Concentration of points on two and three dimensional modular hyperbolas and applications (Q719072)

This paper is concerned firstly with the number \(I_2(M;K,L)\) of integer solutions \((x,y)\) of a congruence \(xy\equiv a\pmod{p}\), lying in specified intervals \(K<x\leq K+M\) and \(L<y\leq L+M\). Here \(K,L,M\) are all assumed to be integers, and \(a\) is an integer not divisible by the prime \(p\). As is well known, one may use Kloosterman sums to show that \(I_2(M;K,L)\) is asymptotic to \(M^2/p\) as soon as \(M\) is larger than about \(p^{3/4}\), but the emphasis here is on upper bounds when \(M\) is smaller than \(p^{3/4}\). Thus it is shown that for any \(\varepsilon>0\) one has \[ I_2(M;K,L)\ll_{\varepsilon} M^{\varepsilon}(M^{4/3}p^{-1/3}+1) \] and \[ I_2(M;L,L)\ll_{\varepsilon} M^{\varepsilon}(M^{3/2}p^{-1/2}+1) \] uniformly in \(K\) and \(L\). In particular one has \(I_2(M;K,L)\ll_{\varepsilon} M^{\varepsilon}\) as long as \(M\leq p^{1/4}\). The idea behind the proof comes from the reviewer's paper [Math. Proc. Camb. Philos. Soc. 83, 357--375 (1978; Zbl 0375.10027)]. The paper goes on to examine \(I_3(M;L)\), defined as the number of integer triples \(x,y,z\in (L,L+M]\) satisfying a congruence \(xyz\equiv a\pmod{p}\). Here it is shown that \(I_3(M;L)\ll_{\varepsilon} M^{\varepsilon}\) provided that \(M\ll p^{1/8}\). The proof of this uses a familiar estimate for the number of solutions of an inhomogeneous binary quadratic equation \(q(x,y)=0\) lying in a square. As a corollary it is shown that for any intervals \(X,Y,Z\) in \(\mathbb{F}_p^{\times}\) with lengths at most \(p^{1/8}\) one has \(\#(XYZ)=\left((\# X)(\# Y)(\# Z)\right)^{1+o(1)}\). A second corollary relates to the number \(J_a(M;K,L)\) of integer solutions \(x\in (K,K+M]\) and \(y\in(L,L+M]\) of the congruence \(y\equiv am^x\pmod{p}\). Here \(m\) is an element of order \(t\), say. For \(M<t\) it is shown that \[ J_a(M;K,L)\ll_{\varepsilon} M^{1/2+\varepsilon}(M^{3/4}p^{-1/4}+1) \] and \[ J_a(M;K,L)\ll_{\varepsilon} M^{1/3+\varepsilon}(Mp^{-1/8}+1) \] improving a result of \textit{T. H. Chan} and \textit{I. E. Shparlinski} [Acta Arith. 142, No. 1, 59--66 (2010; Zbl 1198.11002)].