Asymptotic mean of digits of the $Q_s$-representation of the fractional part of a real number and related problems of fractal geometry and fractal analysis

This paper introduces the concept of the asymptotic mean of digits in the generalized $Q_s$-representation of real numbers and investigates the topological, metric, and fractal properties of sets defined by the existence or specific values of these means, while also exploring their connections to digit frequencies and fractal geometry.

The Gist

Here is an explanation of the paper "Asymptotic Mean of Digits of the Qs–Representation..." using simple language, analogies, and metaphors.

The Big Picture: A New Way to Count

Imagine you have a standard ruler. It's marked in inches, then halves, then quarters, then eighths. This is like our normal base-10 number system (0-9). Every number has a specific "address" in this system.

But what if you built a ruler where the marks weren't equal? What if the first mark was huge, the second was tiny, and the third was somewhere in between? And what if the rules for how those marks were spaced changed depending on which number you were looking at?

This paper introduces a new, flexible way to write numbers called $Q_s$-representation. It's like a "custom ruler" where the spacing of the digits is determined by a set of weights (probabilities) rather than being equal.

The authors, Pratsiovytyi and Klymchuk, are asking a fascinating question: If we look at the infinite string of digits that makes up a number in this custom system, is there an "average" value for those digits?

The Core Concept: The "Asymptotic Mean"

Let's break down the main idea: The Asymptotic Mean.

Imagine you are watching a very long movie.

• The Digits: Every time a character appears on screen, they have a "score" (0, 1, 2, etc.).
• The Mean: You want to know the average score of all characters in the movie.
• The "Asymptotic" part: You don't just look at the first 10 minutes. You look at the entire movie, from start to finish, as it stretches toward infinity.

If you add up all the scores and divide by the total time, does the number settle down to a specific value?

• Yes: The number has an "Asymptotic Mean." (e.g., The average score is exactly 3.5).
• No: The number is chaotic. The average keeps jumping around forever, never settling on a single number.

The authors are studying the set of numbers that do not have this average. They call this set $S$.

The "Chaotic" Numbers (Set $S$)

The paper proves something surprising about the set of numbers that don't have an average digit value.

1. They are everywhere (Everywhere Dense): If you pick any tiny spot on the number line (even a microscopic one), you can find one of these chaotic numbers inside it. They are sprinkled everywhere, like dust in a sunbeam.
2. They are invisible (Zero Measure): Even though they are everywhere, if you tried to "weigh" them or measure their total length on the number line, the weight would be zero. It's like a cloud of dust that fills the room but has no mass.
3. They are "Super-Fractal" (Dimension 1): This is the coolest part. In math, "fractal dimension" measures how "rough" or "complex" a shape is.
   • A smooth line has a dimension of 1.
   • A flat sheet has a dimension of 2.
   • A fractal (like a coastline) might be 1.5.
   • These chaotic numbers are so complex and "rough" that their dimension is 1. They are as "full" as a line can be, even though they have zero weight. The authors call this "Superfractal."

Analogy: Imagine a library where every book is written in a language that makes no sense.

• You can find a nonsense book on every shelf (Everywhere Dense).
• If you tried to stack all the nonsense books, the pile would have no height (Zero Measure).
• But if you looked at the texture of the spines, they would be infinitely jagged and complex (Superfractal).

The "Normal" vs. The "Abnormal"

The paper also touches on Normal Numbers.

• Normal Numbers: These are the "well-behaved" numbers. In a standard system, a normal number has every digit (0-9) appear with equal frequency (10% each). In this custom $Q_s$ system, the digits appear with specific frequencies that match the weights of the system.
• The Result: Almost all real numbers are "Normal" (they have a stable average). The "Abnormal" numbers (the ones without an average) are the rare, chaotic exceptions.

The "Level Sets" (The $M$ Sets)

The authors also look at specific groups of numbers where the "Average" equals the "Frequency" of a specific digit.

• Imagine a group of people where the average height of the group is exactly the same as the height of the tallest person in the group.
• The paper calculates the "fractal dimension" of these specific groups.
  • One group ($M_0$) is very complex (dimension $\approx 0.87$).
  • Another group ($M_1$) is even more complex (dimension $\approx 1.58$).
  • A third group ($M_2$) is so simple it's just a single point (dimension 0).

Why Does This Matter?

You might ask, "Who cares about the average of digits in a weird number system?"

1. Understanding Chaos: It helps mathematicians understand the boundary between order and chaos. Even in a system that seems random, there are deep geometric structures.
2. Fractal Geometry: It shows that "weird" sets (like the ones with no average) aren't just empty holes; they are incredibly rich, complex shapes that fill space in a very specific way.
3. Probability: It connects to how we understand randomness. If you generate random numbers, how likely is it that they will have a stable average? (Answer: Almost 100% likely).

Summary in a Nutshell

The authors invented a new way to look at numbers using a "weighted ruler." They discovered that while most numbers behave nicely and have a stable "average digit," there is a special, invisible, yet infinitely complex group of numbers that never settle on an average. These numbers are everywhere, yet weigh nothing, and their structure is so intricate that they are considered "super-fractals."

It's a study of the beautiful chaos hidden inside the numbers we use every day.

Technical Summary

Here is a detailed technical summary of the paper "Asymptotic Mean of Digits of the $Q_s$–Representation of the Fractional Part of a Real Number and Related Problems of Fractal Geometry and Fractal Analysis" by M. V. Pratsiovytyi and S. O. Klymchuk.

1. Problem Statement and Context

The paper addresses the gap in the metric and fractal theory of real numbers regarding the asymptotic mean of digits in non-standard numeral systems. While traditional $s$-adic representations (like binary or decimal) are well-studied, the authors focus on the $Q_s$-representation, a generalization where the base is defined by a probability vector $Q_s = \{q_0, \dots, q_{s-1}\}$ rather than a single integer base.

Key Definitions:

• $Q_s$-Representation: Any $x \in [0, 1]$ can be represented as $x = \sum_{k=1}^{\infty} \alpha_k \prod_{j=1}^{k-1} q_{\alpha_j}$, where $\alpha_k \in \{0, \dots, s-1\}$.
• Asymptotic Mean ($r(x)$): Defined as the limit of the arithmetic mean of the first $n$ digits:
  $$r(x) = \lim_{n \to \infty} \frac{1}{n} \sum_{i=1}^n \alpha_i(x)$$
  If this limit exists, $r(x)$ is the asymptotic mean; otherwise, it is undefined.
• Digit Frequency ($\nu_i(x)$): The limit of the frequency of a specific digit $i$ appearing in the expansion.

The Core Problem:
The authors investigate the topological, metric, and fractal properties of sets of real numbers where:

1. The asymptotic mean $r(x)$ does not exist.
2. The asymptotic mean $r(x)$ exists and equals the frequency of a specific digit (e.g., $r(x) = \nu_i(x)$).

2. Methodology

The authors employ a combination of fractal geometry, measure theory, and probabilistic number theory.

• Set Construction: They define specific subsets of $[0, 1]$ based on the existence or non-existence of limits for digit sums and frequencies.
• Relationship Analysis: They establish the mathematical link between the asymptotic mean and digit frequencies. For a general $s$, if frequencies exist:
  $$r(x) = \sum_{i=0}^{s-1} i \cdot \nu_i(x)$$
  In the binary case ($s=2$), $r(x)$ coincides exactly with the frequency of the digit 1 ($\nu_1(x)$).
• Dimension Calculation: The Hausdorff–Besicovitch dimension ($\alpha_0$) is calculated for various sets using the entropy formula derived from the constraints on digit frequencies. The formula used for a set defined by frequencies $\tau_i$ is:
  $$\alpha_0 = \frac{\ln(\prod \tau_i^{\tau_i})}{\ln(\prod q_i^{\tau_i})}$$
• Topological Proofs: The authors use cylinder intervals and mapping arguments to prove properties like continuity, density, and discontinuity of the constructed sets.

3. Key Contributions and Results

A. The Set of Numbers Without an Asymptotic Mean ($S$)

The authors define $S = \{x : r(x) \text{ does not exist}\}$.

• Topological Properties: $S$ is a continuous (uncountable), everywhere dense, and everywhere discontinuous set.
• Metric Properties: $S$ has zero Lebesgue measure ($\lambda(S) = 0$). This is because the non-existence of the mean implies the non-existence of at least one digit frequency, and the set of "abnormal" numbers (non-normal) has measure zero.
• Fractal Properties: $S$ is superfractal. The authors prove that its Hausdorff–Besicovitch dimension is $\alpha_0(S) = 1$. This is significant because the set is "small" in terms of measure (measure zero) but "large" in terms of fractal dimension (full dimension of the interval).

B. Sets Where Mean Equals Frequency ($M_i$)

The authors analyze sets $M_i = \{x : r(x) = \nu_i(x)\}$ for $i \in \{0, 1, 2\}$ (specifically in the context of $s=3$).

1. Set $M_0$ ($r(x) = \nu_0(x)$):

   • This imposes the constraint $2\nu_1 + 3\nu_2 = 1$.
   • Result: $M_0$ is continuous, everywhere dense, and has zero Lebesgue measure.
   • Dimension: The Hausdorff dimension is calculated to be approximately 0.8733 (derived by maximizing an entropy function subject to the linear constraint).

2. Set $M_1$ ($r(x) = \nu_1(x)$):

   • This implies $\nu_2 = 0$ and $\nu_1 = 1 - \nu_0$.
   • Result: Continuous, everywhere dense, zero Lebesgue measure.
   • Dimension: The maximum dimension occurs when $\nu_0 = \nu_1 = 0.5$, yielding $\alpha_0(M_1) = \log_2 3 \approx 1.58$ (Note: In the context of the paper's specific $Q_3$ setup, the calculation yields $\log_2 3$ relative to the base 3 structure, indicating a high degree of complexity).

3. Set $M_2$ ($r(x) = \nu_2(x)$):

   • This implies $\nu_1 + \nu_2 = 0$, forcing $\nu_1 = \nu_2 = 0$ and $\nu_0 = 1$.
   • Result: This set consists of numbers with only the digit 0 in their expansion (essentially a single point or a set of measure zero with trivial structure).
   • Dimension: $\alpha_0(M_2) = 0$. The authors term this an "anomalously fractal" set.

4. Significance and Implications

• Generalization of Normality: The paper extends the classical concept of "normal numbers" (where all digits appear with equal frequency) to the concept of numbers having a defined asymptotic mean. It demonstrates that the set of numbers failing to have this mean is "large" in a fractal sense (dimension 1) despite being "small" in a measure-theoretic sense.
• Fractal Geometry of Number Theory: The work provides a rigorous framework for constructing fractal sets based on arithmetic properties of digit expansions in non-standard bases ($Q_s$). It bridges the gap between probabilistic number theory (frequencies) and geometric measure theory (Hausdorff dimension).
• Superfractality: The identification of the set $S$ as "superfractal" (dimension 1 but measure 0) highlights the counter-intuitive nature of fractal sets in real analysis, reinforcing the idea that "size" in mathematics depends heavily on the metric used (Lebesgue vs. Hausdorff).
• Applications: The concepts introduced are applicable to the study of singular distributions, self-similar sets, and the construction of mathematical objects with specific fractal properties, which are relevant in dynamical systems and information theory.

Conclusion

Pratsiovytyi and Klymchuk successfully introduce the asymptotic mean of digits as a fundamental characteristic of real numbers in $Q_s$-representations. They prove that the set of numbers lacking this mean is topologically complex and possesses full Hausdorff dimension, while sets where the mean equals specific digit frequencies exhibit varied fractal dimensions depending on the constraints imposed on the digit frequencies. This work enriches the theory of singular functions and fractal analysis by providing new classes of sets with distinct topological and metric behaviors.