🔢 mathematics

Two dimensional versions of the affine Grassmannian and their geometric description

This paper establishes that two-variable generalizations of the affine Grassmannian are representable by ind-schemes for solvable groups and provides a geometric interpretation of these objects in terms of bundles and trivialization data on a smooth surface relative to a flag of subschemes.

Original authors: Andrea Maffei, Valerio Melani, Gabriele Vezzosi

Published 2026-03-11

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Andrea Maffei, Valerio Melani, Gabriele Vezzosi

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 mathematician trying to understand the shape of space, but instead of looking at a flat sheet of paper (2D) or a line (1D), you are looking at a world made of infinite layers of complexity.

This paper, written by Maffei, Melani, and Vezzosi, is about building a new kind of "map" for a very specific, high-dimensional mathematical world. To understand it, let's break it down using some everyday analogies.

1. The Starting Point: The "One-Dimensional" Map

First, the authors talk about something called the Affine Grassmannian. Think of this as a giant, infinite library.

The Setting: Imagine a single line (like a string).
The Problem: You have a bundle of strings (a "G-bundle") wrapped around this line.
The Twist: You are allowed to cut the string at one specific point and look at the "ends" of the cut.
The Map: The Affine Grassmannian is a catalog that lists every possible way you can arrange these strings, given that you know what they look like everywhere except that one cut point.

In the world of curves (1D), mathematicians have known how to build this catalog for a long time. It's a well-organized library where every book has a specific shelf.

2. The New Challenge: The "Two-Dimensional" Map

Now, the authors ask: What happens if we move from a line to a surface (like a piece of paper)?

Instead of cutting a string at one point, imagine you have a sheet of paper. You draw a line on it (like a river), and then you pick a specific spot on that river (like a bridge).

The Complexity: In 1D, you only had to worry about "left" and "right" of the cut. In 2D, you have "up, down, left, right," and now you are dealing with two variables (like $x$ and $y$ coordinates).
The Confusion: When you try to build the catalog for this 2D surface, things get messy. The "shelves" of the library don't line up neatly anymore. The rules that worked for the line break down.

The paper introduces five different versions of this 2D map. Think of them as five different ways to organize the library:

The Loop Version: Looking at the surface as if you are looping around it.
The Jet Version: Looking at the surface as if you are zooming in infinitely close to the bridge.
The Big Version: Looking at the whole surface at once.
The Local Field Version: A very specific, technical way of looking at the "valley" around the bridge.
The Mixed Version: A combination of the above.

3. The First Big Discovery: "Can We Organize the Mess?"

The authors' first major result (Theorem A) is about Solvable Groups.

The Metaphor: Imagine the "strings" (the mathematical objects) are made of a very flexible, easy-to-fold material (like origami paper). If the material is "solvable" (meaning it follows simple, predictable folding rules), the authors prove that yes, we can organize the library.
The Result: Even though the shelves are weird and infinite, they can still be built as a giant, step-by-step structure (called an ind-scheme). It's like building a skyscraper where each floor is a normal room, but the building goes up forever.
The Catch: If the material is "complicated" (like a tangled ball of yarn that doesn't follow simple rules), they couldn't prove the library can be organized. They had to stick to the "easy" materials for this proof.

4. The Second Big Discovery: "The Universal Blueprint"

The second major result (Theorem B and C) is about Geometric Interpretation.

The Metaphor: Imagine you have a blueprint for a house. Usually, blueprints are abstract. But the authors found a way to say: "This abstract blueprint is exactly the same as a real house built on a specific plot of land."
The Plot of Land: They chose a very specific, simple plot of land: The flat plane ( $\mathbb{A}^2$ ), with a line drawn on it ( $y=0$ ) and a point on that line ( $x=0, y=0$ ).
The Magic: They proved that for almost all their 2D maps, it doesn't matter if you are looking at a complex, curved surface or a simple flat plane. If you look at the "bridge" and the "river" correctly, the catalog of possibilities is identical.
Why it matters: This is huge. It means you don't need to study every possible weird surface in the universe. You can just study this one simple, flat surface, and the answers will apply everywhere else. It's like realizing that the rules of gravity on a mountain are the same as on a beach; you just need to look at the right angle.

5. The "Loophole"

There is one map in their list (the "Loop" Grassmannian) where they couldn't prove the "Universal Blueprint" works for every type of material. They could only prove it works for the "easy" materials (like $GL_n$ or solvable groups) and only for the simplest points. It's like they found a door in the library that is locked for the complicated yarn, but open for the origami paper.

Summary: Why Should You Care?

This paper is a bridge between abstract algebra and geometry.

For the Mathematician: It solves a long-standing problem of how to define these "2D Grassmannians" and proves they are well-behaved (organizable) for a large class of groups.
For the General Audience: It's a story about finding order in chaos. The authors took a confusing, high-dimensional mathematical problem, realized that "simple materials" can be organized into a perfect structure, and discovered that the rules for a complex, curved world are actually the same as the rules for a simple, flat world.

They essentially built a universal translator that allows mathematicians to take problems from complex surfaces and translate them into simple, solvable problems on a flat plane. This is a crucial step for fields like Geometric Representation Theory and the Geometric Langlands Program, which try to connect the shapes of space with the symmetries of numbers.

1. Problem Statement and Motivation

The paper addresses the generalization of the Affine Grassmannian ( $Gr_G$ ) from the setting of algebraic curves to algebraic surfaces.

Context: For a smooth affine algebraic group $G$ over an algebraically closed field $k$ , the classical affine Grassmannian is the fppf sheaf $Gr_G = LG/JG$ , where $LG(R) = G(R((t)))$ is the loop group and $JG(R) = G(R[[t]])$ is the jet group. It admits a geometric interpretation via the Beauville-Laszlo theorem as the moduli space of $G$ -bundles on a curve $C$ trivialized away from a point $x$ .
The Gap: While the one-dimensional case is well-understood (ind-representable, central to geometric Langlands), the two-dimensional case (involving two variables, e.g., $x$ and $y$ ) lacks a unified framework. Various "double loop" quotients exist in the literature (e.g., $G(R((x))((y))) / G(R[[x]]((y)))$ ), but their representability as ind-schemes and their geometric interpretation via bundles on surfaces were not fully established, particularly for general groups and flags.
Goal: The authors aim to:
1. Define several natural two-variable generalizations of the affine Grassmannian.
2. Prove their ind-representability (specifically for solvable groups).
3. Provide a geometric description of these objects in terms of $G$ -bundles on a surface $X$ relative to a flag of subschemes $(D, Z)$ .

2. Methodology

The authors employ a combination of algebraic geometry, ind-scheme theory, and fiber functor formalism.

A. Algebraic and Ind-Scheme Approach (Quotient Grassmannians)

The authors define five distinct presheaf quotients based on different combinations of loop ( $L$ ) and jet ( $J$ ) functors applied to the double loop group $LLG(R) = G(R((x))((y)))$:

$Gr^L_G$ (Loop Group Grassmannian): $LLG / JLG$.
$LGr_G$ (Loop of Affine Grassmannian): $LLG / LJG$.
$Gr^{big}_G$ (Big Grassmannian): $LLG / JJG$.
$Gr^{(2)}_G$ (2D Local Field Grassmannian): $LLG / G^{(2)}$ , where $G^{(2)}$ corresponds to a specific valuation ring in the 2D local field theory.
$Gr^J_G$ (Jet Group Grassmannian): $LJG / JJG$.

Technique for Representability:
To prove these are ind-schemes, the authors rely on finding normal forms (canonical representatives) for elements in the quotients.

They first prove the result for the additive group ( $G_a$ ) and multiplicative group ( $G_m$ ).
They use Lemma 1.10 and Lemma 1.11 to reduce the problem for any solvable group $G$ to the case of $G_m$ and $G_a$ . This involves decomposing solvable groups into unipotent radicals and tori, utilizing the fact that unipotent groups admit sections over affine bases.
For $G_m$ , they construct explicit ind-affine subschemes $\Sigma \subset LLG$ such that every coset in the quotient has a unique representative in $\Sigma$ . This relies on analyzing nilpotent elements in Laurent series rings.

B. Geometric Approach (Geometric Grassmannians)

To provide a geometric interpretation, the authors introduce fiber functors (following the formalism of [9]).

Setup: Let $X$ be a smooth quasi-projective surface, $D \subset X$ an effective Cartier divisor, and $Z \subset D$ a closed subscheme of codimension 1.
Construction: They define fiber functors corresponding to formal completions and punctured formal neighborhoods (e.g., $\hat{Z}$ , $\hat{Z} \setminus D$ , etc.).
Definition: The Geometric Grassmannians are defined as fiber products of bundle stacks:
$Gr_{(F, \hat{F})} = Bun_F \times_{Bun_{\hat{F}}} \{*\}$
where $F$ and $\hat{F}$ are specific fiber functors representing different loci on the surface (e.g., trivialization on a punctured neighborhood vs. a bundle on a formal neighborhood).

C. Comparison Strategy

The core of the paper is establishing an equivalence between the Quotient Grassmannians (algebraic definition) and the Geometric Grassmannians (geometric definition).

Reduction: They prove that for any smooth surface $X$ and flag $(D, Z)$ where $Z$ is a single point, the geometric Grassmannian is isomorphic to the one defined on the affine plane $\mathbb{A}^2_k$ with the standard flag.
Local Analysis: Using étale neighborhoods and properties of formal completions (Lemma 3.4, 3.5), they show that the local geometry of $(X, D, Z)$ is equivalent to $(\mathbb{A}^2, \{y=0\}, \{0\})$ .
Verification: They explicitly compute the $k$ -points and sheafifications for the case $X = \mathbb{A}^2$ to match the algebraic definitions.

3. Key Contributions and Results

Theorem A: Ind-Representability for Solvable Groups

For a solvable group $G$ , all five two-dimensional affine Grassmannians ( $Gr^L_G, LGr_G, Gr^{big}_G, Gr^J_G, Gr^{(2)}_G$ ) are ind-schemes (specifically ind-affine).

Significance: This establishes that these objects are well-behaved geometric objects, not just abstract presheaves. The proof relies on the existence of unique normal forms in the quotients.
Note: Unlike the classical case (where $Gr_G$ is ind-projective for reductive $G$ ), these 2D versions are generally not ind-finite type and are not indexed by countable posets.

Theorem B: Geometric Interpretation on $\mathbb{A}^2$

For $X = \mathbb{A}^2_k$ , $D = \{y=0\}$ , and $Z = \{0\}$ , there are canonical isomorphisms of fppf sheaves:
$LGr_G \simeq LGr_{G; (\mathbb{A}^2, D, Z)}$
$Gr^{big}_G \simeq Gr^{big}_{G; (\mathbb{A}^2, D, Z)}$
$Gr^J_G \simeq Gr^J_{G; (\mathbb{A}^2, D, Z)}$
$Gr^{(2)}_G \simeq Gr^{(2)}_{G; (\mathbb{A}^2, D, Z)}$

Exception: The isomorphism for the loop group Grassmannian $Gr^L_G$ is only proven for $k$ -points (and specific groups like $GL_n, SL_n$ ) in Proposition 2.16, not as a full sheaf isomorphism.

Theorem C: Generalization to Arbitrary Surfaces

For any smooth quasi-projective surface $X$ and a flag $(D, Z)$ where $Z$ is a single smooth point, the geometric Grassmannians are non-canonically isomorphic to the quotient Grassmannians defined on $\mathbb{A}^2$ .

This extends the geometric interpretation from the local model ( $\mathbb{A}^2$ ) to arbitrary surfaces.

Corollary D: Ind-Representability for Geometric Grassmannians

Combining Theorem A and Theorem C, the authors conclude that for solvable $G$ , the geometric Grassmannians on any smooth surface $X$ (with $Z$ a finite set of points) are represented by ind-schemes.

4. Significance and Future Directions

Foundational Work: This paper provides the first rigorous geometric description of 2D affine Grassmannians, linking abstract loop group quotients to concrete moduli problems of bundles on surfaces.
Geometric Langlands: The objects studied are potential candidates for a 2-dimensional version of the Geometric Langlands program. The authors suggest that the 2D local field Grassmannian ( $Gr^{(2)}_G$ ) is a promising candidate for hosting a 2D Geometric Satake equivalence.
Higher Reciprocity: The framework connects to Parshin's higher reciprocity laws on surfaces, suggesting a link between these Grassmannians and number-theoretic properties of surfaces.
Limitations & Open Questions:
- The ind-representability proof is currently restricted to solvable groups. Extending this to reductive groups (like $GL_n$ ) remains an open problem requiring new techniques, as the normal form arguments used for solvable groups do not directly generalize.
- The full geometric interpretation of the Loop Grassmannian ( $Gr^L_G$ ) as a sheaf (beyond $k$ -points) is not yet established.
- The authors plan to study the flag factorization property (analogous to the Beilinson-Drinfeld Grassmannian for curves) in future work, particularly for the jet Grassmannian.

In summary, the paper successfully constructs a bridge between the algebraic theory of double loop groups and the geometric theory of bundles on surfaces, establishing a robust framework for "2D Affine Grassmannians" in the solvable case.