Table of Contents
Fetching ...

On the number of divisors of Mersenne numbers

Vjekoslav Kovač, Florian Luca

TL;DR

This paper investigates the growth of the divisor-sum over Mersenne numbers, $f(n)=\sum_{1\le k\le n} \tau(2^k-1)$. It proves that the doubling ratio $f(2n)/f(n)$ is unbounded in the sense that $\limsup_{n\to\infty} f(2n)/f(n)=\infty$, and it achieves this via a comparison to the function $f'(n)=\sum_{1\le k\le n} 2^{\tau(k)}$ together with highly--composite Mersenne indices. The authors also formulate two conditional results: if either Conjecture 1 (existence of indices of highly--composite Mersenne numbers implying growth of $\tau(2^N+1)$) or Conjecture 2 (a logarithmic bound on $\omega(\Phi_d(2))$) holds, then $f(2n)/f(n)\to\infty$. Extensive numerical data, along with a partially heuristic model based on Gillies' approximation, corroborate the observed large fluctuations in $\tau(2^n-1)$ and support the proposed divergence, providing insight into the complex divisor structure of Mersenne numbers and its connections to cyclotomic polynomials and highly--composite phenomena.

Abstract

Denote $f(n):=\sum_{1\le k\le n} τ(2^k-1)$, where $τ$ is the number of divisors function. Motivated by a question of Paul Erdős, we show that the sequence of ratios $f(2n)/f(n)$ is unbounded. We also present conditional results on the divergence of this sequence to infinity. Finally, we test numerically both the conjecture $f(2n)/f(n)\to\infty$ and our sufficient conditions for it to hold.

On the number of divisors of Mersenne numbers

TL;DR

This paper investigates the growth of the divisor-sum over Mersenne numbers, . It proves that the doubling ratio is unbounded in the sense that , and it achieves this via a comparison to the function together with highly--composite Mersenne indices. The authors also formulate two conditional results: if either Conjecture 1 (existence of indices of highly--composite Mersenne numbers implying growth of ) or Conjecture 2 (a logarithmic bound on ) holds, then . Extensive numerical data, along with a partially heuristic model based on Gillies' approximation, corroborate the observed large fluctuations in and support the proposed divergence, providing insight into the complex divisor structure of Mersenne numbers and its connections to cyclotomic polynomials and highly--composite phenomena.

Abstract

Denote , where is the number of divisors function. Motivated by a question of Paul Erdős, we show that the sequence of ratios is unbounded. We also present conditional results on the divergence of this sequence to infinity. Finally, we test numerically both the conjecture and our sufficient conditions for it to hold.

Paper Structure

This paper contains 12 sections, 3 theorems, 55 equations, 5 figures, 2 tables.

Key Result

Theorem 1

We have i.e., the sequence $(f(2n)/f(n))_{n\geqslant 1}$ is unbounded.

Figures (5)

  • Figure 1: Values of $\tau(2^n-1)$ for $1\leqslant n\leqslant 100$.
  • Figure 2: Numerical values of the ratios $f(2n)/f(n)$ for $1\leqslant n\leqslant 50$, $1\leqslant n\leqslant 200$, and $1\leqslant n\leqslant 603$, respectively.
  • Figure 3: Values of $\omega(\Phi_d(2))$ for $1\leqslant d\leqslant 400$ and $1\leqslant d\leqslant 1206$, respectively.
  • Figure 4: Base $10$ logarithms of the ratios between the approximate and the true values of $\tau(2^n-1)$ for $1001\leqslant n\leqslant 1100$.
  • Figure 5: Numerical values of the approximately evaluated ratios $f(2n)/f(n)$ for $1\leqslant n\leqslant 1000$ (left) and the last $100$ points zoomed (right).

Theorems & Definitions (9)

  • Theorem 1
  • Proposition 2
  • Conjecture 1
  • Conjecture 2
  • Theorem 3
  • proof : Proof of Proposition \ref{['prop']}
  • proof : Proof of Theorem \ref{['thm:limsup']}
  • proof : Proof of Theorem \ref{['thm:cond']} assuming Conjecture \ref{['conj1']}
  • proof : Proof of Theorem \ref{['thm:cond']} assuming Conjecture \ref{['conj2']}