On nefness of the lowest piece of Hodge modules
Ze Yun · 2025
Read On nefness of the lowest piece of Hodge modules by Ze Yun (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-2507-01731.txt.
Introduction
Ze Yun gave lower bounds in 2025 for how positive certain line bundles must stay in Hodge theory, a corner of geometry where positivity can fail. Hodge modules package variation of Hodge structures, and the lowest piece of such a module yields line bundles on quotient spaces. In the nicest case their positivity is guaranteed by a classical theorem of Kawamata.
Yun's solo note moves beyond that case, where the behavior at infinity is not unipotent. The bounds he proves lean on the local twisting at infinity and on intersection numbers with the boundary. The classical theorem falls out as the special case. Geometric examples show the bounds can be met exactly, so nothing is wasted. The route is algebraic, through a vanishing theorem for twisted modules. Readers get a sharp answer to when positivity survives a wilder boundary.
Catalog record
| CREATOR | Ze Yun |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-01731
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On nefness of the lowest piece of Hodge modules Ze Yun Abstract We give degree lower bounds for quotient line bundles of the lowest piece of a Hodge module induced by a variation of Hodge structures outside a simple normal crossing divisor. We also give sufficient conditions on when the lowest piece is nef. Classical results say when X is a curve or the monodromies are unipotent, the lowest piece is a nef vector bundle on X . In dimension greater than or equal to two, if the monodromies at infinity are not unipotent there exist counterexamples in which nefness fails. This note aims to explain the failure of nefness. Our result is a lower bound on degrees of quotient line bundles of the pullback of the lowest piece to curves, in terms of local monodromy and intersection numbers with the boundary divisors. The proof uses twisted Hodge modules, which appear naturally when specializing the lowest piece to divisors. We also give geometric examples to show that the lower bound can be achieved. 1 Introduction The first nonzero step of the Hodge filtration of a variation of Hodge structures, equipped with its Hodge metric, has semi-positive curvature by the work of Griffiths. A common example of the lowest piece of Hodge filtration of a Hodge module is given by R^{j}f_{*}\omega_{X/S} for a map f:X\rightarrow S , and positivity of higher direct images of canonical bundle are very useful in algebraic geometry. On curves, the lowest piece of a Hodge module is always a nef vector bundle, which is proved using asymptotic behavior of Hodge metrics in [ Fuj78 ] , [ Zuc82 ] , and [ Pet84 ] . In the case of SNC discriminant with unipotent monodromy along the discriminant divisor, Kawamata [ Kaw81 , Theorem 5] proved the lowest piece of Hodge filtration of the canonical extension of a variations of Hodge structures across the divisor is a nef vector bundle. In both of their proofs, the analytic results in [ CKS86 ] about the asymptotics of the Hodge metric on the canonical extension, and the curvature properties of the Hodge bundle are essential. In dimensions greater than or equal to two, the example in Fujino-Fujisawa [ FF14 , example 7.9] shows nefness fails in general. We would like to measure how much nefness fails for the lowest piece, by giving degree lower bounds for quotient line bundles of this sheaf pullbacked to curves. The proof of the following theorem is algebraic, by using Saito’s vanishing theorem for twisted Hodge modules, and doesn’t need analysis of the asymptotics of Hodge metric. Let X be a compact complex manifold, D=\sum_{i}D_{i} is a SNC divisor on X , and \mathcal{V} is a polarized variation of Hodge structures on X\backslash D . For a fixed i , the real numbers (rational in the case of quasi-unipotent monodromies) -1<\beta_{ij}\leq 0 are eigenvalues of residues of the Deligne canonical extension of \mathcal{V} along D_{i} . Theorem 1.1 . Given C a smooth projective curve, and a map f:C\rightarrow X , (a) If \mathrm{Im}(C)\nsubseteq D , then f^{*}F_{p}M is a nef vector bundle on C . (b) If \mathrm{Im}(C)\subset D_{(k)}\mathrel{\mathop{:}}=D_{1}\cap\cdots\cap D_{k} , but the image of C doesn’t lie in any (k+1) intersections of boundary divisors, and f^{*}F_{p}M\rightarrow P is a quotient line bundle, then \deg P\geq\min_{\text{over all possible }j_{i}}\big\{\sum_{i=1}^{k}(-\beta_{ij_{i}}\cdot\deg f^{*}\mathcal{O}_{X}(D_{i}))\big\}. When the monodromies are unipotent, we have all \beta_{ij} equal to zero, this implies F_{p}M is a nef vector bundle on X . Therefore our result is a generalization of the semi-positivity theorem of Kawamata [ Kaw81 ] , see also [ FF14 ] , [ FFS14 ] , [ FF19 ] for discussions. Unexpectedly, the theorem also implies if the boundary divisors are nef, then F_{p}M is nef. As suggested by the result, the failure of nefness comes from the locus where the Hodge module is not smooth, which is also shown by the counter examples in [ FF14 , example 7.9] . For curves intersecting the open set underlying the variations of Hodge structures, it can be derived from an argument in [ BS23 , Lemma 2.6] that the pullback of the lowest piece is still a nef vector bundle on the curve. For curves contained in the boundary divisor, we need to restrict the lowest piece to D_{i} and hope to do some kind of induction. The problem is directly restricting F_{p}M to D_{i} doesn’t correspond to an object directly from a Hodge module on D_{i} . Our method relies on the notion of twisted Hodge modules [ SY24 ] [ DV25 ] , which naturally arises when we specialize a Hodge module to divisors. In particular, applying nearby cycles reduces the dimension of support of our (twisted) Hodge module, and eventually we are essentially in the case where C doesn’t meet the singular locus of the (twisted) Hodge module. The proofs uses nearby cycles for complex Hodge modules, see [ SS23 ] and [ SY24 ] , and our method work for complex variations of Hodge structrues which could be even not quasi-unipotent (although for real eigenvalues of residues, the equality may never be achieved; and there’s no effective methods to produce complex variations of Hodge structures not coming from geometry). The lower bound is sharp in general, as shown by various geometric examples. All the variations of Hodge structures are polarized, and they are just abbreviated as VHS. Canonical extension of a VHS will mean Deligne’s canonical extension with eigenvalues of residues in the interval (-1,0] . Nefness of a coherent sheaf \mathscr{F} is defined by the nefness of its relative \mathcal{O}(1) on \mathbb{P}(\mathscr{F}) , or equivalently if for every map f:C\rightarrow X from a smooth projective curve, and every quotient line bundles f^{*}\mathscr{F}\rightarrow L , \deg L\geq 0 . 2 Twisted Hodge modules and nearby cycles One key ingredient of our proof is the notion of twisted Hodge modules [ SY24 ] [ DV25 ] . In our situation, they arise naturally when taking nearby cycles \text{gr}^{W}\text{gr}^{V}M along a divisor, where V is the V -filtration, and W is the monodromy filtration on \text{gr}^{V}M . Let L be a line bundle on a complex manifold X , \alpha\in\mathbb{R} . The ring of \alpha L -twisted differential operators is a quasi-coherent sheaf on X which are locally isomorphic to the usual sheaf of differential operators, \mathscr{D}_{X,\alpha L}|_{U}\cong\mathscr{D}_{U}[t\partial_{t}]/(t\partial t-\alpha)\cong\mathscr{D}_{U}, but with a different gluing to a sheaf on X . The transition map between differential operators under trivializations \phi,\phi^{\prime}:L|_{U}\rightarrow U\times\mathbb{C} is t\partial_{t}\rightarrow t^{\prime}\partial_{t^{\prime}},\quad\partial_{j}\mapsto\partial_{j}+\alpha\cdot g^{-1}\frac{\partial g}{\partial x_{j}}, where \phi^{\prime}\circ\phi^{-1}:U\times\mathbb{C}\rightarrow U\times\mathbb{C} has the form (x,t)\mapsto(x,g(x)t) for some invertible holomorphic function g on U . In our case, if D is a divisor, and L=\mathcal{O}_{X}(D) , then one result in [ SY24 ] says twisted Hodge modules arise as nearby cycles of Hodge modules. Proposition 2.1 (Proposition 3.12 of [ SY24 ] ) . Let \left(\mathcal{M},F_{\bullet}\mathcal{M}\right) be a polarized complex Hodge module on X . For any real number \alpha\in[-1,0) and any integer \ell\in\mathbb{Z} , the object \left(\operatorname{gr}_{\ell}^{W(N)}\operatorname{gr}_{\alpha}^{V}\mathcal{M},F_{\bullet-1}\operatorname{gr}_{\ell}^{W(N)}\operatorname{gr}_{\alpha}^{V}\mathcal{M}\right) is a polarized \alpha L -twisted Hodge module on X . The object \left(\operatorname{gr}_{\ell}^{W(N)}\operatorname{gr}_{0}^{V}\mathcal{M},F_{\bullet}\operatorname{gr}_{\ell}^{W(N)}\operatorname{gr}_{0}^{V}\mathcal{M}\right) is a polarized complex Hodge module (without any twisting). The relation of twisted Hodge modules with our problem is the