On a system of equations with primes (Q2341683): Difference between revisions

Given pairwise coprime integers \(u_1,\ldots,u_n\), a family \({\mathcal D}\) of nonempty subsets of \(\{1,2,\ldots,n\}\) and a function \(\varepsilon: {\mathcal D}\mapsto \{\pm 1\}\), the paper studies whether there exists a prime \(q\) not dividing \(u_1\cdots u_n\) which divides \(\prod_{i\in I} u_i-\varepsilon(I)\) for some \(I\in {\mathcal D}\). The paper answers this question in the affirmative when \(u_1,\ldots,u_n\) are prime powers (\(n\geq 3)\) and \({\mathcal D}\) and \(\varepsilon\) satisfy certain technical conditions. It is also shown that if \(A\) is a set of primes of cardinality at least \(3\) such that every prime factor of \(\prod_{p\in B} p-1\) belongs to \(A\) for all nonempty subsets \(B\) of \(A\), then \(A\) contains all the primes. The proofs are elementary using only congruence and divisibility relations. The deepest result used is Zsigmondy's particular case of the Primitive Divisors Theorem for members of Lucas sequences to the effect that, with some well-understood exceptions, for integers \(a>1\) and \(n>1\), \(a^n-1\) has a prime factor not dividing \(a^k-1\) for any \(1\leq k<n\).