ATLAS and CMS Develop New Tracking for HL-LHC Challenges
The ATLAS and CMS experiments are preparing new tracking algorithms to handle the high particle density of the HL-LHC, achieving over 85% efficiency while
The High-Luminosity Large Hadron Collider (HL-LHC) will create a more crowded environment for particle detectors. To reconstruct charged-particle tracks accurately, the ATLAS and CMS collaborations are upgrading their entire tracking systems and developing faster, more efficient algorithms.
Both experiments are moving to fully silicon-based tracking detectors with finer granularity. According to a paper on arXiv, these new systems offer extended acceptance and updated geometries. The primary goal is to simplify the complex pattern recognition tasks required when charged-particle multiplicity and detector occupancy increase substantially.
Algorithmic Upgrades for Efficiency and Speed
The demanding conditions require algorithms that maintain high tracking efficiency without excessive computational cost. ATLAS and CMS are therefore creating new tracking methods for both offline and online reconstruction. These methods exploit parallelized, vectorized, and heterogeneous computing architectures.
ATLAS is transitioning to a reconstruction chain based on the ACTS framework, which uses the Combinatorial Kalman Filter algorithm. Meanwhile, CMS has established a new baseline for its High-Level Trigger. This approach combines the Patatrack and Line Segment Tracking algorithms for initial seed finding, then uses the mkFit algorithm for the actual track building.
Performance Achievements and Extended Reach
The developed strategies are already showing strong results. They yield tracking efficiencies above 85% across the full phase space, even under high pileup conditions where many simultaneous proton collisions occur. A significant achievement is the extended acceptance for difficult-to-detect tracks.
The algorithms can now reconstruct forward tracks and displaced tracks with transverse displacements of up to tens of centimetres. At the same time, these improvements come with a major reduction in the time required for reconstruction, a critical factor for processing the HL-LHC's immense data flow.
The Role of Machine Learning
Machine-learning techniques are being explored to further enhance the tracking process. The arXiv paper states that these methods are being tested at several stages of reconstruction, including the track building phase itself. This represents a continuing avenue for potential improvement beyond the current state-of-the-art performances.
The presented strategies form the foundation for tracking at the HL-LHC. Both collaborations continue to work on prospects for further improvements as the accelerator upgrade approaches.
