Endoscopic transfer and the wavefront upper bound conjecture

This paper verifies the local analogue of Jiang's conjecture regarding the upper bound of geometric wavefront sets for Arthur type representations of split classical $p$-adic groups with large $p$, thereby establishing related upper bound conjectures by combining Waldspurger's endoscopic transfer with recent wavefront set computations.

The Gist

Imagine you are trying to understand the "shape" of a complex, invisible object floating in a dark room. You can't see the object itself, but you can shine a flashlight on it and see the shadows it casts on the wall. In the world of advanced mathematics (specifically, the study of symmetries in number theory), these "objects" are called representations, and the "shadows" they cast are called wavefront sets.

This paper by Hiraku Atobe and Dan Ciubotaru is like a detective story where the authors figure out exactly how big and what shape the largest shadow of a specific type of object can be.

Here is the breakdown using simple analogies:

1. The Cast of Characters

• The Group (The Machine): Think of a "split classical group" (like $SO$ or $Sp$) as a giant, complex machine with many moving parts. It represents a specific kind of symmetry.
• The Representations (The Employees): Inside this machine, there are different "employees" (mathematical representations) doing specific jobs. Some are "tempered" (steady and calm), while others are more chaotic.
• The A-Parameters (The ID Cards): Every employee has an ID card called an A-parameter. This card tells us exactly what kind of job they are supposed to do. It's like a blueprint or a recipe.
• The Wavefront Set (The Shadow): When an employee works, they cast a shadow on the wall. This shadow is the wavefront set. It tells us the "largest" or most dominant features of that employee's work. The paper is interested in the biggest part of this shadow.

2. The Big Question (The Conjecture)

For a long time, mathematicians had a hunch (a conjecture) about the relationship between the ID Card (the A-parameter) and the Shadow (the wavefront set).

The Hunch: "If you look at the ID card, you can predict the exact size and shape of the largest shadow the employee will cast. Specifically, the largest shadow should be a specific 'dual' version of the shape described on the ID card."

Think of it like this: If your ID card says you are a "Tall, Wide Person," the hunch is that your shadow on the wall will be the "Short, Narrow" version of that shape (mathematically, this is called the Spaltenstein dual). The authors wanted to prove that no matter how you arrange the employees, the biggest shadow never exceeds this predicted size.

3. The Problem

Proving this is incredibly hard because:

1. The "machines" (groups) are abstract and exist in a world of numbers that don't behave like normal numbers (p-adic fields).
2. There are many employees in a single "packet" (a group of related employees), and they all cast shadows that mix together.
3. The "shadows" are made of tiny, invisible geometric shapes (orbits) that are difficult to measure.

4. The Detective Work (The Proof)

The authors used a clever trick called Endoscopic Transfer.

The Analogy of the Translator:
Imagine you have a difficult puzzle in a foreign language (the complex group $H$). You can't solve it directly. But you know a "translator" (Endoscopy) who can translate this puzzle into a much simpler language (the group $GL_m$, which is like a straight line of numbers).

1. Step 1: Translate to Simple Land. The authors took the complex "ID cards" and the "shadows" from the complicated machine and translated them into the simple world of $GL_m$.
2. Step 2: Solve the Simple Puzzle. In this simple world, the rules are well-known. They proved that in the simple world, the hunch is true: the biggest shadow matches the predicted dual shape.
3. Step 3: Translate Back. They then used a "reverse translator" (Waldspurger's work) to bring the result back to the complex machine.

The "Twist":
To make this work, they had to deal with a "twisted" version of the simple world (involving an involution, or a mirror flip). They had to prove that even with this twist, the shadows still behaved predictably. They used a method called induction, which is like climbing a ladder: they proved it for small machines, then used that proof to show it works for slightly bigger machines, and so on, until they covered all sizes.

5. The Result

The authors successfully proved that:

• The Upper Bound is Real: The largest shadow (wavefront set) of any employee in this specific group of machines never exceeds the size predicted by their ID card.
• It's Exact: For at least one employee in the group, the shadow is exactly the predicted size.

Why Does This Matter?

In the real world, this helps mathematicians understand the deep structure of numbers and symmetries. It connects two different ways of looking at the same mathematical object:

1. The Algebraic View: Looking at the "ID card" (parameters).
2. The Geometric View: Looking at the "shadow" (wavefront set).

By proving these two views are perfectly linked, the authors have removed a major piece of uncertainty in the field of representation theory. It's like finally confirming that if you know the blueprint of a building, you can perfectly predict the shape of its shadow at sunset, no matter how complex the building is.

In summary: The paper uses a "translate-to-simpler-world-and-back" strategy to prove that the biggest geometric "shadows" of certain mathematical objects are strictly controlled by their underlying "blueprints."

Technical Summary

Here is a detailed technical summary of the paper "Endoscopic Transfer and the Wavefront Upper Bound Conjecture" by Hiraku Atobe and Dan Ciubotaru.

1. Problem Statement and Context

The paper addresses the relationship between Arthur packets (collections of irreducible admissible representations of reductive $p$-adic groups) and their geometric wavefront sets.

• The Wavefront Set: For an irreducible representation $\pi$ of a $p$-adic group $H(F)$, the Harish-Chandra–Howe local character expansion expresses the character $\Theta_\pi$ as a sum of Fourier transforms of orbital integrals over nilpotent orbits. The wavefront set, denoted $\overline{N}(\pi)_{\max}$, is the set of maximal nilpotent orbits (under the closure ordering) appearing in this expansion with non-zero coefficients.
• The Conjecture: The authors aim to verify the Wavefront Upper Bound Conjecture (Conjecture 1.1) and its local analogue, Jiang's Conjecture (Conjecture 1.3).
  • Conjecture 1.3 (Jiang): Let $\psi$ be an Arthur parameter for a split classical group $H$. For any $\pi$ in the Arthur packet $\Pi_\psi$, the geometric wavefront set $\overline{N}(\pi)$ is bounded above by the Spaltenstein dual of the nilpotent orbit associated with $\psi$. Specifically, if $\mathcal{O}_H \in \overline{N}(\pi)$, then $\mathcal{O}_H \leq d_{H^\vee}(\text{Ad}(H^\vee)N_\psi)$, where $N_\psi$ is the nilpotent element derived from the derivative of $\psi$ on the second $SL_2$ factor, and $d_{H^\vee}$ is the Spaltenstein duality map.
  • Goal: To prove that for split classical groups, the maximal orbit in the wavefront set of any representation in an Arthur packet is exactly the Spaltenstein dual of the orbit defined by the Arthur parameter.

2. Methodology

The proof relies on endoscopic transfer and the comparison of local character expansions. The authors assume the residue characteristic $p$ is sufficiently large ($p \gg 0$).

A. Twisted Endoscopy and $GL_m$

The authors utilize the connection between the classical group $H$ and the general linear group $G = GL_m$ (where $m$ is the dimension of the standard representation of the dual group $H^\vee$).

• They view the Arthur parameter $\psi$ for $H$ as a parameter for $G$.
• They consider the twisted character $\Theta_{\tilde{\pi}_\psi}$ of a representation $\tilde{\pi}_\psi$ on the twisted group $G \rtimes \langle \theta \rangle$, where $\theta$ is a specific involution.
• Key Result (Theorem 2.1): They establish that for $GL_m$, the wavefront set of the twisted representation corresponds exactly to the Spaltenstein dual of the parameter's orbit. This relies on prior work by Konno, Varma, and Mœglin–Waldspurger.

B. Transfer of Orbital Integrals

The core technical step involves relating the character expansion of the stable distribution on $H$ (the sum of characters in the packet) to the twisted character on $G$.

• Theorem 2.2: They prove a transfer result for nilpotent orbital integrals. Specifically, they show that the sum of orbital integrals over a special nilpotent orbit $\mathcal{O}_H$ in $H$ transfers to a sum of orbital integrals over the corresponding orbit $\mathcal{O}_G$ in the twisted $GL_m$.
• Hypothesis 2.3: They introduce a hypothesis regarding the transfer of stable distributions to twisted distributions, which allows them to relate the coefficients of the character expansions.

C. Inductive Argument via Standard Endoscopy

For the case where the Arthur parameter is not "tempered" (or when the packet is not a singleton), the authors use standard endoscopy.

• They decompose the Arthur packet $\Pi_\psi$ using the component group $S_\psi$. If the element $s_\psi \in S_\psi$ is non-trivial, the stable distribution $S\Theta_\psi$ transfers to a tensor product of stable distributions on smaller endoscopic groups $H_1 \times H_2$.
• Waldspurger Map: They utilize Waldspurger's explicit construction of the transfer of nilpotent orbits, denoted $W(\mathcal{O}_{H_1}, \mathcal{O}_{H_2})$.
• Combinatorial Analysis (Section 3): A significant portion of the paper is dedicated to proving Proposition 3.1, which establishes a crucial inequality relating the Waldspurger map $W$ and the Spaltenstein duality $d$. They prove that:
  $$W(\mathcal{O}_{H_1}, \mathcal{O}_{H_2}) \leq d_{H^\vee}(\text{Ad}(H^\vee)\xi(d_{H_1}(\mathcal{O}_{H_1}), d_{H_2}(\mathcal{O}_{H_2})))$$
  This is proven using combinatorial properties of partitions (orthogonal and symplectic) and Lusztig's symbols for Weyl group representations.

3. Key Contributions and Results

Main Theorem (Theorem 1.4)

Let $H$ be a split classical group ($SO_{2n+1}$, $Sp_{2n}$, or $SO_{2n}$) over a $p$-adic field with $p$ sufficiently large (specifically $p > 6n+3$ for odd orthogonal, $p > 6n$ for symplectic and even orthogonal). For any Arthur parameter $\psi$:

1. Existence: There exists a representation $\pi \in \Pi_\psi$ such that the Spaltenstein dual of the parameter's orbit, $d_{H^\vee}(\text{Ad}(H^\vee)N_\psi)$, is contained in the wavefront set $\overline{N}(\pi)$.
2. Upper Bound: For any $\pi \in \Pi_\psi$ and any orbit $\mathcal{O}_H \in \overline{N}(\pi)$, if $\mathcal{O}_H \neq d_{H^\vee}(\text{Ad}(H^\vee)N_\psi)$, then $\dim(\mathcal{O}_H) < \dim(d_{H^\vee}(\text{Ad}(H^\vee)N_\psi))$.
3. Full Conjecture: Under an additional technical hypothesis (Hypothesis 2.3, which is expected to hold generally), the full Conjecture 1.3 (and consequently Conjecture 1.1) is proven. This means the maximal orbit in the wavefront set is exactly the Spaltenstein dual of the parameter's orbit.

Specific Technical Contributions

• Twisted Character Expansion: They successfully adapted the local character expansion to the twisted setting for $GL_m$ to serve as a comparison tool for classical groups.
• Combinatorial Proof: They provided a rigorous combinatorial proof (using partitions and Lusztig symbols) that the Waldspurger map (used in endoscopic transfer) is compatible with the Spaltenstein duality order. This bridges the gap between the geometric transfer of orbits and the representation-theoretic definition of Arthur packets.
• Reduction to Unipotent Case: The proof strategy effectively reduces the general Arthur packet case to the unipotent case (where results by Mason-Brown, Okada, and the second author were already known) via endoscopic induction.

4. Significance

• Verification of Jiang's Conjecture: This work provides the first verification of Jiang's conjecture for split classical groups in the $p$-adic setting (under large $p$ assumptions). It confirms the deep link between the Langlands-Arthur parameters and the singular support of the character distribution.
• Unification of Local and Global: The results support the local analogue of global conjectures regarding the wavefront sets of automorphic representations, reinforcing the consistency of the Langlands program.
• Methodological Advancement: The paper demonstrates a powerful method of using twisted endoscopy (relating classical groups to $GL_m$) combined with standard endoscopy (decomposing packets) to solve problems regarding wavefront sets. This approach bypasses direct computation of character expansions for complex classical groups by leveraging the better-understood structure of $GL_m$.
• Foundation for Future Work: By establishing the upper bound and the existence of the maximal orbit, the paper sets the stage for determining the exact multiplicities of these orbits in the character expansion, which is a next-step problem in the representation theory of $p$-adic groups.

In summary, Atobe and Ciubotaru have successfully verified that the "largest" nilpotent orbit contributing to the character of an Arthur packet is precisely determined by the Arthur parameter via Spaltenstein duality, resolving a major open problem for split classical groups with large residual characteristic.