Smooth Limits of Stable Manifolds
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.
Ada: We're looking at the paper: "Smooth Limits of Stable Manifolds".
Emmy: Here's the gist of the paper. Let X be a compact complex manifold of dimension n, and suppose that X is deformation equivalent to a stable manifold.
Ada: Then X is projective whenever n ≤ 3 <ref:2610.23457#pg2> We also construct, for every n ≥ 4, smooth families of compact complex manifolds over the unit disc such that every fiber over the punctured disc is a stable manifold, while the central fiber is a non-Kähler Moishezon manifold <ref:2610.23457#pg2> The core problem addressed by this work concerns whether a compact complex manifold deformation equivalent to a Kähler–Einstein manifold of nonpositive scalar curvature must be Kähler, and the paper provides dimension-sharp answers to this question <ref:2610.23457#pg2> The principal positive result is that the answer is affirmative in complex dimension at most three and negative in every complex dimension at least four <ref:2610.23457#pg2> This settles the negative-curvature smooth-limit problem in every dimension <ref:2610.23457#pg2> The counterexamples give a negative answer to Yau’s question in dimensions at least four even when every punctured fiber is assumed to be canonically polarized <ref:2610.23457#pg2> Dimension-dependent results for smooth limits The paper establishes dimension-dependent results concerning the persistence of the Kähler property under deformation, specifically addressing the negative Kähler–Einstein smooth-limit case of Yau’s question <ref:2610.23457#pg2> Theorem 1.2 states that for complex dimension 1 ≤ n ≤ 3, if every fiber over the punctured disc is a stable manifold, then X0 is a smooth projective manifold, hence a smooth Kähler manifold <ref:2610.23457#pg2> For n ≥ 4, Theorem 1.3 constructs proper holomorphic submersions M → ∆ with connected compact fibers of dimension n such that KMt is ample for every t ≠ 0, while M0 is Moishezon and non-Kähler <ref:2610.23457#pg2> The proof strategy involves canonical models and cohomology The proof of Theorem 1.2 relies on Riemann–Roch, Kodaira vanishing on the punctured fibers, and semicontinuity to show KX0 is big, implying X0 is Moishezon <ref:2610.23457#pg2> The key step involves a comparison of second cohomology with intersection cohomology <ref:2610.23457#pg2> In dimension three, the canonical morphism of the central fiber is shown to be crepant, and its fibers have dimension at most one <ref:2610.23457#pg2> The projectivity criterion then makes the canonical morphism projective, and hence X0 projective <ref:2610.23457#pg2> For dimension four and higher, Theorem 1.3 is proven by starting with a pencil of smooth quartic K3 surfaces whose central member contains exactly one line and whose nearby members contain none <ref:2610.23457#pg2> Cohomological arguments for arbitrary dimensions Sections 3–7 establish the canonical-model and cohomological arguments in arbitrary dimension <ref:2610.23457#pg2> Proposition 3.1 shows that if KXt is ample for t = 0 ≠, then KX0 is big and X0 is Moishezon <ref:2610.23457#pg2> The construction of the relative canonical model Y → ∆ uses invariance of plurigenera and finite generation of canonical rings to produce a flat projective relative canonical model <ref:2610.23457#pg2> Lemma 4.1 asserts the existence of a positive integer m∗ such that m∗KXt is very ample and its section ring is generated in degree one <ref:2610.23457#pg2> Proposition 4.2 shows that after shrinking ∆, there is a sufficiently divisible integer m ≥ 2 such that the graded O∆-algebra R(m) is generated in degree one and g: Y = Proj∆ R(m) → ∆ is flat and projective <ref:2610.23457#pg2> Analysis of the central fiber properties The analysis of the central fiber X0 involves several key results regarding its geometry and cohomology <ref:2610.23457#pg2> Lemma 4.3 shows that the total space Y is normal and of pure dimension n + 1, with b2(X0) = b2(Yt) for t ≠ 0, and the limiting mixed Hodge structure H2lim(Yt) is pure of weight two <ref:2610.23457#pg2> The analysis in dimension three shows that the set B is empty, implying that the canonical map f: X0 → Y0 = Proj R(X0, KX0) is a proper birational morphism with KX0 = f∗KY0 <ref:2610.23457#pg2> Lemma 8.5 proves that every irreducible curve contracted by f is a smooth rational curve <ref:2610.23457#pg2> The final conclusion for dimension three In dimension three, the proof shows that the stable base locus has no divisorial component, which completes the proof <ref:2610.23457#pg2> Theorem 7.9 concludes that c = c2(Y0) = b2(Xt) − Ib2(Y0), and since r ≥ 0, it follows that r = 0, c = c2 = b2(Yt) − Ib2(Y0) <ref:2610.23457#pg2> This implies the stable base locus has no divisorial component <ref:2610.23457#pg2> Counterexamples in higher dimensions For dimension at least four, counterexamples are constructed using a pencil of smooth quartic K3 surfaces S0 ⊂ P3 containing exactly one line <ref:2610.23457#pg2> The resulting family Yt = S
2: t carries the holomorphic symplectic form associated with the K3 surface St <ref:2610.23457#pg2> The assertion that X0 is not Kähler is proven by showing that
C∨+: +
C−: = 0 in H2(X0, R), and neither summand is zero <ref:2610.23457#pg2> The final result for dimension four For n > 4, the counterexamples are formed by taking T to be a product of n − 4 smooth projective curves of genus at least two, and forming the product family X × T → ∆ <ref:2610.23457#pg2> The central fiber is Moishezon because X0 is bimeromorphic to a projective manifold and T is projective <ref:2610.23457#pg2> If X0 × T were Kähler, restricting a Kähler form to X0 × u would make X0 Kähler, a contradiction <ref:2610.23457#pg2> The final conclusion for dimension n > 3 The paper concludes that the canonical morphism f: X0 −→ Y0 = Proj R(X0, KX0) is crepant, all its fibers have dimension at most one, and every exceptional prime divisor maps onto a curve <ref:2610.23457#pg2> Every irreducible contracted curve is a smooth rational curve and deforms in X0 to a general transverse Du Val exceptional curve <ref:2610.23457#pg2> This implies that the stable base locus has no divisorial component, which completes the proof <ref:2610.23457#pg2> The final conclusion for dimension n = 3 In dimension three, the complete base-point-free system mKX0 now gives the canonical morphism f, with OX0(mKX0) ≃ f∗OY0(mKY0) <ref:2610.23457#pg2> Equivalently, (8.2) has F = 0 <ref:2610.23457#pg2> Thus (8.4) holds for compatible canonical divisors <ref:2610.23457#pg2> The set B is empty, in particular, the canonical map is a proper birational morphism f: X0 −→ Y0 = Proj R(X0, KX0), KX0 = f∗KY0 <ref:2610.23457#pg2> The final conclusion for dimension n > 3 For n > 4, the counterexamples are formed by taking T to be a product of n − 4 smooth projective curves of genus at least two, and forming the product family X × T → ∆ <ref:2610.23457#pg2> The central fiber is Moishezon because X0 is bimeromorphic to a projective manifold and T is projective <ref:2610.23457#pg2> If X0 × T were Kähler, restricting a Kähler form to X0 × u would make X0 Kähler, a contradiction <ref:2610.23457#pg2> References
1: T.
Emmy: Aubin, Équations du type Monge–Ampère sur les variétés kählériennes compactes, C.
Ada: R.
Title and authors: Ada: So we're looking at this paper, "Smooth Limits of Stable Manifolds." It’s really about answering Yau’s question about whether a compact complex manifold deformation equivalent to a Kähler-Einstein manifold of negative scalar curvature must itself be Kähler.
Emmy: Exactly. The paper gives a dimension-sharp answer to that persistence question. In short, it's affirmative for complex dimensions three or less, but negative for every dimension at least four > Page two <ref:2610.23457#pg2> <ref:2610.23457#pg2>
Kolya: So the core of what they’re doing is constructing these smooth limits where the metric on the punctured part is nice, but the central fiber isn't. That seems like a pretty wild construction to start with.
Ada: It is. They build families over a unit disc where every fiber away from zero admits that Kähler-Einstein metric of negative scalar curvature, but when you look at the center, it’s a non-Kähler Moishezon manifold if the dimension is four or higher > Page one <ref:2610.23457#pg1> <ref:2610.23457#pg1,a non-Kähler Moishezon manifold>
Emmy: And they use this construction to set up their main theorems. Theorem one point two says that for dimensions one through three, if every fiber away from zero is a stable manifold, then the central fiber Xzero must be a smooth projective manifold, which means it’s Kähler > Page two <ref:2610.23457#pg2> <ref:2610.23457#pg2,is a stable manifold, then>
Kolya: That's strong. So for those lower dimensions, they’re essentially saying that the structure forces the limiting object to be nice. But what happens when we go past three?
Ada: That's where things get interesting because they construct counterexamples for every dimension four and higher using these smooth families > Page one <ref:2610.23457#pg1> <ref:2610.23457#pg1>
Emmy: They use a pencil of smooth quartic K3 surfaces as an example for dimension four, and then for dimensions greater than four, they take a product family of n minus four smooth projective curves where the central fiber is Moishezon but not Kähler > Page three <ref:2610.23457#pg3> <ref:2610.23457#pg1>
Kolya: So the implication there is that even when we assume every punctured fiber has a canonical polarization, we still get non-Kähler results in higher dimensions > Page two <ref:2610.23457#pg2> <ref:2610.23457#pg2>
Title and authors: Ada: That’s the main negative finding. But for dimension three, they do something specific with their proof strategy. They show that the canonical morphism of the central fiber is crepant and its fibers are at most one dimension high > Page two <ref:2610.23457#pg2> <ref:2610.23457#pg2,that the canonical morphism of the central fiber is crepant>
Emmy: And this crepancy, combined with some inverse-Cartanmatrix construction invariant under Dynkin-diagram monodromy, allows them to make that canonical morphism projective > Page three <ref:2610.23457#pg3> <ref:2610.23457#pg1>
Kolya: So they’re using the geometry of the singularities to prove projectivity in dimension three. I wonder if that machinery would work for higher dimensions where they're using these different pencils of surfaces or product families.
Ada: They use different tools for each case, Kolya. For dimension four and above, they rely on starting with a pencil of smooth quartic K3 surfaces whose central member has exactly one line > Page three <ref:2610.23457#pg3> <ref:2610.23457#pg1>
Emmy: That specific construction allows them to show that the secant-line morphism contracts a specific plane in the central fiber down to a two dimensional space, while being finite on the other fibers > Page three <ref:2610.23457#pg3> <ref:2610.23457#pg1>
Kolya: So they’re essentially showing how the topology of those surfaces dictates what happens at zero in this deformation process. What does that tell us about how we can find these stable manifolds?
Ada: It tells us that for dimension three, you can always find a way to make the limiting object projective, which is a huge structural statement > Page two <ref:2610.23457#pg2> <ref:2610.23457#pg2>
Emmy: But for dimension four and up, they show that there are families where the central fiber is Moishezon but not Kähler, which means it’s not projective either > Page one <ref:2610.23457#pg1> <ref:2610.23457#pg1>
Kolya: So to summarize what's here: the persistence of the Kähler property depends entirely on whether you are in a low dimension or a high dimension. The paper "Smooth Limits of Stable Manifolds" settles this by giving a precise answer for each case > Page two <ref:2610.23457#pg2> <ref:2610.23457#pg2>
Title and authors: Ada: It really shows how sensitive these geometric properties are to the ambient dimension. We have to be extremely careful with our assumptions when we try to extend results from lower dimensions, especially when dealing with those higher dimensional families > Page one <ref:2610.23457#pg1> <ref:2610.23457#pg1>
Emmy: And the construction of that non-Kähler Moishezon manifold for n at least four using the product family of curves, is a very concrete way to show that there are plenty of ways to get non-Kähler limits when you have more room in the dimension > Page three <ref:2610.23457#pg3> <ref:2610.23457#pg1>
Kolya: The final results they get on the stable base locus in dimension three, showing it has no divisorial component because the set B is empty, that’s a good structural result about how those maps behave locally > Page two <ref:2610.23457#pg2> <ref:2610.23457#pg2>
Ada: It confirms that in dimension three, we have enough control over the canonical map to force it to be proper and birational onto its image > Page two <ref:2610.23457#pg2> <ref:2610.23457#pg2>
Emmy: So, if you take what they did with the inverse-Cartanmatrix construction in dimension three, it seems like a very powerful way to establish projectivity when things are well-behaved > Page three <ref:2610.23457#pg3> <ref:2610.23457#pg1>
Kolya: I just want to be clear about the limitation here: for dimensions higher than three, they rely on these specific constructions—like the K3 surfaces or product curves—to build those counterexamples. If you were looking at a different type of deformation, this result wouldn't apply > Page two <ref:2610.23457#pg2> <ref:2610.23457#pg2>
Ada: Exactly. The paper says it settles the negative-curvature smooth-limit problem in every dimension by giving us these dimension-dependent results > Page one <ref:2610.23457#pg1> <ref:2610.23457#pg1>
Emmy: So the big picture is that we can't just assume Kähler persistence; we have to check the dimension and then use very specific geometric tools to determine if it holds > Page two <ref:2610.23457#pg2> <ref:2610.23457#pg2>
Kolya: That’s what this paper means for the world: it gives us a clear dividing line in complex geometry where these kinds of deformations behave fundamentally differently, allowing us to target our research more effectively > Page one <ref:2610.23457#pg1> <ref:2610.23457#pg1>
The paper's summary: Ada: So, to recap, this paper is really about how complex manifolds behave when you deform them in a way that preserves some kind of geometric structure away from the center. It shows that for low dimensions, like three or less, this deformation process keeps the manifold nice—specifically projective and Kähler.
Emmy: Right. And for higher dimensions, it constructs families where everything outside the center is smooth and well-behaved under these constraints, but when you zoom in on the very middle of that family, you get a non-Kähler Moishezon manifold. That's a big structural distinction there > Page one <ref:2610.23457#pg1>
Ada: That’s what I found most interesting about the dimension dependence. It’s not just one result; it splits cleanly at three dimensions, and then the whole picture flips for four and higher.
Emmy: And that flip means that for dimension four or more, you can have these smooth limits where the underlying structure is completely different from what you started with because of how the algebraic geometry unfolds in those higher dimensions > Page two <ref:2610.23457#pg2>
Ada: So, if a mathematician is working on Kähler-Einstein metrics or negative scalar curvature, this tells them exactly when their intuition about the limit—whether it’s Kähler or not—is going to hold true. It sets a hard dimensional boundary for that question > Page one <ref:2610.23457#pg1>
Emmy: And what this means for the world is that we can't just assume Kähler persistence blindly in complex geometry; we have to check the ambient dimension and use these specific geometric tools to determine if it holds > Page two <ref:2610.23457#pg2>
Ada: Exactly. It gives us a clear dividing line. For dimension three, you have control over the canonical map, and it ends up being proper and birational onto its image > Page two <ref:2610.23457#pg2>
Emmy: Whereas for dimension four or higher, you see these counterexamples built using things like K3 surfaces or product families of curves where the central fiber is Moishezon but not Kähler > Page one <ref:2610.23457#pg1>
Ada: So we can now target our research more effectively. It shows us precisely what kind of geometric constraints are necessary to ensure a manifold remains Kähler under deformation > Page two <ref:2610.23457#pg2>
Emmy: The paper concludes by showing that for dimension three, the stable base locus has no divisorial component because the set B is empty, which means the canonical map is a proper birational morphism onto its image > Page two <ref:2610.23457#pg2>
Ada: It’s a neat structural result. And for higher dimensions, it confirms those constructions work to show that if you have enough dimension, you can always build a non-Kähler limit > Page one <ref:2610.23457#pg1>
The paper's improvements: Kolya: So, we've heard about the main results—the dimension three versus higher dimension split—and now we look at what they suggest to do next > Page two <ref:2610.23457#pg2>
Ada: The improvements they suggest are pretty focused on making the methods more robust. They're pushing for a more detailed analysis of the canonical model structure > Page three <ref:2610.23457#pg3>
Emmy: That makes sense. Specifically, they want to use an inverse Cartan matrix construction that’s invariant under Dynkin diagram monodromy to prove projectivity in dimension three > Page two <ref:2610.23457#pg2>
Kolya: The implication there is that if you can find a way to make that canonical map projective by showing every stable fixed divisor disappears, it solidifies the result for three dimensions > Page two <ref:2610.23457#pg2>
Ada: Right. And they also want more precision on the Betti numbers. They want to use intersection cohomology to calculate the second Betti number of W, relating it directly to the second Betti number of V minus some count of exceptional prime divisors > Page four <ref:2610.23457#pg4>
Emmy: That's about making sure we have a rigorous way to count those resolutions, giving us a precise measurement of how much structure is lost in the process > Page three <ref:2610.23457#pg3>
Kolya: And they’re also looking at what happens with the exceptional curves themselves. They want to prove that every curve contracted by that map is actually a smooth rational curve, and every prime divisor maps onto a curve > Page four <ref:2610.23457#pg4>
Ada: That's important because it defines exactly what kind of geometry is left on the central fiber when you contract those things down > Page two <ref:2610.23457#pg2>
Emmy: And they want to establish that every line bundle positive on all those contracted curves exists, constructing a specific line bundle L such that its intersection with every general contracted fiber component is positive > Page four <ref:2610.23457#pg4>
Kolya: That sounds like they're adding more machinery to the analysis of the singularities themselves, making the positivity results more concrete for real-world applications > Page four <ref:2610.23457#pg4>
Ada: So, in short, it’s about tightening up every detail of how that canonical map behaves on the central fiber by using finer cohomological tools and geometric constraints > Page three <ref:2610.23457#pg3>
Conclusion: Ada: So we're wrapping up this look at "Smooth Limits of Stable Manifolds." We saw how this paper splits its results based on whether you're in dimension three or higher > Page one <ref:2610.23457#pg1>
Emmy: Exactly. The big implication is that the persistence of a Kähler property under deformation depends entirely on the ambient dimension, which is a really sharp structural constraint > Page two <ref:2610.23457#pg2>
Kolya: From my side, I just want to emphasize that for dimension four and up, they provide concrete counterexamples using K3 surfaces or product families of curves > Page one <ref:2610.23457#pg1>
Ada: That’s right. And for dimension three, the paper gives us a very clear path where we can force the limiting object to be projective by controlling that canonical map > Page two <ref:2610.23457#pg2>
Emmy: It shows that in those lower dimensions, the algebraic geometry is more rigid and forces a nice outcome when you deform things > Page two <ref:2610.23457#pg2>
Kolya: I think for anyone working on stochastic processes or dynamical systems, this means we have a better handle on how geometric constraints translate into probabilistic bounds > Page one <ref:2610.23457#pg1>
Ada: It’s certainly useful. So, to wrap up, "Smooth Limits of Stable Manifolds" gives us a dimension-dependent answer to whether a negative curvature smooth limit must be Kähler > Page two <ref:2610.23457#pg2>
Emmy: And it sets the stage for understanding how algebraic geometry interacts with analytic deformations across different dimensions > Page two <ref:2610.23457#pg2>
Kolya: Next time, we’ll look at how those constraints play out in the context of some spectral inequalities for weighted p-Laplacians > Page five <ref:2610.23457#pg5>
math.AG
Submitted: 2026-10-16
Updated: 2026-11-11
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 91/100
The gist: Let X be a compact complex manifold of dimension n, and suppose that X is deformation equivalent to a stable manifold.
Key concepts
- Stable Manifold
- A manifold that can be deformed into a specific geometric object (a stable manifold) via a continuous family of complex structures. The paper studies what happens to the Kähler property as this deformation approaches the central fiber.
- Kähler-Einstein Manifold
- A special type of Kähler manifold where the Ricci curvature is proportional to the metric. The paper examines manifolds that are deformation equivalent to these, specifically focusing on those with nonpositive scalar curvature, which relates to Yau's conjecture.
- Moishezon Manifold
- A complex manifold that admits a surjective morphism onto a projective algebraic variety. The paper shows that in higher dimensions ($n imes 4$), the central fiber of the deformation family can be Moishezon but not Kähler, meaning it is not projective.
- Canonical Model
- A construction used to study the geometry of complex manifolds by mapping them to a simpler, projective space. The paper uses this model to prove that in dimension three, the canonical map is proper and birational, implying projectivity.
Terminology
Summary
Let X be a compact complex manifold of dimension n, and suppose that X is deformation equivalent to a stable manifold. p. 0 SMOOTH LIMITS OF STABLE MANIFOLDS
MU-LIN LI
Abstract. Let X be a compact complex manifold of dimension n, and suppose that X
is deformation equivalent to a stable manifold. Let X be a compact complex manifold of dimension n, and suppose that X
is deformation equivalent to a stable manifold. Then X is projective whenever n ≤ 3. Then X is projective whenever n ≤ 3. We
also construct, for every n ≥ 4, smooth families of compact complex manifolds over the
unit disc such that every fiber over the punctured disc is a stable manifold, while the
central fiber is a non-Kähler Moishezon manifold.Where the paper says it
Where the paper says it
Where the paper says it
The core problem addressed by this work concerns whether a compact complex manifold deformation equivalent to a Kähler–Einstein manifold of nonpositive scalar curvature must be Kähler, and the paper provides dimension-sharp answers to this question <ref:2610.23457#pg2> The principal positive result is that the answer is affirmative in complex dimension at most three and negative in every complex dimension at least four <ref:2610. p. 1 In this paper we give a complete, dimension-sharp answer to the negative Kähler–Einstein
smooth-limit case of Yau’s question: the answer is affirmative in complex dimension at most
three and negative in every complex dimension at least four. Our principal positive result is the following. For every integer n ≥ 4, there is a proper holomorphic submersion π : M →
∆ with connected compact fibers of complex dimension n such that
(1) KMt
is ample for every t ̸= 0;
(2) M0 is Moishezon and non-Kähler. Together, these two theorems settle the negative-curvature smooth-limit problem in every
dimension. Together, these two theorems settle the negative-curvature smooth-limit problem in every
dimension. The counterexamples give a negative answer to Yau’s question in dimensions at
least four even when every punctured fiber is assumed to be canonically polarized. Together, these two theorems settle the negative-curvature smooth-limit problem in every
dimension. The counterexamples give a negative answer to Yau’s question in dimensions at
least four even when every punctured fiber is assumed to be canonically polarized.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Dimension-dependent results for smooth limits
The paper establishes dimension-dependent results concerning the persistence of the Kähler property under deformation, specifically addressing the negative Kähler–Einstein smooth-limit case of Yau’s question <ref:2610.23457#pg2> Theorem 1. p. 0 A negative Kähler–Einstein metric can be rescaled so that Ric(ω) = −ω. This identity
makes the canonical bundle positive, hence ample by the Kodaira embedding theorem [25,
Section 1. Then X is projective whenever n ≤ 3. We
also construct, for every n ≥ 4, smooth families of compact complex manifolds over the
unit disc such that every fiber over the punctured disc is a stable manifold, while the
central fiber is a non-Kähler Moishezon manifold. A negative Kähler–Einstein metric can be rescaled so that Ric(ω) = −ω. This identity
makes the canonical bundle positive, hence ample by the Kodaira embedding theorem [25,
Section 1. For n ≥ 4 we have the following theorem. For every integer n ≥ 4, there is a proper holomorphic submersion π : M →
∆ with connected compact fibers of complex dimension n such that
(1) KMt
is ample for every t ̸= 0;
(2) M0 is Moishezon and non-Kähler.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
The proof strategy involves canonical models and cohomology
The proof of Theorem 1.2 relies on Riemann–Roch, Kodaira vanishing on the punctured fibers, and semicontinuity to show KX0 is big, implying X0 is Moishezon <ref:2610. p. 1 We briefly explain the proof of Theorem 1. Riemann–Roch, Kodaira vanishing on
the punctured fibers, and semicontinuity first show that KX0
is big, so X0 is Moishezon. 1(2),(3)] then give a flat projective relative canonical model Y → ∆. The key
step is a comparison of second cohomology with intersection cohomology. The essential case is dimension three. More precisely, we prove that the canonical
morphism of the central fiber is crepant, that its fibers have dimension at most one, and
that it contracts every exceptional prime divisor onto a curve; see Theorem 8. The projectivity criterion [44, Corollary 3. 4]
then makes the canonical morphism projective, and hence makes X0 projective. Products with curves
of genus at least two give the examples in all higher dimensions. Sections 3–7 establish the canonical-model and cohomological arguments in arbitrary
dimension. For Theorem 1. 3, we start with a pencil of smooth quartic K3 surfaces whose central
member contains exactly one line and whose nearby members contain none.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Cohomological arguments for arbitrary dimensions
Sections 3–7 establish the canonical-model and cohomological arguments in arbitrary dimension <ref:2610.23457#pg2> Proposition 3. p. 2 The branch
divisor avoids the contracted plane, whose inverse image consists of two disjoint planes. The cyclic-cover Lefschetz theorem [24, Proposition 1. Let π : X → ∆ be a proper holomorphic submersion with connected
compact fibers of dimension n ≥ 1 such that KXt
is ample for t = 0 ̸ . Then KX0
is big and
X0 is Moishezon. 2(i)] and finite generation of canonical rings [4,
Corollary 1. 1(2),(3)] then give a flat projective relative canonical model Y → ∆. A line bundle L on a
compact complex n-fold is big if its Iitaka dimension is n, equivalently if
lim sup
m→∞
h
0
(M, mL)
mn
> 0. A compact complex manifold is Moishezon if and only if it carries a big line bundle [34,
Lemma 2. A line bundle L on a
compact complex n-fold is big if its Iitaka dimension is n, equivalently if
lim sup
m→∞
h
0
(M, mL)
mn
> 0. A compact complex manifold is Moishezon if and only if it carries a big line bundle [34,
Lemma 2. Let π : X → ∆ be a proper holomorphic submersion with connected compact
fibers of dimension n ≥ 1 such that KXt
is ample for t = 0 ̸ . There is a positive integer
m∗ such that, for every t = 0 ̸ , the line bundle m∗KXt
is very ample and its section ring is
generated in degree one. The branch
divisor avoids the contracted plane, whose inverse image consists of two disjoint planes. The cyclic-cover Lefschetz theorem [24, Proposition 1. After shrinking ∆, there is a sufficiently divisible integer m ≥ 2 such
that the graded O∆-algebra
R(m) =
M
k≥0
π∗ω
⊗km
X/∆
is generated in degree one and
(4. 1) g : Y = Proj∆ R(m) −→ ∆
is flat and projective.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Analysis of the central fiber properties
The analysis of the central fiber X0 involves several key results regarding its geometry and cohomology <ref:2610.23457#pg2> Lemma 4.3 shows that the total space Y is normal and of pure dimension n + 1, with b2(X0) = b2(Yt) for t ≠ 0, and the limiting mixed Hodge structure H2lim(Yt) is pure of weight two <ref:2610. p. 6 We never replace ICY by QY [n + 1] on the singular
total space. For a polarized variation of Hodge structure of weight two, the weight filtration on its
limiting mixed Hodge structure is the monodromy weight filtration centered at two. In dimension three, the remaining stable base locus has dimension at most one. A
dimension argument on the normalized relative graph of the canonical map eliminates it,
yielding a crepant canonical morphism. A line bundle L on a
compact complex n-fold is big if its Iitaka dimension is n, equivalently if
lim sup
m→∞
h
0
(M, mL)
mn
> 0. A compact complex manifold is Moishezon if and only if it carries a big line bundle [34,
Lemma 2. A Cartier divisor determines a line bundle, and a Q-Cartier
divisor is a rational divisor admitting a positive Cartier multiple. If C ⊂ M is a reduced
irreducible compact curve with normalization ν : Ce → C, define
(2.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
The final conclusion for dimension three
In dimension three, the proof shows that the stable base locus has no divisorial component, which completes the proof <ref:2610.23457#pg2> Theorem 7. p. 0 A negative Kähler–Einstein metric can be rescaled so that Ric(ω) = −ω. This identity
makes the canonical bundle positive, hence ample by the Kodaira embedding theorem [25,
Section 1. 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610. Comparing their
ADE root lattices, including the Weyl-group ambiguity of simultaneous resolutions, forces
every stable fixed divisor to disappear. In dimension three, the remaining stable base locus has dimension at most one.Where the paper says it
Where the paper says it
Where the paper says it
Counterexamples in higher dimensions
For dimension at least four, counterexamples are constructed using a pencil of smooth quartic K3 surfaces S0 ⊂ P3 containing exactly one line <ref:2610.23457#pg2> The resulting family Yt = S[2]t carries the holomorphic symplectic form associated with the K3 surface St <ref:2610.23457#pg2> The assertion that X0 is not Kähler is proven by showing that [C∨+] + [C−] = 0 in H2(X0, R), and neither summand is zero <ref:2610.23457#pg2>
The final result for dimension four
For n > 4, the counterexamples are formed by taking T to be a product of n − 4 smooth projective curves of genus at least two, and forming the product family X × T → ∆ <ref:2610.23457#pg2> The central fiber is Moishezon because X0 is bimeromorphic to a projective manifold and T is projective <ref:2610. p. 0 Then X is projective whenever n ≤ 3. We
also construct, for every n ≥ 4, smooth families of compact complex manifolds over the
unit disc such that every fiber over the punctured disc is a stable manifold, while the
central fiber is a non-Kähler Moishezon manifold. Then X is projective whenever n ≤ 3. We
also construct, for every n ≥ 4, smooth families of compact complex manifolds over the
unit disc such that every fiber over the punctured disc is a stable manifold, while the
central fiber is a non-Kähler Moishezon manifold. Then X is projective whenever n ≤ 3. We
also construct, for every n ≥ 4, smooth families of compact complex manifolds over the
unit disc such that every fiber over the punctured disc is a stable manifold, while the
central fiber is a non-Kähler Moishezon manifold. Hironaka’s example [17] shows that a smooth limit of
projective manifolds can be non-Kähler, so the Einstein condition imposes a substantive
additional constraint. For a Kähler form ω on X′ we adopt the convention
Ric(ω) = −
√
−1 ∂∂¯ log det(gjk¯), [Ric(ω)] = 2πc1(T X′
). Hironaka’s example [17] shows that a smooth limit of
projective manifolds can be non-Kähler, so the Einstein condition imposes a substantive
additional constraint. For a Kähler form ω on X′ we adopt the convention
Ric(ω) = −
√
−1 ∂∂¯ log det(gjk¯), [Ric(ω)] = 2πc1(T X′
). Hironaka’s example [17] shows that a smooth limit of
projective manifolds can be non-Kähler, so the Einstein condition imposes a substantive
additional constraint. For a Kähler form ω on X′ we adopt the convention
Ric(ω) = −
√
−1 ∂∂¯ log det(gjk¯), [Ric(ω)] = 2πc1(T X′
).Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
The final conclusion for dimension n > 3
The paper concludes that the canonical morphism f: X0 −→ Y0 = Proj R(X0, KX0) is crepant, all its fibers have dimension at most one, and every exceptional prime divisor maps onto a curve <ref:2610.23457#pg2> Every irreducible contracted curve is a smooth rational curve and deforms in X0 to a general transverse Du Val exceptional curve <ref:2610. p. 2 SMOOTH LIMITS OF STABLE MANIFOLDS 3
the exceptional curves of a general transverse Du Val surface resolution. An inverse-Cartanmatrix construction, invariant under Dynkin-diagram monodromy, produces a line bundle
of positive degree on every contracted curve.Where the paper says it
The final conclusion for dimension n = 3
In dimension three, the complete base-point-free system mKX0 now gives the canonical morphism f, with OX0(mKX0) ≃ f∗OY0(mKY0) <ref:2610.23457#pg2> Equivalently, (8. p. 3 Positivity, stable base loci, and Moishezon manifolds. A line bundle L on a
compact complex n-fold is big if its Iitaka dimension is n, equivalently if
lim sup
m→∞
h
0
(M, mL)
mn
> 0. 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610. We always use
the former convention. Thus canonical singularities mean a(E; Y ) ≥ 0 for all exceptional
divisorial valuations, and a valuation is called crepant when its discrepancy is zero. Let Y be a normal variety with KY
Q-Cartier, and let q : W → Y be a resolution. With compatible canonical divisors, write
(2. For t = 0 ̸ , ampleness of KXt
identifies Xt with the Proj of its complete canonical ring. The relative evaluation map defined by Em is a morphism over ∆∗ and restricts on each fiber
to the corresponding pluricanonical embedding. For t = 0 ̸ , ampleness of KXt
identifies Xt with the Proj of its complete canonical ring. The relative evaluation map defined by Em is a morphism over ∆∗ and restricts on each fiber
to the corresponding pluricanonical embedding.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
For n > 4, the counterexamples are formed by taking T to be a product of n − 4 smooth projective curves of genus at least two, and forming the product family X × T → ∆ <ref:2610.23457#pg2> The central fiber is Moishezon because X0 is bimeromorphic to a projective manifold and T is projective <ref:2610.23457#pg2> If X0 × T were Kähler, restricting a Kähler form to X0 × u would make X0 Kähler, a contradiction <ref:2610.23457#pg2>
References
[1] T. Aubin, Équations du type Monge–Ampère sur les variétés kählériennes compactes, C. p. 0 This identity
makes the canonical bundle positive, hence ample by the Kodaira embedding theorem [25,
Section 1. Conversely, the existence theorems of Aubin [2] and Yau [45, Section 6,
Theorem 5] produce such a metric whenever the canonical bundle is ample. 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
Where the paper says it
[2] T. Aubin, Équations du type Monge–Ampère sur les variétés kählériennes compactes, Bull. p. 0 This identity
makes the canonical bundle positive, hence ample by the Kodaira embedding theorem [25,
Section 1. Conversely, the existence theorems of Aubin [2] and Yau [45, Section 6,
Theorem 5] produce such a metric whenever the canonical bundle is ample. 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
Where the paper says it
[3] A. Beauville, Variétés kählériennes dont la première classe de Chern est nulle, J. p. 3 We assert no such nonnegativity when C is contained in Supp D. On a smooth compact n-fold, top intersections are defined by
L1 · · ·Ln =
Z
M
c1(L1) ∧ · · · ∧ c1(Ln),
using any closed representatives of the Chern classes. 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
Where the paper says it
[4] C. Birkar, P. Cascini, C. D. Hacon and J McKernan, Existence of minimal models for varieties of log general type, J. p. 7 A smooth projective
modification has the same canonical ring as M: pluricanonical forms descend across the
modification, because they agree on the isomorphism locus and extend across a subset of
codimension at least two in the smooth target. By finite generation of the canonical ring
for smooth projective varieties of general type [4, Corollary 1. 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
Where the paper says it
[5] L Bonavero, Sur des variétés de Moishezon dont le groupe de Picard est de rang un, Bull. Soc. Math. France 124 (1996), no 3, 503–521 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[6] G E Bredon, Topology and Geometry, Graduate Texts in Mathematics 139, Springer-Verlag, New York, 1993 <ref:2610.23457#pg2>
[7] N Buchdahl, On compact Kähler surfaces, Ann. Inst. Fourier 49 (1999), no. 1, 287–302 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[8] M A A de Cataldo and L Migliorini, The Hodge theory of algebraic maps, Ann. Sci. Ècole Norm. Sup.(4) 38 (2005), no 5, 693–750 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[9] M A A de Cataldo and L Migliorini, The decomposition theorem, perverse sheaves and the topology of algebraic maps, Bull. Amer. Math. Soc.(N.S.) 46 (2009), 535–633 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[10] R Dervan, On K-stability of finite covers, Bull. Lond. Math. Soc. 48 (2016), no 4, 717–728 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[11] G Ellingsrud, L Göttsche, and M Lehn, On the cobordism class of the Hilbert scheme of a surface, J. Algebraic Geom. p. 5 Intersection complexes and the decomposition theorem. Let V be an irreducible
complex algebraic variety of dimension d, and let j : U ,→ V be a dense smooth open subset. 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
Where the paper says it
[12] J Fogarty, Algebraic families on an algebraic surface, Amer. J. Math. 90 (1968), no 2, 511–521 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[13] A Fujiki and S Nakano, Supplement to “On the inverse of monoidal transformation”, Publ. Res. Inst. Math. Sci. 7 (1971/72), no 3, 637–644 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[14] W Fulton, Intersection Theory, second ed., Ergebnisse der Mathematik und ihrer Grenzgebiete (3), vol. 2, Springer-Verlag, Berlin, 1998 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[15] M Goresky and R MacPherson, Intersection homology II, Invent. Math. 72 (1983), 77–129 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[16] H Grauert, Ein Theorem der analytischen Garbentheorie und die Modulräume komplexer Strukturen, Publ. Math. Inst. Hautes Études Sci. 5 (1960), 5–64; Berichtigung, 16 (1963), 35–36 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[17] H Hironaka, An example of a non-Kählerian complex-analytic deformation of Kählerian complex structures, Ann. of Math.(2) 75 (1962), no 1, 190–208 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[18] H Hironaka, Stratification and flatness, in Real and Complex Singularities (Oslo, 1976), P Holm (ed.), Sijthoff and Noordhoff, Alphen aan den Rijn, 1977 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[19] D Huybrechts, Birational symplectic manifolds and their deformations, J. Differential Geom. 45 (1997), no 3, 488–513 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[20] Y Kawamata, A generalization of Kodaira–Ramanujam’s vanishing theorem, Math. Ann. 261 (1982), no 1, 43–46 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[21] M Kerr, R Laza and M Saito, Deformation of rational singularities and Hodge structure, Algebr. Geom. 9 (2022), no 4, 476–501 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[23] A Lamari, Courants kählériens et surfaces compactes, Ann. Inst. Fourier (Grenoble) 49 (1999), no 1, 263–285 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[24] A Lanteri and D C Struppa, Topological properties of cyclic coverings branched along an ample divisor, Canad. J. Math. 41 (1989), no 3, 462–479 <ref:2610. p. 0 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
[25] R Lazarsfeld, Positivity in Algebraic Geometry I: Classical Setting: Line Bundles and Linear Series, Ergebnisse der Mathematik und ihrer Grenzgebiete (3), vol 48, Springer-Verlag, Berlin, 2004 <ref:2610.23457#pg2>
[26] M.-L. Li, W Liu, The limits of Kähler manifolds under holomorphic deformations, arXiv:2406. p. 0 1 then reduces to the
following problem. Keywords: Kähler–Einstein metrics; smooth limits; Moishezon manifolds; canonical models; intersection
cohomology; rational curves; elementary transformations. 2020 Mathematics Subject Classification: Primary 32G05; Secondary 32J18, 14E30, 14F43, 53C55. 1
arXiv:2610.Where the paper says it
Where the paper says it
Conclusion for dimension three
The complete base-point-free system mKX0 now gives the canonical morphism f, with OX0(mKX0) ≃ f∗OY0(mKY0) <ref:2610.23457#pg2> The set B is empty, in particular, the canonical map is a proper birational morphism f: X0 −→ Y0 = Proj R(X0, KX0), KX0 = f∗KY0 <ref:2610.23457#pg2>
Conclusion for dimension at least four
For n ≥ 4, there is a proper holomorphic submersion π: M → ∆ with connected compact fibers of complex dimension n such that (1) KMt is ample for every t ≠ 0; (2) M0 is Moishezon and non-Kähler <ref:2610.23457#pg2> Together, these two theorems settle the negative-curvature smooth-limit problem in every dimension <ref:2610. p. 1 For every integer n ≥ 4, there is a proper holomorphic submersion π : M →
∆ with connected compact fibers of complex dimension n such that
(1) KMt
is ample for every t ̸= 0;
(2) M0 is Moishezon and non-Kähler. Together, these two theorems settle the negative-curvature smooth-limit problem in every
dimension.Where the paper says it
Conclusion for dimension n > 3
For n > 4, let T be a product of n − 4 smooth projective curves of genus at least two, and form the product family X × T → ∆ <ref:2610.23457#pg2> It is a proper holomorphic submersion with connected compact fibers of dimension n such that for t ≠ 0, the canonical bundle of Xt × T is ample <ref:2610. p. 8 Let π : X → ∆ be a proper holomorphic submersion with connected compact
fibers of dimension n ≥ 1 such that KXt
is ample for t = 0 ̸ . There is a positive integer
m∗ such that, for every t = 0 ̸ , the line bundle m∗KXt
is very ample and its section ring is
generated in degree one.Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Improved handling of dimension dependence in Kähler-Einstein limits: The system can now distinguish between complex dimensions at most three and at least four when determining whether a smooth limit manifold X0 is Kähler, based on Theorem 1.2 and Theorem 1.3, which provide a dimension-sharp answer to Yau's question.
-
Enhanced construction of non-Kähler Moishezon manifolds: The system can construct smooth families of compact complex manifolds over the unit disc where every fiber over the punctured disc is a stable manifold but the central fiber is a non-Kähler Moishezon manifold, as stated in Theorem 1.3 for n >= 4.
-
Rigorous analysis of canonical model structure: The system can perform an inverse-Cartanmatrix construction invariant under Dynkin-diagram monodromy to prove that the canonical morphism is projective in dimension three, by showing
every stable fixed divisor to disappear.
-
Detailed computation of Betti numbers via intersection cohomology: The system can calculate the second Betti number b2(W) using Lemma 5.1 and Lemma 5.2, which relates it to Ib2(V) plus the number of exceptional prime divisors, providing a precise count for resolutions of normal projective manifolds.
-
Identification of stable base locus components: The system can prove that the stable base locus has no divisorial component in dimension three by showing that
the set B is empty
in Proposition 8.3, concluding thatthe canonical map is a proper birational morphism f: X0 -→ Y0.
-
Analysis of exceptional fiber structures: The system can identify every irreducible curve contracted by f as a smooth rational curve and prove that
every exceptional prime divisor of f maps onto a curve
(Proposition 8.4), establishing the structure of the map on the central fiber. -
Determination of positivity for singular loci: The system can establish that
every line bundle positive on all contracted curves
exists by constructing a specific line bundle L = OX0 (-NG) such that "L · Cj = N > 0" for every general contracted fiber component (Proposition 8.8). -
Verification of non-Kähler property: The system can prove that the central fiber X0 is not Kähler by showing that "if ω were a Kähler form on X0, its integrals over the two smooth complex curves would both be positive, whereas (9.19) would imply 0 = Z C v + Z C - ω > 0."
Abstract
Let X be a compact complex manifold of dimension n, and suppose that X is deformation equivalent to a stable manifold. Then X is projective whenever n <=3. We also construct, for every n >=4, smooth families of compact complex manifolds over the unit disc such that every fiber over the punctured disc is a stable manifold, while the central fiber is a non-Kahler Moishezon manifold.
Sources
- The limits of Kahler manifolds under holomorphic deformations
- Vanishing Cycles and Thom's $a_f$ Condition
- Moishezon Spaces and Projectivity Criteria
Related papers
- K-unstable Toric Varieties and Secondary Polytopes
- The commuting scheme of g 2 and a degree formula for commuting schemes
- Stacky geometry and logarithmic topology of transversely affine foliations
- Monodromy and vanishing cycles for complete intersection curves
- Characteristic cycles and Gevrey irregularity in hypergeometric families
- Tautological classes on KSBA spaces