Torsion in profinite completions (Q686031)

Let \(G\) be a residually finite torsion-free group and let \(\widehat{G}\) be the profinite completion of \(G\). It was shown by \textit{M. J. Evans} [J. Pure Appl. Algebra 65, 101-104 (1990; Zbl 0705.20020)] that \(\widehat{G}\) need not be a torsion-free group, and subsequently \textit{A. Lubotzky} [Q. J. Math., Oxf. II. Ser. 44, 327-332 (1993; see the preceding review Zbl 0829.20043)] showed that \(\widehat{G}\) can contain a copy of every finite group. The authors study torsion in \(\widehat{G}\) when \(G\) is soluble. They show that \(\widehat{G}\) is torsion free if \(G\) is an abelian group, a finitely generated abelian by nilpotent or soluble minimax group, but that \(\widehat{G}\) can have torsion if \(\widehat{G}\) is a nilpotent group of class 2. Their main result, proved by a clever and striking construction, is that \(\widehat{G}\) can have torsion if \(G\) is a finitely generated centre by metabelian group.