🔢 mathematics

Collision of orbits for families of polynomials defined over fields of positive characteristic

This paper establishes explicit necessary and sufficient conditions for the infinitude of parameters $\lambda$ in positive characteristic fields where the orbits of two points under the polynomial family $z^d+\lambda$ collide at a specific third point, while also proposing a conjecture for the case where all points lie in a finite field.

Original authors: Shamil Asgarli, Dragos Ghioca

Published 2026-04-02

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Shamil Asgarli, Dragos Ghioca

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: A Cosmic Dance of Numbers

Imagine you are watching a cosmic dance floor. On this floor, there are three dancers: Dancer A, Dancer B, and a Target Spot (let's call it the "Finish Line").

There is a DJ, Lambda ( $\lambda$ ), who controls the music. Every time the DJ changes the song (changes the value of $\lambda$ ), the rules of the dance change slightly. The dancers follow a specific set of steps defined by a polynomial function (a mathematical recipe).

The Dance: Dancer A starts at one spot, and Dancer B starts at another. They take steps according to the DJ's rules.
The Goal: We are looking for a specific song (a specific $\lambda$ ) where both Dancer A and Dancer B eventually land on the Finish Line at the same time (or at different times, but they both get there).
The Question: If we keep changing the DJ's songs, will we find infinitely many songs where this collision happens? Or is it a rare fluke that only happens a few times?

The Setting: A World with a "Loop" (Positive Characteristic)

Most math problems like this are studied in a "normal" world (like the real numbers), where you can keep counting up forever without repeating.

This paper, however, takes place in a Positive Characteristic world. Think of this world like a giant clock or a video game loop.

In a normal world, if you keep adding 1, you get 1, 2, 3, 4... forever.
In this "clock" world, if you keep adding 1, eventually you wrap around. $p + 1$ becomes $1$ again. The numbers are finite and cyclical. This makes the math much trickier because the "steps" the dancers take can get stuck in loops or behave in surprising ways.

The Main Discovery: When Do They Collide Forever?

The authors, Shamil Asgarli and Dragos Ghioca, wanted to know: Under what conditions will there be an infinite number of songs ( $\lambda$ ) where Dancer A and Dancer B both hit the Finish Line?

They found that "unlikely intersections" (finding infinitely many songs where this happens) only occur if the dancers are dynamically related. In plain English, they must be "connected" by the rules of the dance.

There are only two main scenarios where this infinite collision happens:

Scenario 1: The Twins (Condition A)

The Analogy: Imagine Dancer A and Dancer B are identical twins. No matter what song the DJ plays, they take the exact same steps.

The Math: This happens if $A^d = B^d$ .
The Result: If they start in a way that makes them "twins" under the dance rules, they will always merge into the same path. If they ever hit the Finish Line, they do it together. This guarantees an infinite number of songs where they collide.

Scenario 2: The Clock Walkers (Condition B)

The Analogy: Imagine the dance floor is a giant clock face. Dancer A and Dancer B are walking around the clock.

The Math: This only happens if the number of steps in the dance ( $d$ ) is a power of the clock's size ( $p$ ).
The Twist: In this specific "clock world," if the distance between the dancers and the Finish Line is a "simple" number (a number that fits neatly into the clock's cycle), then the dancers are essentially walking in sync with the clock's rhythm.
The Result: Because they are moving in perfect harmony with the clock's loops, there are infinitely many songs where they all land on the Finish Line.

The Surprise: If the dancers are not twins (Scenario 1) and they are not walking in perfect clock harmony (Scenario 2), then finding a song where they both hit the Finish Line is a rare event. You might find a few songs, but you will never find infinitely many. It's like trying to win the lottery every day; eventually, the math says it stops happening.

The "Finite Field" Mystery (The Computer Experiment)

The paper also tackles a weird edge case: What if the dancers, the Finish Line, and the DJ are all living inside a tiny, finite village (a finite field)?

The Intuition: In a tiny village, everything repeats. You might think, "If the village is small, maybe the collision stops happening after a while."
The Conjecture: The authors ran massive computer simulations (over 100 examples). They found that even in these tiny villages, if the dance rules aren't "special" (like the clock scenario), the collisions still happen infinitely often as you look at larger and larger versions of the village.
The Metaphor: It's like saying, "Even if we are in a small town, if we keep building bigger and bigger towns, we will keep finding days where everyone meets at the town square." The computer data suggests this is true, even though the math is too hard to prove it yet.

Why Does This Matter?

This paper is part of a field called Arithmetic Dynamics, which studies how numbers move and interact.

The "Unlikely Intersection" Principle: In math, we often ask, "How often do two things that shouldn't meet, actually meet?" Usually, the answer is "Never, unless there's a hidden reason." This paper proves that in this specific "clock world," the hidden reasons are very specific (Twins or Clock Walkers).
New Rules for New Worlds: Most math is done in "normal" worlds. This paper shows that when you move to "clock worlds" (positive characteristic), the rules change. Things that are impossible in the normal world become possible here, and things that seem random turn out to have hidden structures.

Summary in One Sentence

The paper proves that in a mathematical world where numbers loop like a clock, two paths will collide infinitely often only if the starting points are secretly "twins" or if they are perfectly synchronized with the clock's rhythm; otherwise, such collisions are rare, though computer experiments suggest they might still happen surprisingly often in tiny, finite villages.

1. Problem Statement

The paper investigates the "colliding orbits" problem within arithmetic dynamics over fields of positive characteristic $p$ . Specifically, it considers the family of polynomials $f_\lambda(z) = z^d + \lambda$ parameterized by $\lambda \in L$ , where $L$ is a field of characteristic $p > 0$ .

Given three points $\alpha_1, \alpha_2, \beta \in L$ , the authors define the set:
$C(\alpha_1, \alpha_2; \beta) := \{ \lambda \in L \mid \exists m, n \in \mathbb{N} \text{ such that } f_\lambda^m(\alpha_1) = f_\lambda^n(\alpha_2) = \beta \}$
This set represents the parameters $\lambda$ for which the forward orbits of $\alpha_1$ and $\alpha_2$ collide precisely at the target point $\beta$ under the same polynomial map.

The central question is: Under what conditions is the set $C(\alpha_1, \alpha_2; \beta)$ infinite?
This is framed as an "unlikely intersection" problem. In characteristic 0, such infinite intersections typically only occur if the points are dynamically related (e.g., one is a preimage of the other globally). The authors explore how this principle holds, fails, or manifests differently in positive characteristic, particularly when points lie in finite subfields or when the degree $d$ is a power of $p$ .

2. Methodology

The authors employ a combination of algebraic dynamics, height functions, and Diophantine geometry techniques adapted to positive characteristic.

Transcendence Degree Reduction: They first analyze the algebraic dependence of $\alpha_1, \alpha_2, \beta$ . Using properties of iterates $P_{n,\alpha}(\lambda)$ , they show that if $C(\alpha_1, \alpha_2; \beta)$ is infinite, the points must either be algebraically dependent in a specific way or lie in a finite field.
Canonical Heights (Global and Local): A core tool is the variation of the canonical height $\hat{h}_\lambda(\alpha)$ $\hat{h}_{λ} (α)$ in families. They establish bounds relating $\hat{h}_\lambda(\alpha)$ $\hat{h}_{λ} (α)$ to the Weil height $h(\lambda)$ $h (λ)$ and $h(\alpha)$ $h (α)$ (Proposition 5.2).
- They prove that if the set is infinite, there exists a sequence of parameters $\lambda_k$ such that the canonical heights of $\alpha_1$ and $\alpha_2$ tend to zero (Proposition 6.2).
- This allows them to invoke a theorem (Theorem 6.3, based on [Ghi26]) stating that if canonical heights vanish along a sequence, the points must have equal local canonical heights for all $\lambda$ .
Case Analysis by Degree:
- Case $d \neq p^\ell$ : They utilize the equality of canonical heights to deduce that $\alpha_1^d = \alpha_2^d$ .
- Case $d = p^\ell$ : This is the most complex case. Here, the map $f_\lambda$ commutes with translations by elements of $\mathbb{F}_p$ . The authors derive explicit formulas for iterates (Equation 6.4.1) and analyze the recurrence relations of the differences between orbit points.
Diophantine Equations and Tori: For the converse direction (proving infinitude), they reduce the problem to solving systems of equations over finite fields. In the case where $d$ is not a power of $p$ and points are in $\mathbb{F}_p$ , they use a contradiction argument involving the intersection of a curve with a finitely generated subgroup of a torus (Moosa-Scanlon theorem [MS04]) to show that assuming finiteness leads to a contradiction regarding the growth of degrees of polynomials.
Computational Experiments: For the case where all points lie in a finite field $\mathbb{F}_p$ and $d$ is not a power of $p$ , the theoretical tools are insufficient. The authors develop an algorithm to compute common factors of iterate polynomials over finite fields, providing extensive numerical evidence for a conjecture.

3. Key Contributions and Results

Main Theorem (Theorem 1.1)

Assuming $\alpha_1, \alpha_2, \beta$ are not all contained in a finite subfield of $L$ , the set $C(\alpha_1, \alpha_2; \beta)$ is infinite if and only if exactly one of the following holds:

Condition (A): $\alpha_1^d = \alpha_2^d$ $α_{1}^{d} = α_{2}^{d}$ .
- Implication: The orbits merge after one iteration for all $\lambda$ ( $f_\lambda(\alpha_1) = f_\lambda(\alpha_2)$ ).
Condition (B): $d = p^\ell$ $d = p^{ℓ}$ for some $\ell \in \mathbb{N}$ $ℓ \in N$ , and there exists a finite subfield $\mathbb{F}_q \subset L$ $F_{q} \subset L$ such that $\delta_1 = \alpha_2 - \alpha_1 \in \mathbb{F}_q^*$ $δ_{1} = α_{2} - α_{1} \in F_{q}^{*}$ and $\delta_2 = \beta - \alpha_1 \in \mathbb{F}_q$ $δ_{2} = β - α_{1} \in F_{q}$ . Furthermore, the system:
$\begin{cases} \delta_1 = \sum_{i=0}^{s_1-1} \gamma^{p^{ik\ell}} \\ \delta_2 = \sum_{i=0}^{s_2-1} \gamma^{p^{ik\ell}} \end{cases}$
must have a solution $(\gamma, k, s_1, s_2) \in \mathbb{F}_q^* \times \mathbb{N}^4$ $(γ, k, s_{1}, s_{2}) \in F_{q}^{*} \times N^{4}$ .
- Implication: This relies on the commutativity of $f_\lambda$ with translations in characteristic $p$ .

Converse Results

Theorem 2.4: If $\alpha_1^d = \alpha_2^d$ , then $C(\alpha_1, \alpha_2; \beta)$ is infinite for any $\beta$ .
Theorem 2.2: Provides the precise necessary and sufficient conditions for the infinitude of $C(\alpha_1, \alpha_2; \beta)$ when $d=p^\ell$ and the points satisfy the finite field difference condition. It proves that if the system above has no solution, the set is empty.

The Finite Field Case (Conjecture 9.1)

When $\alpha_1, \alpha_2, \beta \in \mathbb{F}_p$ and $d$ is not a power of $p$ , the authors propose a surprising conjecture based on computational data:

Conjecture 9.1: The set $C(\alpha_1, \alpha_2; \beta)$ is infinite.
Significance: This contradicts the intuition from characteristic 0 (where such intersections are rare without global relations). The authors suggest that in finite fields, the combinatorial structure forces collisions to happen frequently as the field size grows.

Generalization (Conjecture 2.6)

The authors formulate a broad conjecture for arbitrary families of polynomials in normal form, suggesting that infinite collisions occur only if:

The points are dynamically related via a commuting family of polynomials.
The target is in the orbit of a starting point globally.
The polynomial is additive (related to $d=p^\ell$ ).
All points and coefficients lie in a finite field (the "finite field phenomenon").

4. Significance

Positive Characteristic Dynamics: The paper highlights fundamental differences between arithmetic dynamics in characteristic 0 and $p$ . In characteristic $p$ , the Frobenius map and additive polynomials create "unlikely" infinite intersections that do not exist in characteristic 0.
Refinement of Unlikely Intersections: It extends the "unlikely intersection" principle (Zannier's school) to positive characteristic, showing that "unlikely" events can become "likely" when restricted to finite fields or specific characteristic-dependent structures.
New Techniques: The work bridges the gap between height machinery (typically used for transcendence) and finite field combinatorics. The use of the Moosa-Scanlon theorem on tori intersections in characteristic $p$ is a key technical achievement.
Open Problems: The paper opens new avenues for research, particularly regarding the behavior of orbits in finite fields (Conjecture 9.1) and the collision of more than two orbits (Question 9.10).

In summary, this paper provides a complete classification for the collision of orbits for the family $z^d + \lambda$ in positive characteristic, revealing that while global dynamical relations are the primary driver for infinite intersections, the specific arithmetic of finite fields and the Frobenius map can also force such collisions to occur infinitely often.