Semi-rigid stable sheaves: a criterion and examples

This paper introduces the concept of semi-rigidity for stable sheaves on smooth polarized varieties to characterize the existence of stable deformations of direct sums, providing a criterion based on the Yoneda pairing and applying it to line bundles on projective varieties and Lagrangian subvarieties of hyper-Kähler manifolds.

The Gist

Imagine you are an architect studying a very special kind of building made of mathematical "bricks" called sheaves. These aren't physical bricks, but rather complex structures that live on geometric shapes (like spheres, toruses, or higher-dimensional versions of them).

Some of these buildings are stable. This means they are perfectly balanced; if you nudge them slightly, they don't fall apart or change their fundamental shape. They are like a perfectly tuned guitar string that stays in tune no matter how gently you pluck it.

The Big Question: What happens when you stack them?

The authors of this paper ask a fascinating question: What happens if you take two (or more) of these stable buildings and glue them together?

In math, this is called taking a direct sum. If you have a stable building $F$, you can make a new structure $F \oplus F$ (two copies glued together).

Usually, when you glue two stable things together, the result is unstable. It's like stacking two perfect houses on top of each other without a foundation; the whole thing might wobble, and eventually, it might transform into something completely different—a new, unique, "stable" building that wasn't just a simple stack.

However, the authors discovered a special class of these buildings that behave differently. They call them Semi-Rigid.

The "Semi-Rigid" Superpower

A Semi-Rigid building has a superpower: It refuses to change its identity when stacked.

If you take a semi-rigid building $F$ and glue two copies together ($F \oplus F$), the result stays exactly as it is. It doesn't morph into a new, mysterious shape. It remains a simple stack of the original building.

Think of it like Lego bricks.

• Normal Stable Sheaves: If you try to stack two of these, they might melt together and form a weird, new blob that looks nothing like the original bricks.
• Semi-Rigid Sheaves: If you stack two of these, they stay as two distinct, recognizable bricks. They don't melt. They are "rigid" enough to keep their shape, but "semi" because they can still wiggle a little bit (deform) without breaking.

The Secret Test: The "Yoneda Pairing"

How do you know if a building is Semi-Rigid before you try to stack it? You can't just look at it; you have to run a test.

The authors created a mathematical "detector" called the Yoneda Pairing. Imagine this as a stress test machine.

• You put the building into the machine.
• The machine checks for a specific type of "weakness" or "decomposable element" in the kernel (the part of the machine that catches errors).
• The Rule: If the machine finds no decomposable errors, the building is Semi-Rigid. It's safe to stack! If it does find an error, the building will likely melt into something new when stacked.

Where do we find these Semi-Rigid buildings?

The paper explores two main places where these special buildings live:

1. On "Irrational" Landscapes (Projective Varieties)
Imagine a landscape (a geometric shape). If this landscape is "too simple" or "too regular" (mathematicians call this having no "irrational pencils"), then the line bundles (the mathematical equivalent of paint or wallpaper on the landscape) are Semi-Rigid.

• Analogy: Think of a flat, boring desert. If you try to stack two layers of sand there, they just stay as sand. But if you have a complex, hilly terrain with hidden valleys (irrational pencils), stacking might cause a landslide (a deformation into something new).

2. On Hyper-Kähler Manifolds (The "Magic" Spaces)
This is the most exciting part. These are exotic, high-dimensional spaces with special symmetries (like a 4D sphere with extra magic properties).

• The authors found that if you take a special "Lagrangian" surface (a flat sheet floating inside this 4D magic space) and put a line bundle on it, the resulting structure is often Semi-Rigid.
• The Big Example: They looked at the "Fano variety of lines" on a Cubic Fourfold (a shape defined by a specific equation in 5D space). They found a specific component of this shape where the buildings are Semi-Rigid.
  • Why it matters: Usually, when you stack these, you get a mess. But here, because they are Semi-Rigid, the "stacked" version is perfectly predictable. It's like finding a secret room in a chaotic castle where everything is perfectly organized.

The "Split" vs. The "New"

The paper concludes with a surprising discovery about the "Moduli Space" (the map of all possible buildings).

• The Split Component: This is the path where you just stack the buildings ($F \oplus F$). For Semi-Rigid buildings, this path is a solid, unbreakable road. It's an "irreducible component," meaning you can't walk off it onto a different path.
• The New Component: For non-Semi-Rigid buildings, there are hidden paths where the stack melts into a new, stable shape.

The authors showed that on these specific Hyper-Kähler manifolds, the "Split Component" is so strong and rigid that it creates a whole new type of geometric object that cannot be smoothed out or resolved. It's a "primitive symplectic variety"—a shape that is inherently singular (has a sharp point) and cannot be fixed.

Summary in a Nutshell

1. The Problem: When you glue stable mathematical objects together, they usually change into something new and unpredictable.
2. The Discovery: Some objects are Semi-Rigid. They are stubborn enough that when you glue them, they stay exactly as they are.
3. The Test: You can tell if they are Semi-Rigid by checking a specific mathematical "stress test" (the Yoneda pairing) for errors.
4. The Application: This happens on special high-dimensional shapes (Hyper-Kähler manifolds).
5. The Result: Because these objects are Semi-Rigid, the "stacked" versions form a rigid, unchangeable structure in the mathematical map, proving that some mathematical landscapes are more complex and "broken" than we thought, and cannot be fixed.

It's like finding a set of magic Legos that, no matter how many times you stack them, never melt into a blob, but instead form a permanent, unbreakable tower that defines the rules of the entire toy box.

Technical Summary

Here is a detailed technical summary of the paper "Semi-Rigid Stable Sheaves: A Criterion and Examples" by Alessio Bottini and Riccardo Carini.

1. Problem Statement

The paper addresses the local geometry of moduli spaces of stable sheaves on smooth polarised varieties $(X, h)$. Specifically, it investigates the behavior of the direct-sum morphism:
$$\text{Sym}^n \mathcal{M}_v(X, h) \to \mathcal{M}_{nv}(X, h)$$
$$([F_1], \dots, [F_n]) \mapsto [F_1 \oplus \dots \oplus F_n]$$
around the point corresponding to the direct sum of $n$ copies of a stable sheaf $F$, denoted $[F^{\oplus n}]$.

The central question is: Under what conditions does a stable sheaf $F$ admit stable deformations of its direct sum $F^{\oplus n}$?
In other words, does the image of the direct-sum morphism cover a neighborhood of $[F^{\oplus n}]$ in the moduli space $\mathcal{M}_{nv}(X, h)$, or are there "new" stable sheaves (not isomorphic to direct sums) arbitrarily close to $[F^{\oplus n}]$?

2. Methodology

The authors employ a deformation-theoretic approach combined with geometric invariant theory (GIT) and representation theory.

• Formality Assumptions: The analysis assumes the differential graded algebra (DGA) $\text{RHom}^*(F, F)$ is formal and the deformation space $\text{Def}_F$ is smooth. Under these conditions, the deformation theory is governed by the Yoneda pairing (skew-symmetric cup product):
  $$\Upsilon_F: \bigwedge^2 \text{Ext}^1(F, F) \to \text{Ext}^2(F, F)$$
• Local Models:
  • The local structure of the moduli space $\mathcal{M}_{nv}(X, h)$ near $[F^{\oplus n}]$ is modeled by the zero locus of a quadratic map $\mu^{-1}(0)$ inside the tangent space $\text{Ext}^1(F^{\oplus n}, F^{\oplus n})$.
  • The direct-sum morphism corresponds to an inclusion of a commuting scheme $\mathcal{C}(\mathfrak{g}, V)$ (where $\mathfrak{g} = \mathfrak{gl}_n$ and $V = \text{Ext}^1(F, F)$) into the deformation space.
• Algebraic Tools:
  • The authors utilize the Chevalley restriction theorem for commuting schemes. They show that the quotient of the commuting scheme by the group action is isomorphic to the symmetric product of the vector space $V$.
  • They analyze the kernel of the Yoneda pairing, $\ker(\Upsilon_F)$, specifically looking for decomposable elements (elements of the form $e_1 \wedge e_2$).

3. Key Contributions and Definitions

Definition: Semi-Rigidity

A stable sheaf $F$ is defined as semi-rigid if every Jordan–Hölder factor of any sufficiently small deformation of $F^{\oplus 2}$ is a deformation of $F$.

• Geometric Interpretation: $F$ is semi-rigid if and only if the direct-sum morphism $\text{Sym}^2 \mathcal{M}_v \to \mathcal{M}_{2v}$ is locally surjective (a homeomorphism onto a neighborhood) at $2[F]$.
• Consequence: If $F$ is semi-rigid, $F^{\oplus n}$ has no stable deformations in the immediate neighborhood; the only stable sheaves nearby are direct sums of deformations of $F$.

The Main Criterion (Theorem A)

The paper establishes a precise algebraic criterion for semi-rigidity:

Theorem A: A stable sheaf $F$ is semi-rigid if and only if the kernel of the Yoneda pairing, $\ker(\Upsilon_F) \subseteq \bigwedge^2 \text{Ext}^1(F, F)$, contains no non-zero decomposable elements.

This connects the geometric property of the moduli space to the algebraic structure of the Ext-algebra.

4. Key Results

Local Structure of Moduli Spaces (Theorems B & 4.4)

• If $F$ is semi-rigid: The direct-sum morphism $\text{Sym}^n \mathcal{M}_v \to \mathcal{M}_{nv}$ is the embedding of an irreducible component in a neighborhood of $[F^{\oplus n}]$.
• If $F$ is NOT semi-rigid: Any neighborhood of $[F^{\oplus n}]$ in $\mathcal{M}_{nv}$ contains stable sheaves that are not direct sums. This implies the moduli space is reducible or has multiple components meeting at the split locus.
• Stronger Condition: If $\ker(\Upsilon_F) = 0$ (i.e., $\Upsilon_F$ is injective), then the direct-sum morphism is an isomorphism in a neighborhood of the diagonal, and the moduli space is reduced and normal there.

Geometric Criteria for Line Bundles (Proposition C)

For line bundles $L$ on a smooth projective variety $X$, semi-rigidity is equivalent to topological conditions on $X$:

1. For any linearly independent $\eta_1, \eta_2 \in H^1(X, \mathbb{C})$, $\eta_1 \cup \eta_2 \neq 0$.
2. Every morphism $X \to C$ to a curve of genus $g(C) \geq 2$ is constant.

• Implication: Varieties that are Albanese primitive (admit no higher irrational pencils) have semi-rigid line bundles. This includes regular varieties ($q(X)=0$) and many irregular varieties like abelian varieties.

Hyper-Kähler Applications (Proposition D & Theorem E)

The authors extend these results to line bundles supported on Lagrangian subvarieties $Z \subset X$ of hyper-Kähler manifolds.

• Under formality and spectral sequence degeneration assumptions, $L$ is semi-rigid on $Z$ if and only if $i_*L$ is semi-rigid on $X$.
• Main Example (Theorem E): Let $Y$ be a smooth cubic fourfold and $X = F(Y)$ its Fano variety of lines (a hyper-Kähler fourfold of K3$[2]$-type). Let $Z$ be the surface of lines in a hyperplane section. The moduli space component $\mathcal{M}^\circ_v(X, h)$ containing $i_*\mathcal{O}_Z$ parametrizes semi-rigid sheaves.
  • Consequently, for $n \geq 2$, the direct-sum morphism is an isomorphism near the diagonal.
  • This component is a primitive symplectic variety admitting no symplectic resolution.

Reducibility of Moduli Spaces (Corollary 5.14)

The paper demonstrates that moduli spaces on high-dimensional hyper-Kähler manifolds can be reducible, unlike the surface case.

• By constructing a different Lagrangian immersion $j: Z' \to X$ (from planes in a cubic fivefold), they find a stable sheaf $j_*\mathcal{O}_{Z'}$ with Mukai vector $63v$.
• This sheaf lies on a 42-dimensional irreducible component distinct from the 630-dimensional split component arising from the direct sum of 63 copies of the sheaf from Theorem E.
• Thus, $\mathcal{M}_{63v}(X, h)$ is not irreducible.

5. Significance

1. Unification of Deformation Theory: The paper provides a unified framework linking the formality of DGAs, the geometry of commuting schemes, and the topology of the underlying variety to determine the local structure of moduli spaces.
2. New Class of Sheaves: It introduces "semi-rigid" sheaves, a class that generalizes rigid sheaves (like skyscraper sheaves) but allows for non-trivial deformations while maintaining the "split" nature of their direct sums.
3. Counterexamples to Irreducibility: It challenges the intuition that moduli spaces of stable sheaves on hyper-Kähler manifolds are always irreducible, providing concrete examples of reducible moduli spaces in dimension 4 and higher.
4. Geometric Interpretation: It translates abstract algebraic conditions (decomposable elements in kernels) into concrete geometric properties (existence of irrational pencils), bridging deformation theory with classical algebraic geometry (Castelnuovo–de Franchis theorem).
5. Symplectic Geometry: The results have implications for the existence of symplectic resolutions and the classification of irreducible symplectic varieties, showing that certain components of moduli spaces are "primitive" and cannot be resolved.