Nonabelian Cohen-Lenstra moments (Q1734002)

Let $K$ be a quadratic number field, $L/K$ a finite, unramified Galois extension with Galois group $G$ and $\tilde L$ a normal closure of $L$ over $\mathbb Q$. Then $G'$, the Galois group of $\tilde L/\mathbb Q$, is an \textit{admissible} subgroup of the wreath product $G \wr S_2$, where the property to be \textit{admissible} can be described purely group-theoretically. Given a finite group $G$ and an admissible subgroup $G'$ of $G \wr S_2$, the author investigates $E^{\pm} (G,G')$, the proportion of unramified $G$-extensions $L/K$ with $\text{Gal}(\tilde L/\mathbb Q) \simeq G'$, where $K$ runs through the real ($+$) or imaginary ($-$) quadratic fields. She conjectures that if $G'$ is ``good'' one has \[ E^-(G,G') = \frac{|H_2(G',c)[2]|}{|\text{Aut}_{G'}(G)|} \quad \text{and} \quad E^+(G,G') = \frac{|H_2(G',c)[2]|}{|c||\text{Aut}_{G'}(G)|}, \] and otherwise $E^{\pm} (G,G')= \pm \infty$. Here $c$ denotes the unique conjugacy class of order-2 elements of $G'$, which are not in the kernel of $\pi: G' \to S_2$, and $H_2(G',c)[2]$ the 2-torsion of a reduced Schur multiplier. For odd $|G|$, this conjecture agrees with previous ones, and it is known to be true for very few special cases. More generally, the author considers the analogous question for the function field case, replacing $\mathbb Q$ by $\mathbb F_q(t)$, and adapts her investigations to the general global situation. The main result of this paper (Theorem 1.2) proves some analogue of the above conjecture for the function field case, which the author considers as one out of three main motivations for her conjecture. In Section 3, the author defines a new lifting invariant in the global case and proves in Theorem 3.13 that in the function field case this coincides with the invariant given by \textit{J.S. Ellenberg} et al., [Ann. Math. (2) 183, No. 3, 729--786 (2016; Zbl 1342.14055)]. Section 4 gives the proof of Theorem 1.2, by proving the stronger Theorem 4.8. The results needed from group theory are given in Theorem 4.1, and then one counts points on the Hurwitz scheme, as constructed by Ellenberg et al. [loc. cit.]. The final sections give a refinement of the author's conjecture, a further motivation applying the Malle-Bhargava principle, the example for elementary abelian 2-groups and two tables for the values of the invariants for small groups.