🔢 mathematics

A de Rham weight part of Serre's conjecture and generalized mod $p$ BGG decompositions

This paper proposes a geometric formulation of the weight part of Serre's conjecture using de Rham cohomology of Shimura varieties, proving its equivalence to the classical étale cohomology formulation for generic weights and non-Eisenstein eigensystems in the cases of $GL_3$ and $GSp_4$ by establishing generalized mod $p$ BGG decompositions and demonstrating generic concentration in middle degree.

Original authors: Martin Ortiz

Published 2026-09-14

📖 8 min read🧠 Deep dive

CC BY 4.0

Original authors: Martin Ortiz

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

In the vast landscape of modern mathematics, there is a deep and enduring puzzle concerning the hidden patterns that govern numbers. This puzzle sits at the intersection of two seemingly different worlds: the study of symmetry in algebraic structures and the study of shapes that arise from number theory. For decades, mathematicians have sought a precise dictionary to translate between these worlds, a tool that would allow them to predict the behavior of complex numerical systems by looking at their geometric shadows. A central piece of this dictionary is a set of rules known as Serre's conjecture, which attempts to list the specific "weights" or fundamental building blocks that can appear in these systems. While the rules were well understood for simpler, one-dimensional cases, they became murky and difficult to decipher when applied to more complex, multi-dimensional systems. The challenge was to find a new way to see these weights that would work even when the geometry became too intricate for traditional methods.

Martin Ortiz, a mathematician working on this frontier, has proposed a fresh approach to solving this problem by looking at a specific type of geometric data called de Rham cohomology. Think of this as a way of measuring the "holes" and structural features of a shape, but done in a manner that captures information about both its smooth geometry and its underlying arithmetic nature. Ortiz suggests that if we look at these shapes through this specific lens, we can identify the correct building blocks just as reliably as the older, more established methods, but with the added benefit of being able to see connections that were previously invisible. He proposes that this new geometric view is equivalent to the old one, meaning they tell the same story, just in a different language. To prove this, he had to construct a new kind of map, a detailed blueprint that shows how these complex shapes are built from simpler pieces, specifically for two of the most complicated groups of symmetries that have not yet been fully solved: those related to three-dimensional and four-dimensional spaces.

The core of Ortiz's work involves tackling the cases where the symmetry groups are of rank three and four, which are the simplest examples where the old rules break down and new, unexpected phenomena appear. In these higher dimensions, the relationship between the geometric shape and its arithmetic properties becomes tangled. The traditional method of looking at the shape's surface to find its building blocks fails because it cannot distinguish between different layers of complexity that are hidden within the structure. Ortiz's solution was to build a generalized framework, a set of resolutions that act like a scaffolding, allowing him to peel back the layers of these complex shapes one by one. He constructed these scaffolds by extending previous work on how to decompose mathematical objects, creating a new system that works specifically in the context of prime numbers, which is the setting where these conjectures are most relevant.

By applying this new scaffolding to the geometry of Shimura varieties, which are special types of shapes defined by number theory, Ortiz was able to demonstrate that the de Rham approach works perfectly for a wide range of cases. He proved that for a specific type of three-dimensional symmetry group, the building blocks identified by his new method match exactly with those found by the traditional methods, provided the system is in a "generic" state, meaning it is not in a special, degenerate condition. This was a significant step forward because it confirmed that the new geometric perspective is not just a theoretical possibility but a practical tool that yields the correct answers. Furthermore, he showed that this method could be used to upgrade a previous, weaker prediction about how these building blocks relate to one another into a stronger, more definitive statement.

The paper also delves into the four-dimensional case, where the mathematics becomes even more intricate. Here, Ortiz did not just prove a general equivalence but used his new tools to compute specific, explicit decompositions of the shapes. These computations allowed him to prove a new result about how the presence of one type of building block implies the presence of another, a relationship that was previously only known to hold in a weak or uncertain way. This finding is particularly important because it reveals a hidden connection between different parts of the mathematical structure that cannot be seen using the older methods. The work relies on a careful analysis of how these shapes behave when viewed through the lens of prime numbers, using a technique that involves translating the problem into a different geometric setting where the answers become clearer.

One of the most striking aspects of this research is how it bridges the gap between two different ways of thinking about the same problem. The traditional approach relies on looking at the shape's "étale" properties, which are related to how the shape behaves under certain discrete transformations. Ortiz's approach looks at the "de Rham" properties, which are related to the shape's smooth, continuous features. He showed that for the specific cases he studied, these two views are not just compatible but are actually two sides of the same coin. This equivalence is not obvious, because in higher dimensions, the smooth features of a shape can sometimes hide information that the discrete features reveal, and vice versa. By constructing his detailed scaffolding, Ortiz was able to show that the information hidden in the smooth features is exactly what is needed to reconstruct the discrete picture.

The results are not just a confirmation of existing theories but a demonstration of a new way to navigate the complex terrain of number theory. Ortiz's work provides a concrete method for calculating the building blocks of these systems, a task that was previously out of reach for the most complicated cases. He showed that by using his generalized framework, one can determine exactly which weights appear in the system and how they relate to each other. This is a crucial step toward a complete understanding of Serre's conjecture, which aims to provide a universal rule for these patterns. The paper establishes that for the cases of three and four dimensions, the new geometric approach is robust and reliable, offering a path forward for solving even more complex versions of the problem.

In the process of proving these results, Ortiz had to overcome significant technical hurdles. He had to show that certain mathematical maps, which connect different parts of the structure, are injective, meaning they do not lose any information as they pass from one layer to the next. This was a difficult task because these maps behave unpredictably in higher dimensions. By using a combination of geometric intuition and rigorous algebraic computation, he was able to prove that these maps work as expected in the generic cases. This proof was essential for showing that the new method does not introduce any errors or false positives. The work also involved ruling out the possibility that the method would fail for certain specific types of weights, showing that the results hold true as long as the system is not in a special, degenerate state.

The implications of this work extend beyond the immediate results. By providing a clear, geometric way to understand these complex systems, Ortiz has opened up new avenues for research. His methods can be applied to other problems in number theory where similar patterns of symmetry and geometry interact. The ability to translate between the smooth and discrete views of these shapes is a powerful tool that can help mathematicians solve problems that were previously considered intractable. The paper stands as a testament to the power of combining different mathematical perspectives to reveal the underlying unity of complex structures. It shows that even in the most abstract corners of mathematics, there are concrete ways to make sense of the unknown.

Ultimately, this paper is about finding a new language to describe the fundamental building blocks of number theory. Ortiz has shown that by looking at the geometry of these shapes through the lens of de Rham cohomology, we can see patterns that were previously hidden. He has constructed a new set of tools that allow us to break down these complex shapes into their simplest components and understand how they fit together. This work confirms that the new approach is equivalent to the old one, providing a solid foundation for future discoveries. It is a significant step toward a complete understanding of the patterns that govern the universe of numbers, revealing that the key to unlocking these mysteries lies in the careful interplay between geometry and algebra.

Technical Summary: A de Rham Weight Part of Serre's Conjecture and Generalized mod p BGG Decompositions

Problem Statement
The paper addresses the "weight part of Serre's conjecture" for Shimura varieties, specifically seeking a formulation in terms of de Rham cohomology that is equivalent to the classical formulation using étale cohomology and Serre weights.

Classical Formulation: For a Shimura variety $Sh$ and a maximal ideal $\mathfrak{m}$ of the Hecke algebra, the set of modular Serre weights $W(\mathfrak{m})$ consists of irreducible $\mathbb{F}_p$ -representations $F(\lambda)$ of $G(\mathbb{F}_p)$ appearing in the étale cohomology $H^\bullet_{\text{ét}}(Sh_{\overline{\mathbb{Q}}_p}, F(\lambda))_\mathfrak{m}$ .
The Gap: While coherent cohomology (automorphic forms) provides a bridge, for groups of semisimple rank $\ge 2$ (e.g., $GL_3$ , $GSp_4$ ), the irreducible representation $F(\lambda)$ is not always isomorphic to the restriction of the dual Weyl module $V(\lambda)_{\mathbb{F}_p}$ . Consequently, the non-vanishing of the degree 0 coherent cohomology $H^0(Sh^{\text{tor}}, \omega(\lambda+\eta))_\mathfrak{m}$ is insufficient to distinguish which Jordan-Hölder factors of $V(\lambda)_{\mathbb{F}_p}$ are modular.
Proposed Approach: The author proposes studying the de Rham cohomology $H^\bullet_{\text{dR}}(Sh^{\text{tor}}, L(\lambda))_\mathfrak{m}$ , where $L(\lambda)$ is the irreducible algebraic representation. The central conjecture is that the set of de Rham weights $W_{\text{dR}}(\mathfrak{m})$ coincides with the set of modular Serre weights $W(\mathfrak{m})$ .

Methodology
The paper employs a geometric approach combining the de Rham realization functor with explicit representation theory in characteristic $p$ .

De Rham Realization Functor: The author utilizes a functor $\Psi: D^b(\mathcal{O}_{P, \mathbb{F}_p}) \to D^b(C_{Sh/\mathbb{F}_p})$ constructed in a companion paper [Ort26]. This functor maps parabolic Verma modules to vector bundles and maps between them to differential operators (specifically, theta linkage maps).
Generalized mod p BGG Resolutions: The core technical work involves constructing explicit "generalized mod p BGG resolutions" for simple modules $L(\lambda)$ $L (λ)$ in the derived category of a mod $p$ $p$ parabolic category $\mathcal{O}_{P, \mathbb{F}_p}$ $O_{P, F_{p}}$ .
- Unlike the lowest alcove case where BGG resolutions are direct sums of Verma modules, the author constructs filtrations of these resolutions.
- The graded pieces of these filtrations are complexes of Verma modules (or irreducible $M$ -representations) linked by non-trivial maps.
- The construction relies on the specific geometry of $p$ -restricted alcoves for $GL_3$ and $GSp_4$ , the linkage order $\uparrow$ , and the formal characters of $L(\lambda)$ .
Cohomological Analysis: Applying $\Psi$ $Ψ$ to these resolutions yields filtrations on the de Rham complex of $L(\lambda)$ $L (λ)$ . The author analyzes the cohomology of the graded pieces using:
- Vanishing results for higher coherent cohomology of automorphic vector bundles (from [Yan25], [GK18a]).
- Injectivity properties of theta linkage maps (from [Ort26]).
- Serre duality and the Ekedahl-Oort stratification of flag Shimura varieties.
- A novel technique of " $p$ -translating" lowest alcove BGG complexes to analyze higher cohomology groups.

Key Contributions and Results

1. Conjectures on Equivalence and Concentration

Conjecture 1.1: Proposes that $W_{\text{dR}}(\mathfrak{m}) = W(\mathfrak{m})$ (Strong form) and that for generic non-Eisenstein $\mathfrak{m}$ , the generic subsets coincide (Weak form).
Conjecture 1.2: Proposes that for generic $\lambda$ , the de Rham cohomology $H^\bullet_{\text{dR}}(Sh^{\text{tor}}, L(\lambda))_\mathfrak{m}$ is concentrated in the middle degree. The paper proves that if étale cohomology is concentrated in the middle degree (known in many cases), then Conjecture 1.2 implies Conjecture 1.1.

2. Explicit BGG Decompositions for $GL_3$

Theorem 1.3 / Proposition 3.7: For a compact unitary Shimura variety with $G_{\mathbb{Q}_p} = GL_3 \times \mathbb{G}_m$ , the author constructs a 3-step filtration for $L(\lambda_1)$ where $\lambda_1$ is in the upper alcove.
The graded pieces involve:
- A complex $[0 \to \text{Ver}_P W(\lambda_0) \xrightarrow{\phi} \text{Ver}_P W(\lambda_1)]$ where $\phi$ is a non-zero map.
- A term involving the irreducible representation $LM(\mu_1)$ .
- A dual complex involving $\nu_0, \nu_1$ .
Result: This leads to a proof that de Rham cohomology is concentrated in the middle degree for generic weights in the upper alcove (Theorem 4.4). This refines the relation between modular forms and Serre weights, showing that non-lowest-alcove weights can be detected via cokernels of theta linkage maps.

3. Explicit BGG Decompositions for $GSp_4$

Propositions 3.9 & 3.10: The author constructs 4-step filtrations for $L(\lambda)$ in the $C_1$ and $C_2$ alcoves for the Siegel threefold ( $GSp_4$ ).
New Vanishing Results: To handle the $C_2$ alcove, the paper proves new vanishing results for coherent cohomology in degree 2 for non-dominant weights (Propositions 4.10, 4.12) using the Ekedahl-Oort stratification and weight elimination.
Injectivity of Higher Theta Maps: Using a " $p$ -translated" lowest alcove BGG complex and results from Deligne-Illusie [DI87], the author proves that a specific theta linkage map is injective in cohomological degree 1 (Proposition 4.16). This was previously out of reach.

4. The Generic Entailment for $GSp_4$

Theorem 1.6 / Theorem 4.17: The paper upgrades a "weak entailment" from [Ort24] to a full "generic entailment" for $GSp_4$ .
Specifically, for generic $\lambda_0 \in C_0$ and non-Eisenstein $\mathfrak{m}$ :
$F(\lambda_0) \in W(\mathfrak{m}) \implies F(\lambda_2) \in W(\mathfrak{m})$
where $\lambda_2$ is a specific affine Weyl reflection of $\lambda_0$ .
This result is derived entirely from the structure of the BGG decomposition and the cohomological properties of the graded pieces, without relying on the Taylor-Wiles method.

Significance and Claims

Bridging Cohomologies: The paper establishes a concrete link between de Rham cohomology (computable via coherent cohomology and theta maps) and the modularity of Serre weights. It demonstrates that the de Rham framework captures information about Serre weights that is invisible to étale cohomology alone (due to the lack of direct sum decompositions in higher rank) and more refined than degree 0 coherent cohomology.
Explicit Computations: The explicit construction of BGG resolutions for weights in higher alcoves (beyond the lowest one) is a primary contribution. These resolutions reveal non-trivial interactions (via theta linkage maps) between different weights that govern the modularity.
Methodological Advancement: The introduction of " $p$ -translated" BGG complexes and the proof of injectivity for theta maps in higher cohomological degrees provide new tools for studying mod $p$ automorphic forms.
Modesty: The author notes that while the generic concentration of de Rham cohomology is proven for $GL_3$ and specific cases of $GSp_4$ , proving it for all $GSp_4$ weights remains out of reach with current methods. The paper also clarifies that the strong form of the conjecture fails for arbitrary dominant weights (Example 2.3), restricting the scope to $p$ -restricted weights or weights with specific genericity conditions on the local Galois representation.

In summary, the paper provides a geometric and representation-theoretic framework to formulate and partially prove a de Rham version of Serre's conjecture, utilizing explicit mod $p$ BGG decompositions to uncover the modular nature of Serre weights in higher rank groups.

Technical Summary: A de Rham Weight Part of Serre's Conjecture and Generalized mod p BGG Decompositions

More like this