Rave is a toolkit that takes a set of reconstructed particle tracks as its
input and produces reconstructed interaction vertices as its output. It deals
both with finding (pattern recognition) and with fitting (statistical
estimation) of the interaction vertices.  Its main design goals are ease of
use, high integratability in existing software projects, extensibility, and
generality.  To this end the API is defined in an as simple as possible way.
The various algorithms, and optionally, their parameter settings, are
referenced by a simple string. This guarantees that the user code decouples
completely from the toolkit internals. Also, maintaining backwards
compatibility should become a trivial task.  The Rave Toolkit is complemented
by a simple standalone framework, called "Vertigo". Implementation,
verification, and performance analysis of reconstruction algorithms should thus
be possible in a very fast and straightforward manner.  Rave has its roots in
the CMS vertex reconstruction community. The current algorithmic parts of the
toolkit are source code compatible with the original CMS software, but
contributions from "outside" are highly welcome.