Periodic Reporting for period 4 - RECEPT (Real-time precision tests of lepton universality)
Reporting period: 2021-12-01 to 2023-08-31
The scientific objectives of RECEPT are to find particles or forces beyond the SM which can give a hint of this more fundamental theory. Even if no new particles or forces are found, the RECEPT measurements will narrow down the range of possible theories beyond the SM. In either case, humanity will gain a more complete understanding of the world around us, and of the relationship between the smallest and largest building blocks of our reality. In order to enable these fundamental advances, RECEPT researchers will have to develop new and uniquely efficient ways of processing their data in real time. This work goes beyond state-of-the-art in both research and industry, and the new algorithms developed by the RECEPT team may therefore have practical benefits for real-time data processing in society.
One of the SM's most precise predictions is called “lepton universality” : that the six leptons couple with equal strength to the force carriers (photon and the W/Z bosons) of the electroweak force. RECEPT uses the LHCb experiment, based at the Large Hadron Collider (LHC) at CERN, to make the world’s most precise tests of lepton universality by studying the decays of SM particles called beauty mesons. By measuring the rate at which these mesons decay into final states containing electrons or muons, RECEPT researchers are able to indirectly search for the presence of particles and forces beyond the SM, which may break this universality.
RECEPT’s researchers also play a crucial role in the upgrade of LHCb, which will increase the data volume 100 times, allowing much more precise SM tests and searches for new particles and forces. In order to take full advantage of this, LHCb will have to process around 30 million collisions per second, corresponding to around 4 Terabytes of data per second, not only tagging individual LHC collisions as "interesting" but finding the trajectories of particles produced in such collisions and inferring their fundamental physical properties in real time. Such a data volume is equivalent to over one percent of today's global internet traffic, and must be processed in a data centre located close to LHCb, using only around 3000 computer servers. The RECEPT team has developed high-performance algorithms for both CPU and GPU computing architectures which have been shown to be able to meet this challenge. A particular highlight has been the development of "Allen", a complete framework for high-throughput processing on GPUs, which was adopted by the LHCb collaboration in 2020 and put into production by the RECEPT team and our international collaborators in time for the restart of LHCb datataking in 2022. During 2022 and 2023 RECEPT researchers were able to show that the first-level real-time processing of the LHCb experiment, now implemented using Allen on around 400 GPU cards, worked as planned. This is illustrated by the figure showing pairs of particles containing strange quarks, called K-short mesons, reconstructed in real time by the Allen GPU system in 2022 LHCb datataking.
The method used to measure the efficiency with which single electrons can be reconstructed directly in LHCb data had previously been applied to other charged particles, but this is the first time it has been shown to work for electrons. The better than percent level precision shows that the measurements planned by RECEPT over the course of the project will not be limited by our knowledge of electron reconstruction. The RECEPT measurement of electron-muon universality in strange decays of beauty mesons is the first time that multiple lepton universality parameters of this kind were measured coherently in the same analysis, and the most accurate and most sensitive analysis of its kind to date.
The contributions of the RECEPT team to the real-time reconstruction of the upgraded LHCb detector have involved multiple novel algorithms and techniques which go significantly beyond the state of the art. The charged particle reconstruction algorithms developed by RECEPT researchers have both significantly better physics performance and are significantly faster than state of the art algorithms. Allen is the first feature-complete high-throughput GPU trigger in high-energy physics. LHCb's use of Allen is the first time that a major HEP experiment has based its first-level real-time processing entirely on GPU processors, the first time that GPU processors have been used by a HEP experiment in a high-throughput environment, and the highest rate reconstruction of charged hadron trajectories ever achieved by a HEP collider experiment.