On the smooth values of shifted almost-primes

Xiaodong Lü, Zhiwei Wang, Chen BinInternational Journal of Number Theory 2018 · DOI

TL;DR

This paper investigates the distribution of "almost-primes" (numbers with a limited number of prime factors) that have specific properties related to their largest and smallest prime factors. The authors establish a new lower bound for such numbers, contributing to our understanding of prime number distribution in shifted sequences. This work has implications for problems like the twin prime conjecture.

What Problem Does This Paper Address?

The paper tackles a problem in analytic number theory concerning the distribution of prime factors in sequences of integers. Specifically, it focuses on "shifted almost-primes," which are numbers of the form $n-a$ (where $n$ is an almost-prime and $a$ is a fixed integer) that have certain constraints on their largest and smallest prime factors.

The context is the study of prime numbers $p$ where $p-a$ has a large prime factor, or conversely, where $p-a$ has only small prime factors. While the distribution of prime factors of shifted primes has been extensively studied (e.g., the Elliott-Halberstam conjecture), this paper extends these investigations to almost-primes. The authors aim to provide a quantitative estimate (a lower bound) for the number of almost-primes $n \le x$ such that their smallest prime factor is large ($P^-(n) > x^{1/v}$) and their largest prime factor minus a constant is small ($P^+(n-a) < y$, where $y$ is related to $x$). This is a more general problem than dealing with primes, as almost-primes are more numerous.

Methods

The core of the paper's methodology lies in applying sieve methods, specifically the Rosser-Iwaniec sieve, to count these specific types of almost-primes. The authors define a counting function $\Phi(x, y, z)$ for almost-primes $n \le x$ with $P^-(n) > z$ and $P^+(n-a) < y$.

Key steps include:

  1. Applying the Sieve: Lemma 2.1 (the Rosser-Iwaniec sieve) is a crucial tool. It provides upper bounds for sums involving multiplicative functions over sifted sets. The authors use it to estimate $\Phi(x, y, z)$.
  2. Decomposition: The counting function $\Phi(x, y, z)$ is decomposed into a main term $M(x, y, z)$ and an error term $R(x, y, z)$.
  3. Estimating the Error Term: The error term $R(x, y, z)$ is shown to be admissible (i.e., small enough) using Lemma 2.4, which is a form of the Bombieri-Vinogradov theorem for friable numbers. This lemma provides bounds for sums over arithmetic progressions of friable numbers.
  4. Estimating the Main Term: The main term $M(x, y, z)$ is then estimated by changing the order of summation and applying further sieve techniques and results from previous work (specifically, a formula by Liu, Wu, and Xi for sums of multiplicative functions over certain sets of numbers).
  5. Connecting to Known Functions: The analysis involves the Dickman-de Bruijn function $\rho(u)$, which describes the density of smooth numbers (numbers with only small prime factors).

Datasets

Not discussed in the paper. The paper deals with theoretical number distributions rather than empirical data or computational experiments.

Key Insights and Ideas

  • Generalizing to Almost-Primes: The paper successfully extends results previously known for primes to the broader class of almost-primes. This is significant because almost-primes are more abundant, potentially leading to stronger results or new insights.
  • Unconditional Lower Bound: A key achievement is proving an unconditional lower bound for $\Phi(x, y, z)$ under certain conditions on $u$ and $v$. This means the result does not rely on unproven conjectures like the Elliott-Halberstam conjecture, which is a strong result in number theory.
  • Connection to Twin Primes: The paper shows how its main theorem can be applied to study problems related to the distribution of prime factors in consecutive integers, which is closely related to the famous twin prime conjecture. Specifically, it provides a lower bound for the number of integers $n$ where $P^+(n) > P^+(n+2)$.
  • Dual Problem: The authors also investigate the "dual problem" where $P^+(n) > P^+(n+2)$, demonstrating the versatility of their methods.

Key Results

  • Theorem 1: For $x \to \infty$, $y = x^{1/u}$, $z = x^{1/v}$ with $v > 4$ and $1 < u \ll 1$, the paper proves that $\Phi(x, y, z) \ge c_1(u, v) \frac{x}{\log x}$ for some strictly positive constant $c_1(u, v)$. This provides a lower bound for the number of almost-primes with specific prime factor properties.
  • Corollary 1: As an application, for $x \to \infty$ and $0 < \alpha < 1/4$, there exists a strictly positive constant $C_1(\alpha)$ such that $\sum_{n \le x, P^-(n) > x^\alpha, P^+(n) > P^+(n+2)} 1 \ge \{C_1(\alpha) + o(1)\} \frac{x}{\log x}$. This quantifies the density of integers $n$ where the largest prime factor of $n$ is greater than the largest prime factor of $n+2$, given $n$ has no small prime factors.

Limitations and Future Work

  • Parameter Restrictions: The main theorem has restrictions on the parameters $u$ and $v$ ($v > 4$, $1 < u \ll 1$). Extending these ranges, particularly for $u$, would be a natural next step.
  • Constant Factors: The paper provides lower bounds with explicit constants $c_1(u, v)$ and $C_1(\alpha)$, but these constants are not necessarily optimized. Further work could aim to improve these constants.
  • The Case $a=0$: The paper explicitly states it only considers the case $a \neq 0$. The distribution of prime factors for $n$ and $n-0=n$ (i.e., $P^+(n)$ vs $P^+(n)$) is a different problem, and extending the results to $a=0$ would be interesting.
  • Extending Domains: The authors suggest it would be interesting to extend the domain of $\alpha$ in Corollary 1 to $0 < \alpha < 1/2$.

Summary updated

Abstract

Denote by [Formula: see text] (respectively, [Formula: see text]) the largest (respectively, the smallest) prime factor of the integer [Formula: see text]. In this paper, we prove a lower bound of almost-primes [Formula: see text] with [Formula: see text] such that [Formula: see text] for [Formula: see text]. As an application, we study two patterns on the largest prime factors of consecutive integers with one of which without small prime factor.
BibTeX
@article{lu2018smooth,
  author = {Xiaodong Lü and Zhiwei Wang and Chen Bin},
  title = {{On the smooth values of shifted almost-primes}},
  journal = {International Journal of Number Theory},
  year = {2018},
  doi = {10.1142/s1793042118501683},
}
Figure