On cluster primes

From MaRDI portal





\textit{R. Blecksmith}, \textit{P. Erdős} and \textit{J. L. Selfridge} [Cluster primes, Am. Math. Mon. 106, 43--48 (1999; Zbl 0985.11041)] defined a prime \(p>2\) to be a cluster prime if every positive even integer \(2r\leq p-3\) can be written as a difference of two primes \(2r=q-q'\), with \(q'<q\leq p\). It is unknown whether there exist infinitely many cluster primes. If so, it would imply that there exist infinitely many pairs of consecutive primes whose difference is \(\leq 6\). Let \(\pi_C(x)\) be the number of cluster primes \(p\leq x\). In this short note, the author shows that the estimate \[ \pi_C(x)=O\left(\frac {x}{\exp\Bigl( \frac{1} {60} (\log\log x)^2\Bigr)}\right) \] holds for large \(x\), which improves upon the result of R. Blecksmith et al. (loc. cit.) who only proved that for all positive \(s\) one has \(\pi_C(x)=O_s(x/ \log^sx)\). The author also remarks that \(1/60\) might not be the best possible exponent obtainable by his method. The proof is elementary except for the fact that it uses a version of the large sieve.











This page was built for publication: On cluster primes

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4484455)