The class-breadth conjecture revisited. (Q2496873)

Let \(G\) be a finite \(p\)-group. Let \(c=c(G)\) be the nilpotence class of \(G\), and \(p^{b(G)}\) be the size of the largest conjugacy class of \(G\); \(b=b(G)\) is called the breadth of \(G\). \textit{C. R. Leedham-Green, P. M. Neumann} and \textit{J. Wiegold} conjectured [J. Lond. Math. Soc., II. Ser. 1, 409-420 (1969; Zbl 0186.32602)] that \(c\leq b+1\), and proved that \(c\leq p/(p-1)\cdot b+1\). A counterexample to the conjecture was provided by \textit{W. Felsch, J. Neubüser} and \textit{W. Plesken} [J. Lond. Math. Soc., II. Ser. 24, 113-122 (1981; Zbl 0426.20016)]. This is a group of order \(2^{34}\), coclass \(5\) and cobreadth \(7\). If the order of \(G\) is \(p^n\), its coclass is \(cc=n-c\), and its cobreadth is \(cb=n-b\), so the original conjecture claims that \(cb\leq cc+1\). The example relies on the theory of uniserial space groups. In the paper under review, the authors discuss the conjecture in the context of \(p\)-uniserial space groups. They produce an effective algorithm for constructing all \(p\)-uniserial space groups of a given dimension that are potential counterexamples. They find, for instance, that there is a unique counterexample among the \(4\)-dimensional space groups; it has a quotient of order \(2^{19}\), coclass \(4\) and cobreadth \(6\). No counterexamples of odd order are known; the authors show that when \(p=3\) there are none up to dimension \(54\). There is a version of the conjecture for Lie algebras. The authors produce counterexamples over any finite field, using work of the reviewer, \textit{S. Mattarei} and \textit{M. F. Newman} on graded Lie algebras of maximal class [Trans. Am. Math. Soc. 349, No. 10, 4021-4051 (1997; Zbl 0895.17031)]. (Also submitted to MR.)