8
$\begingroup$

The Wikipedia article about Elliott-Halberstam (EH for short) conjecture says that the so-called Bombieri-Vinogradov theorem, which is a weaker form of EH conjecture, is in some sense an averaged form of the Generalized Riemann Hypothesis (that is, the analogue of RH for Dirichlet L-functions). So, would the full EH conjecture follow from GRH?
Thanks in advance.

Edit March 29th 2021: denoting by $\Theta_{EH}:=\sup\{\theta\vert EH(\theta)\}$, EH conjecture is equivalent to $\Theta_{EH}=1$. The proven lower bound for it has increased from $1/2$ (Bombieri-Vinogradov) to a greater value. On the other hand, the proven upper bound for the de Bruijn-Newman constant $\Lambda$ has decreased from $1/2$ to $0.2$ (Platt and Trudgian), while the proof of Newman's conjecture by Rodgers and Tao implies $\Lambda\geq 0$, and RH is equivalent to this constant vanishing. So that, it seems that $\Theta_{EH}$ and $\Lambda$ are "dual", the duality relation being the involution $s\mapsto 1-s$. Is it thus likely that $\theta<1-\Lambda\Longrightarrow EH(\theta)$?

$\endgroup$

1 Answer 1

11
$\begingroup$

The Elliott-Halberstam conjecture is not known to follow from GRH. Even the weak version of EH (which is with $Q=x^{1/2+\epsilon}$ for any fixed $\epsilon>0$) does not follow from GRH. On the other hand, it is known that the Elliott-Halberstam conjecture almost implies the twin primes conjecture, i.e., it implies that there are infinitely many pairs of primes at distance $≤ 16$ (now $\le 12$, see Sylvain's comment). Furthermore, the Bombieri-Vinogradov theorem is indeed an amazingly strong unconditional replacement for the GRH bound (and has as natural strengthening the EH conjecture).

$\endgroup$
2
  • 2
    $\begingroup$ EH even implies that $16$ can be replaced by $12$, as shown by James Maynard. $\endgroup$ Dec 29, 2013 at 19:17
  • 1
    $\begingroup$ I think a big problem in preventing GRH from showing EH is the possible non-uniformity in implicit constants in the error terms. Perhaps a uniform version of GRH would suffice. $\endgroup$ Dec 30, 2013 at 0:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.