Branched covering

In mathematics, "branched covering" is a term mainly used in algebraic geometry, to describe morphisms "f" from an algebraic variety "V" to another one "W", the two dimensions being the same, and the typical fibre of "f" being of dimension 0.

In that case, there will be an open set "W′" of "W" (for the Zariski topology) that is dense in "W", such that the restriction of "f" to "W′" (from "V′" = "f"⁻¹("W′") to "W′", that is) is "unramified". Depending on the context, we can take this as local homeomorphism for the strong topology, over the complex numbers, or as an étale morphism in general (under some slightly stronger hypotheses, on flatness and separability). "Generically", then, such a morphism resembles a covering space in the topological sense. For example if "V" and "W" are both Riemann surfaces, we require only that "f" is holomorphic and not constant, and then there is a finite set of points "P" of "W", outside of which we do find an honest covering

:"V′" → "W′".

The set of exceptional points on "W" is called the ramification locus (i.e. this is the complement of the largest possible open set "W′"); see ramification. In general monodromy occurs according to the fundamental group of "W" − "W′" acting on the sheets of the covering (this topological picture can be made precise also in the case of a general base field).

Branched coverings are easily constructed as Kummer extensions, i.e. by extracting roots of functions in the function field. The hyperelliptic curves are prototypic examples.

An unramified covering then is the occurrence of an empty ramification locus.