Automation of one-loop corrections for multi-particle processes

With the advent of the Large Hadron Collider, we are in a new era in Particle Physics, in which unprecedented energy scales can be probed. Although it is a discovery machine, it has already been shown to be able to produce experimental precisions at the percent level, and so our theoretical calculations must match that, which requires (at least) calculations to next-to-leading order (NLO). In this thesis, we explain and develop new techniques for the evaluation of one-loop integrals, which have historically been the bottleneck in NLO calculations. After introducing Quantum Field Theory and NLO calculations, we explain the process of tensor reduction and the golem95 method for avoiding its numerical instabilities. We follow this by discussing the techniques used to improve the stability of a library of scalar integrals (for two- and three-point integrals), and then we discuss the extension of the golem95 library to include complex internal masses, along with the reasons for doing so. We then bring together the GoSam project with the event generator Sherpa, in order to calculate the process pp -> e+e−μ+μ− by diboson production to NLO, including the (formally higher order) loop-induced process with gluons in the initial state.