Asymptotics of number fields and the Cohen-Lenstra heuristics (Q2642776)

Let \(k\) be a number field and let \(G<S_n\) be a transitive permutation group on \(n\) letters. We say that a finite extension \(K/k\) has Galois group \(G\) if the normal closure \(\widehat{K}\) of \(K/k\) has Galois group isomorphic to \(G\) and \(K\) is the fixed field in \(\widehat{K}\) under a point stabilizer. By abuse of notation we write \(\text{Gal}(K/k)=G\) in this situation. We let \[ Z(k,G;x):=\sharp\{K/k: \text{Gal}(K/k)=G,\;N_{K/k}(d_{K/k})\leq x\} \] be the number of field extensions of \(k\) (inside a fixed algebraic closure \(\overline{{\mathbb{Q}}}\)) of relative degree \(n\) with Galois permutation group isomorphic to \(G\) and norm of the discriminant \(d_{K/k}\) bounded above by \(x\). It is well known, that this number is finite. A conjecture of Malle gives a statement about the asymptotic behavior of this function for \(x\rightarrow \infty\). The author of the paper under review studies this conjecture for dihedral groups \(D_l\) of order \(2l\), where \(l\) is an odd prime. He proves the expected lower bound for those groups. For the upper bounds he shows that there is a connection to class groups of quadratic number fields. The asymptotic behavior of those class groups is predicted by Cohen-Lenstra heuristics. Under the assumption of this heuristic the expected upper bound can be proved.