Mutually orthogonal Latin squares as group transversals (Q6043884): Difference between revisions

A Latin square of order \(n\) is an \(n\times n\) array with every cell being filled with one of \(\{1,2,\ldots ,n\}\) in such a way that each row and each column contains each letter exactly once. Two \(n\times n\) Latin squares \(L\), \(L^{\prime}\) are orthogonal if when the entries are treated as ordered pairs (say first component from \(L\) and second from \(L^{\prime}\)) the \(n^{2}\) ordered pairs produced are all distinct. A set of \(n\times n\) Latin squares any pair of which are orthogonal is called a mutually orthogonal family of Latin squares (MOLS). The size of such a family can fairly easily be shown to be bounded above by \(n-1\): when we have a set of \(n-1\) mutually orthogonal Latin squares, we call it a complete set of MOLS. MOLS have applications in a number of areas including experimental design and coding theory, but in general the question of how many MOLS there are for a given \(n\) is a hard one, though it is known that a complete MOLS exists for \(n\) a prime power. The main contribution of the paper under review is to make a link between MOLS and so-called Frobenius groups. A Frobenius group is a transitive permutation group \(G\), acting on a set \(\Omega\) with the property that only the identity element fixes more than one point. The stabilisers of points in \(\Omega\) are all conjugate by transitivity and are called Frobenius complements. The identity and the elements of \(G\) not in any point stabiliser form a normal subgroup \(N\) of \(G\) called the Frobenius kernel. The link is established through a theorem saying that if \(G\) is a Frobenius group with a Frobenius complement \(H\) and Frobenius kernel \(N\), then \(\{Nh\}\) as \(h\) runs over \(H\), is a Latin square \(T\). Further, if \(T_{i}=\{Nh_{i}\}\) is the Latin square formed as \(h_{i}\) ranges over one choice of Frobenius complement and \(T_{j}=\{Nh_{j}\}\) is the Latin square arising from another choice \(H_{j}\) of Frobenius complement, then \(T_{i}\) and \(T_{j}\) are orthogonal Latin squares. This has the consequence that for any given Frobenius group with Frobenius kernel \(N\) and Frobenius complement \(H\) we can construct \(\vert H\vert\) MOLS of order \(\vert N\vert\). This allows an easy re-proof of the existence of complete sets of MOLS of order an odd prime or any prime power simply by using standard examples of Frobenius groups. It should be noted that there are groups which are not Frobenius which have a complete set of MOLS, an example being given in the paper. Another recent paper, part of which (Section 3) deals with somewhat linked issues is [\textit{R. A. Bailey} et al., Des. Codes Cryptography 90, No. 9, 2069--2080 (2022; Zbl 07578458)].