The oriented graph of multi-graftings in the Fuchsian case (Q2438266)

Let \(S\) be a closed surface of genus \(\geq 2\); a projective structure on \(S\) is the data of an atlas on \(S\) with values in \(\mathbb{CP}^1\), such that the transition maps between two local charts are projective transformations. Such a structure determines a developing map \(D:\widetilde{S}\to \mathbb{CP}^1\) (which agrees with local charts up to projective transformation) and a holonomy map \(\Gamma_g:=\pi_1(S)\to\mathrm{PGL}_2(\mathbb{C})\) which is injective and with discrete image. A projective structure gives \(S\) the structure of a Riemann surface (as \(S\) is diffeomorphic to \(\rho(\Gamma_g)\backslash D(\widetilde{S})\)) which depends only on the holonomy map. A Riemann surface supports a canonical projective structure (called a Fuchsian structure), but the holonomy does not determine the projective structure: indeed, as shown by \textit{D. A. Hejhal} [Acta Math. 135, 1--55 (1975; Zbl 0333.34002)] and W. Thurston, starting from a Fuchsian structure one can construct a lot of projective structures with the same holonomy by the surgical operation known as (multi-) grafting. It was shown by \textit{W. M. Goldman} [J. Differ. Geom. 25, 297--326 (1987; Zbl 0595.57012)] that in fact all projective structures with a given holonomy can be obtained by a grafting on the Fuchsian structure: i.e., if one views the set of all projective structures on a Riemann surface \(S\) (i.e. with the same holonomy) as a directed graph where there is an edge from one projective structure to another one if the latter can be obtained from the former by a multi-grafting, there is an edge from the Fuchsian structure to any other. In this paper the authors prove that given two arbitrary vertices in this graph there is a directed path of length two from the first to the second (Theorem 1.1). Grafting is defined as follows: one chooses a simple curve in \(S\) which maps to a loxodromic element in the holonomy representation, cuts \(S\) open along it and glues in the cylinder (with a projective structure) determined by the holonomy of the curve (note that if the structure is not Fuchsian, not every curve may allow for this operation). A multi-grafting is just a composition of graftings along a collection of disjoint simple curves (a `multicurve'). The main technical result in the paper is Proposition 3.3, which allows to construct from two homotopically transverse multicurves \(\alpha,\beta\) two other multicurves \(\beta'\) (isotopic to \(\beta\)) and \(\gamma\) such that \(\beta'\) is graftable in the structure grafted from the Fuchsian one along \(\alpha\), and the composite of graftings (from the Fuchsian structure) along \(\alpha\) and then \(\beta\) is equal to the grafting of the Fuchsian structure along \(\gamma\). Moreover this process is inversible in the sense that given \(\alpha\) and \(\gamma\) (satisfying appropriate conditions) one can find \(\beta\) which fits in the setting of the previous sentence (Lemma 4.1). Using this it is then not hard to deduce the main result from Goldman's theorem.