Mean representation number of integers as the sum of primes
Gautami Bhowmik (LPP), Jan-Christoph Schlage-Puchta

TL;DR
Under the assumption of the Riemann Hypothesis, the paper derives asymptotic estimates for the average number of ways integers can be expressed as sums of two primes, establishing the optimality of these estimates through an Omega-term.
Contribution
The paper provides the first asymptotic estimates for the mean representation number of integers as sums of two primes under the Riemann Hypothesis, including a proof of the estimate's optimality.
Findings
Asymptotic formulas for mean representation numbers under RH
Proof of the optimality of these asymptotic estimates
Establishment of the best possible bounds through Omega-term
Abstract
Assuming the Riemann Hypothesis we obtain asymptotic estimates for the mean value of the number of representations of an integer as a sum of two primes. By proving a corresponding Omega-term, we prove that our result is essentially the best possible.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Mean representation number of
integers as the sum of primes
Gautami Bhowmik
Université de Lille 1
Laboratoire Paul Painlevé UMR CNRS 8524
59655 Villeneuve d’Ascq Cedex
France
and
Jan-Christoph Schlage-Puchta
Universiteit Gent
Department of Pure Mathematics and Computer Algebra
Krijgslaan 281
Gebouw S22
9000 Gent
Belgium
Abstract.
Assuming the Riemann Hypothesis we obtain asymptotic estimates for the mean value of the number of representations of an integer as a sum of two primes. By proving a corresponding -term, we show that our result is essentially the best possible.
1991 Mathematics Subject Classification:
11P32, 11P55
1. Introduction and Results
When studying the Goldbach conjecture that every even integer larger than is the sum of two primes it is natural to consider the corresponding problem for the von Mangoldt function . Instead of showing that an even integer is the sum of two primes, one aims at showing that is sufficiently large, more precisely, implies the Goldbach conjecture. It is known since long that this result is true for almost all . It is easy to see that if is an increasing function such that the Tchebychev function , then the mean value of satisfies the relation
[TABLE]
If we consider the contribution of only one zero of the Riemann zeta function , an error term of size appears, which, under the current knowledge on zero free regions of , would not be significantly better than . Fujii[3] studied the error term of this mean value under the Riemann Hypothesis (RH) and obtained
[TABLE]
which he later improved [4] to
[TABLE]
with , where the summation runs over all non-trivial zeros of . In fact,the oscillatory term is present even without assuming RH, however, it is necessary for the error estimate above.
In this paper we prove that
Theorem 1.1**.**
Suppose that the RH is true. Then we have
[TABLE]
and
[TABLE]
This confirms a conjecture of Egami and Matsumoto[2, Conj. 2.2]. Recently, Granville[6] used (1) to obtain new characterisations of RH. The innovation of the present work is the idea to use the distribution of primes in short intervals to estimate exponential sums close to the point 0. Note that using the generalised Riemann Hypothesis one could similarly find bounds for the exponential sums in question in certain neighbourhoods of Farey fractions. Such a bound, for example, fixes a gap in the proof of [6, Theorem 1C]. This approach can further be used to study the meromorphic continuation of the generating Dirichlet-series , as introduced by Egami and Matsumoto[2], a topic we deal with elsewhere[1].
The -power in the error term can be improved, but reaching would probably require some new idea.
We would like to thank the referee for suggesting the use of Lemma 2 below, which lead to a substantial improvement.
2. Proofs.
To prove the first part of our theorem, we compute the sum using the circle method. We use the standard notation.
Fix a large real number , set and let
[TABLE]
The following is due to Selberg[8, eq. (13)].
Lemma 1**.**
Assuming RH we have
[TABLE]
The following result is due to Gallagher, confer[7, Lemma 1.9] and put .
Lemma 2**.**
Let be complex numbers, and set . Then
[TABLE]
where
[TABLE]
Our main technical result is the following.
Lemma 3**.**
Suppose the RH. Then we have for the estimate
[TABLE]
Proof.
We put and into Lemma 4. Putting
[TABLE]
we obtain
[TABLE]
In the range we have
[TABLE]
For we have , thus we can apply Lemma 1 to obtain
[TABLE]
Finally for we have . The RH being equivalent to , this implies , and therefore
[TABLE]
Collecting our estimates our claim follows. ∎
Note that no non-trivial unconditional version of Lemma 3 can be proven without better understanding of the zeros of the Riemann -function, since the existence of a single zero close to 1 would already blow up the left-hand side.
Writing as we have
[TABLE]
We claim that the second term yields , and the last one an error of admissible size. In fact, the second term can be written as
[TABLE]
We now insert the explicit formula for , and replace the sum over by an integral to find that the second term is indeed .
We now consider the third term. We split the integral into an integral over and integrals of the form . On each interval we bound by , where is the distance of to the nearest integer, and using Lemma 3. For the first interval this yields
[TABLE]
while for the other intervals we obtain
[TABLE]
There are summands, hence, the contribution of to the whole integral is , and the first part of our theorem is proven.
We now turn to the proof of the -result. To do so we show that , hence, the left hand side of (1) has jumps of order . Since and are continuous, the error term cannot be . By considering the average behaviour of , one can even show that the error term is of order for integral , however, we will only do the easier case of real here.
The idea of the proof is that if an is divisible by many small primes, then should be large. Let be the exceptional modulus for which a Siegel-zero for moduli up to might exist, and be some prime divisor of . For the sake of determinacy we put , if no Siegel zero exists. We now use the following result due to Gallagher[5, Theorem 7].
Lemma 4**.**
We have
[TABLE]
provided that , , is an absolute positive constant, denotes summation over primitive characters modulo , and if there exists an exceptional character, for which a Siegel zero exists, this character has to be left out of the summation.
We put . Then all characters modulo is induced by some primitive character modulo , and
[TABLE]
which is negligible. Hence, it follows from Lemma 2, that
[TABLE]
where is the principal character, provided that for some absolute constant . It follows that for we have
[TABLE]
Now
[TABLE]
On the left we take the average over integers, hence, we obtain
[TABLE]
and our claim follows.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] G. Bhowmik, J.-C. Schlage-Puchta, Meromorphic continuation of the Goldbach generating function, to appear.
- 2[2] S. Egami, K. Matsumoto, Convolutions of von Mangoldt functions and related Dirichlet series, Proceedings of the 4th China-Japan Seminar held at Shandong, 1–23, S. Kanemitsu and J.-Y. Liu eds., World Sci. Publ., Hackensack, NJ, 2007.
- 3[3] A. Fujii, An additive problem of prime numbers, Acta Arith. 58 (1991), 173–179.
- 4[4] A. Fujii, An additive problem of prime numbers, II, Proc. Japan Acad. 67A (1991), 248–252.
- 5[5] P.X. Gallagher, A Large Sieve Density Estimate near σ = 1 𝜎 1 \sigma=1 , Invent. Math. 11 (1970), 329–339. 1 1 1 added reference
- 6[6] A. Granville, Refinements of Goldbach’s Conjecture, and the Generalized Riemann Hypothesis, Functiones et Approximatio 37 (2007), 159–173; Corrigendum 38 (2008), 125–127.
- 7[7] H.Montgomery, Topics in multiplicative number theory , Lecture Notes in Mathematics, Vol. 227, Springer-Verlag, 1971.
- 8[8] A. Selberg, On the normal density of primes in short intervals, and the difference between consecutive primes, Archiv for Mathematik og Naturvidenskab 47 (1943), 87–105.
