Transcendence degree one function fields over a finite field with many automorphisms (Q1703596)

The study of the \(\mathbb{K}\)-automorphisms group \(\operatorname{Aut}_{\mathbb{K}}(\mathbb{L})\), where \(\mathbb{L}:=\mathbb{K}(\mathcal{X})\) is the algebraic function field of a smooth projective curve \(\mathcal{X}\) of genus \(g(\mathbb{L})\geq2\) defined over an algebraically closed field \(\mathbb{K}\) of characteristic \(p\geq0\), is a long standing problem in algebraic geometry. When it comes to a bound on \(|\operatorname{Aut}_{\mathbb{K}}(\mathbb{L})|\), one has the most famous universal bound, the so-called \textit{Hurwitz bound}; \(|\operatorname{Aut}_{\mathbb{K}}(\mathbb{L})|\leq84(g(\mathbb{L})-1)\), given by \textit{A. Hurwitz} [Math. Ann. 41, 403--442 (1893; JFM 24.0380.02)] for zero characteristic and by \textit{P. Roquette} [Math. Z. 117, 157--163 (1970; Zbl 0194.35302)] for positive characteristic \(p>g(\mathbb{L})+1\) with only one exception, namely the hyperelliptic curve \(Y^2Z^{p-2}=X^p-XZ^{p-1},\) which has \(p=2g(\mathbb{L})+1\) and \(2p(p^2-1)\) automorphisms. \textit{K. Oikawa} [Kōdai Math. Semin. Rep. 8, 23--30 (1956; Zbl 0072.07702)] and \textit{T. Arakawa} [Osaka J. Math. 37, No. 4, 823--846 (2000; Zbl 0981.30029)] gave even better upper bounds when the automorphism group fixes (not necessarily pointwise) finite subsets of points on the curve. On the other hand, if \(p\mid |\operatorname{Aut}_{\mathbb{K}}(\mathbb{L})|\) then a much larger automorphism group than Hurwitz bound could happen. For example, a Hermitian curve \(\mathcal{X}(q)\), which is a smooth plane curve of the form \(Y^qZ+YZ^{q}-X^{q+1}=0\), for some \(q=p^{h}\geq3\) has genus \(\frac{1}{2}(q^2-q)\) and \(q^3(q^3+1)(q^2-1)\) automorphisms. Moreover, \textit{H. Stichtenoth} [Arch. Math. 24, 615--631 (1973; Zbl 0282.14007)] showed that a smooth curve \(\mathcal{X}\) satisfying \(|\operatorname{Aut}(\mathbb{L})|\geq16g(\mathbb{L})^4\) is birational equivalent to a Hermitian curve \(\mathcal{X}(q)\). An improvement to \(3(2g(\mathbb{L})^2+g(\mathbb{L}))(3+\sqrt{8g(\mathbb{L})+1})\) was due to \textit{N. Anbar} et al. [J. Pure Appl. Algebra 217, No. 7, 1224--1236 (2013; Zbl 1274.14033)] for the case of smooth plane curves. Now, assume that \(p\) is odd and that \(\mathbb{L}\) is ordinary, i.e. \(g(\mathbb{L})\) is equal to its \(p\)-rank \(\gamma(\mathbb{L})\). In 2004, Guralnick-Zieve announced (without a proof) a sharper bound, namely \(|\operatorname{Aut}_{\mathbb{K}}(\mathbb{L})|\leq c_pg(\mathbb{L})^{8/5}\) for some constant \(c_p\) depending on \(p\). \textit{G. Korchmáros} and \textit{M. Montanucci} [``The geometry of the Artin-Schreier-Mumford curves over an algebraically closed field'', Preprint, \url{arXiv:1612.05912}] proved that \(|G|\leq 68\sqrt{2}g(\mathbb{L})^{3/2}\) for any \(G\), a solvable subgroup of \(\operatorname{Aut}_{\mathbb{K}}(\mathbb{L})\). Accordingly, G. Korchmáros-M. Montanucci-P. Speziali were motivated to study further ordinary function fields \(\mathbb{L}\), for which there exists a solvable subgroup of automorphisms \(G\) whose order is as large as \(g(\mathbb{L})^{3/2}\) up to a positive constant. The article under review considers function fields of the shape \(F_{m,q}=\mathbb{K}(x,y)\) with \(y^q+y=x^m+x^{-m}\), where \(\mathbb{K}\) is the algebraic closure of a finite field \(\mathbb{F}_q\) of odd order \(q=p^h\) and \(m\) is a positive integer coprime to \(p\). Proposition 3.1 gives an explicit representation of the Galois closure \(K_{m,q}\) of \(F_{m,q}\mid \mathbb{K}(x^{m(q-1)})\) as \(\mathbb{K}(x,y,z)\) with \(y^q+y=x^m+x^{-m}\) and \(z^q+z=x^m\). After, one obtains by Lemma 4.1 and Proposition 4.2 that \(K_{m,q}\) has genus \(g(K_{m,q})=(q-1)(qm-1)\), \(p\)-rank \(\gamma(K_{m,q})=(q-1)^2\) (in particular, \(K_{m,q}\) is ordinary if and only if \(m=1\)) and the group \(\operatorname{Aut}_{\mathbb{K}}(K_{m,q})\) has order \(\geq q^2m(q-1)\). Theorem \(5.1\) assures that this is almost the full \(\mathbb{K}\)-automorphisms group; more precisely, \(\operatorname{Aut}_{\mathbb{K}}(K_{m,q})=\Delta\rtimes (C_{m(q-1)}\rtimes\langle\xi\rangle)\), where \(\Delta\) is an elementary abelian normal subgroup of order \(q^2\), \(C_{m(q-1)}\) is a cyclic subgroup of order \(m(q-1)\) and \(\xi\) is an involution (fields of definition for \(\Delta\) and \(D:=C_{m(q-1)}\rtimes\langle\xi\rangle\) are also determined). Thus, \(K_{m,q}\) is hitting the bound \(68\sqrt{2}g(K_{m,q})^{3/2}\) up to a constant, since \(\sqrt{m}|\operatorname{Aut}_{\mathbb{K}}(K_{m,q})|>g(K_{m,q})^{3/2}\) and \(|\operatorname{Aut}_{\mathbb{K}}(K_{m,q})|>g(K_{m,q})^{3/2}\) when \(K_{m,q}\) is ordinary. Finally, they investigate, in section \S 6, when some Galois subcovers of \(K_{m,q}\mid \mathbb{K}(x^{m(q-1)})\) are of the same type, that is, are isomorphic to \(K_{\overline{m},\overline{q}}\mid \mathbb{K}(x^{\overline{m}(\overline{q}-1)})\) for some \(\overline{q}=p^{\overline{h}}\) and a positive integer \(\overline{m}\) coprime to \(p\). Sufficient conditions on \(\overline{q}\) and \(\overline{m}\) ensuring that the Galois closure \(K_{\overline{m},\overline{q}}\) of \(F_{\overline{m},\overline{q}}\mid \mathbb{K}(x^{\overline{m}(\overline{q}-1)})\) be isomorphic to a subfield of \(K_{m,q}^H\) for some \(H\leq\operatorname{Aut}_{\mathbb{K}}(K_{m,q})\), are provided. This in turns shows an example of function fields, different from the well-known Hermitian function field, with large automorphism groups, which also contain some subfields that still have many automorphisms.