5
$\begingroup$

Suppose $r$ integers $n_1, \ldots, n_r$ are given such that $0<|n_i|<N$ for $1 \leq i \leq r$ and for a natural number $N.$ Is it possible to find a prime number $p$ such that all numbers $n_1, \ldots, n_r$ are quadratic nonresidue mode $p$? Can we say $p < N^6?$

I think the answer to my first question is yes, but I don't know how to give an upper bound for $p$?

$\endgroup$
3
  • 2
    $\begingroup$ You must assume that no odd-size product of the $n_i$ is a square; for instance you can't find $p$ for which $2,3,6$ are all quadratic nonresidues (nor, of course, $p$ such that $1$ or $49$ is a quadratic nonresidue). $\endgroup$ Oct 31, 2018 at 2:32
  • $\begingroup$ I'm not sure whether $p<N^6$, but if the integers are chosen to be the primes under $127$, then $p$ is at least $1728061733$, showing that $p<N^{4.3}$ is false. $\endgroup$ Oct 31, 2018 at 7:49
  • $\begingroup$ Also posted to, and then deleted by author from, m.se: math.stackexchange.com/questions/2978501/quadratic-nonresidue $\endgroup$ Oct 31, 2018 at 12:48

1 Answer 1

6
$\begingroup$

I don't think this is true in general, for instance, if there exists $1\leq i <j<k<r$ such that $n_in_j = n_k$, then $(n_i/p)=(n_j/p)=-1\implies (n_k/p)=1$ (where $(a/p)$ is the Legendre symbol). Moreover, if $n_i$ is a perfect square, then trivially you have that for every prime $p$, $n_i$ is a quadratic residue. So, you must have the assumption that, for each $n_i$, there is a prime $p\mid n_i$, such that $p$ appears with an odd exponent in decomposition of $n_i$.

However, if you assume that, say, $(n_i,n_j)=1$ for every $i \neq j$, then yes, you can construct such a prime. Here is one way. For each $n_1,\dots,n_r$, let us consider the sets $P_1,P_2,\dots,P_r$ where $P_i$ is the set of prime divisors of $i $, that appear with an odd exponent (assumed to be non-empty). Now, select primes $p_j\in P_j$ for every $j$, let $\mathcal{P}=\{p_1,\dots,p_r\}$, and let $\mathcal{Q}=\bigcup_{k=1}^r P_k \setminus \mathcal{P}$.

We will construct a prime $q$ such that $q$ is a quadratic residue modulo $p$ for every $p\in\mathcal{P}$, and $q$ is quadratic non-residue, modulo every $p\in \bigcup_{k=1}^r P_k \setminus \mathcal{P}$, using the law of quadratic reciprocity. Take $q\equiv 1\pmod{8}$, and thus, $(-1)^{(p-1)/2 \cdot (q-1)/2}=1$. Now, take $q\equiv n_p\pmod{p}$, where $n_p$ is a quadratic residue, modulo $p$ for every $p\in\bigcup_{k=1}^r P_k \setminus \mathcal{P}$; and take $q\equiv n_p'\pmod{p}$, for every $p\in \mathcal{P}$, where $n_p'$ is a quadratic non-residue. Clearly, such a $q$ exists (from Chinese remainder theorem), and satisfies your claim.

$\endgroup$
6
  • $\begingroup$ I agree with your assumptions and your solution, but can you give an upper bound for $q$? By Chinese remainder, we only get the existence of $q.$ $\endgroup$
    – Ken
    Oct 31, 2018 at 2:40
  • 5
    $\begingroup$ For the upper bound, once you have the arithmetic progression, use Linnik's theorem. en.wikipedia.org/wiki/Linnik%27s_theorem $\endgroup$ Oct 31, 2018 at 4:39
  • 3
    $\begingroup$ Chinese remainder theorem and Linnik’s theorem will give a bound of the form $N^{O(r)}$. I seriously doubt you can bound the result by a fixed power of $N$ independent of $r$. $\endgroup$ Oct 31, 2018 at 14:27
  • 1
    $\begingroup$ The conditions on $n_i$ are somewhat stronger than needed. A sufficient and necessary condition is the one given in a comment by Noam Elkies above: that no product of an odd-size subset of the $n_i$s is a square. $\endgroup$ Oct 31, 2018 at 15:35
  • 1
    $\begingroup$ @Ken No, as is demonstrated by counterexamples both in Noam Elkies’s comment and in this answer. $\endgroup$ Oct 31, 2018 at 18:52

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.