Beyond pair correlation
This paper (and the authors' later more extensive paper [``Primes in short intervals, Commun. Math. Phys. 252, No. 1--3, 589--617 (2004)]) is concerned with the moments for the number of primes in short intervals, \[ \mu_k(X,h) := {1\over X}\int_0^X (\psi(x+h)-\psi(x) -h)^k\, dx, \] where \(\psi(x) = \sum_{n\leq x}\Lambda(n)\) and \(\Lambda(n)\) is equal to \(\log p\) if \(n=p^m\), \(p\) prime, \(m\geq 1\), and is zero otherwise. These papers provide convincing evidence that the values of \(\psi(x+h)-\psi(x)-h\) for \(1\leq x \leq X\) as \(X \to \infty\) are approximately normally distributed with mean \(\sim 0\) and variance \(\sim h\log X/h\), in contrast with the Cramér model where \(\log X/h \) is replaced with the larger \(\log X\). This paper examines the second moment in more detail. The authors have numerically computed \(\mu_k\) for \(1\leq k\leq 6\) with \(X=10^{10}\) and \(h=X^5\). In particular they found with these values \(\mu_2 = 9.0663\ldots \times 10^5\), which is not only much smaller than the Cramér model value of \(23.02\ldots \times 10^5\), but also smaller than the value \(11.51\ldots \times 10^5\) predicted by the variance above. The authors find an explanation for this discrepancy and make the refined conjecture that \[ \mu_2(X,h) = hX\log X/h -(C_0 +\log 2\pi) hX + R(X,h) , \] where \(C_0\) is Euler's constant, and \(R(X,h)\) is a smaller error term. For the data above the main terms here give a value \(9.098\ldots \times 10^5\) in good agreement with the numerics. This refined conjecture is based on the twin prime conjecture together with the singular series average \[ \sum_{k=1}^K (K-k)\mathfrak S(k) = {1\over 2}K^2 -{1\over 2}K\log K + (1-C_0 -\log 2\pi)K +O(K^{1/2+\varepsilon}), \] for which a short simple proof is given. (Here \(\mathfrak S (k)\) is the usual singular series in the Goldbach and twin prime conjecture). As for the error term \(R(X,h)\) in the conjecture above, the authors suggest that it may satisfy \(\ll h^{1/2}(X^{1+\varepsilon}+h^{1+\varepsilon}X^{1/2})\). In addition to the authors' second paper mentioned above, there are two recent results that relate to this work. In his thesis and the paper [``More precise pair correlation of zeros and primes in short intervals, J. Lond. Math. Soc. (2) 68, No. 3, 579--598 (2003; Zbl 1054.11047)] \textit{Tsz Ho Chan} proves an equivalence between a two term pair correlation conjecture and the two term asymptotic second moment formula above. In fact the constant in the lower order term perfectly matches the Riemann-von Mangoldt formula for the number of zeros of the Riemann zeta-function up to a given height, as one would expect if all the zeros of the zeta function are simple. There is also a surprising result of \textit{R. C. Vaughan} [Proc. Lond. Math. Soc. (3) 82, No. 3, 533--553 (2001; Zbl 1037.11060)] concerning the error term in the formula for the singular series average above. Vaughan showed that this error is \(O(K^{1/2}\exp(-c'{(\log 2 K)^{3/5}\over (\log \log 3 K )^{1/5}} ))\) and on the Riemann Hypothesis it is \(O(K^{5/12 +\varepsilon})\). For the entire collection see [Zbl 0999.00015].
- HIGHER MOMENTS OF PRIMES IN SHORT INTERVALS II
- The combinatorics of moment calculations
- MORE PRECISE PAIR CORRELATION OF ZEROS AND PRIMES IN SHORT INTERVALS
- The pair correlation of zeros of the Riemann zeta function and distribution of primes
- A lower bound for the second moment of primes in short intervals
- The tail of the singular series for the prime pair and Goldbach problems
- The error term in the Cesàro mean of the prime pair singular series
- Sums of singular series and primes in short intervals in algebraic number fields
- On a conjecture of Montgomery and Soundararajan
- On pair correlation and discrepancy
- On the singular series for primes in arithmetic progressions
- The combinatorics of moment calculations
- Explicit relations between primes in short intervals and exponential sums over primes
- A singular series average and the zeros of the Riemann zeta-function
- Primes in short intervals
- A Refined Conjecture for the Variance of Gaussian Primes across Sectors
- Sums of singular series with large sets and the tail of the distribution of primes
- On ratios of consecutive prime gaps
- The variance of integers without small prime factors in short intervals
- Arithmetic consequences of the GUE conjecture for zeta zeros
- Sums of singular series along arithmetic progressions and with smooth weights
- Joint distribution of primes in multiple short intervals
- Walking to infinity along some number theory sequences
- The error term in counting prime pairs
- Odd moments in the distribution of primes
- Primes in arithmetic progressions on average I
This page was built for publication: Beyond pair correlation
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4410061)
