🔢 mathematics

Ordinary local representations and $\Ext$ groups

This paper establishes local and global vanishing results for Ext groups associated with admissible unitary representations of $\GL_2(\Q_p)$ arising from ordinary local Galois representations via the $p$ -adic local Langlands correspondence.

Original authors: Debargha Banerjee, Srijan Das

Published 2026-05-18

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Debargha Banerjee, Srijan Das

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 trying to solve a massive, cosmic jigsaw puzzle. In the world of advanced mathematics, this puzzle involves connecting two very different languages: the language of symmetry (represented by groups like $GL_2(\mathbb{Q}_p)$ ) and the language of numbers and shapes (represented by Galois groups).

This paper, written by Debargha Banerjee and Srijan Das, is about proving that certain specific pieces of this puzzle do not fit together in the way some mathematicians might have hoped. They are essentially drawing a "Do Not Enter" sign on a specific section of the mathematical map.

Here is a breakdown of their discovery using simple analogies:

1. The Two Worlds: The "Local" and the "Global"

To understand the paper, you need to know about two different "rooms" where these mathematical objects live:

The Local Room (Drinfeld Towers): Think of this as a microscopic, high-tech laboratory. It's a specific, complex structure (a "tower") used to study how numbers behave in a very small, specific neighborhood (the $p$ -adic numbers).
The Global Room (Shimura Curves): This is a vast, sweeping landscape (like a giant mountain range). It represents the big picture of number theory, connecting many different local neighborhoods together.

2. The Characters: The "Ordinary" and the "Supersingular"

In this story, there are different types of mathematical "characters" (representations):

The Ordinary Characters: These are the "standard" or "predictable" characters. In the paper, they are described as "ordinary" because they follow a specific, orderly pattern (like a well-behaved student).
The Supersingular Characters: These are the "wild" or "exotic" characters. They are chaotic and don't follow the standard rules.

The paper focuses on a specific type of "Ordinary" character, let's call him Mr. Ordinary.

3. The Big Question

Mathematicians have a theory called the Local-Global Compatibility. It's like a rulebook that says: "If you find a character in the Local Room, it should show up in the Global Room in a specific way."

The authors asked: "If we take Mr. Ordinary (who is 'reducible' and 'non-split'—meaning he is a bit broken or split into parts but not fully separated), will he appear in the Global Room (Shimura Curves) or the Local Room (Drinfeld Towers)?"

4. The Discovery: The "Ghost" Theory

The authors proved a surprising result: Mr. Ordinary is a ghost.

The Global Result: They showed that Mr. Ordinary does not exist in the Global Room (the cohomology of Shimura curves). No matter how hard you look at the mountain range, you will never find him there.
The Local Result: They also showed that Mr. Ordinary does not exist in the Local Room (the cohomology of Drinfeld towers). Even in the microscopic lab, he is absent.

The Analogy:
Imagine you are looking for a specific type of rare bird (Mr. Ordinary) in a massive forest (Global) and a specific birdcage (Local).

The "philosophy" of the forest suggests that ordinary birds should live in the "supersingular" (wild) part of the forest.
The authors proved that Mr. Ordinary is nowhere to be found in either the forest or the cage. He simply doesn't inhabit the spaces where these mathematical structures live.

5. The "Extension" Problem (The Sticky Glue)

The paper also looks at Ext groups. In simple terms, an Ext group measures whether you can "glue" two different mathematical objects together to make a new, slightly messy object.

The Question: Can we glue Mr. Ordinary to the "wild" objects found in the Local Room?
The Answer: No. The authors proved that the "glue" doesn't work. There is no way to create a non-trivial connection between Mr. Ordinary and the objects found in the Drinfeld towers. They are too different; they repel each other.

6. How Did They Do It?

They used a powerful tool called a Factorization Theorem (developed by other mathematicians).

Think of this theorem as a magic X-ray machine. It allows them to look inside the complex structures (the towers and curves) and see exactly what ingredients they are made of.
When they used this X-ray, they saw that the ingredients inside were all "wild" (supersingular) or "principal series" (another type of standard character), but never the specific "Ordinary" character they were looking for.
They also used a "block decomposition" strategy. Imagine sorting all the mathematical characters into different boxes. They proved that Mr. Ordinary lives in a box that is completely separate from the boxes containing the objects in the towers and curves. Because the boxes are sealed off from each other, no interaction (or "Ext") can happen between them.

Summary

In everyday language, this paper is a negative result that is actually very helpful. It tells mathematicians:

"Stop looking for this specific type of 'Ordinary' character in the Global and Local cohomology of these curves and towers. You won't find him there, and you can't glue him to the things you do find. He belongs to a different mathematical universe."

This helps refine the map of the mathematical world, ensuring that researchers don't waste time searching for something that, according to this proof, simply isn't there.

Technical Summary: Ordinary Local Representations and Ext Groups

Problem Statement
The paper addresses the structural properties of $p$ -adic local Langlands correspondences for $\text{GL}_2(\mathbb{Q}_p)$ , specifically focusing on the interaction between ordinary local Galois representations and their associated admissible unitary Banach space representations, denoted $\Pi(\rho_p)$ . While the $p$ -adic local Langlands correspondence establishes a bijection for absolutely irreducible representations, the behavior of reducible, non-split ordinary representations within global arithmetic objects remains less understood.

The central problem is to determine whether these ordinary representations $\Pi(\rho_p)$ appear in the cohomology of specific arithmetic varieties, namely finite-level Drinfeld towers and Shimura curves. Furthermore, the authors investigate the vanishing or non-vanishing of extension groups ( $\text{Ext}$ ) between these representations and other components of the cohomology, particularly in the context of local-global compatibility. A specific motivation cited is Chojecki's question regarding the generalization of results from the residually irreducible case to the residually reducible, non-split ordinary case.

Methodology
The authors employ a combination of $p$ -adic Hodge theory, the theory of Drinfeld towers, and the representation theory of $p$ -adic groups. The methodology proceeds through the following steps:

Geometric Setup: The study utilizes the $p$ -adic uniformization of Shimura curves by Drinfeld towers. Specifically, the rigid analytic space of a Shimura curve $Sh_n(U)^{an}$ is related to the Drinfeld tower $M_n$ via the Cerednik-Drinfeld uniformization theorem. This allows the authors to relate the global cohomology of Shimura curves to the local cohomology of Drinfeld towers.
Factorization Theorem: A critical technical tool is the factorization theorem established by Colmez, Dospinescu, and Niziol [19]. This theorem describes the étale cohomology of the Drinfeld tower as a representation of $\text{GL}_2(\mathbb{Q}_p)$ by decomposing it into a direct sum involving the dual of the $p$ -adic local Langlands representation $\Pi(V)'$ , Kisin rings, and Jacquet-Langlands transfers.
Block Decomposition and Mod $p$ Reduction: To analyze extensions, the authors utilize Paskunas's block decomposition of the category of admissible unitary Banach space representations. They reduce the problem from characteristic 0 to characteristic $p$ by examining the mod $p$ reduction of the representations. This involves comparing the $p$ -adic local Langlands correspondence with the mod $p$ local Langlands correspondence (Breuil's classification), distinguishing between supersingular, principal series, and Steinberg representations.
Locally Analytic Representations: The analysis of $\text{Ext}^1$ groups relies on the theory of locally analytic representations developed by Schneider and Teitelbaum. The authors use the faithful flatness of the algebra of locally analytic distributions $D(G, E)$ over the Iwasawa algebra $E[[G]]$ to lift non-split extensions from the smooth/locally analytic setting back to the Banach space setting.

Key Contributions and Results
The paper establishes two primary vanishing theorems for ordinary local representations:

Global Vanishing Theorem: The authors prove that if $\rho_p$ is a reducible, non-split, potentially crystalline ordinary Galois representation (with distinct Hodge-Tate weights), the associated $p$ -adic local Langlands representation $\Pi(\rho_p)$ does not appear as a sub-representation in the étale cohomology of finite-level Shimura curves. Specifically, $\text{Hom}_{G}(\Pi(\rho_p)', H^1_{\text{ét}}(Sh_n(U)^{an}, E)) = 0$ . This result is derived by showing that the cohomology of Shimura curves is built from principal series representations (via the uniformization), while $\Pi(\rho_p)$ for ordinary non-split cases is an extension of principal series representations that does not embed into the cohomology due to the specific structure of the duals involved.
Local Vanishing Theorem: For the local setting, the authors prove that for any sub-representation $W$ of the cohomology of the Drinfeld tower (specifically the part associated with a residually irreducible Galois representation $V$ ), there are no non-trivial extensions between the dual of the ordinary representation $\Pi(\rho_p)'$ and $W$ . This manifests as:
- $\text{Hom}_{G}(\Pi(\rho_p)', W) = 0$
- $\text{Ext}^1_{G}(\Pi(\rho_p)', W) = 0$
  The proof relies on the fact that $\Pi(\rho_p)$ (in the ordinary non-split case) belongs to a different "block" in the category of representations than the supersingular representations that constitute the relevant part of the Drinfeld tower cohomology. Since extensions only exist within the same block, the $\text{Ext}$ groups vanish.

Significance and Claims
The paper claims that these results are consistent with the philosophical expectation that ordinary representations should not appear in the "supersingular" part of the cohomology. The authors emphasize that their work provides a rigorous local and global vanishing result for $\text{Ext}$ functors in the context of ordinary, reducible, non-split representations, a case that was previously less explored compared to the residually irreducible setting.

The authors modestly note that while they prove vanishing for the residually irreducible case, the situation for residually reducible cases may differ. In Remark 5.3, they suggest that instead of vanishing, one might expect finite-dimensionality results for $\text{Ext}$ groups in the residually reducible case, drawing on bounds derived from the mod $p$ theory. They do not claim to have solved the finite-dimensionality question in characteristic 0 for the reducible case but offer a pathway for future investigation using the tools developed.

The work also highlights the utility of the Colmez-Dospinescu-Niziol factorization theorem in describing the cohomology of Drinfeld towers as a $\text{GL}_2(\mathbb{Q}_p)$ -representation, serving as a local analogue to Emerton's local-global compatibility theorems. The results reinforce the structural separation between ordinary and supersingular phenomena in the $p$ -adic Langlands program.