Construction of the mutually orthogonal extraordinary supersquares (Q2440578)

Let \(q=p^n\) be a prime power and \({\mathbb F}_q\) the field of order \(q\). A square of order \(q\) is a \(q\times q\) array with symbols \(\{1,\dots ,q\}\) satisfying the condition that each symbol appears exactly \(q\) times. (Latin squares are special squares.) This article concerns the construction and orthogonality of a certain class of squares, called extraordinary supersquares. The set of locations within a square of order \(q\) can be identified with \({\mathbb F}_q\times {\mathbb F}_q\), and any square of order \(q\) is determined, up to relabeling, by a partition of \({\mathbb F}_q\times {\mathbb F}_q\) into subsets size \(q\). One way to generate a partition is to consider the cosets of an order-\(q\) subgroup of \({\mathbb F}_q\times {\mathbb F}_q\); squares generated in this manner are called supersquares. This method of generating squares is convenient when investigating orthogonality, because two supersquares are orthogonal (i.e., no ordered pair of symbols is repeated in a superposition of the two supersquares) if and only if their generating groups have trivial intersection. The authors are chiefly interested in constructing complete sets of pairwise mutually orthogonal extraordinary supersquares, that is, superquares whose generating subgroups are extraordinary. (If \(G\) is a subgroup of \({\mathbb F}_q\times {\mathbb F}_q\) then \(G\) is extraordinary if \(\det (g_1\;g_2)\) is a zero of \(p(x)=x+x^p+x^{p^2}+\cdots +x^{p^{n-1}}\) for each \(g_1,g_2\in G\).) The authors claim applications of extraordinary supersquares, but these are neither fully described nor adequately referenced in the paper. The authors observe that when \(q\) is prime (i.e., \(n=1\)) a complete set of extraordinary supersquares can be obtain via a collection of ``linear'' subgroups of \({\mathbb F}_q\times {\mathbb F}_q\) (i.e., independent one-dimensional subspaces of \({\mathbb F}_q\times {\mathbb F}_q\)). When \(q\) is not prime the situation is more challenging. The authors construct complete sets of extraordinary supersquares in case \(q=4\), and claim that the case \(q=8\) will appear in another venue. The idea of identifying squares (up to relabeling) by partitions of the set of locations occurs, in the context of sudoku, in [\textit{R. A. Bailey} et al., Am. Math. Mon. 115, No. 5, 383--404 (2008; Zbl 1149.05010)].