The large Davenport constant. II: General upper bounds. (Q392212)

The large Davenport constant \(D(G)\) of a finite group \(G\) (which for Abelian groups coincides with the usual Davenport constant) was introduced by the author and \textit{A. Geroldinger} in the first part of the paper under review [J. Pure Appl. Algebra 217, No. 5, 863-885 (2013; Zbl 1276.20027)], where also an explicit formula for it was obtained in the case of groups having a cyclic subgroup of index \(2\). Now the author proves several upper bounds for \(D(G)\). He shows in particular, when if \(G\) is neither cyclic nor dihedral of order congruent to \(2\bmod 4\), then \(D(G)\leq 3|G|/4\), and if \(p\) is the minimal prime divisor of \(|G|\) and \(G\) is noncyclic, then \(D(G)\leq 2|G|/p\). If \(G\) is a non-Abelian \(p\)-group, then the inequality \(D(G)\leq (p^2+2p-2)|G|/p^3\) is established (note that for Abelian \(p\)-groups an explicit formula for \(D(G)\) was established in 1969 by \textit{J. E. Olson} [J. Number Theory 1, 8-10 (1969; Zbl 0169.02003)], and it is shown that if \(D(G)=2q\) with prime \(q\), then \(G\) is non-Abelian and one has \(|G|=pq\) with prime \(p\) dividing \(q-1\). The proofs depend of combinatorial arguments, utilizing earlier results from group combinatorics and group theory (i.e. the characterization of minimal non-Abelian groups established in 1903 by \textit{G. A. Miller} and \textit{H. C. Moreno} [American M. S. Trans. 4, 398-404 (1903; JFM 34.0173.01)]). The central point of the proof is taken by the highly technical Lemma 3.3, providing a way of construction of a sequence in \(G\) having certain prescribed factorization properties.