Canonical bundle formula and a conjecture on certain algebraic fiber spaces by Schnell

Hyunsuk Kim · 2024

Read Canonical bundle formula and a conjecture on certain algebraic fiber spaces by Schnell by Hyunsuk Kim (2024) 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-2412-19769.txt.

Introduction

Hyunsuk Kim tests what the canonical bundle formula says about a conjecture by Schnell. The conjecture links two big claims about algebraic fiber spaces. One claim says a certain bundle never vanishes; the other bounds forms on the total space. Kim works under two separate extra assumptions to get partway there.

One route weakens the demand on the base by adding a divisor on the branch locus. The other assumes less than the usual good minimal model for a general fiber. Under the second route, Schnell's claim holds when the fiber's class is a rigid current. The note sharpens the map between the two conjectures.

Catalog record

CREATORHyunsuk Kim
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE35 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-19769

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Canonical bundle formula and a conjecture on certain algebraic fiber spaces by Schnell Thanks: The author was partially supported by NSF grant DMS-2301463 and the Simons Collaboration grant Moduli of Varieties Hyunsuk Kim Abstract . We observe what the canonical bundle formula gives towards a conjecture of Schnell on algebraic fiber spaces, a question concerning the equivalence between the non-vanishing conjecture and the Campana–Peternell conjecture. As a result, we give a partial result on Schnell’s conjecture under two independent assumptions. One weakens Schnell’s assumption of the pseudo-effectivity of the canonical bundle of the base by adding some effective divisor supported on the ramification locus. The other is analogous to results on algebraic fiber spaces where the existence of good minimal models of a general fiber is assumed, but we use a priori a weaker assumption. More precisely, we prove Schnell’s conjecture when the canonical class of the general fiber is represented by a rigid current . 2020 Mathematics Subject Classification 14D06, 14E30, 14J15, 32J25 1. Introduction In [ 20 ] , Schnell attempts to prove an equivalence between two conjectures, namely the non-vanishing conjecture and the Campana–Peternell conjecture (on Kodaira dimension, see [ 5 ] *page 43). Both conjectures are special cases of the abundance conjecture. The non-vanishing conjecture predicts that if X is a smooth projective variety whose canonical bundle K_{X} is pseudo-effective, then a positive multiple of K_{X} is effective. The Campana–Peternell conjecture says the following. Conjecture 1.1 (Campana–Peternell) . Let X be a smooth projective variety and let D be an effective divisor. Suppose that m_{0}K_{X}-D is pseudo-effective for some m_{0}>0 . Then \kappa(X)\geq\kappa(X,D) . In particular, if we put D=0 in the conjecture above, we get the non-vanishing conjecture. In this sense, the Campana–Peternell conjecture seems to be a priori stronger than non-vanishing. In [ 20 ] , it is shown that assuming non-vanishing, the Campana–Peternell conjecture reduces to proving the following problem on algebraic fiber spaces. Conjecture 1.2 . Let f\colon X\to Y be an algebraic fiber space between smooth projective varieties with \kappa(F)=0 , where F is the general fiber. If there is an ample divisor H on Y such that m_{0}K_{X}-f^{*}H is pseudo-effective, then \kappa(X)=\dim Y . This formulation is slightly different than [ 20 ] *Conjecture 10.1. However, one can reduce to the case above by following the lines of [ 20 ] *§9. We prefer this formulation so that the canonical bundle formula naturally comes into the picture. Schnell proves his conjecture when K_{Y} is pseudo-effective, using the result of Păun–Takayama [ 18 ] on semi-positive metrics on relative pluri-adjoint line bundles. The aim of this article is to observe what canonical bundle formula gives toward Conjecture 1.2 . As a result, we are able to weaken Schnell’s assumption on the pseudo-effectivity of K_{Y} , and also observe that rigid currents in the sense of [ 15 ] naturally plays a role in this conjecture. 1.1. Canonical bundle formula We introduce the canonical bundle formula, following [ 11 , 8 , 13 ] , in the set-up of Conjecture 1.2 . We also refer to [ 12 ] for analytic aspects of the canonical bundle formula using fiberwise integrals, which is related to the content appearing in the later part of this paper. First, following [ 20 ] *§5, we note that this conjecture is insensitive upon taking birational models of X and Y . By taking suitable resolutions, we may and will assume that we have an SNC divisor Q=\sum_{l\in L}Q_{l} on Y such that f is smooth over Y_{0}=Y\setminus Q , and f^{-1}(Q)_{\mathrm{red}}=\sum_{j\in J}P_{j} is an SNC divisor on X . We can write K_{X} in the following way: (1) K_{X}\sim_{\mathbf{Q}}f^{*}(K_{Y}+B_{Y}+M_{Y})+\Delta, so that (1) f_{*}\mathcal{O}_{X}(\left\lfloor i\Delta^{+}\right\rfloor)=\mathcal{O}_{Y} for all i>0 , (2) \codim_{Y}f(\supp\Delta^{-})\geq 2 , and (3) H^{0}(X,mK_{X})=H^{0}(Y,m(K_{Y}+B_{Y}+M_{Y})) for sufficiently divisible m (see [ 8 ] *Theorem 4.5). Here, \Delta^{+} and \Delta^{-} are the effective divisors sharing no irreducible components and satisfying \Delta=\Delta^{+}-\Delta^{-} . Roughly speaking, the divisor B_{Y} measures the singularity of the fibers of f , and M_{Y} measures the Hodge-theoretic variation of the fibers. In particular, B_{Y} is always effective and is supported in the discriminant locus of f , and M_{Y} is nef. For a precise description of the divisors B_{Y} and \Delta , we refer to § 2.2 . We return to the setup of Conjecture 1.2 . Suppose we have some m_{0}>0 such that m_{0}K_{X}-f^{*}H is pseudo-effective. Hence K_{X}-\epsilon f^{*}H is pseudo-effective for \epsilon>0 small enough. By taking birational models of both X and Y , we assume that f\colon X\to Y satisfies the SNC assumption as above. In order to show that \kappa(X)=\dim Y , it is equivalent to showing that K_{Y}+B_{Y}+M_{Y} is big, as the pluricanonical sections of X descend to Y . Hence, it is enough to show that K_{Y}+B_{Y}+M_{Y}-\epsilon H is pseudo-effective for some \epsilon>0 . Rewriting the canonical bundle formula, we have (K_{X}+\Delta^{-}-\epsilon f^{*}H)-\Delta^{+}\sim_{\mathbf{Q}}f^{*}(K_{Y}+B_{Y}+M_{Y}-\epsilon H). Note that K_{X}+\Delta^{-}-\epsilon f^{*}H has a semi-positively curved singular hermitian metric on X . Hence, we see that \Delta^{+} is the only obstacle for descending the metric positivity of K_{X}+\Delta^{-}-\epsilon f^{*}H down to Y . In our main result, we give two separate ways to deal with this \Delta^{+} . 1.2. Main results We state the main results. Theorem 1.3 . As in the setup of Conjecture 1.2 and § 1.1 , suppose that one of the two following conditions holds: (1) K_{Y}+(1-\epsilon)B_{Y} is pseudo-effective for some \epsilon>0 , or (2) for the general fiber F , the class \{K_{F}\} is rigid. Then \kappa(X)=\dim Y . For a compact Kähler manifold X , a class \alpha\in H^{1,1}(X)(\simeq H_{\mathrm{BC}}^{1,1}(X)) is rigid if it contains exactly one closed positive (1,1)-current T such that \{T\}\in\alpha . For varieties of Kodaira dimension zero, the rigidity of the canonical class is tightly related to the existence of good minimal models. For a systematic treatment of rigid currents in terms of birational geometry, we refer to [ 15 ] . Here, we mention the easy direction that if F admits a good minimal model, then the cohomology class \{K_{F}\} is rigid [ 15 ] *Lemma 3.9. We point out that the proofs for the two cases are independent. Also, due to technical reasons concerning the volume asymptotics, we were not able to handle the case when \epsilon=0 for the first condition. 1.3. Comparison with known results We compare Theorem 1.3 with previously known results. The first result can be understood as a strengthening of [ 20 ] *Theorem 12.1, where he assumes that K_{Y} is pseudo-effective. Indeed, adding an effective divisor (1-\epsilon)B_{Y} in the first case of Theorem 1.3 can be thought of as a correction term coming from the singularity of the fibers of f . Our initial motivation was to overcome the non-pseudo-effectivity of K_{Y} by considering the contribution from the singular fibers, since heuristically, a projective morphism from a variety with ‘positive’ canonical bundle to a variety with ‘negative’ canonical bundle tends to have a lot of singular fibers (see [ 6 ] *Question 4.1, [ 14 ] , [ 25 ] , [ 4 ] , [ 17 ] for related results). However, K_{Y}+(1-\epsilon)B_{Y} is not always pseudo-effective, especially when the singular fibers are mildly singular,

Read the complete text