Symmetric complete sum-free sets in cyclic groups (Q5915549): Difference between revisions

Let us say here that a subset \(S\) of the abelian finite cyclic group \(\mathbb{Z}_n\) is \textit{admissible} if it is symmetric (\(-S = S\)), complete (\((S+S) \cup S = G\)) and sum-free (\((S+S)\cap S = \emptyset\)). The authors study the structure and the number of such admissible sets. They first recall main results in this direction. They write \(S = [n-2s+1, 2s-1]\cup(s+T)\cup (-s-T)\); the heart of the paper is to give a complete characterization of sets \(T\) such that \(S\) is admissible and to count their number, a clever combinatorial piece of work. A first consequence is that for any \(c<1/42\), the number of admissible subsets of \(\mathbb{Z}_n\) is larger than \(2^{cn}\) when \(n\) is large enough. As is usual, the authors obtain more precise results when \(n\) is a prime number. They also prove a conjecture of P. J. Cameron that for any \(\alpha\) in \([0, 1/3]\) and any positive \(\varepsilon\), when \(n\) is large enough, there exists an admissible \(S\) in \(\mathbb{Z}_n\) with cardinality in \([(\alpha-\varepsilon)n, (\alpha+\varepsilon)n]\). They finally give some applications of their results to graph theory and dioid partitions of graphs. [For the preliminary version see EuroComb 2017, Electronic Notes in Discrete Mathematics 61, 585--591 (2017; Zbl 1404.11009).