Non-PORC behaviour of a class of descendant \(p\)-groups. (Q1931696)

\textit{G.~Higman}'s PORC conjecture [Proc. Lond. Math. Soc., III. Ser. 10, 566-582 (1960; Zbl 0201.36502)] states that for a fixed \(n\), the number \(f(p,n)\) of \(p\)-groups of order \(p^n\) is Polynomial On Residue Classes (PORC), that is, is given by a polynomial in \(p\), whose coefficients depend on the residue class of \(p\) modulo some fixed integer \(N\). In the paper under review the authors start with a \(p\)-group \(G_p\) of order \(p^9\) and nilpotence class \(2\) studied by the first author [J. Reine Angew. Math. 549, 1-21 (2002; Zbl 1001.20032)], which encodes the elliptic curve \(y^2=x^3-x\). They show that the function counting the number of immediate descendants of \(G_p\) of order \(p^{10}\) is not PORC. We recall that the lower \(p\)-central series in a group \(G\) is defined by \(G_1=G\), and \(G_{i+1}=[G_i,G]G_i^p\) for \(i\geq 1\), and that if \(G_c\neq\{1\}=G_{c+1}\), then we say that \(G\) is an immediate descendant of \(G/G_c\). The authors discuss the implication of this result for the PORC conjecture. In the cases where Higman was able to exhibit PORC behaviour, he counted groups by adding up the numbers of groups in certain subclasses, each of which was PORC. In the example discussed in the paper under review, one of those summands is shown not to be PORC. It might well be that other non-PORC summands give a PORC ``grand total'', but the authors consider this unlikely. We refer to the paper, which is very clearly written, for further details.