Algorithms@UCL

The UCL group, enhanced by activities in the Centre for Data Intensive Science and Industry, is heavily involved in the underlying algorithms that reconstruct the physic objects created in the detector, in particular:

• Tracks: The reconstruction of charged particle trajectories (tracks). The UCL group specializes on tracks in jets and developing machine learning based track reconstruction.
• Jets: The reconstruction of very boosted (high momentum) streams of particles, or jets. When very boosted particles decay, their decay products will typically overlap creating a very high energy jet. The UCL group pioneered the study of the structure of such large jets, which is vital for studying highly boosted final states, and introduced both physics-inspired and transformer-based machine learning models to identify them in ATLAS collisions. We also developed novel machine-learning-based calibration methods and particle-flow reconstruction.
• Flavour Tagging: Many physics processes of interest contains b-quarks in their decays. These manifest themselves as jets in the detector. However, due to the relatively long-lifetime of the b-quark they will typically travel several millimetres before decaying, leaving a distinct signature that we can reconstruct using advanced algorithmic and machine learning techniques. The UCL group transformed jet flavour tagging with transformer-based machine learning models and pioneers cutting-edge methods.

UCL is a leader in all these areas, which are vital to fully exploiting the data produced by the LHC in our quest to answer fundamental questions about the Universe. See the individual webpages linked above for further details on each of the topics.

For more information, contact Jon Butterworth, Mario Campanelli, Gabriel Facini, Andreas Korn, or Tim Scanlon.