🔢 mathematics

Lagrangian Intersections, Symplectic Reduction and Kirwan Surjectivity

This paper establishes that $G$ -invariant Lagrangians in a holomorphic symplectic variety induce Lagrangians in the symplectic quotient with an isomorphism between their fiber product and the shifted cotangent bundle of their intersection, leading to derived Ext group computations and a generalized Kirwan surjectivity theorem interpreted as the commutativity of symmetry and reduction in the 3d B-model.

Original authors: Naichung Conan Leung, Ying Xie, Yu Tung Yau

Published 2026-02-13

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Naichung Conan Leung, Ying Xie, Yu Tung Yau

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

Imagine you are a master architect working in a universe governed by perfect, invisible rules of balance and symmetry. This universe is a Symplectic Variety (let's call it "The Grand Hall").

In this Grand Hall, there are special, flat surfaces called Lagrangians. Think of these as giant, perfectly smooth sheets of fabric floating in the air. They are special because they follow the specific laws of the Hall's geometry.

Now, imagine a giant, invisible force (a Group $G$ ) that spins, stretches, and rotates the entire Grand Hall. Because the Hall is so perfectly balanced, this spinning doesn't tear the fabric; it just moves the sheets around in a synchronized dance.

This paper is about what happens when we take a snapshot of this dance, freeze the spinning, and look at the "shadow" or "reduced" version of the Hall.

Here is the breakdown of the paper's big ideas using simple analogies:

1. The Setup: The Dance and the Shadow

The Grand Hall ( $X$ ): A complex, high-dimensional space with a special "symplectic" structure (a rule that defines how things move and interact).
The Spinners ( $G$ ): A group of symmetries (like a rotation or a scaling) acting on the Hall.
The Sheets ( $C_1, C_2$ ): Two specific Lagrangian sheets floating in the Hall. They are "invariant," meaning if you spin the Hall, the sheets just move along with it; they don't get crumpled.
The Intersection ( $B$ ): Where these two sheets cross each other. In a perfect world, they might cross cleanly (like two roads intersecting at a flat roundabout) rather than tangling messily.

2. The Magic Trick: Symplectic Reduction

The authors ask: What happens if we "squash" the Grand Hall down to remove the spinning motion?

In math, this is called Symplectic Reduction. Imagine taking a 3D sculpture of a spinning dancer and pressing it flat against a wall to get a 2D shadow.

The Grand Hall becomes a smaller, simpler space called the Symplectic Quotient ( $X // G$ ).
The two floating sheets ( $C_1$ and $C_2$ ) become two new sheets in this smaller shadow world ( $C_1/G$ and $C_2/G$ ).

The Big Question: If the original sheets crossed each other in a specific, "clean" way, do their shadows also cross in a predictable, clean way?

3. The Discovery: The "Ghost" Intersection

The paper proves a beautiful result: Yes, they do.

If the original sheets crossed cleanly, their shadows in the reduced world also cross cleanly. But here is the twist: the intersection in the shadow world isn't just a simple point or line. It's a "fuzzy" or "derived" object.

The Analogy:
Imagine two roads crossing. In the real world, they meet at a flat intersection. In the "shadow" world (after reduction), the intersection looks like a road that has a ghostly, extra dimension attached to it.

The authors show that this "fuzzy" intersection is mathematically identical to the cotangent bundle of the original crossing point.
In plain English: The complex geometry of the intersection in the reduced world is perfectly described by the "slopes" and "directions" available at the place where the original sheets met.

4. Counting the Intersections (The "Ext" Groups)

Mathematicians love to count things. They want to know: How many ways can these two sheets interact?
In the language of the paper, they calculate something called Ext groups.

Without Reduction: Counting interactions in the big, spinning Hall is hard.
With Reduction: The paper shows that counting interactions in the shadow world is actually easier. It turns out that the number of interactions is exactly equal to the cohomology (a fancy way of counting holes, loops, and shapes) of the intersection point, but with a special "twist."

The "Twist" (Local Systems):
Sometimes, the sheets have a hidden "spin" (like a top spinning on its own axis). If the sheets are "spin" sheets, the intersection gets a special "color" or "label" (a local system). The paper shows that even with this color, the counting formula still works perfectly. It's like saying, "Even if the roads are painted different colors, the number of ways cars can cross is still predictable."

5. The Kirwan Surjectivity: The "One-Way Street"

The final part of the paper deals with a concept called Kirwan Surjectivity.

Imagine you have a map of the whole Grand Hall ( $X$ ) and a map of the "safe zone" or "stable zone" ( $X^{ss}$ ) where the spinning is well-behaved.
There is a natural map that takes information from the whole Hall and projects it onto the safe zone.
The Result: The authors prove that this map is surjective.
The Metaphor: Imagine you have a library with every book ever written (the whole Hall). You want to know if you can find a copy of every book in the "Bestsellers" section (the safe zone).
- Surjectivity means: Yes. Every "Bestseller" in the safe zone has a corresponding book in the main library. You don't lose any information when you move from the complex, spinning world to the simpler, stable world. You can always "lift" a solution from the simple world back to the complex world.

Why Does This Matter?

This paper connects three huge fields:

Symplectic Geometry: The study of shapes and motion.
Algebraic Geometry: The study of equations and shapes.
Physics (String Theory): Specifically, the "3D B-model," which describes how particles (branes) interact.

The Takeaway:
The authors found a universal rule. If you have two symmetrical, spinning objects that cross cleanly, you can predict exactly how their "shadows" will interact in a reduced world. Furthermore, you can count these interactions using simple topological tools (like counting holes), even if the objects have complex "spins."

It's like discovering that no matter how complex the dance of the universe gets, if you look at the footprints left on the floor, they follow a simple, beautiful, and predictable pattern. This helps physicists and mathematicians build better models of the universe and the "Fukaya category" (a mathematical structure used to organize these shapes).

1. Problem Statement and Context

The paper addresses the computation of Ext groups between $G$ -invariant Lagrangian subvarieties within the context of symplectic reduction and derived algebraic geometry.

Mathematical Setting: Let $(X, \omega)$ be a smooth holomorphic symplectic variety equipped with a Hamiltonian action of a complex linear reductive group $G$ . Let $C_1, C_2 \subset X$ be smooth, connected, $G$ -invariant Lagrangian subvarieties.
The Core Question: How can one compute the derived intersection and the associated Ext groups $\text{Ext}^\bullet_{X//G}(\mathcal{O}_{C_1/G}, \mathcal{O}_{C_2/G})$ in the symplectic quotient stack $X//G$ (also denoted $X \mathbin{/\mkern-6mu/}_c G$ )?
Motivation: This problem is central to:
1. Constructing the Fukaya category of complex Lagrangians.
2. Understanding the 3d B-model 2-category proposed by Kapustin, Rozansky, and Saulina.
3. Generalizing the Kirwan surjectivity theorem (which relates equivariant cohomology of a space to its semi-stable quotient) to the context of Lagrangian intersections and Ext groups.

2. Methodology and Framework

The authors employ a synthesis of derived algebraic geometry, symplectic geometry, and Geometric Invariant Theory (GIT).

Derived Fiber Products: Instead of treating intersections as classical schemes, the authors utilize the derived fiber product $C_1 \times_X C_2$ . They analyze the structure of this derived stack, specifically looking for a "fibration structure" where the derived intersection is isomorphic to the total space of a shifted cotangent bundle (or a perfect complex).
Excess Bundles and Splitting Conditions: A crucial technical condition is the splitting of the conormal sequence. For the intersection $B = C_1 \cap C_2$ , the sequence $0 \to N^\vee_{B/X} \to \Omega_X|_B \to \Omega_B \to 0$ must split as $G$ -equivariant bundles. This condition ensures that the derived intersection behaves like a vector bundle stack, allowing for explicit calculations.
Symplectic Reduction as a 0-Shifted Symplectic Stack: The symplectic quotient $X//G$ is treated as a 0-shifted symplectic Artin stack. The $G$ -invariant Lagrangians $C_i$ induce Lagrangian morphisms $C_i/G \to X//G$ .
Local Systems and Torsion: The authors investigate cases where the determinant of the normal bundle, $\det(N_{B/C_2})$ , is a torsion line bundle. This allows the construction of local systems (specifically $n$ -torsion $\mathbb{C}$ -local systems) which twist the cohomology calculations.
Twisted Kirwan Map: The paper extends the classical Kirwan map (from equivariant cohomology of $X$ to $X^{ss}$ ) to a twisted Kirwan map involving local coefficients. The surjectivity of this map is proven using the HKKN (Hesselink-Kempf-Kirwan-Ness) stratification of the unstable locus.

3. Key Contributions and Results

A. Geometry of Derived Lagrangian Intersections (Theorem 1.1 / Theorem 4.5)

The primary geometric result establishes the structure of the derived intersection in the quotient.

Result: If the conormal sequence of $B = C_1 \cap C_2$ splits, the derived fiber product in the symplectic quotient is isomorphic to the $(-1)$ -shifted cotangent stack of the quotient intersection:
$C_1/G \times_{X//G} C_2/G \cong \text{Tot}_{B/G}(L_{B/G}[-1])$
where $L_{B/G}$ is the cotangent complex of the quotient stack $B/G$ .
Significance: This reduces the complex derived intersection to a manageable geometric object (a shifted cotangent bundle), facilitating cohomological computations.

B. Computation of Ext Groups (Theorem 1.1 / Corollary 4.7)

The paper provides explicit formulas for the Ext groups in terms of equivariant cohomology.

General Case: Under the splitting condition, the Ext groups are isomorphic to a direct sum of cohomology groups involving the exterior powers of the cotangent complex and the determinant of the normal bundle.
Torsion Case: If $\det(N_{B/C_2})$ is a torsion line bundle (defining a local system $L^\delta$ ), and assuming the Hodge-to-de Rham spectral sequence degenerates (guaranteed if $B$ is proper-over-affine and $h^0(B, \mathcal{O})^G < \infty$ ), the Ext groups simplify to:
$\text{Ext}^\bullet_{X//G}(\mathcal{O}_{C_1/G}, \mathcal{O}_{C_2/G}) \cong H^\bullet_G(B, \det(N_{B/C_2})^\delta)$
Spin Case (Theorem 1.2 / Theorem 4.9): If the Lagrangians admit square roots of their canonical bundles (equivariant spin structures), the Ext groups between the square roots of the structure sheaves are computed via a 2-torsion local system arising from the spin structures and the normal bundle.

C. Generalized Kirwan Surjectivity (Theorem 1.3 / Theorems 5.5–5.7)

The authors prove that the restriction map from the full symplectic quotient to the semi-stable symplectic quotient is surjective on Ext groups.

Setup: Let $X^{ss}$ be the semi-stable locus defined by an ample $G$ -linearization. There is a natural restriction map:
$\text{res}: \text{Ext}^\bullet_{X//G}(\mathcal{O}_{C_1/G}, \mathcal{O}_{C_2/G}) \to \text{Ext}^\bullet_{X^{ss}//G}(\mathcal{O}_{C_1^{ss}/G}, \mathcal{O}_{C_2^{ss}/G})$
Result: If the conormal sequence splits and the relevant cohomological conditions are met, this map is surjective.
Interpretation: This is interpreted as a "symmetry commutes with reduction" problem in the 3d B-model. When $C_1 = C_2$ , this recovers the classical Kirwan surjectivity theorem but interpreted through the lens of Lagrangian intersections and Ext algebras.

D. Specific Examples

The theory is applied to:

Cotangent Bundles ( $X = T^\vee M$ ): The zero section and graphs of closed 1-forms.
Morse-Bott Functions: Where the intersection is the critical locus of a function.
Conormal Varieties: Intersections of conormal bundles to subvarieties.

4. Significance and Impact

Unification of Geometry and Physics: The paper bridges the gap between the mathematical construction of the Fukaya category (via derived intersections) and the physical 3d B-model (via Ext groups and Lagrangian branes). It provides a rigorous mathematical framework for the "symmetry commutes with reduction" principle in quantum field theory.
Extension of Kirwan Surjectivity: By generalizing Kirwan surjectivity to Ext groups and twisted coefficients (local systems), the authors open new avenues for studying the topology of symplectic quotients and the structure of their associated categories.
Handling Non-Transverse Intersections: The use of derived algebraic geometry allows the authors to handle "clean" but non-transverse intersections systematically, identifying them with shifted cotangent bundles. This is essential for applications in mirror symmetry where transversality is often not guaranteed.
Spin Structures and Local Systems: The explicit identification of 2-torsion local systems arising from spin structures on Lagrangians connects the algebraic geometry of the intersection to the physical requirement of spin structures for worldsheet fermions in string theory.

In summary, the paper provides a powerful computational toolkit for equivariant Lagrangian intersections, proving that under natural splitting conditions, these complex derived objects are governed by the equivariant cohomology of the intersection locus, twisted by specific local systems, and that these structures behave well under symplectic reduction.