Field of moduli versus field of definition for cyclic covers of the projective line (Q988070)

Let \(k\) be a perfect field and \(\overline{k}\) an algebraic closure. If \(X\) is a \(\overline{k}\)-curve (proper, smooth, connected), then \(K\subset \overline{k}\) is a \textit{field of definition} for \(X\) if there exists a \(K\)-curve \(X'\) and an isomorphism \(X'\times_K\overline{k}\cong X\). Set \(H:=\{\sigma\in\text{Gal}(\overline{k}|k) \;| \;X\cong_{\overline{k}}{}^\sigma\!X\}.\) Then the \textit{field of moduli} of \(X\) (relative to \(\overline{k}|k\)) is the fixed field \(\overline{k}^H\). If the field of moduli is a field of definition for \(X\), then it is the unique minimal field of definition containing \(k\). However, \(X\) need not be defined over its field of moduli if \(g(X)\geq 2\). The paper under review studies curves \(X\) that are \(q\)-cyclic covers of the projective line for \(q\) prime, and asks when such an \(X\) can be defined over its field of moduli. This generalizes and draws on the work of \textit{B. Huggins} [Math. Res. Lett. 14, No. 2, 249--262 (2007; Zbl 1126.14036)] which handles the hyperelliptic case \(q=2\). Let \(X\rightarrow \mathbb{P}^1_{\overline{k}}\) be a \(q\)-cyclic cover, with Galois group \(C_q\). When \(C_q\) is normal in \(\text{Aut}_{\overline{k}}(X)\), the strategy is to look at the reduced automorphism group \(\text{Aut}_{\overline{k}}(X)/C_q\), which is a finite subgroup of \(PGL(2, \overline{k})\). In the Artin-Schreier case (\(q=p\)), the Galois group \(C_q\) is normal in \(\text{Aut}_{\overline{k}}(X)\) except in three exceptional cases dealt with in Theorem 1.1(2). In the Kummer case (\(q\neq p\)), in order to ensure the normality of \(C_q\), the author restricts to \textit{strongly branched} covers, i.e. those with at least \(2q+1\) branch points. Using the classification of finite subgroups of \(PGL(2,\overline{k})\) as in Huggins' paper cited above, the author shows that \(X\) is defined over its field of moduli if the reduced automorphism group is not cyclic of order prime to \(p\) (Theorem 1.1(1)). A key ingredient in the argument is Proposition 4.1, which asserts that under certain conditions the rational subfield \(\overline{k}(x)\subset \overline{k}(X)\) corresponding to the \(q\)-cyclic cover \(X\rightarrow \mathbb{P}^1_{\overline{k}}\) is the unique rational subfield \(\overline{k}(t)\subset \overline{k}(X)\) with the property that \(\overline{k}(X)|\overline{k}(t)\) is a \(q\)-cyclic Galois extension. Unfortunately, the proof of this result in the strongly branched Kummer case is incomplete -- the argument doesn't reach a contradiction as announced. However, the proof can be completed as follows: using the notation of Proposition 4.1, a simple argument shows that the \(2q\) places \(Q_{i,\nu}\) are in fact the \textit{only} ramified places in the extension \(F|F_0\). But this contradicts the assumption that \(F|F_0\) is strongly branched. Part (3) of Theorem 1.1 addresses the case of Artin-Schreier covers with reduced automorphism group cyclic of order prime to \(p\), stating that under certain group-theoretic assumptions, \(X\) can be defined over its field of moduli. The paper concludes with an example of a \(q\)-cyclic cover of \(\mathbb{P}^1_{\mathbb{C}}\) with field of moduli \(\mathbb{R}\) but not defined over \(\mathbb{R}\).