🔢 mathematics

On nefness of the lowest piece of Hodge modules

This paper establishes degree lower bounds for quotient line bundles of the lowest piece of Hodge modules induced by complex variations of Hodge structures with non-unipotent monodromies, demonstrating the failure of nefness in such cases while recovering Kawamata's semi-positivity theorem for the unipotent setting through an algebraic proof involving a vanishing theorem for twisted Hodge modules.

Original authors: Ze Yun

Published 2026-05-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ze Yun

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: A Map with Rough Edges

Imagine you are a cartographer trying to draw a perfect map of a landscape (a mathematical space called a manifold). In the middle of this landscape, there is a smooth, beautiful region where everything is well-behaved. However, the edges of your map are jagged, marked by a "Simple Normal Crossing" (SNC) divisor. Think of these edges as a fence made of intersecting walls ( $D_1, D_2, \dots$ ).

On the smooth part of the land, you have a special kind of data flowing through it, called a Variation of Hodge Structures (VHS). You can think of this data as a complex, multi-layered fabric that changes slightly as you move around, but it follows strict rules.

Mathematicians want to know: Is this fabric "nice" (or "nef") everywhere, even near the jagged fence?

In math-speak, "nef" roughly means "non-negative" or "stable." If a bundle is nef, it behaves well under stretching and pulling. If it's not nef, it might tear or collapse in weird ways.

The Problem: When the Fence is "Twisted"

For a long time, mathematicians knew that if the data near the fence was "unipotent" (a technical way of saying the data behaves very predictably, like a gentle breeze), the fabric was always "nef." It was smooth and stable.

However, the author, Ze Yun, investigates what happens when the data near the fence is not unipotent. Imagine the wind near the fence isn't just blowing gently; it's swirling, spinning, or twisting violently. In this case, the fabric might not be "nef" anymore. It might have a negative degree, meaning it's "unstable."

The paper asks: If the fabric isn't perfectly stable, how bad is it? Can we put a number on exactly how much it fails to be stable?

The Solution: The "Twisted" Lens

To answer this, the author uses a new mathematical tool called Twisted Hodge Modules.

The Analogy:
Imagine you are looking at a twisted piece of wire through a special pair of glasses.

Standard View: You see the wire, but you can't quite tell how much it's twisting because the view is too complex.
Twisted View (The Author's Method): The author puts on "twisted glasses" (Twisted Hodge Modules). These glasses allow you to peel back the layers of the problem. Instead of looking at the whole messy fabric at once, you can look at it layer by layer, right up against the fence.

By using these glasses, the author can see that the "badness" (the failure to be nef) comes directly from the monodromy (the twisting/spinning) of the data as it hits the fence.

The Main Discovery: A Formula for "Badness"

The paper proves a specific formula. It says that if you take a slice of this fabric (a line bundle) and pull it onto a curve (a path), its "degree" (a measure of its stability) cannot be lower than a specific number.

This number depends on two things:

How much the data twists at the fence: Represented by numbers called $\beta$ (eigenvalues of residues). If the twist is zero (unipotent), the number is zero, and the fabric is perfectly stable. If the twist is strong, the number is negative.
How the fence intersects the path: Represented by how many times your path hits the fence walls.

The Formula in Plain English:

"The stability of your fabric is at least the sum of (how much the wind twists) $\times$ (how hard you hit the fence)."

If the wind twists a lot (large negative $\beta$ ) and you hit the fence hard (large intersection number), the fabric becomes very unstable (very negative degree). But the paper proves it never gets worse than this calculated limit.

Why This Matters (According to the Paper)

It Explains the Failure: Previous examples showed that the fabric could be unstable. This paper explains why and how much. It shows that the instability isn't random; it's a direct result of the twisting at the boundary.
It Recovers Old Truths: If the wind is gentle (unipotent), the formula gives zero, confirming the old theorem that the fabric is always stable in that case.
It's Algebraic, Not Analytic: Usually, to prove things about these fabrics, mathematicians have to do very difficult calculus (analyzing how the fabric stretches and shrinks infinitely). The author's method is "algebraic." It's like solving a puzzle using logic and rules rather than measuring physical forces. This makes the proof cleaner and more robust.
It's Sharp: The author provides examples (like a branched cover of a projective bundle) where the fabric hits this "badness limit" exactly. This proves the formula is the best possible answer; you can't make the bound tighter.

Summary of the "Twisted" Magic

The core innovation is the use of Twisted Hodge Modules.

Normal Math: Tries to look at the whole picture and gets confused by the singularities (the fence).
This Paper: Says, "Let's specialize." It takes the data and pushes it onto the fence, then onto the intersection of fences, using "twisted" versions of the data at each step.
The Result: By breaking the problem down into these twisted layers, the author can use a powerful "vanishing theorem" (a rule that says certain messy parts disappear) to prove the lower bound on stability.

The Takeaway

Ze Yun's paper is like a weather report for a mathematical landscape. It tells us that while the "fabric" of our data might get stormy near the edges (the divisors), we can now calculate the exact severity of the storm based on how much the wind twists. We know the fabric won't collapse completely; it will only fail up to a specific, calculable limit. This gives mathematicians a precise tool to handle complex, non-smooth situations that were previously mysterious.

Technical Summary: On Nefness of the Lowest Piece of Hodge Modules

Problem Statement
The paper addresses the nefness (numerical effectiveness) of the lowest piece of the Hodge filtration, denoted $F^pM$ , for a Hodge module $M$ induced by a polarized complex variation of Hodge structures (VHS) on a complex manifold $X$ minus a simple normal crossing (SNC) divisor $D$ . While it is a classical result (Griffiths, Fujino, Zucconi, Peters) that $F^pM$ is a nef vector bundle when the image of a curve lies outside the discriminant locus, or when the monodromies at infinity are unipotent (Kawamata), the general case where monodromies are non-unipotent remains problematic. Specifically, examples by Fujino and Fujisawa demonstrate that $F^pM$ fails to be nef in the presence of non-unipotent monodromy. The central problem is to quantify this failure: specifically, to establish degree lower bounds for quotient line bundles of the pullback of $F^pM$ to curves that intersect or lie within the boundary divisor $D$ .

Methodology
The author employs a purely algebraic approach, avoiding the analytic study of Hodge metric asymptotics used in previous works (e.g., Cattani-Kaplan-Schmid, Fujino-Fujisawa). The core methodology relies on:

Twisted Hodge Modules: Utilizing the theory of twisted Hodge modules (developed by Schnell-Yun and D'Addezio-Voigt) to handle the specialization of Hodge modules to divisors. When restricting a Hodge module to a divisor with non-unipotent monodromy, the resulting object is naturally a twisted Hodge module, where the twisting is determined by the eigenvalues of the residues of the Deligne canonical extension.
Iterative Nearby Cycles: The proof strategy involves iteratively applying the nearby cycle functor along the components of the SNC divisor. This reduces the dimension of the support of the Hodge module, eventually restricting the problem to curves.
Vanishing Theorems: The argument hinges on Saito's vanishing theorem for twisted Hodge modules. By establishing a vanishing theorem for the first cohomology of the dual of the lowest piece twisted by a line bundle of negative degree, the author derives lower bounds for the degrees of quotient line bundles.
Algebraic Reduction: The proof reduces the general case to the case of curves by pulling back the VHS to a curve $C$ , extending it to a Hodge module, and analyzing the filtration of the restricted sheaf via the $V$ -filtration and monodromy weight filtration.

Key Results
The main theorem (Theorem 1.1) provides a precise lower bound for the degree of any quotient line bundle $P$ of $f^*F^pM$ , where $f: C \to X$ is a map from a smooth projective curve.

Case 1 (Transversal): If the image of $C$ is not contained in $D$ , $f^*F^pM$ is a nef vector bundle.
Case 2 (Contained in Boundary): If the image of $C$ lies in the intersection of $k$ boundary divisors $D_1 \cap \dots \cap D_k$ (but not in higher-order intersections), and $f^*F^pM \to P$ is a quotient line bundle, then:
$\deg P \ge \min_{\{j_i\}} \left\{ \sum_{i=1}^k (-\beta_{i, j_i} \cdot \deg f^* \mathcal{O}_X(D_i)) \right\}$
Here, $-\beta_{i, j_i}$ are the eigenvalues of the residues of the Deligne canonical extension along $D_i$ (normalized to lie in $(-1, 0]$ ).

Corollaries and Special Cases

Recovery of Kawamata's Theorem: If all monodromies are unipotent, all residue eigenvalues $\beta$ are zero. The lower bound becomes 0, recovering Kawamata's semi-positivity theorem which states $F^pM$ is nef.
Nef Boundary Divisors: If the boundary divisors $D_i$ are themselves nef, then $F^pM$ is nef, regardless of the monodromy. This recovers results for complex tori (where effective divisors are nef) previously established by Popa, Schnell, and others via generic vanishing theory.
Complete Intersections: The paper proves that if a curve $C$ is a smooth complete intersection (even if it lies inside the boundary divisor), the pullback $f^*F^pM$ remains nef. This is surprising as it holds even when the curve is contained in the singular locus of the VHS.

Significance and Claims
The paper claims to provide a sharp, algebraic explanation for the failure of nefness in the non-unipotent case.

Algebraic vs. Analytic: The work offers an alternative to analytic proofs relying on the growth of Hodge metrics and Lelong numbers. It attributes the failure of nefness directly to the "jumping indices" of the $V$ -filtration (residue eigenvalues) and the intersection numbers with the boundary.
Sharpness: The author provides geometric examples (generalizing the counter-example in Fujino-Fujisawa [FF14]) to demonstrate that the derived lower bound is sharp and can be achieved.
Utility of Twisted Hodge Modules: The paper highlights twisted Hodge modules as a natural and effective tool for studying the restriction of Hodge modules to divisors, particularly when monodromies are not unipotent.

The author explicitly notes that the results are limited to polarized variations of Hodge structures and that the lower bounds depend on local monodromies and intersection numbers, without claiming broader applications to non-polarized cases or other geometric contexts not explicitly discussed.