On the profinite regular inverse Galois problem (Q1011986)

The present paper is devoted to the regular inverse Galois problem for profinite groups. Let \(k\) be a field and let \(\overline k\) be a fixed separable closure of \(k\). A \(k\)-curve means a smooth, projective, geometrically connected \(k\)-scheme of dimension \(1\). Given a \(k\)-curve \(X\), a \(k-G\)-extension of \(k(X)\) is a Galois extension \(E/k(X)\) regular over \(k\) (i.e. such that \(k\) is algebraically closed in \(E\)). Then, given a (pro)finite group \(G\), the regular inverse Galois problem for \(G\) over \(X\) asks whether or not there exists a \(k\)-\(G\)-extension of \(E/k(X)\) with group \(G\). A weaker version of this problem can be stated for fields of moduli: whether or not there exists a \(k\)-curve \(X_G\) and a \(k-G\)-extension \(E/\overline k(X_G)\) with group \(G\) and field of moduli \(k\). In the paper under review the author analyzes the constraints imposed by the existence of an abelian \(G\)-cover \(f : Y \to X\) defined over \(k\) on \(X\) and the ramification divisor \(t\) of \(f\); these are called by the author the abelian constraints for \(X \setminus t\). The first type is the etale constraint which is related to a certain Jacobian variety of \(X\) over \(k\). The second type is the ramification constraint which is related to the existence of enough roots of unity in a certain residue field at a point of \(Y\). The author's main result is that these constraints produce an obstruction to the weak regular inverse Galois problem for a whole class of profinite groups, called \(p\)-obstructed, when \(k\) is a finitely generated field of characteristic \(\neq p\). More precisely, if a profinite group \(G\) has an open subgroup \(U\subset G\) with an epimorphism \(U\twoheadrightarrow\mathbb Z_p\), then the weak regular inverse Galois problem for fields of moduli has a negative answer.