🔢 mathematics

Finite length for unramified $\mathrm{GL}_2$ : beyond multiplicity one, non-semisimple case

This paper proves that smooth mod $p$ representations of $\mathrm{GL}_2(K)$ arising from Shimura curve cohomology are of finite length even without the multiplicity one assumption, specifically addressing the non-semisimple case of the associated local Galois representation.

Original authors: Lucrezia Bertoletti

Published 2026-07-28

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Lucrezia Bertoletti

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 the universe of numbers not as a flat line of counting sticks, but as a vast, multi-layered city where every building has its own unique architecture. In this city, mathematicians are constantly trying to map the hidden connections between different neighborhoods. One of the most famous neighborhoods is the world of "prime numbers," the indivisible atoms of arithmetic. For decades, a group of explorers known as number theorists have been trying to understand how these primes behave when they interact with geometric shapes called "Shimura curves." Think of these curves as intricate, multi-dimensional roller coasters that twist through the landscape of numbers.

To navigate this city, the explorers use a special toolkit called "representation theory." If a number is a brick, a representation is the blueprint showing how that brick fits into a larger wall. Sometimes, these blueprints are simple and clean, like a single-story house. Other times, they are complex, multi-story structures with hidden rooms and tangled wiring. A key question in this field has been: "How big and complicated can these blueprints get before they fall apart?" For a long time, mathematicians knew the answer for the simplest, most orderly blueprints (called "semisimple" cases). They proved that these structures have a finite size, meaning they are made of a limited number of building blocks. But what about the messy, tangled blueprints? What happens when the wiring is crossed and the rooms are stacked in unexpected ways? This is the mystery that has kept the explorers up at night.

The Paper's Mission: Taming the Tangled Blueprints

In this paper, the author, Lucrezia Bertoletti, steps into the chaotic, tangled part of the city to see if the rules still hold. She focuses on a specific type of blueprint used by a group of symmetries called $GL_2(K)$ , which acts like a master key for unlocking patterns in these number cities. Specifically, she looks at the "non-semisimple" cases—the messy, non-repeating structures where the usual "one-to-one" matching rules (known as "multiplicity one") don't apply. In the simple cases, every room in the blueprint corresponds to exactly one unique key. But in these messy cases, multiple keys might fit the same room, or the rooms might be stacked in a way that makes it hard to tell where one ends and another begins.

The paper asks a terrifying question for a mathematician: "If we stop assuming the blueprint is simple, does the whole structure collapse into infinity, or does it still have a finite, countable size?"

The Discovery: Finite Length in the Chaos

Bertoletti proves that even in this messy, tangled world, the structures do not explode into infinity. They have a finite length.

To understand what this means, imagine a Russian nesting doll. In a simple case, you open the doll, find one smaller doll inside, then another, and eventually you reach the tiny core. You can count them: 1, 2, 3, 4. The "length" is 4. In the messy cases Bertoletti studied, the dolls are glued together in weird ways, and some might be stuck inside others sideways. It looked like you might be able to keep opening them forever, finding an infinite chain of dolls.

However, the paper demonstrates that no matter how tangled the blueprints get (under certain "generic" conditions, which are like saying the city isn't built on a weird, broken foundation), you will always hit a bottom. You can break the structure down into a finite list of irreducible pieces. The paper provides a specific formula to calculate the maximum possible number of these pieces, showing that the size is bounded by a number related to the complexity of the field (specifically, $r \cdot (f + 1)$ , where $r$ is a measure of how many times a pattern repeats and $f$ is a measure of the field's size).

How They Did It: The "Diagram" Detective Work

To solve this, the author didn't just look at the finished building; she looked at the scaffolding. She used a tool called a "diagram," which is like a simplified map of the blueprint's most important connections. She showed that even when the blueprint is messy, this map follows a strict, predictable pattern.

She also used a technique called "dévissage," which is a fancy French word for "taking apart." She showed that if you can prove the smaller, simpler pieces of the puzzle are finite, then the whole messy puzzle must be finite too. It's like proving that a giant, chaotic pile of LEGOs is finite by showing that every single brick in the pile is a standard, countable size.

The Verdict

The paper doesn't just guess; it proves. It establishes that these complex, non-simple representations are well-behaved. They are not infinite monsters. They are finite, manageable structures that can be counted and understood. This extends a previous discovery that only worked for the "clean" blueprints, proving that the rules of finiteness are robust enough to survive even the most tangled, non-repeating configurations.

In short, the paper says: "Even when the math gets messy and the usual shortcuts don't work, the structures we are studying still have a definite, finite size. The chaos is contained."

Technical Summary: Finite length for unramified GL2: beyond multiplicity one, non-semisimple case

Problem Statement
This paper investigates the structure of smooth mod $p$ representations of $GL_2(K)$ , where $K$ is a finite unramified extension of $\mathbb{Q}_p$ . Specifically, it focuses on representations $\pi(V^v)$ arising from the cohomology of Shimura curves associated with a global Galois representation $r$ . The central problem is to establish that these representations have finite length as $GL_2(K)$ -modules under "genericity" assumptions on $r$ , but crucially without assuming the "multiplicity one" property (i.e., allowing the multiplicity $r \ge 1$ to be greater than 1) and without assuming that the local restriction $r|_{Gal(F_v/F_v)}$ is semisimple.

Previous work by Calegari, Emerton, and others (specifically [10]) established finite length in the "minimal case" where multiplicity is one ( $r=1$ ). A companion article [2] extended these results to higher multiplicity ( $r \ge 1$ ) but retained the assumption that the local Galois representation is semisimple. This paper addresses the remaining, more difficult case: the non-semisimple local Galois representation with arbitrary multiplicity $r \ge 1$ .

Methodology
The authors employ a local-global strategy, reducing the global problem to the study of local representations satisfying a specific set of four hypotheses (Section 2.3). The methodology relies heavily on the interplay between:

Diagram Theory: Utilizing the diagram $D(\rho)$ associated with the Galois representation $\rho = r^\vee|_{Gal(K/K)}$ . The paper generalizes the structure of subrepresentations of $\pi$ in terms of these diagrams, specifically analyzing the $K_1$ - and $I_1$ -invariants.
$(\phi, \Gamma)$ -modules: Using the functor $D^\vee_\xi$ to associate cyclotomic $(\phi, \Gamma)$ -modules to the representations. The rank of these modules serves as a key numerical invariant.
Commutative Algebra: Analyzing the structure of the dual representation $\pi^\vee$ as a module over the completed group algebra $\Lambda = \mathbb{F}[[I_1/Z_1]]$ . The authors investigate the Cohen-Macaulay property and the grade of these modules.
Extension of [27]: A significant portion of the technical work involves extending the results of Paškūnas [27] (which treated the $r=1$ case) to the case $r \ge 1$ . This requires careful bookkeeping of "mixed terms" and ensuring that constructions involving the vector space $V = \mathbb{F}^r$ (representing the multiplicity) remain compatible with the diagram structures.

Key Contributions and Results

Finite Length Theorem (Theorem 1.1.1):
The main result proves that for a generic Galois representation $r$ (satisfying specific bounds on the exponents of the fundamental characters), the representation $\pi(V^v)$ has finite length as a $GL_2(K)$ -representation. This holds regardless of the multiplicity $r$ or the prime-to- $v$ level.
Structure of Subquotients (Theorem 1.1.2):
The paper describes the precise structure of the representation $\pi$ (and its subquotients $\pi(V^v, \sigma^v_p)$ ). It is shown that $\pi$ fits into an exact sequence:
$0 \to V \otimes \pi_0 \to \pi \to \pi' \to 0$
where $\pi_0$ and $\pi_f$ are principal series representations, and $\pi'$ is a subrepresentation whose Jordan-Hölder constituents are all supersingular. The length of $\pi'$ is bounded above by $r \cdot (f-1)$ .
Cohen-Macaulay Property (Theorem 1.1.3 & Proposition 4.1.9):
The authors prove that for any non-zero subrepresentation $\pi_1 \subseteq \pi$ , the linear dual $\pi_1^\vee$ is a Cohen-Macaulay $\Lambda$ -module of grade $2f$ . This generalizes previous results from the semisimple case to the non-semisimple, higher multiplicity setting.
Rank Formula for $(\phi, \Gamma)$ -modules (Theorem 3.2.13 & Corollary 3.2.15):
A crucial technical achievement is the computation of the rank of the $(\phi, \Gamma)$ -module $D^\vee_\xi(\pi_1)$ associated with a subrepresentation $\pi_1$ . The rank is given by the formula:
$\text{rk}_A D_A(\pi_1) = \sum_{\ell=0}^f r(\pi_1, \ell) \binom{f}{\ell}$
where $r(\pi_1, \ell)$ are numerical invariants derived from the filtration of the diagram associated with $\pi_1$ . This formula generalizes [27, Theorem 9.2] to $r \ge 1$ .
Characterization of Subrepresentations (Theorem 4.2.1):
The paper establishes that the inclusion of subrepresentations $\pi_1 \subseteq \pi_0$ can be detected entirely at the level of $K_1$ -invariants (or equivalently, via the inclusions of the vector subspaces $V(\pi_1, \ell) \subseteq V(\pi_0, \ell)$ ). This allows for a combinatorial control over the lattice of subrepresentations.

Significance and Claims
The authors claim that this work extends the foundational results of [10] to the "higher multiplicity" and "non-semisimple" regimes, which were previously out of reach.

Generalization: The results generalize the finite length property and the Cohen-Macaulayness of the duals from the minimal case ( $r=1$ , semisimple) to the general case ( $r \ge 1$ , non-semisimple).
Technical Barrier: The paper explicitly notes that the non-semisimple case introduces unique difficulties, such as the fact that the natural injection $\pi_1^{K_1}/\pi_0^{K_1} \to (\pi_1/\pi_0)^{K_1}$ is not surjective in general. The authors develop new machinery (specifically the "weight cycling" arguments adapted for $r \ge 1$ and the extension of [27]) to overcome these obstructions.
Limitations: The authors remain modest regarding the precise structure of the "middle" part $\pi'$ . While they provide an upper bound on its length ( $r(f-1)$ ) and prove its constituents are supersingular, they state that proving the exact multiplicity of irreducible constituents (ideally $r$ ) "seems currently out of reach." Similarly, while they expect the rank formula (3) to hold for subquotients $\pi_2$ , they lack a proof for the general non-semisimple case at the time of writing.

In summary, the paper provides a rigorous framework for understanding the finite length and structural properties of mod $p$ representations of $GL_2(K)$ in the most general setting currently accessible, bridging the gap between the minimal case and the complex non-semisimple, higher multiplicity scenario.

More like this