🔢 mathematics

On the existence of a morphism between certain Artin-Schreier curves

This paper investigates the converse of a known result regarding morphisms between Artin-Schreier curves over $\F_p$ , proving that the existence of a morphism from $\mathcal{X}: y^p+cy=x^{p^k+1}$ to $\mathcal{Y}:y^p+cy=x^{p^\ell+1}$ implies that the exponent of the target curve divides that of the source curve under specific hypotheses, while addressing both Galois and non-Galois cases.

Original authors: Beatriz Barbero Lucas, Stefano Lia, Gary McGuire

Published 2026-02-18

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Beatriz Barbero Lucas, Stefano Lia, Gary McGuire

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 an architect designing a series of unique, complex buildings called Artin-Schreier curves. These aren't made of brick and mortar, but of mathematical equations. Specifically, these buildings have a specific shape defined by the formula $y^p + cy = x^{power}$ .

The "power" in the exponent is the most important feature of the building. Let's call the power of the first building $M$ and the power of the second building $N$ .

The Big Question: Can You Shrink a Building?

Mathematicians have long known a simple rule: If you have a building with a huge power ( $M$ ) and another with a smaller power ( $N$ ), and $N$ divides $M$ (meaning $M$ is a perfect multiple of $N$ , like 12 and 4), you can easily build a bridge (a morphism) from the big building to the small one. It's like taking a large, detailed map and folding it down to fit a smaller, simpler map. The structure holds up perfectly.

The paper asks the reverse: If you can build a bridge from a big building to a small one, does that guarantee that the small number must divide the big number?

In the world of these specific curves, the authors suspect the answer is YES. They believe that if a bridge exists, the numbers must be related by division. If they aren't, the bridge simply cannot be built, no matter how hard you try.

The Investigation: Two Types of Bridges

The authors investigate this by looking at two specific types of bridges:

The "Standard" Bridge (Non-Galois): This is a general connection. The authors prove that if you build a bridge that doesn't get "stuck" or "twisted" in a specific way (totally ramified at the top), and the bridge isn't made of "p-sized" blocks, then the rule holds: the smaller number must divide the larger one.
- Analogy: Imagine trying to pour water from a giant bucket (Big Curve) into a small cup (Small Curve). If the water flows perfectly without spilling or getting stuck, the size of the cup must be a perfect divisor of the bucket's capacity.
The "Symmetrical" Bridge (Galois): This is a bridge where the big building has a lot of internal symmetry, and the small building is just the big one "folded" perfectly by a group of symmetries. This is a much stricter condition.
- Analogy: Imagine a kaleidoscope. If you look through a kaleidoscope with a complex pattern (Big Curve) and see a simpler pattern (Small Curve), the simpler pattern must be a perfect, symmetrical slice of the complex one.

The Surprising Findings

The authors found some very interesting "impossible" scenarios where you might think a bridge should exist, but math says NO.

The "Even Power" Trap: They proved that if the power of the big building is an even number (like $2^a$ ), you can never build a symmetrical bridge to a smaller building of the same family, even if the math of their "blueprints" (L-polynomials) suggests they should match.
- Metaphor: It's like having a puzzle piece that looks like it fits the hole perfectly, but when you try to snap it in, it just bounces off. The shapes are "too even" to fit together in this specific symmetrical way.
The "Square Root" Rule: They found that if the power of the big building is less than the square of the small building's power (e.g., Big=100, Small=11), a symmetrical bridge is impossible unless the numbers are in that perfect division relationship we mentioned earlier.
- Metaphor: You can't shrink a skyscraper down to a bungalow just because they look similar from a distance. The scale difference is too massive to bridge with a symmetrical fold.

Why Does This Matter?

In the world of mathematics, there is a famous rule called the Kleiman-Serre Theorem. It says: "If the blueprints of Building A contain the blueprints of Building B, then you can build a bridge from A to B."

The authors of this paper found a massive family of curves where this rule fails. They showed cases where the blueprints match perfectly (the L-polynomials divide each other), but no bridge can be built.

The Takeaway:
This paper is like a detective story in the world of shapes. The authors are proving that for these specific mathematical curves, the "divisibility" rule is much stricter than we thought. You can't just have similar shapes; the numbers governing their size must be perfectly aligned. If they aren't, the connection is impossible, no matter how much the rest of the math looks like it should work.

They have essentially drawn a "No Entry" sign for a huge number of potential mathematical bridges, clarifying exactly when these complex shapes can talk to each other and when they are destined to remain strangers.

1. Problem Statement

The paper investigates the existence of non-constant morphisms between specific families of algebraic curves defined over finite fields.

Context: Let $K$ be the algebraic closure of $\mathbb{F}_p$ (where $p$ is an odd prime). Consider the family of curves $\mathcal{C}$ defined by the affine equation $y^p + cy = x^m$ , where $c \neq 0$ and $\gcd(m, p)=1$ .
Known Fact: If $n$ divides $m$ , there exists a non-constant morphism $C_m \to C_n$ . This is induced by the function field extension $K(x^{m/n}, y) \subset K(x, y)$ .
The Converse Question: The authors ask if the converse holds: If a morphism $C_m \to C_n$ exists, must $n$ divide $m$ ?
Specific Focus: The paper narrows the scope to a subfamily $\mathcal{B}$ where the exponents are of the form $p^k + 1$ . Specifically, they study curves $B_k$ defined by $y^p + cy = x^{p^k+1}$ .
Conjecture: The authors conjecture that a morphism $B_k \to B_\ell$ exists if and only if $p^\ell + 1$ divides $p^k + 1$ . (Note: By number theory, $p^\ell + 1 \mid p^k + 1$ if and only if $\ell \mid k$ and $k/\ell$ is odd).

2. Methodology

The authors employ a combination of algebraic geometry, function field theory, and elementary number theory to analyze the existence of morphisms.

Riemann-Hurwitz Formula: Used to establish upper bounds on the degree $d$ of a morphism $\phi: B_k \to B_\ell$ . The formula relates the genera of the curves and the ramification indices: $2g(B_k) - 2 = d(2g(B_\ell) - 2) + \Delta$ .
Function Field Analysis:
- The authors analyze the pullback of generators of the function field $K(B_\ell)$ into $K(B_k)$ .
- They utilize the structure of Riemann-Roch spaces $L(nP_\infty)$ for the curves $B_k$ . A basis for these spaces is known (Theorem 2.3), consisting of monomials $x^i y^j$ with specific pole order constraints.
- By expressing the pullback of the generators in terms of this basis, they derive constraints on the degrees of polynomials involved in the morphism.
Galois vs. Non-Galois Morphisms:
- Non-Galois Case: The authors prove the converse for morphisms that are totally ramified at the point at infinity and have a degree coprime to $p$ .
- Galois Case: They restrict the study to $c = \pm 1$ (Artin-Schreier curves) and assume the morphism is Galois, meaning $B_\ell \cong B_k / G$ for some subgroup $G \leq \text{Aut}(B_k)$ . They utilize the precise description of the automorphism group $\text{Aut}(B_k)$ (from [HKT08]), which is a semi-direct product of a cyclic group and an elementary abelian $p$ -group.
Genus Computation: For the Galois case, they compute the genus of the quotient curve $B_k/G$ using the Hilbert Different Formula to calculate the different $\Delta$ . This allows them to equate the genus of the quotient to the genus of $B_\ell$ .
Number Theoretic Lemmas: The core of the non-existence proofs relies on solving Diophantine equations derived from the genus equality. They use properties of $\gcd(p^a+1, p^b+1)$ and bounds on exponents to show that certain equations have no integer solutions.

3. Key Contributions and Results

A. Non-Galois Morphisms (Theorem 1.1 / Theorem 3.5)

The paper provides a definitive answer for a specific class of morphisms:

Existence: If $p^\ell + 1 \mid p^k + 1$ , a morphism exists with degree $d = \frac{p^k+1}{p^\ell+1}$ .
Non-Existence (Converse): If a morphism $\phi: B_k \to B_\ell$ $ϕ : B_{k} \to B_{ℓ}$ exists such that:
1. $\phi$ maps the point at infinity to the point at infinity ( $P_k \to P_\ell$ ).
2. $\phi$ is totally ramified at $P_k$ .
3. $\deg(\phi)$ is coprime to $p$ .
  Then, it must be true that $p^\ell + 1 \mid p^k + 1$ and $\deg(\phi) = \frac{p^k+1}{p^\ell+1}$ .

Significance: This establishes that for this specific family, the divisibility of the exponents $p^k+1$ is a necessary condition for the existence of "well-behaved" morphisms.

B. Galois Morphisms (Theorems 1.2, 1.3, 4.2)

The authors investigate when $B_\ell$ can be a quotient of $B_k$ by a subgroup of automorphisms.

General Non-Existence: They prove that if $k = 2^a$ (a power of 2), there is no Galois morphism from $B_k$ to any $B_\ell$ (Corollary 7.1).
Specific Non-Existence: There is no Galois morphism $B_{2\ell} \to B_\ell$ for any integer $\ell > 2$ (Corollary 7.3).
Condition for Existence: A Galois morphism $B_k \to B_\ell$ exists only if $p^\ell + 1 \mid p^k + 1$ (which implies $\ell \mid k$ and $k/\ell$ is odd) AND the subgroup $G$ satisfies specific structural constraints.
Counter-Example to Kleiman-Serre: The paper highlights a significant phenomenon regarding the Kleiman-Serre theorem, which states that if a morphism exists, the $L$ $L$ -polynomial of the target must divide the $L$ $L$ -polynomial of the source.
- It is known that for $c=-1$ , $L_{B_\ell}(t) \mid L_{B_k}(t)$ if and only if $\ell \mid k$ .
- However, the authors show that even if $\ell \mid k$ (so the $L$ -polynomials divide), a morphism does not necessarily exist (e.g., when $k/\ell$ is even, or specifically for $B_{2\ell} \to B_\ell$ ).
- This provides an infinite family of explicit examples where the converse of the Kleiman-Serre theorem fails.

C. Technical Lemmas

The paper develops several lemmas (6.1–6.5) involving the equation:
$m(p^\ell t + 1) = t(p^{k-r} + 1)$
derived from the genus equality of the quotient. These lemmas prove that under the constraints of the problem (e.g., $k < \ell^2$ , $k/\ell$ even), this equation has no integer solutions, thereby proving the non-existence of the corresponding Galois covers.

4. Significance

Resolution of a Converse Problem: The paper makes significant progress on the converse of the standard morphism existence result for Artin-Schreier curves, proving it true under strong conditions (totally ramified, degree coprime to $p$ ).
Clarification of the Kleiman-Serre Theorem: The results demonstrate that the divisibility of $L$ -polynomials is a necessary but not sufficient condition for the existence of morphisms between these curves. This refines the understanding of the relationship between arithmetic invariants (L-polynomials) and geometric structures (morphisms).
Classification of Galois Covers: The paper provides a near-complete classification of when one curve in this family can be a Galois quotient of another, identifying specific infinite subfamilies (like $k=2^a$ ) where no such morphisms exist.
Methodological Contribution: The combination of Riemann-Roch basis analysis with precise genus calculations for quotient curves offers a robust framework for studying morphisms between curves with large automorphism groups.

In summary, the paper proves that for the family of curves $y^p + cy = x^{p^k+1}$ , the existence of a morphism is tightly constrained by the arithmetic of the exponents $p^k+1$ , specifically requiring $p^\ell+1 \mid p^k+1$ for "nice" morphisms, and showing that the divisibility of L-polynomials alone is insufficient to guarantee geometric maps.