Products of characters and finite \(p\)-groups. (Q1879643)

Let \(G\) be a finite \(p\)-group (\(p\) prime) throughout this review. The main result of this paper is the following Theorem A. Let \(\chi\), \(\psi\) be irreducible complex characters of \(G\) both faithful. Let \(\chi\psi=\sum^t_{i=1}a_i\eta_i\) with \(a_i\in\mathbb{N}\) and \(\eta_i\in\text{Irr}(G)\), with \(i=1,\dots,t\). Then \(2t>p\) or \(t=1\). The proof depends on a few easy lemmas and one main lemma. Main Lemma (under the assumptions made above): Suppose \(2t\leq p\). Then \(\chi(g)=0=\psi(g)\) for all \(g\in G\setminus Z(G)\). It is remarkable that the proof of the main lemma requires thirteen pages in this paper! The bound \(2t>p\) from Theorem A is sharp in the sense that there exist groups \(G\) with \(\chi\in\text{Irr}(G)\) such that \(\chi^2=\sum^t_{i=1}a_i\eta_i\) where \(2t-1=p\). As a consequence of Theorem A, the following is shown. Theorem B (under the notations as above): Let \(\psi=\overline\chi\). Let \(n\in\mathbb{N}\). Then there exists a finite set \(F\) of positive integers such that for any \(G\) and any \(\chi\in\text{Irr}(G)\) with \(\chi\overline\chi=\sum^n_{i=1}a_i\eta_i\), we have that \(\chi(1)\in F\). The same conclusion is valid if \(G\) is replaced by any finite nilpotent group. The reviewer remarks that the results of this paper possibly might be related to the work of Arad and Fisman from around 1983, in the sense that these authors worked on products of irreducible characters having two distinct irreducible constituents in the decomposition into irreducible characters of such a product (i.e., for which \(t=2\) features). Maybe also the theory of table algebras might be of use.