Rank 4 stable vector bundles on hyperkähler fourfolds of Kummer type

Imagine you are an architect trying to build a perfect, symmetrical city. In the world of mathematics, this "city" is a special kind of geometric shape called a Hyperkähler (HK) fourfold. These are incredibly complex, four-dimensional objects that are hard to visualize, let alone describe.

For a long time, mathematicians knew these shapes existed, but they were like "ghost cities"—we knew they were there, but we couldn't draw a clear blueprint for them. We needed a way to construct them explicitly.

This paper, written by Kieran G. O'Grady, is about finding the perfect key to unlock the blueprint for a specific type of these geometric cities, known as "Kummer type."

Here is the story of how he did it, explained in everyday terms:

1. The Problem: The Ghost City

Imagine you have a collection of these complex 4D shapes. You want to describe them all in a family, like a catalog. But to do that, you need a specific "fingerprint" that is unique to each shape.
In the past, mathematicians found that on simpler versions of these shapes (called K3 surfaces), there was a special, rigid vector bundle (think of this as a "fabric" or a "net" stretched over the shape) that acted as a perfect fingerprint. If you had this net, you could reconstruct the shape.

The goal of this paper is to find that same "perfect net" for the more complex 4D Kummer shapes.

2. The Solution: The "Rigid Net"

O'Grady proves that for these specific 4D shapes, there is indeed a unique, rigid net (a vector bundle of rank 4) that fits perfectly.

"Rigid" means the net doesn't wiggle or change shape. It's locked in place.
"Unique" means there is only one way to tie this net onto the shape. No other net fits the criteria.
"Stable" means the net is strong and doesn't fall apart under pressure.

He shows that if you have a shape with certain mathematical properties (specific numbers related to its size and curvature), you can build this net, and it will be the only one of its kind.

3. How He Built It: The "Shadow" Trick

Building this net directly on the 4D shape is like trying to sew a suit while blindfolded. It's too hard. So, O'Grady used a clever trick involving shadows and mirrors.

The Setup: He started with a simpler shape, an Abelian Surface (think of a 2D torus, like a donut).
The Map: He created a map (a function) that takes points from a "donut" and projects them onto the complex 4D shape.
The Construction: Instead of sewing the net directly on the 4D shape, he sewed a simpler net on the "donut" first. Then, he used his map to "project" or "cast the shadow" of that net onto the 4D shape.
The Result: This projection created the complex, rigid net he was looking for.

4. The "Lagrangian Fibration": The Train Tracks

To prove that his net was truly unique and stable, he had to test it under different conditions. He imagined the 4D shape as a train station with tracks (called Lagrangian fibers) running through it.

He checked if the net stayed strong and didn't tear when the train (the shape) moved along the tracks.
He proved that even when the tracks got bumpy or had singularities (kinks), the net remained stable, unless the train hit a very specific, rare obstacle.
This testing proved that the net is robust and works for the "general" case, which is what mathematicians care about most.

5. Why Does This Matter?

Why do we care about a rigid net on a 4D donut?

The Blueprint: Just as a unique net can define a K3 surface, this unique net defines the 4D Kummer shape.
Explicit Description: Before this, we couldn't write down the exact equations for these shapes. Now, because we have this "net," we can describe the entire family of these shapes explicitly. It's like going from knowing a city exists to having the actual street map and building codes.
The "Mukai" Connection: The author compares this to how we describe K3 surfaces using "Mukai models." He hopes this discovery will lead to similar explicit descriptions for these 4D shapes, opening the door to understanding a whole new class of geometric universes.

The Analogy Summary

Imagine you have a mysterious, shifting cloud (the 4D shape). You want to take a photo of it, but it keeps changing.

Old Way: You tried to guess what it looked like, but you couldn't get a clear picture.
O'Grady's Way: He found a special, unbreakable glass mold (the rigid vector bundle) that fits perfectly around the cloud.
The Magic: Because the mold is unique and rigid, the moment you snap it into place, the cloud freezes into a perfect, defined shape. You can now look at the mold and say, "Ah, that is exactly what the cloud looks like."

This paper provides the instructions for making that glass mold, finally allowing mathematicians to see and describe these elusive 4D geometric worlds clearly.

Here is a detailed technical summary of the paper "Rigid stable rank 4 vector bundles on HK fourfolds of Kummer type" by Kieran G. O'Grady.

1. Problem Statement and Motivation

Context:
The paper addresses the existence and uniqueness of slope-stable, rigid vector bundles on Hyperkähler (HK) manifolds. While such bundles are well-understood on K3 surfaces and HK varieties of type $K3^{[n]}$ (Hilbert schemes of points on K3 surfaces), the case of HK fourfolds of Kummer type (deformations of generalized Kummer varieties $K_2(A)$ where $A$ is an abelian surface) presents distinct challenges due to the different topology and automorphism groups.

Specific Goal:
The author aims to prove the existence and uniqueness of a specific rank 4 vector bundle $F$ on a general polarized HK fourfold $(M, h)$ of Kummer type, subject to specific numerical constraints on the polarization $h$ . The ultimate motivation is to use this unique bundle to construct an explicit description of a locally complete family of polarized HK fourfolds of Kummer type, analogous to "Mukai models" for K3 surfaces.

Numerical Constraints:
Let $q_M$ be the Beauville–Bogomolov–Fujiki (BBF) quadratic form and $\text{div}(h)$ be the divisibility of the polarization class $h$ . The paper focuses on two cases:

$q_M(h) \equiv -6 \pmod{16}$ and $\text{div}(h) = 2$ .
$q_M(h) \equiv -6 \pmod{144}$ and $\text{div}(h) = 6$ .

Target Bundle Properties:
The bundle $F$ must satisfy:

Rank $r(F) = 4$ .
First Chern class $c_1(F) = h$ .
Discriminant $\Delta(F) := 8c_2(F) - 3c_1(F)^2 = c_2(M)$ .
Rigidity: $H^1(M, \text{End}_0(F)) = 0$ .

2. Methodology

The proof strategy involves constructing the bundle explicitly on a specific model ( $K_2(A)$ ) and then extending it to the general deformation.

A. Construction via Isogenies and Resolutions

Setup: Let $f: B \to A$ be an isogeny of abelian surfaces of degree 2. This induces a rational map $\rho: K_2(B) \dashrightarrow K_2(A)$ defined by $[Z] \mapsto [f(Z)]$ .
Resolution: The map $\rho$ is not regular; its indeterminacy locus is a codimension-2 subvariety $V(f)$ . The author resolves this by blowing up $V(f)$ to obtain a smooth variety $X$ and a regular map $\tilde{\rho}: X \to K_2(A)$ .
Bundle Definition: For a line bundle $L$ on $X$ , the vector bundle is defined as the pushforward $E(L) := \tilde{\rho}_* L$ . This results in a rank 4 torsion-free sheaf on $K_2(A)$ .

B. Modularity and Stability Analysis

Modularity: The author determines precise conditions on the Chern class of $L$ (specifically relating coefficients $x$ and $y$ in the decomposition of $c_1(L)$ ) such that $E(L)$ is a modular sheaf. A sheaf is modular if its discriminant satisfies a specific relation with the BBF form.
Key Result: $E(L)$ is modular if and only if $y=x$ or $y=x+1$ . In these cases, the discriminant is exactly $\Delta(E(L)) = c_2(K_2(A))$ .
Local Freeness: Under the condition $y=x$ , the sheaf is shown to be locally free (a vector bundle).
Bridgeland–King–Reid (BKR) Equivalence: A significant portion of the paper (Section 5, based on referee insights) utilizes the BKR equivalence between the derived category of the symmetric product and the Hilbert scheme. This identifies $E(L)$ with a semi-homogeneous vector bundle on the kernel of the summation map $A^3 \to A$ . This identification provides a conceptual proof of the vanishing of cohomology groups for the traceless endomorphism bundle ( $H^p(K_2(A), \text{End}_0(E(L))) = 0$ ).

C. Restriction to Lagrangian Fibrations

To prove stability on the general deformation, the author analyzes the restriction of $E(L)$ to Lagrangian fibers of a fibration $\pi_A: K_2(A) \to \mathbb{P}^2$ (induced by an elliptic fibration on $A$ ).

Smooth Fibers: The restriction to a smooth Lagrangian fiber is proven to be slope-stable. This relies on the fact that the restriction is a simple semi-homogeneous bundle with vanishing discriminant.
Singular Fibers: The analysis of singular fibers (which are non-reduced and reducible) is the most technically demanding part. The author proves that the restriction to a general singular fiber is not destabilized by any subsheaf of integer rank. This involves detailed geometric analysis of the components $V(A)_{D_0}$ and $\Delta(A)_{D_0}$ and their intersection.

D. Deformation and Uniqueness

Existence: Using the openness of stability and the fact that the moduli space of Kummer type varieties is irreducible, the existence of such a bundle on a general $(M, h)$ follows from its existence on the specific model $K_2(A)$ .
Uniqueness: The uniqueness is established by showing that any two such bundles must be isomorphic when restricted to a general Lagrangian fiber. This relies on the monodromy action on the 2-torsion of the fiber and the fact that the bundle is rigid ( $H^1(\text{End}_0(F))=0$ ).

3. Key Contributions and Results

Main Theorem (Theorem 1.1):
For a general polarized HK fourfold $(M, h)$ of Kummer type satisfying the specific congruence conditions ( $q_M(h) \equiv -6 \pmod{16}$ or $144 $), there exists a **unique** (up to isomorphism) slope-stable vector bundle$ F $with rank 4,$ c_1(F)=h $, and$ \Delta(F) = c_2(M) $. Furthermore, this bundle is rigid ($ H^1(M, \text{End}_0(F)) = 0$).
Explicit Construction:
The paper provides an explicit construction of these bundles via the pushforward of line bundles on a resolution of the map between generalized Kummers associated with a degree-2 isogeny.
Modularity of Pushforwards:
The author establishes precise criteria for when the pushforward of a line bundle under the resolution of the isogeny map yields a modular sheaf, a result of independent interest for the study of vector bundles on HK manifolds.
Stability on Singular Fibers:
The paper develops a robust framework for analyzing the stability of vector bundles on non-reduced, reducible Lagrangian fibers, proving that the constructed bundle remains stable (or at least not destabilized by integer-rank subsheaves) in the singular limit.
Connection to BKR Equivalence:
The paper links the geometric construction of $E(L)$ to the representation-theoretic construction of semi-homogeneous bundles via the BKR equivalence, providing a deeper understanding of the bundle's structure and cohomological properties.

4. Significance

Explicit Moduli Spaces: The primary significance lies in the potential to describe a locally complete family of polarized HK fourfolds of Kummer type explicitly. Just as Mukai used rigid bundles on K3 surfaces to embed K3 surfaces into Grassmannians, O'Grady suggests that the projectivization of this rank 4 bundle could provide an explicit embedding of the Kummer type HK fourfolds into a projective space or Grassmannian.
Extension of K3[2] Results: It successfully extends the theory of rigid stable bundles from the $K3^{[2]}$ case (Hilbert schemes) to the Kummer case, which has a different deformation type and automorphism group structure.
Test for Franchetta Conjecture: The existence of such a unique, rationally defined bundle provides a concrete test case for the generalized Franchetta conjecture regarding Chow groups on moduli spaces of HK varieties.
Technical Advancement: The methods developed for handling singular Lagrangian fibers and the interplay between isogenies and Hilbert schemes offer new tools for the study of vector bundles on higher-dimensional hyperkähler manifolds.

In summary, O'Grady establishes the existence and uniqueness of a specific rigid rank 4 bundle on Kummer-type HK fourfolds, providing a crucial step toward the explicit geometric realization of these moduli spaces.