2
$\begingroup$

Let $p$ be an odd prime number which large enough. I am interested in the study of the sums of reciprocals in the field $\mathbb{F}_p$.
In particular, I have the following question:
which primes $p$ do satisfy \begin{equation*} \sum_{j\equiv 1\pmod{5}}^{\frac{p-1}{2}}j^{-1}\not\equiv\sum_{j\equiv 3\pmod{5}}^{\frac{p-1}{2}}j^{-1}\pmod{p}\quad ? \end{equation*} Thanks for any comments or any helpful references.

$\endgroup$
6
  • $\begingroup$ Have you done any numerical experiments? $\endgroup$ Commented Jan 27, 2018 at 10:21
  • $\begingroup$ According to my calculations, the equality fails e.g. for $p=17$ and $p=1867$. $\endgroup$ Commented Jan 27, 2018 at 12:59
  • $\begingroup$ This is true for almost all primes up to $10^{14}$ using pariGP. $\endgroup$
    – Zakariae.B
    Commented Jan 27, 2018 at 13:07
  • 2
    $\begingroup$ Are there any other exclusions? Can you list them, if so? $\endgroup$ Commented Jan 27, 2018 at 13:10
  • 3
    $\begingroup$ A reasonable guess might be that the residues of the the left and right hand sides modulo $p$ act like independent, uniformly-distributed random variables. If so, we should expect that there are infinitely many $p$ for which the two sides are congruent, and that the number of $p\leq x$ for which there is congruence grows like $\log\log x$. But questions like this are hard, and I expect it will be difficult to prove that there are infinitely many $p$ for which congruence holds, or even that there are infinitely many $p$ for which congruence does not hold. $\endgroup$ Commented Jan 27, 2018 at 16:19

0

You must log in to answer this question.

Browse other questions tagged .