The number of rational iterated preimages of the origin under unicritical polynomial maps

Basic Information

• Paper ID: 2510.14397
• Title: The number of rational iterated preimages of the origin under unicritical polynomial maps
• Author: Kaoru Sano (NTT Institute for Fundamental Mathematics)
• Classification: math.NT (Number Theory), math.AG (Algebraic Geometry), math.DS (Dynamical Systems)
• Publication Date: October 17, 2025
• Paper Link: https://arxiv.org/abs/2510.14397v1

Abstract

This paper investigates rational iterated preimages of the origin under unicritical maps $f_{d,c}(x) = x^d + c$. Building upon earlier work by Faber-Hutz-Stoll and Hutz-Hyde-Krause establishing finiteness and conditional bounds in the quadratic case, we prove that for $d=2$ and $c \in \mathbb{Q}\setminus\{0,-1\}$, the origin has no rational quartic preimages; for all $d \geq 3$, there are no rational quadratic preimages except in trivial cases. The proof relies on geometric analysis of preimage curves, the elliptic Chabauty method, and Diophantine reduction. Consequently, we determine the number of rational iterated preimages of the origin under $f_{d,c}$ for all $d \geq 2$.

Research Background and Motivation

Problem Description

The core problem studied in this paper is to determine the number of rational iterated preimages of the origin under the unicritical polynomial map $f_{d,c}: x \mapsto x^d + c$. Specifically, for a given number field $L$, we are concerned with the cardinality of the set $\bigcup_{N \geq 1} f_{d,c}^{-N}(a)(L) = \{x \in A^1(L) | f_{d,c}^{\circ N}(x) = a \text{ for some } N \geq 1\}$ where $f_{d,c}^{\circ N}$ denotes the $N$-fold iterate of $f_{d,c}$.

Research Significance

1. Dynamical Systems Theory: This problem is closely related to the distribution of rational points in polynomial dynamical systems and represents a fundamental question in algebraic dynamics.
2. Arithmetic Geometry: It involves the distribution of rational points on higher-degree algebraic curves, connecting number theory and algebraic geometry.
3. Galois Representation Theory: From the perspective of tree-like representations, this relates to the action of the absolute Galois group.

Limitations of Existing Work

• Previous work by Faber-Hutz-Stoll and others focused primarily on the quadratic case ($d=2$)
• Results on the non-existence of quartic preimages previously relied on the Birch-Swinnerton-Dyer conjecture
• Systematic study of cases with $d \geq 3$ was lacking

Core Contributions

1. Unconditional Proof: We prove that for $d=2$ and $c \in \mathbb{Q}\setminus\{0,-1\}$, $f_{2,c}$ has no rational quartic preimages of the origin, removing the previous dependence on the BSD conjecture.
2. Complete Resolution of Higher Degree Cases: We prove that for $d \geq 3$, there are no rational quadratic preimages of the origin except in trivial cases.
3. Complete Classification Result: We provide exact formulas for the number of rational iterated preimages for all $d \geq 2$.
4. Methodological Innovation: We combine the elliptic Chabauty method, geometric analysis, and Diophantine reduction in a unified approach.

Detailed Methods

Main Results

Theorem 1.3:

• (i) For rational $c \in \mathbb{Q}\setminus\{0,-1\}$, the map $f_{2,c}$ has no rational quartic preimages of the origin.
• (ii) For integers $d \geq 3$ and rational $c \in \mathbb{Q}$, under appropriate conditions, the map $f_{d,c}$ has no rational quadratic preimages of the origin.

Technical Approach

Proof Strategy for the Quadratic Case ($d=2$)

1. Geometric Reduction: Utilizing a degree-2 covering map $\pi: X^{\text{pre}}(2,4,0) \to X^{\text{pre}}(2,3,0) \cong E$ where $E: v^2 = u^3 - u + 1$ is an elliptic curve.
2. Ramification Analysis: Analyzing ramification over $E[2]$ and constructing appropriate base changes to obtain finite flat double coverings.
3. Curve Reduction: Reducing the problem to lower-genus curves $C_D$, where $D$ is classified by classes $[D] \in R^\times/(R^\times)^2$, with $R = \mathbb{Z}[1/2, 1/23]$.

Application of the Elliptic Chabauty Method

Theorem 1.5: For $D = (-1)^{\varepsilon_0} 2^{\varepsilon_1} 23^{\varepsilon_2}$ ($\varepsilon_i \in \{0,1\}$), consider the affine plane curve $C_D: D^2 y^4 = x^3 - x + 1$

We prove that $C_D(\mathbb{Q})$ is non-empty if and only if $D = \pm 1$, and $C_1(\mathbb{Q}) = C_{-1}(\mathbb{Q}) = \{(0,\pm 1), (\pm 1, \pm 1)\}$

Proof for Higher Degree Cases ($d \geq 3$)

We utilize the Darmon-Merel theorem, which is based on the modularity of elliptic curves: Theorem 4.1: When $n \geq 4$, the equation $x^n + y^n = z^2$ has no non-trivial primitive solutions.

Through clever algebraic transformations, we convert the existence problem of rational quadratic preimages into such Diophantine equations.

Technical Innovations

1. Avoiding the BSD Conjecture: The elliptic Chabauty method avoids dependence on the Birch-Swinnerton-Dyer conjecture.
2. Combining Geometry and Arithmetic: Skillfully combining covering theory from algebraic geometry with rational point problems from number theory.
3. Computational Feasibility: The method has lower computational cost compared to generalized $n$-descent and requires no additional conjectures.

Experimental Setup

Computational Tools

Using the Magma computer algebra system for:

1. Computing Mordell-Weil groups of elliptic curves
2. Determining rational points
3. Implementing the Chabauty method

Key Computations

1. Elliptic Curve $E: v^2 = u^3 - u + 1$:
   • Mordell-Weil group: $E(\mathbb{Q}) \cong \mathbb{Z}$, generated by $Q_0 = (1,-1)$
   • 10 known rational points in $X^{\text{pre}}(2,4,0)$
2. Number Field $K = \mathbb{Q}(\theta)$ (where $\theta^3 - \theta + 1 = 0$):
   • Discriminant: $-23$
   • Class number: 1
   • Unit group: Generated by $\pm 1$ and $\theta$

Experimental Results

Main Results

Corollary 1.4:

• For even $d \geq 3$: $\kappa(d,0,\mathbb{Q}) = 3$
• For odd $d \geq 3$: $\kappa(d,0,\mathbb{Q}) = 1$
• For $d = 2$: $\kappa(2,0,\mathbb{Q}) = 6$

Specific Counts of Rational Iterated Preimages

For different parameter values $c$, we provide exact counts of rational iterated preimages:

Even $d \geq 3$: