Jet Clustering and Momentum Regression

In high energy particle physics, jets are an important observable for theory testing as they connect the detected final state with the processes at the collision vertex. Due to their versatile employment, jets play a central role in most of the analysis conducted in HEP. Beginning in 2007 with the introduction of the $k_t$ clustering algorithm, significant improvements in the performance and accuracy of jet clustering algorithms for pp collisions where achieved. The aims of this project are two-fold. During the first part I investigated the anti-$k_t$ clustering algorithm, representing the established jet reconstruction scheme used for pp collisions. Specifically, I reproduced the jet clustering done through CMS Software components (CMSSW) using the pyjet library, before attempting the same task using my own Python implementation of the anti-$k_t$ algorithm. In the second part of the project, I introduce a novel jet momentum regression method, which employs correction factors to the $p_t$ of the jet constituents, deduced by a Deep Neural Network. The comparison with an established regression method shows that with the currently trained model, the new method is not competitive.