🔢 mathematics

Local-global principles for the existence of Levi factors

This paper investigates local-global principles for the existence of Levi factors in linear algebraic groups over one-variable function fields, providing counterexamples for disconnected groups and establishing a strong local-global principle when Levi descent holds.

Original authors: David Harbater, Julia Hartmann, George McNinch

Published 2026-03-30

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: David Harbater, Julia Hartmann, George McNinch

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

The Big Picture: The "Levi Factor" Puzzle

Imagine you have a complex machine (a Linear Algebraic Group). This machine is made of two distinct parts glued together:

The "Messy" Part (Unipotent Radical): Think of this as a pile of tangled, floppy rubber bands. It's unstable and hard to work with.
The "Sturdy" Part (Levi Factor): Think of this as a solid steel frame. It's rigid, reliable, and gives the machine its shape.

In a perfect world (specifically, in Characteristic 0, which is like our standard math universe), you can always easily pull the "Sturdy" steel frame out of the "Messy" rubber bands. The machine splits perfectly into two separate pieces.

The Problem:
The paper looks at a weird, twisted version of math called Positive Characteristic (think of a universe where numbers wrap around like a clock, but with a prime number $p$ ). In this universe, the "Messy" part and the "Sturdy" part are glued together so tightly that sometimes, you cannot pull them apart. The "Sturdy" frame might not even exist as a separate piece within the machine.

The Question: The "Local-Global" Test

The authors ask a very specific question about these machines over Function Fields (which are like maps of a landscape where every point has a number attached to it).

The Question:
"If I take this machine to every single small neighborhood (a 'local' view) of the landscape, and in every neighborhood I can successfully pull the 'Sturdy' frame out of the 'Messy' part... does that mean I can pull the frame out of the machine in the whole landscape (the 'global' view)?"

In math terms: If a Levi factor exists everywhere locally, does it exist globally?

The Bad News: The "Local-Global" Principle Fails

The authors say: No, not always.

They built a specific counter-example (a "trap" machine).

The Analogy: Imagine a puzzle that looks solvable if you look at it through a magnifying glass in any single room of a house. In every room, the pieces fit together perfectly.
The Twist: However, when you step back and look at the whole house, the pieces are actually locked in a way that prevents them from separating.
How they did it: They used a mathematical tool called an Artin-Schreier polynomial (think of it as a secret code). They created a situation where the code is "broken" (solvable) in every local neighborhood, but the "global" code remains unbroken. Because the code is unbroken globally, the "Sturdy" frame cannot be separated from the "Messy" part.

So, just because something works everywhere locally, it doesn't guarantee it works globally. This is a surprise because in many other areas of math (like solving quadratic equations), the local-global principle does work.

The Good News: When It Does Work (Levi Descent)

The authors don't just stop at "it fails." They found a special condition where the principle does hold, and it holds very strongly.

They call this condition Levi Descent.

The Analogy: Imagine you have a blueprint for a building. If you can build a perfect version of the building in a small, temporary shed (a local view), and the rules of construction are "descent-friendly," then you are guaranteed to be able to build the real, permanent skyscraper on the main site.
The Condition: If the machine follows certain symmetry rules (mathematically, if the group satisfies "Levi descent"), then the local success does guarantee global success.

The Strong Principle:
If the machine satisfies these rules, you don't even need to check every neighborhood. You only need to find one specific type of neighborhood (a "divisorial" valuation, which is like checking a major intersection on the map). If the "Sturdy" frame can be pulled out there, you know for a fact it can be pulled out everywhere, including the whole landscape.

Summary of the Journey

The Setup: We are trying to separate a "Sturdy" part from a "Messy" part in a mathematical machine.
The Hope: We hoped that if we can separate them in every small neighborhood, we can separate them everywhere.
The Reality Check: In the weird world of "Positive Characteristic," this hope is usually false. The authors built a machine that works locally but fails globally.
The Solution: However, if the machine follows specific "Levi Descent" rules, the hope is restored. In fact, checking just one special spot is enough to prove it works everywhere.

Why does this matter?
This helps mathematicians understand the hidden structures of complex systems. It tells us when we can trust our local observations to tell us the truth about the whole system, and when we need to be careful because the whole might be more complicated than the sum of its parts.

1. Problem Statement

The paper investigates the local-global principle regarding the existence of Levi factors for linear algebraic groups over one-variable function fields.

Context: Let $F$ be a field and $G$ a linear algebraic group over $F$ . A Levi factor (or Levi subgroup) is a reductive subgroup $M \subseteq G$ that maps isomorphically to the quotient $G/R_u(G)$ , where $R_u(G)$ is the unipotent radical. Equivalently, it is a splitting of the short exact sequence $1 \to R_u(G) \to G \to G/R_u(G) \to 1$ .
Characteristic Dependence:
- In characteristic 0, Levi factors always exist (Mostow's Theorem) and are unique up to conjugation.
- In positive characteristic, Levi factors may fail to exist even if $R_u(G)$ is defined over $F$ . Furthermore, even if they exist, they may not be conjugate.
The Core Question: If $F$ is a one-variable function field over a field $K$ , and $G$ is a linear algebraic group over $F$ (with $R_u(G)$ defined over $F$ ), does the existence of a Levi factor over every completion $F_v$ (at discrete valuations $v$ ) imply the existence of a Levi factor over $F$ ?

2. Methodology

The authors employ a combination of algebraic group theory, Galois cohomology, and arithmetic geometry (specifically the theory of function fields and valuations).

Counterexample Construction (Sections 2–3):
- The authors utilize specific constructions of non-reductive groups involving Artin-Schreier extensions (polynomials of the form $Z^p - Z - f$ ).
- They rely on a result from McNinch ([McN25]) linking the existence of a Levi factor for a specific family of groups $G_f$ to the solvability of an Artin-Schreier polynomial $P(Z) = Z^p - Z - f$ in the base field.
- To violate the local-global principle, they construct a function field $F$ and an element $f \in F$ such that the extension defined by $P(Z)$ is non-trivial globally (no root in $F$ ) but trivial locally (has a root in every completion $F_v$ ).
- They utilize properties of Hensel's Lemma and the structure of complete local rings (specifically $k[[t, x]]$ ) to demonstrate that local solvability holds despite global non-solvability.
Positive Results via Descent (Section 4):
- The authors introduce the concept of Levi descent. A group $G$ satisfies Levi descent if the existence of a Levi factor over a finite Galois extension $L/E$ implies the existence of one over $E$ (under specific degree constraints).
- They utilize Artin Approximation Theorem and properties of excellent discrete valuation rings to lift local sections (splittings) defined over completions to the henselization, and subsequently to a finite separable extension of the base field.
- By combining this with the assumption of Levi descent, they bridge the gap between local existence and global existence.

3. Key Contributions and Results

A. Failure of the Local-Global Principle (General Case)

The paper proves that the local-global principle for Levi factors fails in general for function fields in positive characteristic.

Proposition 3.3: Constructs a specific counterexample. Let $K = k((t))$ and $F$ be a degree $p$ Galois extension of $K(x)$ defined by $Y^p - Y = \frac{t}{x} + \frac{t}{x-1}$ . There exists a linear algebraic group $G$ over $F$ such that:
- $G$ has no Levi factor over $F$ .
- $G$ has a Levi factor over $F_v$ for every discrete valuation $v$ of $F$ .
- Mechanism: The obstruction is an Artin-Schreier extension that is non-trivial globally but splits locally at all valuations.
Proposition 3.5: Generalizes the failure to elliptic curves. For all but finitely many elliptic curves $C$ over an algebraically closed field $K$ of characteristic $p > 0$ , there exists a nontrivial Artin-Schreier extension of the function field of $C$ that splits over every completion. Consequently, the local-global principle fails for groups associated with these extensions.

B. A Strong Local-Global Principle (Under Levi Descent)

The paper establishes a strong positive result when the group satisfies the condition of Levi descent.

Theorem 4.1: Let $F$ be a one-variable function field over $K$ , and let $G$ be a linear algebraic group over $F$ with $R_u(G)$ defined over $F$ . If $G$ satisfies Levi descent, then the following are equivalent:
1. $G$ has a Levi factor over $F$ .
2. $G$ has a Levi factor over $F_v$ for every discrete valuation $v$ of $F$ .
3. $G$ has a Levi factor over $F_v$ for some divisorial discrete valuation $v$ of $F$ .
Significance of the Result: This is a "strong" principle because condition (c) requires only the existence of a Levi factor at a single specific type of valuation (divisorial) to guarantee global existence, provided the descent condition holds.

C. Conditions for Levi Descent

The authors clarify when Levi descent is known to hold (Remark 4.2), ensuring the applicability of Theorem 4.1:

If the unipotent radical $U$ is split over $F$ , Levi descent holds away from the characteristic.
Specific cohomological conditions involving the invariant group scheme $U^{M_\ell}$ and cohomology sets $H^1_{coc}$ guarantee Levi descent (citing McNinch [McN25]).
Currently, no counterexamples to Levi descent are known for connected groups where the unipotent radical is defined over the base field.

4. Significance

Resolution of a Fundamental Question: The paper definitively answers that the naive local-global principle for Levi factors is false in positive characteristic function fields, contrasting with classical results like Hasse-Minkowski for quadratic forms.
Refinement of Theory: By introducing the Levi descent condition, the authors recover a powerful local-global principle. This shifts the focus from the geometry of the field alone to the specific structural properties of the algebraic group.
Methodological Innovation: The work demonstrates how arithmetic properties of function fields (specifically the behavior of Artin-Schreier extensions under valuation) can be used to construct counterexamples in the theory of linear algebraic groups.
Bridge to Cohomology: The results highlight the role of Galois cohomology and descent theory in understanding the structure of non-reductive groups, providing a framework for future investigations into the classification of such groups over global fields.

In summary, the paper establishes that while local solvability does not generally imply global solvability for Levi factors in positive characteristic, it does so under the natural and often satisfied condition of Levi descent, thereby delineating the precise boundary between failure and success in this arithmetic-geometric context.