On the real nerve of the moduli space of complex algebraic curves of even genus (Q351753)

The authors consider smooth, irreducible, real, projective, algebraic curves. Such a curve has three topological invariants, the number of connected components \(k\), the algebraic genus \(g\), and the separability character \(\epsilon\). The complexification maps the curves of genus \(g\) into the moduli space \(\mathcal{M}_g\). The real locus, which is the image of the map, \(\mathcal{M}_g^\mathbb{R}\), is covered by the subsets \(\mathcal{M}_g^{k,\epsilon}\) proceeding from curves with given \(g\), \(k\) and \(\epsilon\). The subsets \(\mathcal{M}_g^{k,\epsilon}\) and \(\mathcal{M}_g^{k',\epsilon'}\) overlap if there is a curve with two real forms of types \((k,\epsilon)\) and \((k',\epsilon')\). Using the functorial equivalence between these curves and the connected compact Riemann surfaces, the paper is written under this latter language. Then, a symmetry of a Riemann surface \(X\) is an antiholomorphic involution \(\sigma\), which is separating if \(X\backslash \text{Fix}(\sigma)\) is disconnected, and non-separating otherwise. The set of fixed points of \(\sigma\) consists of \(k\) curves called ovals. Then \((k,\epsilon)\) is called the topological type of \(\sigma\), where \(\epsilon=+1\) or \(-1\) according to \(\sigma\) be separating or not. The goal of the paper is to study the real nerve \(\mathcal{N}(g)\). This is a simplicial complex whose vertices are the topological types \((k,\epsilon)\). A sequence \(((k_0,\epsilon_0),\dots,(k_n,\epsilon_n))\) is an \(n\)-simplex in \(\mathcal{N}(g)\) if there exists a surface having \(n+1\) symmetries of types \((k_0,\epsilon_0),\dots ,(k_n,\epsilon_n)\). Some results are already known for \(\mathcal{N}(g)\), for instance it has \([(3g+4)/2]\) vertices, and it is connected. The main results of the paper under review concern the dimension of \(\mathcal{N}(g)\) for even values of \(g\). They are obtained using the theory of NEC groups, which are discrete co-compact subgroups of the group of all isometries (allowing orientation-reversing) of the hyperbolic plane. First, concerning geometrical dimension, Theorem 3.5 proves that for even \(g \geq 2\), the geometrical dimension of \(\mathcal{N}(g)\) is \(3\). And second, the authors deal with the homological dimension, that is to say, the greatest \(n\) such that \(\text{H}_n(\mathcal{N}(g), \mathbb{Z}) \neq 0\). In Theorem 4.4 they prove that for even \(g \geq 6\), the homological dimnesion of \(\mathcal{N}(g)\) is \(3\). For the remaining even numbers, the authors obtain \(\mathcal{N}(2) = 0\), and they announce that \(\mathcal{N}(4) = 1\). The final Section 5 of the paper is devoted to global geometric properties of \(\mathcal{N}(g)\). Necessary and sufficient conditions for the existence of a 3-simplex in \(\mathcal{N}(g)\) whose vertices come from non-separating commuting symmetries are obtained. Those conditions are determined in terms of the number of ovals of the symmetries. All in all, the paper is appealing and very well-written, and the usage of NEC groups is made in a very transparent way.