String theory, with its richness of dynamical scenarios, represents a microscopic tool of utmost relevance to jointly describe all the fundamental forces of nature and simultaneously it involves a wide and fascinating spectrum of geometrical implications. Because of its nature, in many cases research in string theory cannot be carried through on a purely locally-based analysis, but it should often be supplemented with intrinsically global investigations and consistency checks. The mathematical apparatus needed for such a study, however, turns out to be usually much more sophisticated than the one which is su cient for eld theory-like computations, and, unfortunately, it is not always really familiar to the physics community. The arguments discussed in this thesis constitute manifest examples of this situation...
