Non-archimedean connected Julia sets with branching

Dvij Bajpai, Robert L. Benedetto, Ruqian Chen, Edward Kim, Owen Marschall, Darius Onul, Yang Xiao

TL;DR

This paper provides the first known examples of rational functions over non-archimedean fields (like the $p$-adics) whose Julia sets are connected but possess a "branching" structure, similar to the dendrites seen in complex dynamics. Furthermore, the authors demonstrate that for these maps, the measure-theoretic entropy can be strictly smaller than the topological entropy, resolving a long-standing open question in the field.

What Problem Does This Paper Address?

In the study of dynamical systems, Julia sets represent the regions where a function's behavior is chaotic. While Julia sets over the complex numbers ($\mathbb{C}$) are famous for their beautiful, fractal-like branching (dendrites), their counterparts in non-archimedean dynamics—studied on the Berkovich projective line—have historically appeared much simpler.

Until this paper, every known example of a non-archimedean Julia set was either a single point, a simple line segment, or a totally disconnected set (like a Cantor set). Mathematicians wondered: can non-archimedean Julia sets be connected and "branchy"?

Additionally, there is a fundamental question regarding entropy, which measures the complexity or "chaos" of a system. In 2010, mathematicians Favre and Rivera-Letelier asked if the measure-theoretic entropy ($h_\mu$) of such a map could be strictly less than its topological entropy ($h_{top}$) when the Julia set is connected. This paper sets out to prove that the answer is "yes" on both counts.

Methods

The authors employ a mix of arithmetic geometry and symbolic dynamics. Their approach is centered on two major theoretical results:

  1. Theorem A (The Branching Engine): They establish criteria for a rational function $\phi$ to have a Julia set that is connected and contains a specific "finite tree" structure. This involves finding maps that "stretch" specific intervals across the branches of a tree in the Berkovich space.
  2. Theorem B (The Entropy Calculator): They develop a method to compute entropy by partitioning the Julia set into a countable collection of pieces. This allows them to translate the continuous dynamics of the rational function into a topological Markov shift on a countable set of symbols.

To prove their point, they construct a specific "degree six" example: $$\phi(z) = \frac{az^6 + 1}{az^6 + z^3 - z}$$ By analyzing this map in a field with residue characteristic $p=3$, they can explicitly track how points move across the branches of the Berkovich tree.

Datasets

Not discussed in the paper. As a work of pure mathematics, the "data" consists of formal proofs and the construction of specific algebraic examples rather than experimental datasets.

Key Insights and Ideas

  • Dendrites in the Non-Archimedean World: The authors show that by carefully choosing coefficients (specifically using a parameter $|a|_v < 1$), they can force the map to "fold" and "stretch" the Berkovich line in a way that creates a dense set of branching points.
  • Symbolic Dynamics as a Bridge: The most powerful "aha moment" is the realization that the complex geometry of a branched Julia set can be perfectly captured by a transition matrix of a Markov shift. This turns a difficult calculus problem into a combinatorial one involving the roots of polynomials.
  • Entropy Gap: The paper proves that the "natural" invariant measure $\mu$ (associated with the Laplacian of the Green's function) is not always the measure of maximal entropy. In their degree 6 example, the system's "topological" capacity for chaos outpaces what the standard measure can "see."

Key Results

The paper culminates in the explicit calculation of entropies for their degree 6 map:

  • Measure-Theoretic Entropy ($h_\mu$): $$h_\mu(\phi) = \log 2 + \frac{5}{11} \log 3 \approx 1.1925$$
  • Topological Entropy ($h_{top}$): $$h_{top}(\phi) = \log \lambda \approx 1.3496$$ where $\lambda$ is the largest real root of the polynomial $t^3 - 4t^2 - t + 6$.

Since $1.1925 < 1.3496$, they successfully demonstrate a strict inequality, providing the first example of a connected non-archimedean Julia set where these two values diverge.

Limitations and Future Work

The authors note that their primary example relies on the field having a residue characteristic of $p=3$. While they mention that a degree 4 example for $p=2$ was in preparation at the time of writing, the general behavior for arbitrary characteristics or higher-degree maps remains an open area for exploration.

Furthermore, while they have shown that branching can occur, a full classification of which rational functions produce branched vs. linear Julia sets is still missing. The techniques used here—mapping to countable Markov shifts—might be applicable to a wider class of "tame" non-archimedean maps, potentially leading to a broader theory of non-archimedean "dendrite" dynamics.

Summary updated

Abstract

We construct the first examples of rational functions defined over a non-archimedean field with certain dynamical properties. In particular, we find such functions whose Julia sets, in the Berkovich projective line, are connected but not contained in a line segment. We also show how to compute the measure-theoretic and topological entropy of such maps. In particular, we show for some of our examples that the measure-theoretic entropy is strictly smaller than the topological entropy, thus answering a question of Favre and Rivera-Letelier.
BibTeX
@inproceedings{bajpai2015non,
  author = {Dvij Bajpai and Robert L. Benedetto and Ruqian Chen and Edward Kim and Owen Marschall and Darius Onul and Yang Xiao},
  title = {{Non-archimedean connected Julia sets with branching}},
  booktitle = {Ergodic},
  year = {2015},
  doi = {10.1017/etds.2015.42},
}

Loading related content...

Figure