satisfies two extra conditions, then the shapes are indeed birationally equivalent.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The surjectivity proof for the flow map g in Theorem 5.9 is incomplete: a limit point of a leaf need not lie in the leaf; compactness of the h-fiber supplies the missing completeness argument.","rationale":"The main theorem is a converse to Theorem 4.1, and its proof strategy is sound: use the volume-detecting subspace to produce h and a meromorphic vector field v tangent to the h-fibers, then flow inside compact h-fibers to trivialize the family birationally. The crucial point is that the flow exists over a uniform disk. The written proof tries to prove this via a limiting argument in the paragraph beginning 'We claim that g is surjective.' That argument relies on an invalid local inference: a sequence lying on a leaf and converging to a regular point does not force the limit point to lie on the same leaf, nor does it produce a point of that leaf over the limiting base value. Local transversality at x̄ only describes the leaf through x̄. Thus, as written, the proof does not strictly establish that the integral curves are defined over all of Δ. The gap is repairable: the fiber H of h over h(x0) is compact and avoids D, so v|_H is holomorphic and complete, making γ(t) defined for all t∈Δ. The paper should state this explicitly. I therefore do not see a counterexample to the theorem, but the main proof needs this repair. This is a distinct concern from the reader's emphasis on strictness in Lemma 5.6, although the reader did list 'surjectivity of g' as a step to expand; hence partial agreement. The verdict remains conditional.","tokens_in":17372,"tokens_out":42076,"duration_ms":462666,"concrete_test":"Verify the surjectivity step of Theorem 5.9 by replacing the limit-point inference with the standard completeness argument: for x0∈X0\\h_0^{-1}(Z), show the fiber H=h^{-1}(h(x0)) is compact, smooth, and disjoint from D (since h(x0)∉h(D)∪sing(h)), so v|_H is a holomorphic vector field on a compact manifold; conclude its flow is defined for all t∈Δ, yielding γ(t)∈F0 and g(F0)=Δ. If this replacement succeeds, the proof goes through; if not, the current 'it easily follows' sentence does not justify surjectivity.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the proof of Theorem 5.9, after choosing x0 in X0\\h_0^{-1}(Z) and letting F0 be the F_v-leaf through x0, the argument that g:F0→Δ is surjective runs: if t̄ is a boundary point, take x_n∈F0 with t_n→t̄; choose a limit point x̄; since x̄ lies in the closure of F0 and F_v is regular and transverse to f^{-1}(t̄) at x̄, 'it easily follows' that F0 meets f^{-1}(t̄). This step is invalid as written. A point in the closure of a leaf need not belong to the leaf; local transversality of F_v at x̄ concerns the leaf through x̄, not F0. So g(F0) could a priori be a proper open interval even though every boundary point is a regular point of F_v. The missing ingredient is that H=h^{-1}(h(x0)) is a compact smooth fiber of h (because h(x0) is outside Z, which contains h(D) and the singular values of h) and v is holomorphic on H. A holomorphic vector field on a compact complex manifold is complete, so the integral curve γ(t) through x0 is defined for all t∈Δ and γ(t)∈F0; hence g is surjective. Without this compactness/completeness argument, the existence of the flow over the whole disk—and therefore the birationality of the fibers—is not established. The proof is also hard to follow because X0 was just defined as the central fiber, yet the containment 'closure of F0 ⊂ h^{-1}(h(x0))∩X0' is only plausible if X0 denotes the total space over Δ.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the birational analogue of the classical statement that vanishing Kodaira--Spencer class implies that general fibers of a fibration are isomorphic. Theorem 4.1 proves that if the general fibers of a fibration are birational, then the general Kodaira--Spencer class is supported on an effective non-movable divisor. The main result, Theorem 5.9, gives a converse under additional hypotheses: a volume detecting subspace U, a relatively non-movable divisor D, generic fiberwise support of the Kodaira--Spencer class on D, and F-invariance of D. The proof constructs a meromorphic vector field lifting the base vector field, uses the volume detecting forms to build a morphism h and a foliation F, and then uses the flow of the vector field on the h-fibers to obtain bimeromorphic maps between general fibers. Section 6 provides local and global examples testing the necessity of the hypotheses, and derives a global Volumetric Theorem as an application.","tokens_in":17729,"tokens_out":6751,"duration_ms":70486,"significance":"If the main theorem is correct, it fills a natural gap in the literature by giving a birational isotriviality criterion in terms of supported deformations, complementing Theorem 4.1. The proof strategy, combining a meromorphic vector field with a volume-detecting subspace and a foliation, is original and plausible. The paper also contains explicit examples (Examples 6.1, 6.2, 6.3, 6.5) that illustrate why the hypotheses of non-movability and F-invariance are needed, and the final applications to the Volumetric Theorem are relevant. However, two load-bearing arguments are currently not fully justified: the surjectivity of the flow map in Theorem 5.9 and the passage from a rational map to a morphism in Proposition 5.5. These gaps appear fixable, but they need to be addressed before the main theorem can be accepted as proven.","major_comments":[{"comment":"The claim that the map g: F0 → Δ is surjective is not justified as written. From a limit point ar x of a sequence x_n ∈ F0 with f(x_n) → ar t, one can only conclude that ar x lies in the closure of F0; the local transversality of F_v at ar x concerns the leaf through ar x, not the leaf F0 through x0. The sentence \"it easily follows\" therefore does not prove that f^{-1}(ar t) meets F0. The missing ingredient is that the h-fiber H = h^{-1}(h(x0)) is compact and v is holomorphic on H, so the holomorphic vector field v is complete on H and the integral curve through x0 is defined over all of Δ. This completeness argument is load-bearing: without it, g(F0) could be a proper open interval, and the holomorphic map X0 \\ h_0^{-1}(Z) → X_t used to conclude birationality is not obtained.","section":"Section 5.3, proof of Theorem 5.9"},{"comment":"The proof of Proposition 5.5 is too terse at two central points. First, the \"standard local argument\" showing that the forms s_i are pullbacks of meromorphic forms on Y' and that dim Y' = k is only indicated by a citation. Second, the passage from a rational map X ⇢ Y' to a surjective morphism h: X → Y over a normal k-dimensional variety of Albanese general type via the generalized Castelnuovo-de Franchis theorem needs a precise statement or proof. Theorem 5.9 uses h and F = T_{X/Y} essentially, so this step should be fully justified rather than summarized.","section":"Section 5.1, Proposition 5.5"},{"comment":"In the proof of Proposition 5.3, the deduction of equation (5.3) from the commutativity of Diagram (5.2) and the assumption h^0(X_b, O_{X_b}(D_b)) = 1 is stated as \"immediately get\", but it involves a base-change/divisorial argument that is not written out. Since the equality s_1 ∧ ⋯ ∧ s_{k+1} = Σ ω_i ∧ f^*σ_i is one of the main ingredients used later to construct the foliation, a fuller explanation of this step would improve the accessibility and rigour of the paper.","section":"Section 5.1, after Diagram (5.2)"}],"minor_comments":[{"comment":"The notation X_0 is used inconsistently: in Section 5.1 it denotes f^{-1}(B_0), while in the proof of Theorem 5.9 it denotes the central fiber over Δ. This creates confusion in the containments involving closures of leaves; please use different notation for the central fiber, e.g. X_c or X_0^{\\mathrm{cent}}.","section":"Section 5.1 and Section 5.3"},{"comment":"Lemma 5.6 states the conclusion s_i(v) = 0 for i = 1, ..., n+1, but there are only k+1 forms s_i; the index range should be i = 1, ..., k+1. The same typo appears in the last sentence of the proof and in Remark 5.14, where \"Corollary 5.6\" should read \"Lemma 5.6\".","section":"Lemma 5.6"},{"comment":"The sentence \"Note that \\bar x is in \\overline{F_0} ⊂ h^{-1}(h(x_0)) ∩ X_0\" is only correct if X_0 denotes the whole space f^{-1}(Δ), not the central fiber; moreover, the inclusion of the closure in a fiber of h should be justified by the fact that h is continuous and constant on the leaves of F_v.","section":"Section 5.3, proof of Theorem 5.9"},{"comment":"In the displayed formula for the integral curve, the branch of the square root is specified on C \\setminus \\mathbb{R}_{<0}, but the expression \\sqrt{2t + \\lambda_1^2} requires that 2t + \\lambda_1^2 avoids that locus; this is true for the stated choices but might be worth a brief remark to avoid ambiguity.","section":"Section 6.1, Example 6.1"}],"recommendation":"major_revision","confidential_remarks":"The main theorem is plausible and the overall strategy is sound, but the proof of Theorem 5.9 contains a genuine logical gap in the surjectivity of the flow map, and Proposition 5.5 is too terse for a central construction. Both issues appear fixable by adding a completeness argument for the vector field on compact h-fibers and by expanding the morphism construction. The examples in Section 6 are valuable and support the necessity of the hypotheses. If the authors address the two load-bearing points, I would be willing to support acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main thing to know: this is a plausible and useful converse to their Theorem 4.1, and it deserves a referee. The genuinely new content is Theorem 5.9 plus the global Volumetric Theorem 6.8, with Section 6 examples showing the hypotheses are needed. The authors also say plainly that 4.1 is possibly well-known with no reference found, and they identify the gap they are filling. That is honest.\n\nWhat the paper does well: the volume-detecting subspace framework from [RZ1, RZ2, PZ] is deployed carefully; Proposition 3.5 and Lemma 5.6 give the meromorphic vector field with s_i(v)=0. Section 6 is genuinely useful: the local examples in 6.1 and 6.2 display exactly why F-invariance is needed, and the double-cover constructions show the non-movable and invariance conditions are not just technical. The paper is seriously written.\n\nWhere it is soft: the stress-test note is right. In the proof of Theorem 5.9, the argument that g:F_0 -> Delta is surjective contains an invalid step. A limit point of a leaf need not lie on the leaf; local transversality at the limit point concerns the leaf through that point, not F_0. The phrase \"it easily follows\" does not do the work claimed. The fix is available: because h(x_0) is outside Z, the h-fiber H is a compact manifold, v is holomorphic on H, D does not meet H, and a holomorphic vector field on a compact manifold is complete. So the integral curve through x_0 is defined on all of Delta, and g is surjective. I would ask the authors to replace that paragraph with this completeness argument. There is also a notational slip: X_0 is introduced as the central fiber, then used as if it were the total space over Delta in the same paragraph. That should be cleaned up.\n\nTwo smaller points. Proposition 5.5's \"standard local argument\" and the broad Castelnuovo-de Franchis application are terse; probably fine, but a referee should ask for a sentence or two of expansion. The missing reference for Theorem 4.1 is acknowledged, so I would not count it against the paper. The examples and computations in Section 6 look solid.\n\nBottom line: the main theorem is plausible, the proof has a real but local gap that is fixable, and the paper is not a paradigm shift but it is a solid contribution to deformation theory and birational isotriviality. I would send it to a serious referee. If I worked in that area I would cite it once the flow argument is patched.","headline":"A solid, honestly written paper with a real but fixable gap in the main flow argument; worth sending to a serious referee.","tokens_in":18267,"tokens_out":3246,"would_cite":true,"duration_ms":36711,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14D06","14E05","14J40","32M25"],"pacs":[],"model":"deepseek-v4-flash","headline":"General fibers are birational when the Kodaira-Spencer class is supported on a foliation-invariant divisor.","keywords":["birational isotriviality","supported deformations","Kodaira-Spencer class","volume detecting subspace","foliation","meromorphic vector field","fibrations of general type","semistable fibration"],"falsifier":"Construct a semistable fibration of general-type fibers that satisfies all the hypotheses of Theorem 5.9—a volume-detecting subspace, $h^0(X_b,\\mathcal{O}_{X_b}(D_b))=1$, Kodaira-Spencer supported on $D_b$, and $D$ invariant under $F$—but whose general fibers are not birational. Concretely, in local coordinates compute the radius of definition of the integral curves of the meromorphic vector field $v$; the theorem predicts a uniform positive lower bound independent of the starting point, so any example satisfying the hypotheses with radii shrinking to zero as the starting point approaches $D$ would disprove it.","tokens_in":17174,"feed_emoji":"🌀","tokens_out":9998,"duration_ms":95858,"temperature":0.7,"pith_summary":"The paper studies when the general fibers of a holomorphic family of complex projective varieties are birational, a property called birational isotriviality. It proves a converse to the known fact that birational fibers force the Kodaira-Spencer class—the first-order derivative of the complex structure—to be supported on a divisor. The main theorem says that if the supportedness is witnessed by a volume-detecting subspace of relative closed one-forms, if the supporting divisor is non-movable, and if the divisor is invariant under the natural foliation, then the general fibers are birational. The mechanism is a meromorphic vector field whose flow moves points between fibers; the hypotheses ensure the flow is defined on a disk of fixed radius, so it yields a birational identification. This matters because it gives a deformation-theoretic criterion for birational isotriviality in cases where the classical vanishing-of-Kodaira-Spencer argument does not apply.","feed_headline":"Supported deformations force birational general fibers","feed_subtitle":"A fibration whose Kodaira-Spencer class lives on a divisor, with two natural conditions, must have birational fibers.","key_machinery":"The load-bearing object is the meromorphic vector field $v$ on $f^{-1}(\\Delta)$ with poles on $D$ that lifts the base vector field $\\partial/\\partial t$ (Proposition 3.5). It is controlled by the volume-detecting subspace $U=\\langle\\eta_1,\\ldots,\\eta_{k+1}\\rangle$: a subspace of the local system of de Rham closed relative one-forms whose wedge product $\\eta_1\\wedge\\cdots\\wedge\\eta_{k+1}$ vanishes and which is strictly independent on the general fiber. These forms yield closed one-forms $s_i$ with $s_1\\wedge\\cdots\\wedge s_{k+1}=0$, which define a surjective morphism $h\\colon X\\to Y$ and a foliation $F=T_{X/Y}$. Lemma 5.6, via the injectivity of the morphism $\\beta'$ in diagram (5.2), shows that the contractions $s_i(v)$ vanish, so the leaves of the flow foliation $F_v$ lie inside the fibers of $h$ and avoid $D$. Compactness of $X$ then gives a uniform disk of definition for the flow, and the general-type hypothesis upgrades the resulting holomorphic maps to bimeromorphic, hence birational, maps.","core_discovery":"The central claim is Theorem 5.9. Let $f\\colon X\\to B$ be a semistable fibration whose fibers are of general type, and suppose there is a volume-detecting subspace $U=\\langle\\eta_1,\\ldots,\\eta_{k+1}\\rangle$ of the local system of de Rham closed relative $1$-forms, with an associated effective horizontal divisor $D$ contained in the common zero locus of the relative forms $\\omega_i$. Assume $h^0(X_b,\\mathcal{O}_{X_b}(D_b))=1$ for general $b$, that the Kodaira-Spencer class $\\xi_b$ is supported on $D_b$ for general $b$, and that $D$ is invariant under the foliation $F$ induced by the morphism $h\\colon X\\to Y$ obtained from $U$. Then the general fibers of $f$ are birational. Here \"supported on a divisor\" means the class maps to zero under $H^1(X_b,T_{X_b})\\to H^1(X_b,T_{X_b}(D_b))$, i.e. the first-order deformation is carried by the divisor. The proof produces a meromorphic vector field $v$ with poles on $D$ lifting $\\partial/\\partial t$, uses the volume-detecting condition to show its flow foliation is contained in the fibers of $h$ and avoids $D$, and then extends the resulting holomorphic maps between fibers to birational maps using the general-type hypothesis.","pith_inferences":["The three hypotheses in Theorem 5.9—supportedness, non-movability, and foliation invariance—are exactly the traces that birationality leaves on the Kodaira-Spencer class, so the theorem can be read as a deformation-theoretic characterization of birational isotriviality that avoids constructing an explicit birational map.","A quantitative version suggested by the proof is that the obstruction to birational triviality is governed by the rate at which the flow disks of $v$ shrink as initial points approach the polar divisor; this rate is controlled by foliation invariance and could be estimated in explicit families, as in the examples of Section 6.","The same flow-and-foliation mechanism should apply beyond volume-detecting subspaces, for instance whenever a rational map to a variety of general type controls the meromorphic vector field and its polar divisor is invariant; the Bogomolov sheaf case in Section 5.4 is evidence for this."],"forward_implications":["Under the hypotheses of Theorem 5.9, the general fibers of the fibration are birational, so the family is birationally isotrivial.","By Corollary 5.11, after passing to a finite cover of a Zariski open subset of the base, the family becomes birational to a product $X_0\\times B'$, with $X_0$ a projective model of the fiber.","Theorem 6.8 gives a global, non-smooth version of the Volumetric Theorem: a semistable family with a morphism to an abelian variety whose general fiber maps generically one-to-one, and with a general pullback subspace that is fiberwise Massey trivial, has birational general fibers.","Theorem 6.9 reproves the original local Volumetric Theorem using the same flow-and-foliation mechanism.","The same methods apply to Bogomolov sheaves (Theorem 5.15): a Bogomolov sheaf whose associated map is of general type and whose divisorial pole locus is foliation-invariant forces birational fibers."],"supporting_citations":[{"why":"Supplies the Volumetric Theorem and the strictness/Massey-triviality notions that motivate the volume-detecting condition and control the flow.","marker":"[PZ]"},{"why":"Provides the definition of strictness and the generalized Castelnuovo-de Franchis theorem used to construct the morphism h.","marker":"[Cat]"},{"why":"Establishes the local system of de Rham closed relative one-forms and the lifting results that produce the closed forms s_i.","marker":"[RZ2]"},{"why":"Theorem 1.1 on birationally isotrivial fiber spaces is used in Theorem 4.1 and in Corollary 5.11 to pass to a product after base change.","marker":"[BBG]"},{"why":"The classical flow argument showing that vanishing Kodaira-Spencer class gives local triviality; the paper adapts this to a meromorphic vector field.","marker":"[KS, KM]"},{"why":"Theorem 2 extends the holomorphic map between fibers to a bimeromorphic map under the general-type hypothesis.","marker":"[KO]"},{"why":"Gives the final step that bimeromorphic projective varieties are birational.","marker":"[Ch]"}],"fun_headline_variants":["Divisor-supported deformations yield birational fibers","When Kodaira-Spencer lives on a divisor, fibers become birational","Supported deformations force birationality of fibers","Birational isotriviality from divisor-supported deformations"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the polar divisor D is invariant under the foliation given by the fibers of h, together with the strictness of U on the general fiber; these are what keep the integral curves of the meromorphic vector field away from the poles and defined on a common disk.","fun_headline_variants_meta":{"raw":{"variants":["Divisor-supported deformations yield birational fibers","When Kodaira-Spencer lives on a divisor, fibers become birational","Supported deformations force birationality of fibers","Birational isotriviality from divisor-supported deformations"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000207,"raw_usage":{"total_tokens":1366,"prompt_tokens":879,"completion_tokens":487,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":495,"completion_tokens_details":{"reasoning_tokens":420}},"tokens_in":495,"tokens_out":487,"duration_ms":4468,"temperature":1.0,"reasoning_tokens":420,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T05:43:31.258581+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct a semistable fibration of general-type fibers that satisfies all the hypotheses of Theorem 5.9—a volume-detecting subspace, $h^0(X_b,\\mathcal{O}_{X_b}(D_b))=1$, Kodaira-Spencer supported on $D_b$, and $D$ invariant under $F$—but whose general fibers are not birational. Concretely, in local coordinates compute the radius of definition of the integral curves of the meromorphic vector field $v$; the theorem predicts a uniform positive lower bound independent of the starting point, so any example satisfying the hypotheses with radii shrinking to zero as the starting point approaches $D$ would disprove it.","supporting_citations":[],"review_version":1}