The difference of consecutive primes (Q2646694): Difference between revisions

Let \(p_n\) denote the \(n\)-th prime, and let \(A=\liminf {p_{n+1}-p_n \over \log n}\). \textit{Hardy} and \textit{Littlewood} proved a few years ago, by using the Riemann Hypothesis (= R.H.) that \(A\leq {2\over 3}\) (not yet published), and \textit{R. A. Rankin} [Proc. Camb. Philos. Soc. 36, 255--266 (1940; Zbl 0025.30702)] recently proved, again by using R.H. that \(A \leq {3 \over 5}\). Depending on Brun's method the author proves without R.H. that \(A<1-c\) for a certain \(c > 0\). Denote by \(q_1 < q_2 \cdots < q_\nu\) the primes not exceeding \(n\). Then the author enunciates the following conjecture: \(\sum_{i=1}^{\nu-1} (q_{i+1}-q_i)^2 = O(n\log n)\). This is, if true, a stronger result than that of \textit{H. Cramér} [Acta Arith. 2, 23--46 (1936; Zbl 0015.19702)].