5
$\begingroup$

Fix some nonzero integers $a_1, \dots, a_k$, with $k \geq 2$, and some real numbers $c_1, \dots, c_k \in (0,1)$.

For every positive integer $m$, let $N(m)$ be the number of $k$-tuple of integers $(x_1, \dots, x_k)$ such that $$a_1 x_1 + a_2 x_2 + \cdots + a_k x_k \equiv 0 \pmod m $$ and $0 \leq x \leq c_i m$ for $i=1,\dots,k$.

I am interested in the asymptotic behavior of $N(m)$ as $m \to \infty$.

My guess is that, for $m$ relatively prime with $a_k$, we have that $N(m)$ has order about $c_1 \cdots c_k m^{k-1}$. Since for $i=1,\dots,k-1$ we have about $c_i m$ choices for each $x_i$, while $x_k \equiv -a_k^{-1}(a_1 x_1 + \cdots + a_{k-1}x_{k-1}) \pmod m$ has "probability" about $c_k$ of being in $[0, c_k m]$. If $\gcd(m, a_k) > 1$ one can get a similar heuristic after reducing by the common divisor.

Are there some asymptotic formulas or good upper/lower bounds for $N(m)$ as $m \to \infty$ ?

Thanks for your help

$\endgroup$
5
  • $\begingroup$ Number of solutions of a linear congruence in a box. In the years following Deligne's work on the Weil conjectures, there were several papers on solutions of systems of congruences (points on varieties) in small boxes. Some of the arguments used in those papers might go through in the linear setting, and without the need for the heavy machinery Deligne needed. $\endgroup$ Commented Dec 27, 2021 at 15:03
  • $\begingroup$ You might have a look at Edward B Burger, Small solutions to systems of linear congruences over number fields, Rocky Mountain Journal of Mathematics, Vol. 26, No. 3 (SUMMER 1996), pp. 875-888. It deals with a more general problem, but the first page gives a useful theorem of Thue and Aubry, and several references to the literature that might be helpful. $\endgroup$ Commented Dec 28, 2021 at 19:39
  • $\begingroup$ Making any progress, Erik? $\endgroup$ Commented Dec 31, 2021 at 2:23
  • 1
    $\begingroup$ @GerryMyerson Thanks for your suggestions. I looked at the papers but it seems to me that they are still quite far from what I have in mind. On the other hand, I've found the paper "Number of small solutions of a homogeneous linear congruence" of Semaev, which seems relatively close to what I have in mind. But the fact that his result is for "almost" all moduli, makes me think that I might be asking too much. $\endgroup$
    – Erik4
    Commented Dec 31, 2021 at 9:26
  • $\begingroup$ I think you're more likely to find a result for all primes, or all but finitely many primes, than for all moduli. $\endgroup$ Commented Dec 31, 2021 at 22:09

0

You must log in to answer this question.