On the Ratat-Goormaghtigh equation and integer points close to the graph of a smooth function

Basic Information

• Paper ID: 2510.11252
• Title: On the Ratat-Goormaghtigh equation and integer points close to the graph of a smooth function
• Author: Tomohiro Yamada (Center for Japanese Language and Culture, Osaka University)
• Classification: math.NT (Number Theory)
• Publication Date: October 13, 2025
• Paper Link: https://arxiv.org/abs/2510.11252

Abstract

This paper proves that for a given integer $N$, among the integer solutions to the Diophantine equation $(x^m-1)/(x-1)=N$ with $x,m\geq 2$, the sum of reciprocals of all solutions except the smallest is less than 5.9037. When $x$ is restricted to primes, this sum is less than 0.73194.

Research Background and Motivation

Problem Origin

This research originates from the celebrated Ratat-Goormaghtigh equation: $N = \frac{x^m-1}{x-1} = \frac{y^n-1}{y-1}$ where $x,y\geq 2$ and $m,n\geq 3$ are integers.

This equation has a long history:

1. Ratat (1916) discovered the solution $(x,m,y,n,N) = (2,5,5,3,31)$
2. Goormaghtigh (1917) discovered another solution $(x,m,y,n,N) = (2,13,90,3,8191)$
3. It is currently conjectured that these are the only two solutions

Research Motivation

Although the complete Ratat-Goormaghtigh equation is difficult to study, the distribution of solutions to the single equation $(x^m-1)/(x-1) = N$ is equally important and more tractable. Known results include:

• Loxton: The number of solutions is at most $\log^{1/2+o(1)} N$
• Luca: When $x$ is prime, more precise upper bounds on the number of solutions exist

Innovation

This paper employs a geometric approach, transforming the problem into studying the distribution of integer points near the graph of a smooth function, contrasting with the arithmetic methods used by Luca and others.

Core Contributions

1. Main Theorem 1: Proves that the sum of reciprocals of all solutions except the smallest satisfies $\sum_{i\geq 2} \frac{1}{x_i} < 5.9037$
2. Main Theorem 2: When $x$ is restricted to primes, the reciprocal sum satisfies $\sum_{i\geq 2} \frac{1}{q_i} < 0.73194$
3. Asymptotic Properties: Proves that the reciprocal sum tends to 0 as $N\to\infty$
4. Methodological Innovation: Establishes a connection between Diophantine equations and integer point counting near smooth functions

Methodology Details

Problem Formulation

Given a positive integer $N$, find all integer pairs $(x,m)$ satisfying $(x^m-1)/(x-1) = N$ with $x\geq 2$ and $m\geq 2$.

Core Idea Transformation

Through logarithmic transformation, the original equation is converted to: $m \log x - \log(x-1) - \log N = \log\left(\frac{x^m}{x^m-1}\right)$

This yields the key inequality: $0 < m - f_N(x) < \frac{1}{N\log x}$ where $f_N(x) = \frac{\log N + \log(x-1)}{\log x}$

Geometric Approach

The problem is transformed into studying the set: $\left\{x \in \mathbb{Z}_{\geq 2} : \|f_N(x)\| < \frac{1}{N\log x}\right\}$ where $\|t\|$ denotes the distance from $t$ to the nearest integer.

Technical Tools

1. Integer Point Counting Near Smooth Functions

Using results by Huxley and others (Lemma 2.1): For a $k$-times differentiable function $f$, if the $k$-th derivative satisfies $\lambda \leq |f^{(k)}(x)| \leq c\lambda$, then $R(f,M,\delta) \leq \alpha M\lambda^{2/(k^2+k)} + 4k$

2. Lower Bounds for Logarithmic Linear Forms

Using Matveev's theorem to obtain lower bounds for the second smallest solution $x_2$: $x_2 > \log^{0.33479} N$

3. Analysis of the Function $f_N(x)$

Detailed computation of expressions and estimates for $f_N^{(k)}(x)$: $f_N^{(k)}(x) = \frac{(-1)^k}{\log^{k+1} x}\left(P_{k,k}(\log x)\frac{\log N + \log(x-1)}{x^k} - \sum_{r=1}^k \frac{P_{k,r}(\log x)}{x^r(x-1)^{k-r}}\right)$

Experimental Setup

Numerical Computation Strategy

1. Interval Partitioning: Different values of $k$ are selected based on the magnitude of $N$ to optimize estimates
2. Constant Optimization: Optimal constants $C_k$, $\gamma_k$, $\tau_k$ are determined through numerical computation
3. Boundary Cases: Special treatment for small values with $N < 10^{100000}$

Parameter Selection

• For different values $k = 1,2,\ldots,6$, choose $M_k = \min\{2^n M_6 : 2^n M_6 \geq \log^{2/k} N\}$
• $M_6 = \min\{10^5, \log^{0.33479} N\}$

Experimental Results

Main Results

Theorem 1.1: $\sum_{i\geq 2} \frac{1}{x_i} < 5.9037$

Theorem 1.2: When $x$ is restricted to primes, $\sum_{i\geq 2} \frac{1}{q_i} < 0.73194$$\prod_{i\geq 2} \frac{q_i}{q_i-1} < 2.07913$

Technical Estimates

Through refined analysis, the following table of key constants is obtained:

Asymptotic Behavior

It is proved that $\sum_{i\geq 2} \frac{1}{x_i} \to 0$ as $N \to \infty$.

Related Work

Historical Development

1. Classical Results: History of the Ratat-Goormaghtigh conjecture
2. Modern Progress:
   • Loxton's general upper bounds
   • Luca's results for the prime case
   • Bennett and others' results for small parameters

Methodological Comparison

• Arithmetic Methods: Luca employs elementary number theory and sieve methods
• Geometric Methods: This paper uses integer point counting near smooth functions
• Analytic Methods: Uses lower bounds for logarithmic linear forms

Conclusions and Discussion

Main Conclusions

1. Provides explicit numerical bounds for the distribution of solutions to the one-sided Ratat-Goormaghtigh equation
2. Proves boundedness and asymptotic properties of the reciprocal sum of solutions
3. Establishes new connections between Diophantine equations and geometric integer point counting

Limitations

1. The constant 5.9037 may not be optimal
2. The method primarily applies to equations of the form $(x^m-1)/(x-1) = N$
3. Generalization to general exponential Diophantine equations is limited

Future Directions

1. Improve the constant bounds
2. Generalize to more general exponential Diophantine equations
3. Study multivariate cases

In-Depth Evaluation

Strengths

1. Methodological Innovation: Novel approach to geometrizing Diophantine equations
2. Precise Results: Provides explicit numerical constants
3. Technical Rigor: Detailed function analysis and error estimates
4. Theoretical Completeness: Complete theoretical framework from method to application

Weaknesses

1. Computational Complexity: Involves extensive numerical computation and case analysis
2. Constant Optimization: Some constants may have room for improvement
3. Generalizability: Limited scope for generalization of the method

Impact

1. Theoretical Contribution: Provides new perspective for Diophantine equation research
2. Technical Value: New application of integer point counting methods
3. Practical Utility: Provides concrete numerical bounds for related problems

Applicable Scenarios

1. Study of solution distribution in exponential Diophantine equations
2. Integer point counting problems near smooth functions
3. Explicit estimation problems in number theory

References

The paper cites important literature in the field, including Matveev's theory of logarithmic linear forms, Huxley's integer point counting results, and recent advances by Luca and Bennett on the Ratat-Goormaghtigh equation.

Summary of Technical Highlights: This paper cleverly transforms the Diophantine equation problem into a problem of integer point distribution near the graph of a smooth function through logarithmic transformation, employing modern analytic number theory techniques for integer point counting and lower bounds for logarithmic linear forms to obtain precise numerical results for this classical problem. This geometric approach opens new technical pathways for research on related Diophantine equations.