Difference between consecutive primes (Q1906587)

The main theorem of this paper is that `almost all' intervals \([X, X+ X^{1/14+ \varepsilon} ]\) contain a prime. It is stated that similar methods prove that in any interval \([X, X+ X^{1/12+ \varepsilon} ]\) `almost all' even integers are the sum of two primes; and that the range \([X, X+ X^{23/ 588+ \varepsilon} ]\) contains at least one integer that is a sum of two primes. In all these results \(\varepsilon\) is a fixed positive constant, and \(X\) is sufficiently large. The first result has been obtained independently by \textit{N. Watt} [Acta Arith. 72, No. 2, 131-167 (1995; Zbl 0832.11030)]\ and also by \textit{Li Hongze} (unpublished). Moreover, in work to appear the author has shown that the exponent \(1/20+ \Sigma\) is admissible. While Watt's approach used a mean-value bound derived from `Kloostermania', the present argument depends solely on the classical mean and large values estimates for Dirichlet polynomials, building on ideas that go back to \textit{G. Harman} [Math. Z. 180, 335-348 (1982; Zbl 0482.10040)]. The sieve decompositions are however somewhat more intricate, involving a splitting into 20 separate regions.