English

On a system of equations with primes

Number Theory 2015-02-02 v5

Abstract

Given an integer n3n \ge 3, let u1,,unu_1, \ldots, u_n be pairwise coprime integers 2\ge 2, D\mathcal D a family of nonempty proper subsets of {1,,n}\{1, \ldots, n\} with "enough" elements, and ε\varepsilon a function D{±1} \mathcal D \to \{\pm 1\}. Does there exist at least one prime qq such that qq divides iIuiε(I)\prod_{i \in I} u_i - \varepsilon(I) for some IDI \in \mathcal D, but it does not divide u1unu_1 \cdots u_n? We answer this question in the positive when the uiu_i are prime powers and ε\varepsilon and D\mathcal D are subjected to certain restrictions. We use the result to prove that, if ε0{±1}\varepsilon_0 \in \{\pm 1\} and AA is a set of three or more primes that contains all prime divisors of any number of the form pBpε0\prod_{p \in B} p - \varepsilon_0 for which BB is a finite nonempty proper subset of AA, then AA contains all the primes.

Keywords

Cite

@article{arxiv.1212.0802,
  title  = {On a system of equations with primes},
  author = {Paolo Leonetti and Salvatore Tringali},
  journal= {arXiv preprint arXiv:1212.0802},
  year   = {2015}
}

Comments

13 pp, to appear in Journal de Th\'eorie des Nombres de Bordeaux. Fixed a number of typos (particularly, in the proof of Theorem 1.3). Abridged the part on the lifting-the-exponent lemma. Slightly simplified the formulation of Question 1. Added a new question (viz., Question 4 in this version). "Generalized" Theorem 1.3 (a hypothesis in the old statement was "evidently" useless)

R2 v1 2026-06-21T22:48:39.716Z