On singular fibers of parabolic fibrations
Yiming Zhu · 2025
Read On singular fibers of parabolic fibrations by Yiming Zhu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-09962.txt.
Introduction
Yiming Zhu studies special fibers in a type of map between geometric spaces, in work from 2025. The map is a parabolic fibration, a structured way one space can sit over another. Zhu works in the case where two linked divisor terms vanish in a numeric sense. The abstract states this aim alone, so the note gives a short account of the setup and the class of fibers described.
Catalog record
| CREATOR | Yiming Zhu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-09962
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On singular fibers of parabolic fibrations Yiming Zhu Abstract . We describe the singular fibers of a parabolic fibration f : X → Y f:X\to Y whose moduli divisor M Y M_{Y} is numerically trivial and whose discriminant divisor B Y B_{Y} is zero. 2020 Mathematics Subject Classification 14E30 1. Introduction Let f:S\to C be an elliptic fibration over a curve. By successively contracting (-1) -curves on fibers, we get a birational morphism \pi:S\to S^{\prime} to a smooth surface S^{\prime} , so that the resulting elliptic fibration f^{\prime}:S^{\prime}\to C has no (-1) -curves on its fibers. Kodaira’s canonical bundle formula of f^{\prime} writes: \displaystyle K_{S^{\prime}}=f^{\prime*}(K_{C}+L+\sum_{P}(1-{1}/{m_{P}})P), where the sum runs over all points P such that the fiber f^{\prime*}P is a multiple fiber of multiplicity m_{P} , and L is a divisor associated to the line bundle f^{\prime}_{*}O_{S^{\prime}}(K_{S^{\prime}/C}) . Assume that \deg L=0 and \sum_{P}(1-{1}/{m_{P}})P=0 . Then f^{\prime} is locally trivial by [ 2 , Chapter 3, Theorem 18.2] . In particular, every fiber f^{\prime*}P is reduced and irreducible. Now, consider the original fibration f:S\to C . Write K_{S}=\pi^{*}K_{S^{\prime}}+B . Then B is an effective divisor that does not dominate C . Let f^{*}P=\sum_{i}a_{i}Q_{i} be a fiber of f , where the Q_{i} are prime divisors, and let f^{*}P\setminus B:=\sum_{i:Q_{i}\not\subset B}a_{i}Q_{i} . Then f^{*}P\setminus B is reduced and irreducible. In this note, we consider the higher-dimensional case. Following Ueno, we call a fibration parabolic if its general fiber has Kodaira dimension zero. For parabolic fibrations, we have the following canonical bundle formula of Fujino and Mori. Theorem 1.1 ( [ 5 ] ) . Let f:X\to Y be a parabolic fibration between smooth projective varieties and let F be its general fiber. Let b=\min\{k\in\mathbb{Z}_{>0}\mid\dim H^{0}(kK_{F})>0\} . Then there are \mathbb{Q} -divisors B_{Y} (the discriminant divisor), M_{Y} (the moduli divisor), and B_{X} such that \displaystyle bK_{X}=f^{*}b(K_{Y}+M_{Y}+B_{Y})+B_{X}, with the following properties: (1) If we write B_{X}=B_{X}^{+}-B_{X}^{-} as a difference of two effective \mathbb{Q} -divisors without common components, then f_{*}O_{X}(\lfloor mB_{X}^{+}\rfloor)=O_{Y} for all positive integers m , and \operatorname{codim}f(B_{X}^{-})\geq 2 . (2) B_{Y}=\sum_{P}(1-t_{P})P , where t_{P}=\operatorname{lct}(X,-\frac{1}{b}B_{X},f^{*}P) , the log canonical threshold of f^{*}P with respect to the pair (X,-\frac{1}{b}B_{X}) near the generic point of P . The following is the first main result of this note. Theorem 1.2 . Let f:X\to Y be a parabolic fibration between smooth projective varieties. Assume that M_{Y} is numerically trivial and B_{Y}=0 . If X has a good minimal model X^{\prime} over Y , and f^{\prime}:X^{\prime}\to Y is the induced fibration with general fiber F^{\prime} , then there is a finite étale cover Y^{\prime}\to Y such that X^{\prime}\times_{Y}Y^{\prime} is isomorphic to Y^{\prime}\times F^{\prime} over Y^{\prime} . Without the good minimal model assumption, we prove the following. Theorem 1.3 . Let f:X\to Y be a parabolic fibration between smooth projective varieties. Assume that M_{Y} is numerically trivial and B_{Y}=0 . Then (1) B_{X} is effective, (2) any prime divisor Q on X with \operatorname{codim}f(Q)\geq 2 is a component of B_{X} , and (3) for any prime divisor P on Y , the divisor f^{*}P\setminus B_{X} is reduced and irreducible. The proof uses the following theorem, which is a consequence of the Clemens–Schmid sequence. Theorem 1.4 . Let f:X\to C be a semistable fibration from a smooth projective variety X to a smooth projective curve C . Let \sum_{i}X_{i} be a singular fiber and let X_{t_{0}}=f^{-1}(t_{0}) be a general fiber. If \deg f_{*}O_{X}(K_{X/C})=0 , then \sum_{i}\dim H^{0}(K_{X_{i}})=\dim H^{0}(K_{X_{t_{0}}}) . Parabolic fibrations with M_{Y} numerically trivial and B_{Y}=0 are closely related to Ueno’s conjecture. Conjecture 1.5 (Ueno’s conjecture K) . Let X be a smooth projective variety of Kodaira dimension zero. Let f:X\to Y be the Albanese morphism and let F be a general fiber of f . Then (1) f is a fibration, (2) \kappa(F)=0 , and (3) there is a finite étale cover Y^{\prime}\to Y such that X\times_{Y}Y^{\prime} is birational to Y^{\prime}\times F over Y^{\prime} . Part (1) was proved by Kawamata [ 10 , Theorem 1] . Part (2) follows as a corollary of Cao-Păun’s solution to the Iitaka conjecture in the case where the base is an abelian variety [ 4 , Theorem 1.1] . Let f be the parabolic fibration appearing in Conjecture 1.5 , and let M_{Y} and B_{Y} denote the moduli divisor and the discriminant divisor of f , respectively. By [ 9 , Theorem 5.2] , we have that M_{Y} is numerically trivial and B_{Y}=0 (cf. [ 17 , Lemma 2.1] ). Consequently, Theorems 1.2 and 1.3 apply to this fibration. Notation and conventions We work over the field \mathbb{C} of complex numbers. A surjective projective morphism between normal varieties with connected fibers is called a fibration. A fibration is called parabolic if its general fiber has Kodaira dimension zero. Let f:X\to Y be a fibration between normal varieties. Let D be an effective \mathbb{Q} -divisor. Then we can write D=D^{h}+D^{v} , where D^{v} does not dominate Y , and every component of D^{h} dominates Y . We call D a vertical divisor (resp. horizontal divisor) if D=D^{v} (resp. D=D^{h} ). If D is vertical, it is called exceptional over Y if \operatorname{codim}f(D)\geq 2 , and very exceptional over Y if for any prime divisor P\subset Y , there exists a prime divisor Q\subset X such that Q dominates P but Q\not\subset D . 2. Proof of Theorem 1.2 Lemma 2.1 ( [ 7 , Lemma 2.15] ) . Let f:X\to Y be a fibration between normal varieties. Assume that Y is smooth. Let D be a vertical effective \mathbb{Q} -divisor. If (f_{*}O_{X}(\lfloor mD\rfloor))^{\wedge}=O_{Y} (where \wedge denotes the reflexive hull) for all sufficiently divisible positive integers m , then D is very exceptional over Y . Proof. If D is not very exceptional over Y , then there is a prime divisor P\subset Y , an effective divisor E that is exceptional over Y , and a positive integer N , such that f^{*}P\leq N(E+D) . We may assume that N(E+D) is a \mathbb{Z} -divisor. Then O_{Y}(mP)\subset f_{*}O_{X}(mN(E+D))\subset(f_{*}O_{X}(mND))^{\wedge}=O_{Y} for all sufficiently divisible positive integers m , which is a contradiction. ∎ Lemma 2.2 . Let f:X\to Y be a fibration between normal varieties with a general fiber F . Assume that Y is smooth. Let D_{1} be an effective horizontal \mathbb{Z} -divisor on X such that \dim H^{0}(D_{1}|_{F})=1 , let D_{2} be an effective very exceptional \mathbb{Q} -divisor on X , and let L be a \mathbb{Q} -divisor on Y . Then f_{*}O_{X}(\lfloor f^{*}L+D_{1}+D_{2}\rfloor)=O_{Y}(\lfloor L\rfloor) . Proof. Let U\subset Y be an affine open subset, and let \phi\in H^{0}(f^{-1}U,f^{*}L+D_{1}+D_{2})\subset\mathbb{C}(X) be a section. Then \displaystyle\operatorname{div}_{X}(\phi)+f^{*}L+D_{1}+D_{2}|_{f^{-1}U}\geq 0. By restricting to F , we have \displaystyle\operatorname{div}_{X}(\phi)+D_{1}|_{F}\geq 0. Since \dim H^{0}(D_{1}|_{F})=1 , \phi|_{F} is a constant, which means that \phi=f^{*}\psi for some \psi\in\mathbb{C}(Y) . Then by \displaystyle f^{*}\operatorname{div}_{Y}(\psi)+f^{*}L+D_{1}+D_{2}|_{f^{-1}U}\geq 0, we have \displaystyle\operatorname{div}_{Y}(\psi)+L|_{U}\geq 0, so \psi\in H^{0}(U,\lfloor L\rfloor) . Hence f_{*}O_{X}(\lfloor f^{*}L+D_{1}+D_{2}\rfloor)\subset O_{Y}(\lfloor L\rfloor) . On the other hand, we have O_{Y}(\lfloor L\rfloor)=f_{*}O_{X}(f^{*}\lfloor L\rfloor)\subset f_{*}O_{X}(\lfloor f^{*}L\rfloor)\subset f_{*}O_{X}(\lfloor f^{*}L+D_{1}+D_{2}\rfloor). Hence the Lemma follows. ∎ Lemma 2.3 . Let f:X\to Y be a parabolic fibration between smooth projective varieties. If M_{Y}\equiv 0 and B_{Y}=0 , then (1) bM_{Y} is a \mathbb{Z} -divisor, (2) O_{Y}(mbM_{Y})=f_{*}O_{X}(mbK_{X/Y}) for all m\in\mathbb{Z}_{>0} , and (3) B_{X} is an effective \mathbb{Z} -divisor. Proof. Consider the non-zero