The $L$-polynomials of van der Geer--van der Vlugt curves in characteristic $2$

This paper establishes an explicit formula for the $L$-polynomials of van der Geer--van der Vlugt curves in characteristic 2 by leveraging the structure of Heisenberg groups and Lang torsors, ultimately enabling the construction of examples that attain the Hasse--Weil bound.

The Gist

Imagine you are an architect trying to design the most efficient, symmetrical building possible. In the world of mathematics, specifically number theory, these "buildings" are called curves. But these aren't curves you draw on a piece of paper; they are complex, multi-dimensional shapes defined by equations over "finite fields" (which are like tiny, self-contained universes of numbers).

The goal of this paper is to understand the "blueprints" (called L-polynomials) of a specific family of these mathematical buildings, known as van der Geer–van der Vlugt curves.

Here is the story of what the authors did, explained without the heavy math jargon.

1. The Problem: A Missing Piece of the Puzzle

Mathematicians have known how to calculate the blueprints for these curves when the numbers they are built with are "odd" (like 3, 5, 7). However, there was a stubborn gap: they didn't know how to do it when the numbers were even (specifically, characteristic 2, which is like a world where $1 + 1 = 0$).

Think of it like having a perfect recipe for a cake that works for chocolate and vanilla, but when you try to make a "binary" cake (using only 0s and 1s), the oven behaves completely differently, and the old recipe fails. The authors set out to write a new recipe specifically for this "binary oven."

2. The Key Players: The Heisenberg Group and the "Magic Mirror"

To solve this, the authors looked at the hidden symmetries of these curves. They found that these curves are guarded by a mathematical structure called a Heisenberg group.

• The Analogy: Imagine the curve is a castle. The Heisenberg group is the secret society of knights that can rotate and flip the castle without breaking it.
• The Twist: In the "odd" world, these knights move in a simple, predictable way. In the "even" (characteristic 2) world, their movements are more chaotic and complex. They discovered that in this specific world, the knights' movements are governed by something called Witt vectors.

Witt vectors are like a special "magic mirror" that takes a simple number and reflects it into a more complex, layered version of itself. The authors realized that to understand the curve in characteristic 2, they had to stop looking at the curve directly and start looking at it through this "Witt vector mirror."

3. The Breakthrough: The "Lang Torsor" Bridge

The authors built a bridge between the complex curve and a simpler object using a tool called a Lang torsor.

• The Analogy: Imagine you are trying to count the number of people in a massive, foggy stadium (the curve). It's impossible to see everyone. But, you notice that the stadium is built on top of a simple, flat parking lot (a line).
• The Method: Instead of counting the people in the stadium directly, the authors found a way to map the stadium down to the parking lot. They realized that the "fog" (the complex cohomology) could be broken down into simple, manageable chunks using the "magic mirror" (Witt vectors).

By doing this, they could calculate the Frobenius eigenvalues.

• What are these? Think of the Frobenius eigenvalue as the "fingerprint" of the curve. It tells you exactly how many "rooms" (points) the building has when you visit it with different keys (different finite fields). If you know the fingerprint, you know the building's entire structure.

4. The Result: A New Formula

The authors successfully derived an explicit formula for these fingerprints in the "even" world.

• The Formula: It looks complicated, but essentially, it says: "To find the fingerprint of this complex curve, you just need to look at a specific character (a type of code) on the Witt vector group."
• Why it matters: Before this, if you wanted to know the properties of these curves in characteristic 2, you were stuck guessing or doing brute-force calculations. Now, there is a clear, step-by-step instruction manual.

5. The Application: Building "Perfect" Curves

The most exciting part of the paper is what they did with this new formula. They used it to construct Maximal Curves.

• The Goal: In coding theory (which helps send data over the internet without errors), mathematicians want to build "perfect" curves. A maximal curve is one that has the absolute maximum number of points possible for its size. It's the most efficient building you can possibly design.
• The Trick: The authors found a way to take a "minimal" curve (one with the fewest possible points) and "twist" it.
• The Analogy: Imagine you have a small, inefficient house. By applying a specific mathematical "twist" (adding a specific term to the equation), they transformed it into a massive, ultra-efficient skyscraper that hits the theoretical limit of efficiency.

They showed that by twisting a minimal curve in a specific way (using a parameter $t$ where the "trace" is 1), you instantly get a maximal curve. This is a huge deal for creating better error-correcting codes for digital communication.

Summary

In short, this paper is about:

1. Solving a mystery: Figuring out how to calculate the properties of a specific family of mathematical curves when the numbers behave like binary code ($1+1=0$).
2. Using a new tool: Instead of fighting the complexity, they used a "Witt vector mirror" to simplify the problem.
3. Building better tech: Using their new math to design "perfect" mathematical shapes that can be used to make our digital communications faster and more reliable.

They took a problem that was stuck in the dark for the "even" number world and turned on the light, showing us exactly how these complex shapes behave.

Technical Summary

Here is a detailed technical summary of the paper "The L-polynomials of van der Geer–van der Vlugt curves in characteristic 2" by Ito, Takeuchi, and Tsushima.

1. Problem Statement

The paper addresses the explicit computation of L-polynomials for a specific class of algebraic curves known as van der Geer–van der Vlugt curves in characteristic 2.

• Context: These curves, denoted $C_R$, are defined over a finite field $\mathbb{F}_q$ (where $q$ is a power of 2) by the affine equation $y^p - y = xR(x)$, where $R(x)$ is an $\mathbb{F}_p$-linearized polynomial of degree $p^e$ ($e \ge 1$).
• The Gap: Previous work by the authors (specifically in [15]) provided an explicit formula for the L-polynomials of these curves in odd characteristic ($p_0 \neq 2$). In that case, the Frobenius eigenvalues could be computed by factoring the covering map through an intermediate curve defined by a quadratic equation, reducing the problem to quadratic Gauss sums.
• The Obstacle: The method used for odd characteristics relies on the fact that the maximal abelian subgroups of the associated Heisenberg groups are annihilated by the prime $p_0$. In characteristic 2, this condition fails (elements have order 4, not 2), rendering the previous factorization technique invalid. The paper aims to develop a new method to compute these eigenvalues specifically for the characteristic 2 case.

2. Methodology

The authors employ a geometric and representation-theoretic approach, leveraging the structure of Heisenberg groups and Lang torsors over Witt vectors.

• Heisenberg Group Structure: The curve $C_R$ admits an action by a Heisenberg group $H_R$. The authors analyze the maximal abelian subgroups $A \subset H_R$. In characteristic 2, these subgroups exhibit 4-torsion, distinguishing them from the odd characteristic case.
• Factorization via Intermediate Curves: The core innovation is proving that the covering $C_R \to \mathbb{A}^1_{\mathbb{F}_q}$ factors through an intermediate curve $C_S$:
  $$C_R \to C_S \to \mathbb{A}^1_{\mathbb{F}_q}$$
  Here, $C_S$ is a van der Geer–van der Vlugt curve associated with a binomial $S(x) = x^p + s_0 x$.
• Witt Vectors and Lang Torsors: The authors identify the Galois group of the covering $C_S \to \mathbb{A}^1_{\mathbb{F}_q}$ with the group of Witt vectors of length 2, $W_2(\mathbb{F}_p)$. They utilize the Lang torsor over $W_2$ (defined by the map $(x, y) \mapsto (x^p, y^p) - (x, y)$) to construct the relevant $\ell$-adic sheaves.
• Cohomological Decomposition: Using the factorization, they decompose the $\ell$-adic cohomology of $C_R$ into sheaves on $\mathbb{A}^1_{\mathbb{F}_q}$. They show that the sheaf associated with a character $\xi$ is isomorphic to a pullback of a Lang torsor sheaf twisted by specific polynomials.
• Trace Formula: The Frobenius eigenvalues are computed using the Grothendieck–Lefschetz trace formula applied to these specific sheaves, relating them to additive characters of the Witt vector group.

3. Key Contributions

A. Explicit Formula for Frobenius Eigenvalues (Theorem 1.1)

The paper provides an explicit formula for the Frobenius eigenvalue $\tau_{R, \xi, q}$ acting on the cohomology of $C_R$.

• The eigenvalue is expressed in terms of a character $\xi_q$ of the Witt vector group $W_2(\mathbb{F}_q)$ and a specific element $c_{R, \xi} \in \mathbb{F}_q$.
• Formula:
  $$\tau_{R, \xi, q} = \xi_q(c_{R, \xi}, 0)^{-1} \cdot (-1 - \sqrt{-1})^{[\mathbb{F}_q : \mathbb{F}_2]}$$
  where $\sqrt{-1}$ is a primitive 4th root of unity derived from a faithful character of $W_2(\mathbb{F}_2)$.
• This formula replaces the quadratic Gauss sums used in odd characteristic with values derived from the geometry of Witt vectors.

B. Relation Between Twists (Theorem 1.2)

The authors establish a precise relationship between the Frobenius eigenvalues of a curve $C_R$ and its twist $C_{Rt}$ (defined by $R(x) + F^*(t)^2 x$).

• Result: The eigenvalue of the twist is the product of the original eigenvalue and a specific 4th root of unity:
  $$\tau_{Rt, \xi_t, q} = \xi_q(t, t^2 + c_{R, \xi_t}) \cdot \tau_{R, \xi, q}$$
• This allows for the systematic generation of new curves with known spectral properties from existing ones.

C. Construction of Maximal Curves (Theorem 1.3)

As a major application, the paper constructs explicit examples of maximal curves (curves attaining the Hasse–Weil upper bound on the number of rational points).

• By choosing specific parameters $t$ such that $\text{Tr}_{q/2}(t) = 1$, the authors show that twisting a minimal curve (attaining the lower bound) yields a maximal curve.
• They provide concrete examples of such curves, including quotients of the form $z^2 + z = xQ_a(x)$, which are proven to be $\mathbb{F}_{q^2}$-maximal.

4. Key Results

1. Geometric Factorization: For any van der Geer–van der Vlugt curve $C_R$ in characteristic 2, there exists a finite étale Galois covering $C_R \to C_S$ where $C_S$ is defined by $y^p + y = x^{p+1} + s_0 x^2$. The Galois group of $C_S \to \mathbb{A}^1$ is $W_2(\mathbb{F}_p)$.
2. Sheaf Isomorphism: The $\ell$-adic sheaf $Q_\xi$ associated with the covering $C_R \to \mathbb{A}^1$ is isomorphic to $L_{\xi_{W_2}}(s, \alpha s^2 + \beta_\xi s)$, where $L$ denotes a sheaf derived from the Lang torsor over $W_2$.
3. Supersingular Elliptic Quotients: The paper demonstrates that supersingular elliptic curves naturally appear as quotients of these van der Geer–van der Vlugt curves. Specifically, quotients of $C_S$ by certain subgroups of $W_2(\mathbb{F}_p)$ yield elliptic curves with equations of the form $z^2 + z = x^3 + (c^2+c+1)x^2$.
4. Maximality Criterion: The paper proves that if $C_R$ is minimal over $\mathbb{F}_{q^2}$, then a specific twist $C_{Rt}$ is maximal over $\mathbb{F}_{q^2}$, provided the trace condition on $t$ is met.

5. Significance

• Resolution of a Characteristic 2 Case: The paper fills a critical gap in the arithmetic geometry of Artin–Schreier curves by solving the characteristic 2 case, which was previously intractable using methods designed for odd characteristics.
• New Geometric Tools: It introduces the use of Witt vector torsors and the specific geometry of $W_2$ to study $\ell$-adic cohomology in characteristic 2, offering a new toolkit for similar problems.
• Explicit Maximal Curves: The construction of explicit maximal curves is of significant interest in coding theory (e.g., for constructing algebraic geometry codes with optimal parameters) and number theory. The ability to generate these curves via twisting provides a systematic construction method rather than relying on ad-hoc examples.
• Local Langlands Correspondence: As noted in the introduction, understanding the Frobenius eigenvalues of these curves sheds light on the explicit local Langlands correspondence, particularly regarding the reductions of affinoids in Lubin–Tate space.

In summary, this paper successfully extends the theory of van der Geer–van der Vlugt curves to characteristic 2 by developing a novel geometric framework involving Witt vectors, resulting in explicit formulas for L-polynomials and the construction of new families of maximal curves.