🔢 number theory

Arithmetic Bias in the Distribution of Mersenne Prime Exponents and the Divisor Structure of p-1

This paper proposes that the divisor structure of the exponent $p-1$ introduces a detectable arithmetic bias in the distribution of Mersenne primes, leading to a refined Wagstaff heuristic where exponents with higher normalized divisor complexity $S(p)$ are statistically more likely to yield prime Mersenne numbers.

Original authors: Jesus Dominguez

Published 2026-07-30✓ Author reviewed ⓘ

📖 4 min read🧠 Deep dive

Original authors: Jesus Dominguez

✨ This is an AI-generated explanation of the paper below. It is not written by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are a detective trying to solve the mystery of the universe's most elusive numbers: prime numbers. These are the building blocks of arithmetic, numbers like 2, 3, 5, and 7 that can only be divided by themselves and 1. Among these, there is a special club called "Mersenne primes." These are primes that look like a specific recipe: take the number 2, raise it to a prime power (like $2^3$ or $2^5$ ), and subtract 1. If the result is also a prime, you've found a Mersenne prime.

For a long time, mathematicians have had a very good rule of thumb for guessing how often these special numbers appear. It's called the Wagstaff heuristic. Think of it like a weather forecast for prime numbers: it says the chance of finding a Mersenne prime depends almost entirely on how big the exponent (the power you raise 2 to) is. The bigger the number, the harder it is to find a prime, but the rule gives a reliable average probability. However, just like a weather forecast might miss a sudden, localized storm, this rule doesn't look at the tiny, hidden details of the numbers involved. This paper asks a fascinating question: Is there a secret pattern in the "family tree" of the exponent that makes some numbers more likely to be prime than others, even if they are the same size?

The author, Jesús Domínguez, investigates whether the "divisor structure" of the exponent $p$ (specifically, the number of ways you can break down $p-1$ into smaller pieces) acts like a hidden filter. To understand this, imagine $p-1$ as a complex machine with many gears. Some machines have simple gears; others have a tangled web of many small gears working together. The paper introduces a score called $S(p)$ to measure how "tangled" or complex this machine is. A higher score means $p-1$ has a rich, complicated structure with many divisors.

The study looks at all the known Mersenne prime exponents and compares them to a control group of similar-sized prime numbers that aren't Mersenne primes. The results are striking: the Mersenne prime exponents consistently have higher $S(p)$ scores. In other words, the "machines" behind the Mersenne primes tend to be more complex and have more divisors than the machines behind the non-prime numbers.

The paper proposes a new model to explain this. It suggests that the complex structure of $p-1$ creates a series of "cyclotomic layers"—think of these as invisible sieves or filters. When you try to break down the number $2^p - 1$ into factors, these filters make it much harder for the number to be composite (breakable) if the structure is rich. It's as if a complex machine has so many internal safety locks that it's less likely to fall apart, making it more likely to stay whole (prime).

The author builds a mathematical model that refines the old Wagstaff rule. Instead of just looking at the size of the number, the new rule says: "The probability of being prime depends on the size and the complexity score $S(p)$ ." The model predicts that numbers with higher complexity scores get a "boost" in their chances of being prime. When the author tests this against the known data, the prediction matches reality very well. For example, in the range of numbers up to $10^8$ , the model predicts that about 33 out of 47 Mersenne primes should be in the "high complexity" group, and the actual observed number is 32 out of 47. This is a very close match.

However, the paper is careful to state that this is a "heuristic" model—a smart, educated guess based on patterns and simulations, not a proven mathematical theorem. It doesn't claim to have solved the mystery of prime numbers forever, nor does it say the old Wagstaff rule is wrong. Instead, it suggests that the old rule is a great average, but if you want to know the specific odds for a single number, you need to look at its internal "gears." The paper concludes by predicting that future discoveries of Mersenne primes will continue to show this bias toward complex structures, offering a new way to hunt for these mathematical treasures.

Technical Summary: Arithmetic Bias in the Distribution of Mersenne Prime Exponents and the Divisor Structure of $p-1$

Problem Statement
The classical heuristic framework for Mersenne primes, primarily the Wagstaff heuristic, posits that the probability of $M_p = 2^p - 1$ being prime depends almost exclusively on the magnitude of the exponent $p$ , following the asymptotic form $P(M_p \text{ prime}) \sim C \frac{\log p}{p}$ . This model treats secondary arithmetic features of the exponent, specifically the divisor structure of $p-1$ , as irrelevant to the aggregate probability scale.

This paper investigates whether the divisor structure of $p-1$ induces detectable, finite-scale variations in the distribution of known Mersenne prime exponents. The central hypothesis is that the algebraic constraints imposed by the cyclotomic decomposition of $2^{p-1}-1$ create a "filtering mechanism" that biases the occurrence of Mersenne primes toward exponents with specific divisor complexities.

Methodology

Structural Parameterization:
The author introduces a normalized divisor parameter, $S(p)$ , to quantify the complexity of the divisor lattice of $p-1$ :
$S(p) = \frac{\log \tau(p-1)}{\log \log p}$
where $\tau(n)$ is the divisor function. This normalization scales the divisor count relative to the typical multiplicative growth of integers, allowing for statistical comparison across different exponent sizes.
Algebraic and Cyclotomic Framework:
The analysis relies on the cyclotomic decomposition:
$2^{p-1} - 1 = \prod_{d|(p-1), d>1} \Phi_d(2)$
The paper argues that the divisors $d$ of $p-1$ generate "cyclotomic layers." Prime divisors of the values $\Phi_d(2)$ impose modular constraints on the potential factors of $M_p$ . Specifically, if $M_p$ is composite, its factors $q_i$ must satisfy $q_i \equiv 1 \pmod{2p}$ . The algebraic identity relating the factors to the symmetric polynomials of the parameters $k_i$ (where $q_i = 2pk_i + 1$ ) implies that the factorization must satisfy congruence conditions modulo the prime divisors of $\Phi_d(2)$ .
Empirical Analysis:
Using known Mersenne prime exponents (excluding the smallest cases), the author compares the values of $S(p)$ against a control set of nearby prime exponents of comparable size. Statistical procedures include local percentile comparisons, stratified conditional logistic regression, and permutation tests.
Reduced Coarse-Grained Structural Model:
To interpret the empirical bias, a heuristic model is developed. It treats the cyclotomic constraints as a filtering process on the space of admissible residue assignments for the factors of $M_p$ .
- Residue Concentration: The model analyzes the "concentration" of factors into specific residue classes modulo the cyclotomic moduli. Configurations where factors concentrate in fewer residue classes (diagonal configurations) are assigned higher "filtering factors."
- Effective Modulus Count: The model defines an effective count of cyclotomic moduli $L(p)$ (primes $a < p$ dividing $2^{p-1}-1$ ) and establishes an empirical linear relation $L(p) \approx k \tau(p-1)$ .
- Conditioned Probability: The model derives a conditioned probability law $P(M_p \text{ prime} | S)$ that refines the Wagstaff scale by a structural factor $C(p, S)$ .

Key Contributions and Results

Empirical Detection of Bias: The study finds that known Mersenne prime exponents exhibit significantly elevated values of $S(p)$ compared to control primes. For a comparison window of $W=5000$ , the mean percentile rank of Mersenne exponents is $\pi_{5000} = 0.648$ , with a standardized mean difference (Cohen's $d$ ) of approximately $0.56$. This suggests that exponents with richer divisor structures (higher $\tau(p-1)$ ) are overrepresented among Mersenne primes.
Structural Interpretation: The paper proposes that a richer divisor structure generates a larger family of cyclotomic layers, which in turn imposes a broader system of modular constraints. The model suggests that these constraints act as a finite-scale filter that preferentially allows primes to survive when the divisor complexity is higher.
Quantitative Model Calibration:
- A structural coefficient $\theta \approx 0.691$ is calibrated independently of the Mersenne prime distribution, based on the effective modulus count and the geometry of residue assignments.
- A marginal normalization factor $\beta \approx 0.621$ is derived to ensure the model preserves the aggregate Wagstaff probability scale when averaged over the distribution of $S$ .
Predictive Validation: The model predicts a redistribution of Mersenne primes toward the region of higher $S(p)$ . When comparing "Wagstaff-balanced" regions (dividing the exponent space into two sets with equal classical Wagstaff mass), the model predicts a ratio of approximately $33/14$ for the higher- $S$ to lower- $S$ regions. The observed distribution of known Mersenne primes is $32/15$ , showing close quantitative agreement.

Significance and Claims

The paper claims to provide a heuristic, finite-scale structural interpretation of the observed arithmetic bias, rather than a derivation from first principles or a modification of the classical asymptotic Wagstaff scale.

Refinement, Not Replacement: The proposed framework does not replace the classical Wagstaff heuristic. Instead, it refines it by redistributing the aggregate probability according to the structural parameter $S$ . The model asserts that the classical scale is recovered only after marginalizing over the distribution of $S$ .
Finite-Scale Nature: The author explicitly states that the model is "finite-scale" and "coarse-grained." It relies on the resolution of moduli $a < p$ and assumes that contributions from larger moduli are unresolved or dilute the signal at the exponent scale $p$ .
Falsifiability: The paper offers two prospective tests:
1. Future Mersenne prime exponents should continue to exhibit higher average $S(p)$ values than comparable primes.
2. The model makes specific quantitative predictions for the distribution of future Mersenne primes across Wagstaff-balanced structural regions (e.g., predicting a 76.25% allocation to the higher- $S$ region for the interval $10^8 \le p < 10^9$ ).

The work is presented as an experimental number theory investigation that links the algebraic structure of cyclotomic polynomials to the statistical distribution of Mersenne primes, offering a potential explanation for why certain exponents are more likely to yield primes than others of similar magnitude.

Technical Summary: Arithmetic Bias in the Distribution of Mersenne Prime Exponents and the Divisor Structure of p−1p-1p−1

More like this

Technical Summary: Arithmetic Bias in the Distribution of Mersenne Prime Exponents and the Divisor Structure of $p-1$