Optimisation of the software-based muon identification at the LHCb experiment

The LHCb experiment performs precision tests of the standard model of particle physics. A significant part of the physics programme of the experiment is based on final states containing muons. In this work, the performance of the LHCb experiment's muon identification in Run~I is examined. Using simulated $B^0\to K^{*0}\mu^+\mu^-$ candidates, the relative inefficiency of the first stage of the software trigger with respect to the offline reconstruction is evaluated. For single muon tracks, an inefficiency of $(9.4 \pm 0.4)\,\%$ which stems only from the reconstruction software is found. Thanks to an upgraded computing farm, more CPU time can be spent on reconstruction. Through removal of simplifications and other optimisations, the software-based inefficiency is reduced by $(8.43 \pm 0.30)\,\%$ for Run~II. This means, that $(89.7 \pm 5.0)\,\%$ of the software-based inefficiency are removed. The underlying software is modularised such that in the online farm the identical algorithm from offline processing can be used. The software developed in this thesis is used in all future processing and triggering aiming to identify muon. As a side effect, the offline reconstruction is now roughly 2.5 times faster.