🔢 mathematics

Perfect Sets of Liouville Numbers with Controlled Self-Powers

This paper constructs a perfect set of Liouville numbers with continuum cardinality that is closed under finite sums, finite products, and the self-power operation $x \mapsto x^x$ , demonstrating that rich algebraic and topological structures persist within the Liouville universe under these arithmetic transformations.

Original authors: Sidney A. Morris, Marcelo O. Ribeiro, Diego Marques

Published 2026-03-27

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Sidney A. Morris, Marcelo O. Ribeiro, Diego Marques

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: The "Wild" and the "Tame"

Imagine the world of numbers as a vast ocean.

Rational numbers (like 1/2 or 3/4) are like calm, predictable islands. You can write them down exactly.
Irrational numbers (like $\pi$ or $\sqrt{2}$ ) are like deep, mysterious waters. You can't write them down exactly, but they usually behave "normally."
Liouville numbers are the monsters of this ocean. They are a special type of irrational number that is infinitely close to rational numbers. If you try to approximate them with a fraction, you can get closer and closer, much faster than you can with any other irrational number. They are "wild" because they break the usual rules of how numbers approximate each other.

For a long time, mathematicians knew these monsters existed, but they were hard to find in groups. They were scattered like single, lonely monsters.

The Main Question: What happens when a monster eats itself?

The paper asks a tricky question: If you take one of these wild Liouville monsters ( $x$ ) and calculate its "self-power" ( $x^x$ ), what do you get?

Does it turn into a normal, tame number?
Does it stay a wild monster?
Does it become something completely new?

Previously, we knew that for some of these monsters, $x^x$ becomes a "transcendental" number (a very complex, non-repeating number). But the authors wanted to know: Can we build a whole family of these monsters where $x^x$ stays a monster (a Liouville number)?

The Solution: Building a "Monster Zoo"

The authors didn't just find one or two examples; they built a Perfect Set. In math, a "perfect set" is like a Cantor set (think of a fractal dust that never ends). It's a collection of numbers that is:

Uncountably large: There are as many of them as there are points on a line (the "continuum").
Dense and connected: You can't pick one out without finding infinitely many neighbors right next to it.

They constructed a "zoo" of these numbers with three amazing properties:

1. The "Self-Power" Trick

They found a specific group of Liouville numbers where, if you take any number $x$ from the group and calculate $x^x$ , the result is still a Liouville number.

Analogy: Imagine a group of chameleons. Usually, if you paint a chameleon, it changes color. But in this specific group, no matter how you paint them (by raising them to their own power), they stay the exact same color. They are "self-preserving."

2. The "Algebraic Safety Net"

The group is "algebraically Liouville-closed." This is a fancy way of saying: If you take any two (or more) numbers from this group and mix them together using standard math operations (addition, multiplication, or plugging them into a polynomial equation), the result is either a simple fraction or another Liouville monster.

Analogy: Imagine a club where the members are all made of a special, unstable material. If you smash two members together, they don't turn into normal steel; they either turn into a harmless pebble (a rational number) or they explode into more unstable material (another Liouville number). They never accidentally become "normal" by mistake.

3. The "Pairwise Power" Superpower

The most impressive part is the final section. They refined their zoo to create a smaller, even more special group ( $W$ ). In this group, if you take any two numbers $x$ and $y$ (they can be different!), and calculate $x^y$ (x to the power of y), the result is still a Liouville number.

Analogy: In most groups, if you mix two different monsters, you might get a normal animal. But in this specific group, mixing any two monsters together always results in another monster. It's a "monster-only" ecosystem.

How Did They Do It? (The "Tuned" Construction)

The authors used a construction method similar to building a fractal (like the famous Cantor set).

The Blueprint: They started with a standard fractal pattern where numbers are built by adding tiny pieces (digits) in a specific base (base 3).
The "Spiffy" Constants: They created a special rule for how these pieces are added. They called these "spiffy constants." These numbers are designed to be incredibly close to fractions, but in a very specific, controlled way.
The "Tuning": They added a "tuning" mechanism. Imagine you are building a tower of blocks. Usually, the blocks might wobble. These authors added "tuned spiffy constants" which act like shock absorbers. They ensured that when you calculate $x^x$ , the "wobble" (the error) is so small and the "denominator" (the size of the fraction) grows in just the right way that the result remains a Liouville number.

Why Does This Matter?

This paper is a tour de force in Topological Number Theory.

Topological Size: It shows that these "monsters" aren't just rare, isolated freaks. They can form massive, structured, infinite families (perfect sets).
Algebraic Structure: It proves that the "wild" behavior of Liouville numbers is robust. Even when you perform complex operations like self-powers or mixing different members, the "wildness" survives.
Counter-Intuition: Usually, complex operations tend to "smooth out" wild numbers. This paper shows that with the right construction, you can keep the chaos alive and well, even in a structured, infinite set.

Summary in One Sentence

The authors built a massive, infinite family of "wild" numbers (Liouville numbers) that are so perfectly constructed that no matter how you mix them, add them, or raise them to their own power, they never lose their "wild" nature and always remain Liouville numbers.

1. Problem Statement and Context

The paper investigates the arithmetic behavior of self-powers ( $x^x$ ) when $x$ is a Liouville number.

Liouville Numbers ( $\mathcal{L}$ ): A real number $x$ is Liouville if it can be approximated by rationals $p/q$ extremely well, specifically $|x - p/q| < q^{-N}$ for any $N \in \mathbb{N}$ . The set $\mathcal{L}$ is a dense $G_\delta$ set with Lebesgue measure zero.
The Gap: While it is known that $\mathcal{L}$ $L$ is algebraically rich (e.g., sums and products of Liouville numbers can cover $\mathbb{R}$ $R$ ), the behavior of the map $x \mapsto x^x$ $x \mapsto x^{x}$ on $\mathcal{L}$ $L$ was unclear.
- A recent result (2023) showed that for "strong" Liouville numbers, $x^x$ is transcendental.
- However, it was unknown whether $x^x$ could remain a Liouville number for any $x \in \mathcal{L}$ , and whether there exist large (perfect) sets of Liouville numbers closed under this operation.
Goal: To construct perfect sets (cardinality $\mathfrak{c}$ ) of Liouville numbers where the self-power $x^x$ (and even pairwise powers $x^y$ ) retains the Liouville property, while also characterizing sets where $x^x$ is transcendental but not Liouville.

2. Methodology

The authors employ a combination of Cantor-type constructions, Diophantine approximation, and topological arguments.

A. "Spiffy" and "Tuned" Constants

The core construction relies on a specific subset of the middle-third Cantor set.

Spiffy Constants: Defined via a series $x_a = \sum_{n=1}^\infty a_n 3^{-e_n}$ where $a_n \in \{0, 2\}$ and the exponents grow super-exponentially ( $e_{n+1} = 3e_n$ ). These numbers are shown to be $\epsilon$ -strong Liouville numbers (a subclass with even faster approximation rates).
Tuned Spiffy Constants: A refinement where the authors impose additional constraints on the rational approximations of $\log(x)$ and $x \log x$ . This involves selecting specific blocks of digits to ensure that $x \log x$ admits rational approximations $U_j/V_j$ with controlled denominator growth, allowing $e^{x \log x} = x^x$ to be approximated by rationals with Liouville-quality error bounds.

B. Bi-Lipschitz Analysis

To handle the function $f(x) = x^x$ , the authors analyze its derivative $f'(x) = x^x(\log x + 1)$ .

They restrict their domain to intervals $[\delta, 1]$ where $\delta > e^{-1}$ . In this region, $f'(x)$ is bounded away from zero, making $f$ bi-Lipschitz.
This property is crucial for transferring topological properties (like density and $G_\delta$ status) and measure-theoretic properties (Hausdorff dimension) from the domain to the image.

C. Probabilistic and Dimensional Arguments

To show the existence of Liouville numbers whose self-powers are not Liouville, they use a Hausdorff dimension argument. They construct a Cantor set $C$ of $\epsilon$ -strong Liouville numbers and show that the set of points mapping to "bad" approximations (non-Liouville) has small Hausdorff dimension, leaving a residual set of continuum cardinality where $x^x$ is transcendental but not Liouville.

D. Synchronized Truncations

For the construction of sets closed under pairwise powers ( $x^y$ ), the authors synchronize the truncation schedules of all elements in the set. By forcing all $x, y$ in the set to share the same "anchor" rational approximations at specific sparse levels, they ensure that $y \log x$ remains uniformly close to a rational anchor, facilitating the Liouville approximation of $x^y$ .

3. Key Contributions and Results

A. Existence of Perfect Sets with Liouville Self-Powers

Theorem 2.4: There exists a perfect set of tuned spiffy constants such that for every $x$ in this set, $x^x$ is a Liouville number.
Theorem 5.5: There exists a perfect set $T \subset \mathcal{C}$ (Cantor set) of cardinality $\mathfrak{c}$ that is algebraically Liouville-closed. This means for any $x_1, \dots, x_t \in T$ and any non-constant polynomial $P \in \mathbb{Z}[X_1, \dots, X_t]$ , $P(x_1, \dots, x_t)$ is either rational or a Liouville number. Furthermore, for every $x \in T$ , $x^x \in \mathcal{L}$ .

B. Existence of Perfect Sets with Non-Liouville Self-Powers

Theorem 3.1: There exist continuum many $\epsilon$ -strong Liouville numbers $x$ such that $x^x$ is transcendental but not a Liouville number. This demonstrates a dichotomy: the self-power map can either preserve the Liouville property or push the number into the "transcendental but not Liouville" regime.

C. Topological Structure of the Sets

Theorem 4.1 & 4.3: The set of $x > 0$ such that $x^x$ is transcendental is a dense $G_\delta$ set of full Lebesgue measure. Similarly, the set of $x \in \mathcal{L}$ such that $x^x \in \mathcal{L}$ is a dense $G_\delta$ subset of $(0, \infty)$ .
This confirms that the "Liouville-preserving" behavior is topologically generic within the space of positive reals, despite the arithmetic thinness of $\mathcal{L}$ .

D. Pairwise Power Closure

Theorem 6.1: The authors construct a perfect set $W$ of cardinality $\mathfrak{c}$ such that for all pairs $x, y \in W$ , the power $x^y$ is a Liouville number.
Corollary 6.1: Inside $W$ , there exists a strictly descending chain of perfect sets $\{W_t\}_{t \in (0,1)}$ of length $\mathfrak{c}$ , all retaining the pairwise power-Liouville property.

4. Significance

Resolution of Open Questions: The paper resolves the question of whether $x^x$ can be Liouville for $x \in \mathcal{L}$ , proving that not only does such an $x$ exist, but an entire perfect set of them does.
Algebraic Closure: It establishes the existence of "algebraically Liouville-closed" perfect sets. This is a strong structural result, showing that the Liouville universe is robust enough to support complex algebraic operations (polynomials) and transcendental operations (self-powers) without escaping the class of Liouville numbers (except for rational outcomes).
Interplay of Topology and Arithmetic: The work highlights a fascinating tension: while Liouville numbers are "arithmetically thin" (measure zero), they are "topologically large" (dense $G_\delta$ ). The authors show that within this topological largeness, one can find highly structured subsets that maintain specific arithmetic properties under non-linear maps like $x \mapsto x^x$ .
Methodological Innovation: The introduction of "tuned spiffy constants" and the use of synchronized truncation schedules provide new tools for constructing sets with precise Diophantine properties, potentially applicable to other problems involving transcendence and approximation.

In summary, the paper demonstrates that the class of Liouville numbers is not just a collection of pathological examples but contains rich, perfect substructures that are stable under addition, multiplication, and the self-power map $x \mapsto x^x$ .