Integral automorphisms of affine spaces over finite fields (Q2364636): Difference between revisions

A classical problem in geometry is to construct large point sets in \(\mathbb{R}^n\) with pairwise integral Euclidean distances. Indeed, Ulam posed in 1945 the problem of whether there is a dense set of points in the real Euclidean plane so that all pairwise Euclidean distances between the points of \(S\) are rational. In [Discrete Comput. Geom. 39, No. 4, 786--790 (2008; Zbl 1145.52010)], \textit{T. Kreisel} and \textit{S. Kurz} gave two configurations of seven points in the plane, no three points in a line, no four points on a circle with pairwise integral distances. But it is still unknown if there exists an 8-point set in the plane with pairwise integral distances, with no three collinear and no four concircular. The subject of this paper is to consider similar questions in affine spaces over finite fields. See \textit{S. Kurz} [Australas. J. Comb. 43, 3--29 (2009; Zbl 1250.52011)] for more on this problem. The authors of the paper under review are interested in \textit{integral automorphisms} of the affine space \(\mathrm{AG}(n,q)\). These are permutations of the points of \(\mathrm{AG}(n,q)\) which preserve integral Euclidean distances. To be more specific, we say that \(x\) and \(y\) have integral distance if \(d(x,y)\) is a square in \(\mathbb{F}_q\) (and \(d\) denotes the usual Euclidean metric). Integral automorphisms in the plane case (\(n=2\)) were determined in two papers: Kurz [loc. cit.] and \textit{I. Kovács} and \textit{J. Ruff} [Finite Fields Appl. 27, 104--114 (2014; Zbl 1300.51003)]. In contrast with the plane, this paper shows that the integral automorphisms of \(\mathrm{AG}(n,q)\) for \(n\geq 3\), are semiaffine transformations. Thus, by the work of \textit{S. Kurz} and \textit{H. Meyer} [J. Comb. Theory, Ser. A 116, No. 6, 1120--1139 (2009; Zbl 1193.05163)] on integral semiaffine transformations, the integral automorphisms are completely known.