The closure of Teichmüller space (Q1174673)

An interesting problem in Teichmüller theory is to study the closure of various Teichmüller spaces (roughly spaces of conformal structure of various objects, such as Riemann surfaces, modulo natural equivalences). The reasons are manifold and well known. In this paper the author offers a new characterization of the closure of Bers' model of Teichmüller spaces in terms of a class of homeomorphisms of planar simply connected domains (the approach also leads to new models for other Teichmüller spaces). This is important as only Bers' model provides the universal Teichmüller space \({\mathcal T}\). The author is motivated by \textit{K. Astala} and \textit{F. W. Gehring's} characterisation of the closure of Becker space in the Bloch space [J. Anal. Math. 46, 16-57 (1986; Zbl 0628.30026)]. They used a class of homeomorphisms which are, in a sense, locally almost Euclidean isometries. The author studies this more general problem using a class of homomorphisms which are, in a sense, locally almost Möbius transformations. More precisely, a homeomorphism \(f: \Omega\to \Omega'\) is \(\delta\)-almost-Möbius if for any disk \(D\), with \(2D\subset\Omega\), and any four distinct points \(z_ 1,\dots,z_ 4\) in \(D\), the distance between cross ratios in the hyperbolic metric \(\tau\) of \(\mathbb{C}\setminus\{0,1\}\) is less than \(\delta\), that is, \(\tau(| z_ 1,z_ 2,z_ 3,z_ 4|,| f(z_ 1),f(z_ 2),f(z_ 3),f(z_ 4)|)<\delta\), where \(| x,y,z,w|\) is the (any) cross ratio. The main theorem then asserts that if \(f\) is analytic and univalent on the unit disk \(\mathbb{D}\) and if \(\varphi\) is the Schwarzian derivative of \(f\), then \(\varphi\) is in \(\text{cl}({\mathcal T})\) if and only if for every \(\delta>0\) there is a \(\delta\)-almost-Möbius mapping of \(f(\mathbb{D})\) onto a quasidisk. This is quite natural since, if \(\varphi\) were in \({\mathcal T}\), \(f(\mathbb{D})\) would be a quasidisk. The result asserts that, after a small perturbation by a nice map, this is the case. The author then uses this result and the methods developed to characterize the isolated points of \({\mathcal T}\) [the existence of which was established by \textit{W. P. Thurston}, in The Bieberbach conjecture (West Lafayette. Ind., 1985), 185-197, Am.Math. Soc., Providence, RI (1986)]. Roughly, \(\varphi\) is not isolated if for every \(\delta\) there is a \(\delta\)-almost-Möbius map of \(f(\mathbb{D})\) which does not have a quasiconformal extension to the plane (recall that the area of \(\mathbb{C}\setminus f(\mathbb{D})\) must be zero for isolation).