Upper bounds for prime \(k\)-tuples of size \(\log N\) and oscillations (Q1880186): Difference between revisions

Let \({\mathcal A}=\{a_1,\ldots,a_k\}\) be a set of positive integers satisfying \({\mathcal A}\subset[1,N]\) with a large real number \(N\), and write \(E_{\mathcal A}(N)\) for the number of positive integers \(n\) up to \(N\) such that the numbers \(n+a_i\) are simultaneously prime numbers for all \(i\) with \(1\leq i\leq k\). In this paper, it is proved that if \(k\geq c\log N\) with a fixed positive constant \(c\), then one has the upper bound \[ E_{\mathcal A}(N)\leq2N\exp\bigl(-(1/4+o(1))\log N \log\log\log N/\log\log N\bigr), \] as \(N\rightarrow\infty\). It is also pointed out that the same bound holds for the case of prime \(k\)-tuples of the type \(n-a_i\) for \(1\leq i\leq k\). The proof uses the large sieve methods of Montgomery and of Gallagher, together with an estimate due to Vaughan for a sum appearing in the former large sieve. A bound of this strength had been proved only in the special case of the type \(n-2^i\) with \(1<2^i<n\), by \textit{R. C. Vaughan} ``Some applications of Montgomery's sieve'', [J. Number Theory 5, 64--79 (1973; Zbl 0266.10041)]. And for the general case, the above result improves the bound \(E_{\mathcal A}(N)\ll N\exp(-C(\log N)^{1/2})\) with a positive constant \(C\) depending on \(c\) which was implicitly obtained by \textit{A. Hofmann} and \textit{D. Wolke} ``On additive decompositions of the set of primes'' [Arch. Math. 67, 379--382 (1996; Zbl 0868.11045)]. It is also shown that if the Extended Riemann Hypothesis for Dedekind zeta functions is true, then there is no positive integer \(n\leq N\) such that \(n+2^i\) are simultaneously primes for all \(i\) with \(1\leq i\leq 2.7\log N\). In this case, \(2^i\) may be as large as \(N^{1.9}\) roughly. This theorem is derived from the fact that if such an \(n\) would exist, then \(n\) must be a multiple of each prime \(p\), with \(2^{p-1}<n\), for which \(2\) is a primitive root modulo \(p\). The proof requires a conditional result of Hooley on the asymptotic formula for the density of the primes \(p\) such that \(2\) is a primitive root modulo \(p\), and this is the reason why the Extended Riemann Hypothesis is assumed. Concerning the latter theorem, the author further discusses considerable oscillations of \(E_{\mathcal A}(N)\) according to \({\mathcal A}\), when \(k\) is as large as \(c\log N\) with a fixed positive constant \(c\).