Hyperbolic lengths and conformal embeddings of Riemann surfaces (Q1976608)

Let \(R\) and \(R'\) denote noncompact Riemann surfaces and assume there is some homeomorphism \(h\) of \(R\) onto \(R'\). Then, in general, \(h\) will not respect the conformal structure of the surfaces, i.e. it is nowhere expected to be holomorphic. The author asks for conditions which allow to call such a mapping \(h\) to be ``good'' in the sense of function theory; this will be the case if \(h\) is homotopic to some holomorphic injection \(f:R\to R'\) (of course it would be too much restrictive to require \(f\) as a biholomorphic function). For hyperbolic surfaces \(R\), \(R'\) we can consider the hyperbolic length of a closed curve \(\gamma\) on \(R\) and also the infimum \(l_R(c)\) of these numbers where the curves run in the free homotopy class \(c\) of \(\gamma\) (the curvature of the hyperbolic metric should be normalized). Now, if \(h\) is a ``good' homeomorphism, then we obtain the Schwarz-Pick-type inequality \(l_{R'} (h^*(c))\leq l_R(c)\), where \(h^*(c)\) is the image of \(c\) under the induced map \(h^*\) of the homotopy classes. One main result of the article is that the converse is not true in general: a homeomorphisms which fulfills this arclength-inequality is not necessarily a ``good'' one. But, if we replace the hyperbolic length in the definition above by the length, then at least in special cases (if \(R\), \(R'\) are one-punctured tori) the inequality is sufficient for \(h\) to be a ``good'' mapping. The techniques of the proofs are essentially based on results of \textit{M. Shiba} [Holomorphic functions and moduli I, Publ. Math. Sci. Res. Inst. Berkeley 10, 237-246 (1986; Zbl 0653.30028)]. An interesting walk along the border between geometry and function theory.