Links of surface singularities and CR space forms (Q580554): Difference between revisions

The authors define a ``CR space form'' to be a compact, non-degenerate CR manifold which is locally homogeneous, i.e., whose universal cover is homogeneous under CR automorphisms. In this paper, they give a classification of such CR space forms in dimension 3 where ``non- degenerate'' means ``strictly pseudoconvex''. They then apply this information to classify normal surface singularities (X,x) which have a locally homogeneous link \((=smooth\) boundary of a neighbourhood of x). These links are CR space forms, and most types appear this way. The CR space forms naturally split into two sorts: those which are locally isomorphic to \(S^ 3\subset {\mathbb{C}}^ 2\), called spherical, and those which are not, called aspherical. The classification of spherical CR space forms is contained in the paper of \textit{D. Burns} jun. and \textit{S. Shnider}, Invent. Math. 33, 223-246 (1976; Zbl 0357.32012). For the aspherical case, the following theorem of \textit{E. Cartan} [``Sur la géométrie pseudoconforme des hypersurfaces de deux variables complexes. I.'' Oeuvres II.2., 1231-1304 (1954; Zbl 0058.083)] can be applied: A homogeneous 3-dimensional aspherical CR manifold has a 3- dimensional CR automorphism group. Thus, classification of aspherical 3- dimensional CR space forms reduces to determining left invariant strictly pseudoconvex CR structures on 3-dimensional Lie groups, which the authors carry out.