Local, global, divergent - which gauge symmetries are redundant?

Abstract

A local gauge symmetry can not possibly be a mere redundancy of the description, provided that 1. charge is nonzero. 2. we also consider divergent gauge transformations, whose brackets with local transformations contain global charge operators. If these conditions hold, unitarity requires the existence of gauge anomalies. We describe the relevant multi-dimensional generalization of affine Kac-Moody algebras, explain why it only arises if the observer's trajectory is explicitly introduced, and contrast it to the Mickelsson-Faddeev algebra which pertains to chiral-fermion type anomalies.