Abstract:The author sharpens a result of Jia (1996), showing that the interval $[n, n+n^{\frac{1}{21.5}+\varepsilon}]$ contains prime numbers for almost all $n$. Watt's mean value bound, a delicate sieve decomposition and more accurate estimates for integrals are used to good effect.
What problem does this paper attempt to address?
### The problems the paper attempts to solve
This paper aims to solve a well - known problem in number theory, that is, to find prime numbers in almost all short intervals. Specifically, the author attempts to prove that for almost all \(n\), the interval \([n, n + n^{1/21.5+\varepsilon}]\) contains prime numbers.
#### Background and existing results
- **Cramér's conjecture**: In 1937, Cramér [7] conjectured that for some \(f(n)\to1\) as \(n\to\infty\), every interval \([n, n + f(n)(\log n)^2]\) contains prime numbers.
- **Riemann hypothesis**: According to the Riemann hypothesis, for all sufficiently large \(n\), the interval \([n, n + n^{\theta}]\) contains approximately \(\frac{n^{\theta}}{\log n}\) prime numbers, where \(\frac{1}{2}+\varepsilon\leq\theta\leq1\).
- **Hoheisel's result**: In 1930, Hoheisel [19] proved that when \(\theta = 1-\frac{1}{33000}\), the above asymptotic formula holds.
- **Subsequent improvements**: Through the efforts of many mathematicians, such as Heilbronn, Chudakov, Ingham and Montgomery, etc., Huxley [20] proved in 1972 that when \(\theta>\frac{7}{12}\), the above asymptotic formula holds. Recently, Guth and Maynard [9] improved this result to \(\theta>\frac{17}{30}\).
#### Progress in almost all intervals
If only "almost all" intervals are considered instead of "all" intervals, then the lengths of these intervals can be shorter. For example:
- **Selberg's result**: In 1943, under the Riemann hypothesis, Selberg [44] proved that Cramér's interval contains prime numbers for almost all \(n\) if and only if \(f(n)\to\infty\) as \(n\to\infty\). He unconditionally proved that the interval \([n, n + n^{19/77+\varepsilon}]\) contains prime numbers for almost all \(n\).
- **Montgomery's result**: In 1970, Montgomery [37] improved the exponent \(\frac{19}{77}\) to \(\frac{1}{5}\) and gave an asymptotic formula.
- **Huxley's result**: Huxley [20] used the zero - density estimate to further improve the exponent to \(\frac{1}{6}\) and gave an asymptotic formula.
- **The strongest result**: The currently strongest result is the exponent \(\frac{2}{15}\) proved by Guth and Maynard [9].
#### The main contributions of this paper
This paper, by further developing sieve methods, especially by combining Watt's theorem, proves the following theorem:
**Theorem 1.1**: The interval \([n, n + n^{1/21.5+\varepsilon}]\) contains prime numbers for almost all \(n\). Specifically, assume that \(B\) is a sufficiently large positive constant, \(\varepsilon\) is a sufficiently small positive constant, and \(X\) is a sufficiently large positive integer. Then for \(n\in[X, 2X]\), except for \(O\left(\frac{X}{(\log X)^B}\right)\) exceptional values, the interval \([n, n + n^{1/2}\)