🔢 mathematics

On the dynamical Manin-Mumford conjecture for plane polynomial maps

The paper proves the dynamical Manin-Mumford conjecture for regular polynomial maps on the affine plane and irreducible curves that avoid super-attracting orbits at infinity over any field of characteristic zero.

Original authors: Romain Dujardin, Charles Favre, Matteo Ruggiero

Published 2026-06-19

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Romain Dujardin, Charles Favre, Matteo Ruggiero

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 Game of Musical Chairs with Infinite Players

Imagine a giant, infinite dance floor (the mathematical plane). On this floor, there is a set of rules (a polynomial map) that tells every dancer exactly where to move next.

If you follow the rules, you might move in a circle forever (a periodic point).
You might eventually land on a spot that starts a circle (a preperiodic point).
Or, you might wander off into chaos, never repeating a pattern.

Mathematicians have long been interested in a specific question: If you find a straight line or a curved path on this dance floor that contains an infinite number of dancers who eventually start dancing in circles, does that entire path have to be part of the choreography itself?

In other words, if a path is "infested" with dancers who repeat their steps, is the path itself a repeating loop?

This is the Dynamical Manin-Mumford Conjecture. It's a bit like asking: "If a river is filled with fish that swim in perfect circles, is the river itself a giant circular track?"

The Problem: The "Super-Trap" Exception

For a long time, mathematicians knew the answer was "Yes" for simple cases, but there were tricky exceptions.

The Trap: Imagine a spot on the dance floor that acts like a super-magnet. Once a dancer gets close, they get sucked in and spin faster and faster until they vanish into the center. This is called a super-attracting point.
The Confusion: Sometimes, a path could lead into this super-magnet. The path would be filled with dancers who eventually get sucked in and start their loops inside the magnet. But the path itself wasn't a loop; it was just a slide into the trap. This broke the "Yes" answer.

The authors of this paper, Romain Dujardin, Charles Favre, and Matteo Ruggiero, wanted to prove that if you avoid these super-magnets, the answer is always "Yes."

The Main Discovery: The "No Super-Trap" Rule

The paper proves a specific version of this conjecture for 2-dimensional spaces (like a flat sheet of paper).

The Rule: If you have a curved path on this sheet, and that path has an infinite number of dancers who eventually repeat their steps, AND the path doesn't end up in a "super-magnet" at the edge of the world (infinity), then the path itself must be a repeating loop.

They call this Theorem A.

How Did They Solve It? (The Detective Work)

To prove this, the authors acted like detectives using two different toolkits:

1. The "Height" Meter (The Arithmetic Toolkit)

Imagine every dancer has a "height score."

If a dancer repeats their steps (is preperiodic), their score is zero.
If they wander chaotically, their score is positive.

The authors used a special mathematical tool called a canonical height. They showed that if a path has infinitely many dancers with a score of zero, the "average height" of the whole path must also be zero. This implies the path is special.

However, just knowing the average is zero isn't enough to prove the path is a loop. They needed to look closer at the edge of the world (infinity).

2. The "Local Map" (The Geometry Toolkit)

The authors zoomed in on the edge of the dance floor (the "line at infinity").

They looked at the dancers who live right on the edge.
They checked if any of these edge-dancers were stuck in a super-magnet (super-attracting).
The Assumption: The paper assumes the path touches the edge at a point that is not a super-magnet. It might be a regular loop, or a point that pushes dancers away (repelling), or a point that lets them drift slowly (parabolic).

Once they confirmed the edge point wasn't a super-magnet, they used a technique called Graph Transforms.

The Analogy: Imagine you have a piece of string (the path) and a rubber sheet (the dance floor). If you pull the rubber sheet over and over, the string gets stretched and twisted.
The authors showed that if the string has enough "zero-score" dancers on it, and the edge isn't a super-magnet, the stretching process forces the string to align perfectly with a specific "stable track" (a super-stable manifold).
Once the string aligns with this track, it can't wiggle anymore. It must be a repeating loop.

The "Specialization" Trick (The Time Traveler)

The paper also had to handle the fact that the dance floor could be defined using different types of numbers (not just the standard real numbers we use in daily life, but also complex algebraic numbers).

To solve this, they used a Specialization Argument:

Imagine the dance floor is part of a huge family of dance floors, parameterized by a dial.
They proved that if the rule holds for a "typical" setting (a number field), it must hold for the specific setting they are studying, provided the dancers don't suddenly "collide" or disappear when you turn the dial.
They carefully checked that the infinite number of repeating dancers wouldn't suddenly shrink into a finite bunch just because they changed the dial. This ensured the logic held up for any field of numbers.

The Conclusion

In simple terms, the paper says:
"On a 2D plane, if a curve is packed with dancers who eventually dance in circles, and the curve doesn't lead into a 'black hole' at the edge of the universe, then the curve itself is a giant, eternal dance circle."

They proved this for a very wide class of mathematical maps, effectively solving a major puzzle in the field of arithmetic dynamics, provided you stay away from the "super-attracting" traps.

Technical Summary: On the Dynamical Manin-Mumford Conjecture for Plane Polynomial Maps

Problem Statement
The paper addresses the dynamical Manin-Mumford conjecture for regular polynomial endomorphisms of the affine plane $\mathbb{A}^2$ over a field $k$ of characteristic zero. The conjecture, originally formulated by S.-W. Zhang, posits that if a subvariety $Z$ of a polarized dynamical system contains a Zariski dense set of preperiodic points, then $Z$ must be preperiodic or "special" (contained in a subvariety invariant under a commuting polarized endomorphism).

Previous work has settled the conjecture for Abelian varieties (Raynaud) and for product maps on $(\mathbb{P}^1)^k$ (Ghioca, Nguyen, Ye). However, the case of genuinely higher-dimensional dynamics, specifically regular polynomial maps on $\mathbb{A}^2$ that are not of product type, remained open. The authors note that while counterexamples exist for the original formulation on products of elliptic curves (via Lattès constructions), no such counterexamples are known for regular maps on the birationally equivalent model $\mathbb{P}^2$ . The specific obstruction to the conjecture in this setting would be the existence of a preperiodic curve $C$ for a map $\psi$ commuting with $f$ , where $C$ is not preperiodic for $f$ . The authors argue that known classifications of commuting pairs suggest such obstructions do not arise for regular polynomial maps, leading to the expectation that the conjecture holds unconditionally.

Main Result
The authors prove the following theorem (Theorem A):
Let $f$ be a regular polynomial endomorphism of $\mathbb{A}^2$ of degree $d \ge 2$ , defined over an arbitrary field $k$ of characteristic 0. Let $C \subset \mathbb{A}^2$ be an irreducible algebraic curve containing infinitely many preperiodic points of $f$ . If the closure of $C$ in $\mathbb{P}^2$ contains a point $p$ on the line at infinity $L_\infty$ that is not eventually super-attracting for the restriction $f_\infty$ , then $C$ is preperiodic under $f$ .

A corollary (Corollary B) states that if $f$ has no super-attracting points on $L_\infty$ at all, then the dynamical Manin-Mumford conjecture holds for $f$ unconditionally.

Methodology
The proof employs a synthesis of arithmetic geometry and complex/non-Archimedean dynamics. The strategy is divided into two main phases:

The Number Field Case ( $k$ is a number field):
- Canonical Heights: The authors utilize the canonical height function $h_f$ associated with $f$ . By Zhang's inequality, if a curve $C$ contains infinitely many preperiodic points, its canonical height $h_f(C)$ must be zero. This implies that the restriction of the dynamical Green function $g_v$ to the analytification of $C$ is harmonic on the set where $g_v > 0$ .
- Local Analysis at Infinity: The proof focuses on a point $p \in C \cap L_\infty$ $p \in C \cap L_{\infty}$ that is not super-attracting. Depending on the multiplier $\lambda$ $λ$ of $f_\infty$ $f_{\infty}$ at $p$ $p$ , two cases are analyzed:
  - Repelling/Parabolic ( $|\lambda|_v > 1$ or $\lambda$ is a root of unity): The authors construct local super-stable manifolds $W^s_{loc}(p)$ . In the repelling case, they use rescaling arguments; in the parabolic case ( $\lambda=1$ ), they employ graph transform techniques adapted from Hakim's work on parabolic points.
  - Contradiction via Normality: Assuming $C$ does not locally coincide with $W^s_{loc}(p)$ , the authors show that the pull-backs of $C$ under iterates of $f$ would intersect the stable manifold transversely. This leads to a contradiction: the sequence of iterates would fail to be a normal family (due to the Inclination Lemma) while simultaneously being normal (due to the harmonicity of the Green function derived from $h_f(C)=0$ ).
- Non-Archimedean Extension: The construction of super-stable manifolds and the graph transform estimates are extended to the non-Archimedean setting (Berkovich spaces) to handle places $v$ where $|\lambda|_v > 1$ .
The Arbitrary Field Case (Specialization):
- To extend the result from number fields to any field of characteristic 0, the authors use a specialization argument. They view the map $f$ and curve $C$ as part of an algebraic family parameterized by a variety $S$ .
- Control of Preperiodic Points: A key technical challenge is ensuring that an infinite set of preperiodic points on the generic fiber does not collapse into a finite set on a specific fiber. The authors adapt the Shub-Sullivan theorem to control the multiplicity of periodic points and analyze the splitting of local preimages of super-attracting cycles (referencing Chio and Roeder).
- Reduction: They prove that if the conjecture holds for the generic fiber (a number field case), it holds for the specific fiber, provided the condition on non-super-attracting points at infinity is preserved.

Key Contributions

Resolution of the Conjecture for Regular Maps: The paper establishes the dynamical Manin-Mumford conjecture for a wide class of 2-dimensional dynamical systems (regular polynomial maps on $\mathbb{A}^2$ ) where the dynamics are genuinely higher-dimensional, distinct from the product-type cases previously solved.
Local Super-Stable Manifolds in Non-Archimedean Settings: The authors provide a unified construction of local super-stable manifolds for analytic maps in both Archimedean and non-Archimedean fields, including cases where the fixed point is parabolic or super-attracting in a specific direction.
Graph Transform Estimates: New estimates for graph transforms in the presence of parabolic points are developed, borrowing from Hakim's work but adapted to the dynamical context of the Green function.
Specialization Argument for Preperiodic Sets: The paper refines specialization techniques to handle infinite sets of preperiodic points, addressing the potential "shrinking" of these sets on special fibers, a problem not fully addressed in previous works dealing with periodic points.

Significance and Scope
The authors claim that their result confirms the expectation that the dynamical Manin-Mumford conjecture holds for regular polynomial maps on $\mathbb{A}^2$ , as the known obstructions (commuting endomorphisms inducing special curves) do not arise in this setting. The proof relies on the assumption that the curve intersects the line at infinity at a point that is not eventually super-attracting. The authors explicitly note that the case where all points at infinity are super-attracting remains open, as the key step involving the Julia set of $f_\infty$ fails in that scenario.

The work is presented as a significant step in arithmetic dynamics, bridging the gap between the 1-dimensional (product) cases and higher-dimensional dynamics, utilizing a combination of Arakelov geometry (canonical heights) and complex/non-Archimedean analysis. The authors acknowledge that alternative approaches using Moriwaki's height theory exist but emphasize their reliance on the interplay between global arithmetic properties and local dynamical behavior.