On a decomposition of $p$-adic Coxeter orbits

This paper establishes that for classical unramified reductive groups over non-archimedean local fields, specific $p$-adic Deligne–Lusztig spaces $X_w(b)$ associated with basic elements $b$ and Coxeter elements $w$ decompose into disjoint unions of translates of integral $p$-adic Deligne–Lusztig spaces, while also extending results on rational conjugacy classes of unramified tori and proving a loop version of Frobenius-twisted Steinberg's cross section.

The Gist

Imagine you are trying to understand the shape of a vast, invisible city built in a strange, mathematical dimension. This city is called a p-adic Deligne–Lusztig space. It's not a city made of brick and mortar, but of pure algebra and geometry, designed to help mathematicians decode the "music" of symmetry groups (specifically, groups of numbers that behave like p-adic numbers).

For a long time, these spaces were like foggy, sprawling mazes. Mathematicians knew they existed and that they held the keys to understanding deep secrets of number theory (like the Langlands program), but they were too complex to map out clearly. They were "infinite-dimensional," meaning they had too many directions to count, making them impossible to draw or analyze with standard tools.

The Big Breakthrough
In this paper, the author, Alexander Ivanov, acts like a master urban planner who finally finds the blueprint. He proves that for a specific, very important class of these spaces (those associated with "Coxeter" elements, which are like the most efficient, rotating gears in the machine), the fog clears.

He discovers that these massive, confusing spaces aren't actually one giant, messy blob. Instead, they are made of disjoint Lego bricks.

Here is the core idea broken down with simple analogies:

1. The "Foggy City" vs. The "Lego Set"

• The Problem: Imagine trying to describe a city that stretches infinitely in every direction, where the streets keep branching into new dimensions. It's hard to study because you can't see the whole picture at once.
• The Solution: Ivanov shows that this "city" is actually just a collection of identical, smaller, and much simpler neighborhoods (which he calls integral p-adic Deligne–Lusztig spaces).
• The Analogy: Think of a giant, complex tapestry. For years, people thought it was a single, woven piece of fabric that was too intricate to understand. Ivanov proves that if you look closely, the tapestry is actually just a grid of identical, simple squares stitched together. If you understand one square, you understand the whole tapestry.

2. The "Translation" Trick

The paper shows that you can slide these simple squares around to cover the entire space.

• The Metaphor: Imagine a floor covered in a complex, swirling pattern. Ivanov proves that this pattern is actually just a single, simple tile that has been stamped over and over again in different positions.
• Why it matters: These "tiles" are affine schemes. In the language of math, this means they are "nice" and "manageable." They are like a standard, flat sheet of paper, whereas the original space was a crumpled, infinite-dimensional ball. By breaking the ball down into flat sheets, mathematicians can finally do calculations, measure things, and predict how the space behaves.

3. The "Ghost" and the "Shadow"

The paper also deals with a concept called rational conjugacy classes.

• The Analogy: Imagine you have a group of identical twins (the "stable" class). To the naked eye, they look exactly the same. However, if you look at them through a specific "lens" (the rational conjugacy), you can tell them apart because they have different "shadows" or histories.
• The Discovery: Ivanov shows that his new "Lego brick" decomposition allows the space to "see" these differences. It can distinguish between the twins that were previously indistinguishable. This is crucial because in the world of number theory, these tiny differences often correspond to different solutions to major equations.

4. The "Cross Section" Shortcut

The paper also introduces a new version of a famous mathematical tool called Steinberg's cross section.

• The Metaphor: Imagine you are trying to cross a wide, turbulent river. Usually, you have to build a massive bridge or swim the whole way. Steinberg's cross section is like a magical, invisible stepping-stone path that lets you walk straight across the center of the river without getting wet.
• The Innovation: Ivanov creates a "loop version" of this stepping-stone path. It allows him to cut through the complex algebraic river of his problem, proving that the "Lego bricks" fit together perfectly without any gaps or overlaps.

Why Should You Care?

You might ask, "Who cares about these infinite-dimensional Lego cities?"

These spaces are the bridge between two giant worlds of mathematics:

1. Geometry: The study of shapes and spaces.
2. Number Theory: The study of prime numbers and equations.

By proving that these spaces are made of simple, manageable pieces, Ivanov has given mathematicians a new toolkit. This allows them to:

• Calculate Cohomology: This is a fancy way of counting the "holes" and "loops" in these shapes, which reveals hidden symmetries.
• Solve the Langlands Program: This is often called the "Grand Unified Theory" of mathematics. It tries to connect the world of numbers to the world of symmetry. Ivanov's work provides a clearer map for navigating this connection, specifically for a class of groups that are fundamental to physics and cryptography.

In a Nutshell:
Alexander Ivanov took a mathematical object that was too big and messy to understand, and showed that it is actually just a neat, repeating pattern of simple, clean shapes. He didn't just describe the pattern; he gave everyone the instructions on how to build it, measure it, and use it to unlock deeper secrets of the universe of numbers.

Technical Summary

Here is a detailed technical summary of the paper "On a decomposition of p-adic Coxeter orbits" by Alexander B. Ivanov.

1. Problem Statement and Context

The Object of Study:
The paper investigates the geometry of p-adic Deligne–Lusztig spaces, denoted $X_w(b)$ and their covers $\dot{X}_{\bar{w}}(b)$. These spaces were introduced by the author in a previous work (Crelle, 2023) attached to an unramified reductive group $G$ over a non-archimedean local field $k$.

• Definition: $X_w(b)$ is defined as the intersection of the loop space of a Schubert variety $LO(w) \subseteq L(G/B)$ with the graph of the twisted Frobenius map $b\sigma: L(G/B) \to L(G/B)$.
• Nature: Unlike classical Deligne–Lusztig varieties (which are schemes of finite type over finite fields), these p-adic spaces are sheaves for the arc-topology on perfect algebras. Their representability is subtle; while often ind-representable, it was an open question whether they are schemes in specific cases.

The Specific Challenge:
The author focuses on the case where:

1. $G$ is a classical group (types $A_n, B_n, C_n, D_n$, and their twisted forms).
2. $c \in W$ is a (twisted) Coxeter element.
3. $b \in G(\breve{k})$ is a basic element (its Newton point factors through the center).

The central problem is to understand the geometric structure of $X_c(b)$ and $\dot{X}_{\bar{c}}(b)$. Specifically, the author aims to prove that these spaces admit a decomposition into simpler, well-understood components, thereby resolving Conjecture 1.1 from the author's previous work, which posited that these spaces are schemes (specifically, disjoint unions of affine schemes) in the Coxeter case.

2. Methodology

The proof combines several advanced techniques from arithmetic geometry, representation theory, and the theory of isocrystals:

• Decomposition via Parahoric Models: The strategy involves decomposing the infinite-dimensional spaces $X_c(b)$ into disjoint unions of translates of integral p-adic Deligne–Lusztig spaces ($X^G_x$). These integral spaces are defined using parahoric models $G_x$ associated with the unique fixed point $x$ of the affine transformation $b\sigma$ on the apartment.
• Loop Versions of Steinberg's Cross Section: A crucial technical tool is the proof of a "loop version" of Steinberg's cross section. This establishes an isomorphism between products of loop spaces of unipotent subgroups and a larger loop space, allowing the reduction of the complex space $X_c(b)$ to a more tractable form involving the intersection $cU \cap U^-$.
• Newton Polygons and Isocrystals: To prove that the decomposition covers the entire space (i.e., that no points are "missing" from the integral level), the author employs a detailed analysis of Newton polygons. By interpreting the defining equations of the spaces as conditions on isocrystals, the author derives valuation estimates for the coefficients of the elements in the unipotent radical. This ensures that the elements lie within the required parahoric subgroup.
• Rational Conjugacy Classes: The paper extends the work of DeBacker and Reeder on the parametrization of rational conjugacy classes of unramified tori. It establishes a bijection between the connected components of the covering space $\dot{X}_{\bar{c}}(b)$ and the rational conjugacy classes of unramified Coxeter tori in the inner form $G_b$.
• Descent Arguments: The proof utilizes $v$-descent (a topology finer than the étale topology) to descend results from the adjoint group to the general reductive group and from the absolutely almost simple case to products.

3. Key Contributions and Results

A. Main Theorem (Theorem 1.1)

The paper proves that for a classical unramified group $G$, a Coxeter element $c$, and a basic element $b$, the spaces admit the following equivariant decompositions:

1. For the base space:
   $$X_c(b) \cong \bigsqcup_{\gamma \in G^{\text{ad}}_b(k) / G^{\text{ad}}_{x,b}(\mathcal{O}_k)} \gamma X^{\text{ad}}_{x, c, b}$$
2. For the covering space:
   $$\dot{X}_{\bar{c}}(b) \cong \bigsqcup_{\gamma \in G_b(k) / G_{x,b}(\mathcal{O}_k)} \gamma \dot{X}^{G_x}_{\bar{c}, b}$$

Here, $X^{\text{ad}}_{x, c, b}$ and $\dot{X}^{G_x}_{\bar{c}, b}$ are integral level Deligne–Lusztig spaces.

• Significance: These integral spaces are affine schemes. Consequently, $X_c(b)$ and $\dot{X}_{\bar{c}}(b)$ are shown to be disjoint unions of affine schemes, proving they are schemes (resolving Conjecture 1.1).

B. Steinberg's Cross Section (Section 5)

The author proves a loop-theoretic analogue of Steinberg's cross section theorem. Specifically, for classical groups and special Coxeter elements, the map:
$$\alpha_b: L(cU \cap U) \times L(cU \cap U^-) \to L(cU), \quad (x, y) \mapsto x^{-1}y\sigma_b(x)$$
is an isomorphism. This result is non-trivial because the standard Ax–Grothendieck theorem (used in finite field settings) is not applicable in the loop setting where the spaces are not schemes of finite type. The author provides a new argument using filtrations of root subgroups and properties of the Frobenius map on Witt vectors.

C. Parametrization of Rational Conjugacy Classes (Section 3)

The paper extends the parametrization of rational conjugacy classes of unramified maximal tori to extended pure inner forms.

• It establishes that the set of non-empty pieces of the cover $\dot{X}_{\bar{c}}(b)$ is in canonical bijection with the set of rational conjugacy classes of unramified Coxeter tori in the inner form $G_b$.
• This clarifies how the p-adic Deligne–Lusztig space "detects" the difference between various rational conjugacy classes that share the same stable class (determined by $c$).

D. Estimates via Newton Polygons (Section 7)

The most technical part of the paper involves a case-by-case analysis (Types $A_n, B_n, C_n, D_n, {}^2A_n, {}^2D_n$) to prove that the Lang map restricted to the relevant spaces factors through the integral parahoric model.

• The author constructs explicit isocrystals associated with the elements of the space.
• By analyzing the slopes of these isocrystals and applying Lemma 7.3 (relating Newton polygons to the valuations of coefficients in the characteristic polynomial), the author proves that the coefficients $a_i$ of the unipotent elements satisfy specific valuation bounds ($\text{ord}_\varpi(a_i) \geq \dots$).
• These bounds confirm that the elements lie in the parahoric subgroup $G_x$, ensuring the surjectivity of the decomposition.

4. Significance and Impact

1. Resolution of Conjecture: The paper definitively proves that p-adic Deligne–Lusztig spaces for Coxeter elements in classical groups are schemes, specifically disjoint unions of affine schemes. This removes a major ambiguity in the theory.
2. Bridge to Local Langlands: By establishing that these spaces are schemes and providing an explicit decomposition, the paper facilitates the computation of their cohomology. This is a critical step toward understanding the local Langlands correspondence for $p$-adic groups, as the cohomology of these spaces is expected to realize representations of $G(k)$.
3. Connection to Affine Deligne–Lusztig Varieties: The decomposition result bears a striking resemblance to decomposition theorems for affine Deligne–Lusztig varieties (He–Nie–Yu). The author suggests a deeper relationship, noting that $X_c(b)$ can be viewed as an inverse limit of affine Deligne–Lusztig varieties, generalizing previous results for $GL_n$.
4. Methodological Innovation: The development of a "loop version" of Steinberg's cross section without relying on the Ax–Grothendieck theorem provides a new toolkit for studying infinite-dimensional spaces in $p$-adic geometry. The use of Newton polygon estimates to control integral structures is also a powerful technique for future applications in the study of Rapoport–Zink spaces and Shimura varieties.

In summary, Ivanov's work provides a rigorous geometric foundation for p-adic Deligne–Lusztig theory in the Coxeter case, transforming these objects from abstract sheaves into concrete geometric objects (schemes) amenable to cohomological analysis and representation-theoretic applications.