On the distribution of rational functions along a curve over \(\mathbb F_p\) and residue races (Q557561)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | On the distribution of rational functions along a curve over \(\mathbb F_p\) and residue races |
scientific article |
Statements
On the distribution of rational functions along a curve over \(\mathbb F_p\) and residue races (English)
0 references
30 June 2005
0 references
Each \(x\in\mathbb F_p=\mathbb Z/p\mathbb Z\), where \(p\) is a prime, can be given the rational representation \(t(x)=x/p\) on the torus \(\mathbb T=\mathbb R/\mathbb Z=[0,1)\), and \(\mathbb F_p\) can be identified with \(p\mathbb T\cap\mathbb Z\); for \(s\geq1\), each \({\mathbf x}=(x_1,\ldots,x_s)\in\mathbb F_p^s\) then has the representation \(t({\mathbf x})=(x_1/p,\ldots,x_s/p)\in\mathbb T^s\). Now the points on a given curve \({\mathcal C}\) over \(\mathbb F_p\) often seem to be uniformly distributed over~\(\mathbb T^s\) when the natural embedding is imposed, although there may also be certain rational maps associated with \({\mathcal C}\) that will embed the points into a geometrically identifiable object. In particular, points on \({\mathcal C}\) can be injected into a proper linear subspace, which then rules out uniform distribution within~\(\mathbb T^s\). However, if one considers \({\mathcal C}\) to be embedded inside the smallest such subspace then one can ask whether the points are uniformly distributed therein. Roughly speaking the main result of the paper states that there is uniform distribution in such a situation under reasonable conditions. The authors take \({\mathcal C}\) to be an absolutely irreducible curve over~\(\mathbb F_p\) embedded in an affine space of dimension~\(r\), and call \({\mathbf h}=(h_1,\ldots,h_s)\) a suitable rational map from \({\mathcal C}\) to an embedded space of dimension~\(s\), with \(h_j=f_j/g_j\) and \(f_j,g_j\in \mathbb F_p[X_1,\ldots,X_r]\), if \({\mathcal C}\) is not contained in the hyper-surface \(g_j=0\) for \(1\leq j\leq s\). Thus the set \({\mathcal C}(\mathbb F_p)\) of points on~\({\mathcal C}\) is a subset of~\(\mathbb T^r\), while the image \(t({\mathbf h}({\mathcal C}(\mathbb F_p)))\) is a subset of~\(\mathbb T^s\). For a region~\(\Omega\subset\mathbb T^s\) with a smooth boundary and under certain conditions involving linear independence imposed on the components of a suitable rational map~\({\mathbf h}\), the ratio \[ \mu_{{\mathcal C},{\mathbf h}}(\Omega)={^\sharp\{ {\mathbf x}\in{\mathcal C}(\mathbb F_p)\,| \,t({\mathbf h}({\mathbf x}))\in\Omega \} \over^\sharp{\mathcal C}(\mathbb F_p)} \] is expected to have the measure of~\(\Omega\) as its asymptotic value when \(p\to\infty\). Special cases of the problem have been investigated by others; see, for example, \textit{J.~Beck} [Period. Math. Hung. 44, 147--155 (2002; Zbl 1017.11043)], \textit{C.~Cobeli} and \textit{A.~Zaharescu} [Acta Arith. 99, 321--329 (2001; Zbl 1025.11021)], \textit{M.~Vajaitu} and \textit{A.~Zaharescu} [Monatsh. Math. 136, 81--86 (2002; Zbl 1029.11022)]. The authors generalise such results for suitable rational maps from \({\mathcal C}\) to regions in proper subspaces and their translates; the precise statements of the theorems require too much further technical definitions to be given here. The proof of the main result involves the estimation of exponential sums in finite fields; in particular it makes use of a theorem of \textit{E.~Bombieri} [Am. J. Math. 88, 71--105 (1966; Zbl 0171.41504)]. Letting \(p\to\infty\), the set of values for \(\mu_{{\mathcal C},{\mathbf h}}(\Omega)\) has a limiting spectrum, and there is a theorem on this when \({\mathcal C}\) runs over the set of lines. The paper also contains a discussion on the discrepancies. In a `residue race' we seek to determine the number of \(x\in\mathbb F_p\) for which \(t(h_1(x))<\cdots<t(h_s(x))\). Under reasonable conditions, the number concerned should be asymptotic to~\(p/s!\) as \(p\to\infty\). The special cases with \(h_i(x)=a_ix\) and \(h_i(x)=1/(x+a_i)\) for the problem have been investigated by \textit{A.~Granville, D.~Shiu} and the reviewer [Ramanujan J. (to appear)], and \textit{C.~Cobeli} and \textit{A.~Zaharescu} [Mathematika 47, 87--108 (2000; Zbl 1031.11058)], respectively. Some results corresponding to suitable rational maps from \({\mathcal C}\) are now corollaries to the theorems.
0 references
affine curves
0 references
distribution on the torus
0 references
discrepancy
0 references
0 references