ATLAS Experiment, Energy Calibration, Gradient Boosting, Decision Trees, Online Filtering
Abstract
In experimental high-energy physics, dealing with a large volume of data is essential to produce relevant information, as a significant portion of the data comes from background noise, complicating the characterization of phenomena of interest. A complex sequential online event selection process, known as trigger, is crucial to address this challenge. In the ATLAS experiment at the Large Hadron Collider (LHC), the trigger system operates in two sequential stages: first-level and high-level. In the case of electrons, pivotal as messengers of new physics, the trigger system heavily relies on calorimeters, which measure the energy of incident particles. This work proposes an energy calibration method based on gradient-boosted decision trees (GBDT) to enhance the accuracy of energy estimation in the ATLAS high-level trigger. This method can reduce computational requirements and increase efficiency in selecting particles, including electrons, even in scenarios with pileup presence.
