Simulation truth for optimal CMS pixel tracking

The CMS detector will undergo several key upgrades for its Phase-2 in the upcoming High-Luminosity Large Hadron Collider (HL-LHC), including the replacement of the entire silicon tracking system, which will consist of an inner tracker based on silicon pixel modules and an outer tracker made from silicon modules with strip and macro-pixel sensors. In addition to a new detector, reconstruction algorithms will have to deal with a much denser hadronic environment, which poses major challenges for tracking, especially in the High-Level Trigger (HLT), where the available computational time per event is limited. Therefore, it is inevitable to adapt and retune the current pixel tracking algorithm, which is based on cellular automata, in order to maximize the performance in terms of physics reconstruction while keeping the computational time low. The approach presented in this contribution introduces several new Monte Carlo (MC) truth-based objects, which represent the ideal target connections between reconstructed hits. Adding this level of truth information not only significantly improves the understanding of the algorithm and its validation, but can also directly point out its limitations and bottlenecks in the configured settings. This poster will focus on how MC simulated truth information can be used effectively to optimize existing algorithms by gaining more insight into the individual steps. The improved performance of the CMS HLT pixel tracking at the future HL-LHC will serve as a showcase to demonstrate the potential of this approach.