Suppose we consider primes of the form $p = 1 \text { mod } 4$, so that $p = a^2 + b^2$, $a$ and $b$ being integers. Considering only the first quadrant, all $(a,b)$ pairs will be of the form (odd,even) or (even,odd).

Now if we consider $q = (a + n)^2 + (b + n)^2 = p + 2na + 2nb + 2n^2$, $n$ an integer, then $q = 1 \text { mod } 4$.

$1 \text { mod } 4$ as an arithmetic progression contains an infinite number of primes, and while every $q$ may not turn out to be a prime, one would still expect an infinite number of prime pairs $(p,q)$.

Has any work been done related to this? My original motivation was that this could give very close Gaussian primes, differing by a distance of just $n\sqrt2$, when $n$ is small.

Update: I came across a recent paper (Bounded gaps between Gaussian primes) where instead of $(a + n,b + n)$, the author considers $(a + n,b + m)$, with $(n,m)$ either $\text (odd,odd)$ or $\text (even,even)$, and then proves it for the case $(0,m)$, $m>=246$.

| cite | improve this question | | | | |
  • 3
    $\begingroup$ Check out Bunyakovsky conjecture. $\endgroup$ – Wojowu Dec 26 '15 at 15:24
  • 3
    $\begingroup$ This is a particular case of a natural conjecture that would generalize Dirichlet's theorem to the Gaussian integers, but this generalization is not known. I doubt this particular case is known. It is a famous open problem whether there are infinitely many primes in another arithmetic progression, $n+i$, or equivalently, infinitely many primes in the integers of the form $n^2+1$. $\endgroup$ – Douglas Zare Dec 26 '15 at 15:26
  • 1
    $\begingroup$ I am pretty sure that Bunyakovsky conjecture is open for any specific nonlinear polynomial. And probably (less sure here) it is not even proved that there exists a nonlinear polynomial for which it holds. $\endgroup$ – Fedor Petrov Dec 26 '15 at 15:54
  • 1
    $\begingroup$ Friedlander and Iwaniec's paper (arxiv.org/pdf/math/9811185.pdf) mentions that the problem in one variable is generally more difficult than that in two. So it's possible this problem is more tractable than Bunyakovsky's. $\endgroup$ – valar123 Dec 26 '15 at 18:05
  • $\begingroup$ @valar123 this question is equivalent to the specific case of Bunyakovsky conjecture, is not it? $\endgroup$ – Fedor Petrov Dec 27 '15 at 14:54

An infinite number of primes on a ray of slope $45^{\circ}$ would not imply close Gaussian primes any more than ordinary primes $0^{\circ}.$ I didn't read the paper you link, but $246$ is the current record for gaps between ordinary primes. You might enjoy reading about Guassian moats.

| cite | improve this answer | | | | |
  • $\begingroup$ The result for ordinary primes by Maynard, Tao and Polymath was in 2013-14, while the paper I linked seems to be very recent (Nov 2015). I found the link on the Polymath project page (michaelnielsen.org/polymath1/…), so it seems credible. The author has adapted the original machinery for Gaussian primes. Maybe the first question doesn't have a elementary solution. I read about Gaussian moats, and the subject is indeed fascinating :) $\endgroup$ – valar123 Dec 27 '15 at 19:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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