On hyperelliptic modular curves over function fields (Q1024166)

Let \(A={\mathbb F}_q[T]\) be the polynomial ring over the finite field \({\mathbb F}_q\) with \(q\) elements, \(F={\mathbb F}_q(T)\) its quotient field, and \(F_{\infty}={\mathbb F}_q((1/T))\) its completion at the place \(1/T\). Let \(R\) be a finite, non-empty set consisting of an even number of places in \(A\), and \(D\) the quaternion algebra over \(F\) that ramifies exactly at the places from \(R\). Finally, let \(\Lambda\) be a maximal \(A\)-order in \(D\) (known to be unique up to conjugation by a theorem of Eichler) and \(\Gamma=\Lambda^*\) its unit group. Then \[ \Gamma\subset D^*(F)\subset D^*(F_{\infty})\cong \text{GL}_2(F_{\infty}), \] so \(\Gamma\) acts on the Drinfeld upper half-plane over \(F_{\infty}\). The quotient is the rigid-analytic space of a smooth, projective, geometrically irreducible curve \(X^R\) that is already defined over \(F\). These curves are the function field analogues of Shimura curves and can be interpreted as the moduli schemes of \({\mathcal D}\)-elliptic sheaves introduced by \textit{G. Laumon, M. Rapoport} and \textit{U. Stuhler} [Invent. Math. 113, No. 2, 217--338 (1993; Zbl 0809.11032)] The author shows that for fixed \(q\) only finitely many of these curves are hyperelliptic. If \(q\) is odd, he proves more precisely that \(X^R\) is hyperelliptic if and only if \(R\) consists of exactly two places, one linear and one of degree \(2\). The strategy of the proof (and the result) is similar to the one for classical Shimura curves used by \textit{J.-F.~Michon} [Bull. Soc. Math. Fr. 109, 217--225 (1981; Zbl 0505.14024)] and \textit{A.~Ogg}, [Arithmetic and geometry, Pap. dedic. I. R. Shafarevich, Vol. I: Arithmetic, Prog. Math. 35, 277--307 (1983; Zbl 0531.14014)], namely counting rational points on a suitable reduction, a genus formula for \(X^R\) from the author's previous work [Arch. Math. 92 No. 3, 237--250 (2009; Zbl 1194.11067)], and controlling the fixed points of (Atkin-Lehner) involutions of \(X^R\). The latter explains why the explicit determination of all hyperelliptic \(X^R\) in characteristic \(2\) would require extensive extra treatment.