Perverse-Hodge complexes for Lagrangian fibrations

Imagine you are looking at a complex, multi-layered sculpture. From the front, it looks like a smooth, flowing wave. From the side, it looks like a jagged mountain range. From above, it looks like a perfect circle.

For a long time, mathematicians studying a specific type of geometric shape called a Lagrangian fibration (think of it as a giant, multi-dimensional "bundle" of shapes) knew a surprising secret: the "wave" view and the "mountain" view were mathematically identical, even though they looked completely different. This was a numerical coincidence, like finding that the number of red bricks equals the number of blue bricks in two different piles.

This paper, by Junliang Shen and Qizheng Yin, asks a deeper question: Is there a deeper reason why these two views are the same? Can we build a bridge between them that explains how they are identical, not just that they are?

Here is the breakdown of their discovery using simple analogies.

1. The Setting: The Great Bundle

Imagine a giant, magical library (the space $M$ ). Inside this library, there are millions of books arranged in rows.

The Fibers: If you look at a single row, it's a perfect, smooth circle (a torus).
The Base: If you look at the whole library from above, you see a map (the base $B$ ) showing where all these rows are located.
The Twist: In some parts of the library, the rows get squished or broken (singular fibers). This makes the math very hard because the rules for smooth circles don't apply to broken ones.

2. The Old Discovery: Counting Bricks

Previously, mathematicians found a rule called the "Perverse = Hodge" identity.

Hodge Numbers: These count the "holes" and "loops" in the library's structure (like how many tunnels go through the building).
Perverse Numbers: These count the complexity of the rows, especially the broken ones.

They found that if you count the holes in a specific way, the number of "row-complexities" exactly matches the number of "library-holes." It was like saying, "The number of red bricks in the wall equals the number of blue bricks in the floor." It was true, but it felt like magic.

3. The New Discovery: The "Perverse-Hodge" Complexes

Shen and Yin propose that these aren't just numbers; they are entire objects.

Imagine that instead of just counting bricks, you have a 3D hologram of the library.

The "Perverse" Hologram: Shows the structure of the broken rows.
The "Hodge" Hologram: Shows the structure of the smooth holes.

The authors propose a Symmetry: There is a way to rotate or flip the "Perverse" hologram so that it becomes the "Hodge" hologram perfectly. They call these objects Perverse-Hodge Complexes.

The Big Conjecture:
They guess that for any part of the library, the "Perverse" view and the "Hodge" view are actually the same object, just seen from a different angle.

Analogy: Imagine a chameleon. From the left, it looks green. From the right, it looks blue. The old math said, "The amount of green equals the amount of blue." The new math says, "The chameleon is both green and blue simultaneously; it's a single object that changes appearance based on how you look at it."

4. How They Proved It (The Three Levels)

Since the library is huge and has broken parts, they couldn't prove it everywhere at once. So, they built a ladder of proof:

Level 1: The Smooth Library (The Easy Case)
First, they looked at a library where every row is perfect and smooth. Here, the symmetry is obvious. It's like looking at a perfect sphere; you can spin it, and it looks the same. They used a known mathematical tool (the symplectic form, which is like the "glue" holding the library together) to show the two views are identical.
Level 2: The Broken Library (The Hard Case)
Real libraries have broken rows. The authors used a powerful new tool called Hodge Modules. Think of this as a "super-microscope" that can look at broken, jagged shapes and still see the smooth patterns underneath.
- They tested this on Hilbert Schemes (which are like collections of points that can move around). They showed that even when the points crash into each other (creating singularities), the symmetry holds.
- The Magic Trick: They realized that the "broken" parts of the library actually contribute to the symmetry in a very specific way, balancing out the smooth parts.
Level 3: The Whole Picture (Global Cohomology)
Finally, they looked at the entire library as a whole. They used a giant algebraic machine called the LLV Algebra (named after Looijenga, Lunts, and Verbitsky).
- Analogy: Imagine the library has a hidden "symmetry engine" (the LLV algebra) that rotates the entire building. This engine has a special switch that swaps the "Perverse" view with the "Hodge" view.
- Because this engine exists for all compact symplectic varieties, the symmetry must exist for the total count of holes. This proves their conjecture is true "on average" or "globally."

5. Why Does This Matter?

This paper is a step toward Categorification.

Old Math: "A equals B" (Numbers match).
New Math: "A is isomorphic to B" (The structures themselves are the same).

By proving that these complex objects are symmetric, the authors are building a bridge between two different worlds of geometry. This helps mathematicians:

Predict the unknown: If we know the structure of the "broken" rows, we can instantly know the structure of the "smooth" holes, and vice versa.
Understand the universe: These shapes appear in string theory and physics. Understanding their hidden symmetries might help us understand the fundamental laws of the universe.

Summary

Shen and Yin took a mysterious numerical coincidence (where two different ways of counting geometric features gave the same answer) and upgraded it to a deep structural truth. They showed that the "broken" and "smooth" aspects of these geometric shapes are actually two sides of the same coin, connected by a beautiful, hidden symmetry that works even when the shapes are messy and complex.

Here is a detailed technical summary of the paper "Perverse–Hodge complexes for Lagrangian fibrations" by Junliang Shen and Qizheng Yin.

1. Problem Statement and Context

The paper addresses the categorification of a known cohomological identity relating the topology of Lagrangian fibrations to the Hodge theory of the total space.

Background: Let $M$ be a compact irreducible symplectic variety (hyper-Kähler manifold) of dimension $2n $with a Lagrangian fibration$ \pi: M \to B$. The Decomposition Theorem (BBD) provides a decomposition of the derived pushforward of the constant sheaf:
$R\pi_* \mathbb{Q}_M[2n] \simeq \bigoplus_{i=-n}^n P_i[-i]$
where $P_i$ are perverse sheaves on the base $B$ .
The "Perverse = Hodge" Identity: Previous work (Shen-Yin, 2022) established an identity connecting the cohomology of these perverse sheaves to the Hodge numbers of $M$ :
$h^{j-n}(B, P_{i-n}) = h^{i,j}(M)$
This identity relies heavily on global cohomological properties of compact manifolds and does not explain the underlying sheaf-theoretic mechanism.
The Core Question: Can this cohomological identity be lifted to an isomorphism in the derived category of coherent sheaves on the base $B$ ? Specifically, can one construct objects (complexes) that categorify the perverse sheaves $P_i$ and the Hodge filtration such that a symmetry exists between them?

2. Methodology and Definitions

The authors introduce Perverse–Hodge complexes ( $G_{i,k}$ ) constructed via Saito's theory of Hodge modules.

Hodge Modules: They utilize the decomposition theorem in the category of Hodge modules. The trivial Hodge module $\mathbb{Q}^H_M[2n]$ decomposes as:
$\pi_+ \mathbb{Q}^H_M[2n] \simeq \bigoplus_{i=-n}^n \mathcal{P}^H_i[-i]$
where $\mathcal{P}^H_i = (P_i, F_\bullet)$ consists of a regular holonomic $D_B$ -module $P_i$ and a good filtration $F_\bullet$ .
Construction of $G_{i,k}$ : The authors define the perverse–Hodge complexes by taking the associated graded pieces of the de Rham complex of these Hodge modules:
$G_{i,k} := \text{gr}^F_{-k} \text{DR}(P_{i-n})[n-i] \in D^b\text{Coh}(B)$
Here, $i$ represents the perverse degree, and $k$ represents the Hodge degree.
The Conjecture (Conjecture 1.2): The central proposal is a Perverse–Hodge Symmetry:
$G_{i,k} \simeq G_{k,i} \quad \text{in } D^b\text{Coh}(B)$
This conjecture suggests that the roles of the perverse degree and the Hodge degree are interchangeable at the level of complexes, not just their cohomology.

3. Key Contributions and Results

The paper verifies this conjecture in three distinct settings, employing different mathematical tools for each case.

A. Smooth Morphisms (Theorem 1.3)

Context: When $\pi: M \to B$ is a smooth morphism (a family of abelian varieties).
Method: The authors utilize the variation of Hodge structures (VHS) associated with the fibration. They leverage the Donagi–Markman cubic condition, which relates the symplectic form $\sigma$ on $M$ to the Gauss–Manin connection.
Result: They construct an explicit term-by-term isomorphism between the complexes $G_{i,k}$ and $G_{k,i}$ . The symmetry arises from the symmetry of the morphism $\theta: V^{1,0} \to \Omega^1_B \otimes \Omega^1_B$ induced by the symplectic form and polarization.
Significance: This establishes the conjecture in the "generic" case where singular fibers are absent, showing the symmetry is rooted in the geometry of the VHS.

B. Hilbert Schemes of Points (Theorem 1.4)

Context: The paper considers Lagrangian fibrations induced by elliptic fibrations of symplectic surfaces, specifically the Hilbert schemes of points $S^{[n]} \to B^{(n)}$ . This includes cases like $K3^{[n]}$ and Hitchin systems.
Method: The authors use a decomposition strategy involving:
1. Symmetric Products: They analyze the decomposition of the pushforward for symmetric products of the surface.
2. Compatibility: They prove that the Perverse–Hodge Symmetry (PHS) property is preserved under:
  - Closed embeddings and finite morphisms (Proposition 4.1).
  - External products of varieties (Proposition 4.3).
  - Symmetric products (Proposition 4.6).
Result: By reducing the problem to the base case of the elliptic surface ( $n=1$ , verified via duality) and applying the compatibility lemmas, they prove the conjecture holds for all $n \geq 1$ .
Significance: This extends the symmetry to cases with singular fibers and non-trivial support on the boundary of the base, demonstrating that the symmetry is robust even when the decomposition theorem involves contributions from lower-dimensional strata.

C. Global Cohomology and LLV Algebras (Theorem 1.5)

Context: For general Lagrangian fibrations where $M$ is a compact irreducible symplectic variety.
Method: The authors work at the level of global cohomology using the Looijenga–Lunts–Verbitsky (LLV) Lie algebra $\mathfrak{g}(M) \cong \mathfrak{so}(b_2(M)+2)$ $g (M) ≅ so (b_{2} (M) + 2)$ .
- They define a subalgebra called the Perverse–Hodge algebra $\mathfrak{g} \subset \mathfrak{g}(M)$ , generated by the standard Kähler classes and the classes defining the fibration. This algebra is isomorphic to $\mathfrak{so}(6)$ .
- They identify the cohomology groups $H^*(B, G_{i,k})$ with specific weight spaces of the LLV algebra representation on $H^*(M, \mathbb{C})$ .
Result: The symmetry $H^*(B, G_{i,k}) \simeq H^*(B, G_{k,i})$ follows from the action of the Weyl group of $\mathfrak{so}(6)$ . Specifically, an element of the Weyl group exchanges the weights corresponding to the perverse and Hodge degrees while fixing the total degree.
Significance: This proves the conjecture cohomologically for the general case, providing a deep structural explanation for the "Perverse = Hodge" identity. It shows that the identity is a shadow of a larger symmetry encoded in the LLV algebra.

4. Significance and Implications

Categorification: The paper successfully lifts the numerical identity $h^{j-n}(B, P_{i-n}) = h^{i,j}(M)$ to an isomorphism of complexes in the derived category. This provides a "sheaf-theoretic" explanation for the identity, moving beyond global cohomological calculations.
Matsushita's Theorem: The conjecture specializes to and generalizes Matsushita's theorem on higher direct images of the structure sheaf ( $R^i \pi_* \mathcal{O}_M \simeq \Omega^i_B$ ). Specifically, the case $k=2n$ of the conjecture is equivalent to Matsushita's result.
New Computational Tools: The authors suggest that the Perverse–Hodge symmetry can be used as a "recipe" to calculate higher direct images of $\Omega^k_M$ by trading contributions from different perverse degrees, potentially simplifying calculations in complex geometric settings.
Geometric Insight: The work connects three major areas of algebraic geometry:
1. Hodge Theory (via Hodge modules and VHS).
2. Representation Theory (via LLV algebras and Weyl groups).
3. Symplectic Geometry (via Lagrangian fibrations and the Donagi–Markman cubic condition).

In summary, Shen and Yin propose a profound symmetry between the perverse and Hodge filtrations of Lagrangian fibrations, verifying it through smooth families, Hilbert schemes, and global cohomological structures, thereby unifying disparate results in the study of hyper-Kähler manifolds.