The fraction of the bijections generating the near-ring of 0-preserving functions. (Q2580942)

Let \(G\) be an additively written finite group which may not be Abelian, and let \(0\) denote the identity element of \(G\). The article under review concerns the zero-symmetric nearring \(M_0(G)=\{f\colon G\to G\mid f(0)=0\}\). Denote by \(B_0(G)\) the set of all bijections on \(G\), and \(\text{Inv}(G)\) the set of all involutions (elements of order~\(2\)) in \(M_0(G)\). A mapping \(f\in B_0(G)\) may or may not generate \(M_0(G)\) as a nearring. Let \(m=|B_0(G)|\) and \(n=|\{f\in B_0(G)\mid f\) generates \(M_0(G)\}|\), and define \(p(G)=n/m\). Also put \(q(G)=1-p(G)\). The main theorem asserts that (1) \(q(G)\geq|\text{Inv}(G)|/2(|G|-1)\), and (2) there is an \(n_0\geq 0\) such that if \(|G|\geq n_0\), then \(q(G)\leq\text{Inv}(G)/(|G|-1)+3|G|^{-2/3}\). Note that a mapping \(f\in B_0(G)\) can generate \(M_0(G)\) only if it is not contained in a maximal subnearring of \(M_0(G)\). Thus, the proof of the theorem makes use of a previous article of the author [Commun. Algebra 33, No. 8, 2499-2518 (2005; Zbl 1081.16050)] on the characterization of the maximal subnearrings of \(M_0(G)\).