On the conformal order representation of Riemannian surfaces and a characterization of separating families of curves (Q851111)

Let \(C\) be a real, projective, smooth and geometrically irreducible algebraic curve. Let \(\sigma:C(\mathbb C)\to C(\mathbb C)\) be the involution induced by complex conjugation on the set of \(\mathbb C\)-rational points of \(C\), and let \(C(\mathbb R)\) be its fixed point set. The curve \(C\) is said to be separating if the difference \(C(\mathbb C)\setminus C(\mathbb R)\) is disconnected. If we interpret \(\sigma\) as an antianalytic involution on the compact Riemann surface \(C(\mathbb C)\) this is equivalent to say that the quotient \(C(\mathbb C)/\langle\sigma\rangle\) is an orientable topological surface [see e.g. \textit{E. Bujalance} and \textit{D. Singerman}, Proc. Lond. Math. Soc. (3) 51, 501--519 (1985; Zbl 0545.30032)]. The author presents an elementary but deep proof of the following theorem, due to \textit{L. V. Ahlfors} [Comment. Math. Helv. 24, 100--134 (1950; Zbl 0041.41102)]. Theorem. The curve \(C\) is separating if and only if there exists an \(\mathbb R\)-saturated morphism \(f:C\to \mathbb P^1\), that is, a morphism satisfying that \(f^{-1}(p)\subset C(\mathbb R)\) for every \(p\in \mathbb P^1(\mathbb R)\). Let \(g\) be the genus of \(C\) and let \(r\) be the number of connected components of \(C(\mathbb R)\). It is easily seen that \(r\leq \deg(f)\), and Ahlfors proved that \(\deg (f)\leq g+1\). The article under review, which is extremely well written and contains many enlightening drawings and examples, presents two main contributions: (i) The author provides a new proof of Ahlfors's result, whose main ingredient is Abel's Theorem, which reduces the argument to prove that fixed a basis \(\{\gamma_1, \ldots, \gamma_g\}\) of the homology \(H_1(T^g)\) of the torus \(T^g=(\mathbb R/\mathbb Z)^g\), each point \(p\in T^g\) can be written as a sum \(p=p_1+\cdots+p_g\) where each \(p_j\in \gamma_j\). (2) Secondly, the author improves dramatically the upper bound for \(\deg(f)\), by getting \[ \deg(f)\leq \frac{r+g+1}{2}. \] Notice that both Ahlfors and Gabard upper bounds just coincide whenever \(C\) is an \(M\)-curve, that is a curve whose real part \(C(\mathbb R)\) attains its maximum number \(g+1\) of connected components. The question of sharpness of the bound \(\deg(f)\leq \frac{r+g+1}{2}\) remains open.