In recent years, tensions between measurements and Standard Model predictions in the decays of b-hadrons have hinted at the possible violation of lepton universality, specifically in observables in b→sll and b→clυ transitions. Among them is the ratio of branching fractions R(D*)τμ = B→ D*τυ/ B→ D*μυ. I will discuss the first measurement of the ratio R(D*)eμ = B→ D*eυ / B→ D*μυ at the LHCb experiment at CERN, which will lead towards a combined measurement of all three lepton species in the future.
The LHCb experiment is currently being upgraded for Run 3 of the LHC to record more statistics and therefore reduce the uncertainties of observables testing for example lepton flavor universality. After this upgrade, LHCb will run without a hardware level trigger, resulting in the complete detector being read out at the full bunch-crossing rate of 30 MHz and a maximum data rate of 40 Tbit/s. Events of interest are selected with a software-only trigger in two stages. This allows unprecedented flexibility for trigger selections but at the same time poses a significant computing challenge.
In this seminar, I will also present the first complete high throughput trigger implemented entirely on graphics processing units (GPUs) for an HEP experiment. In LHCb’s High Level Trigger 1 (HLT1) charged particle trajectories and decay vertices are reconstructed to select pp-collisions of interest and reduce the event rate by a factor 30-60. The full HLT1 will be processed on about 200 state of the art GPUs from 2022 onwards. I will discuss the software framework, reconstruction algorithms and performance of the GPU HLT1, as well as ongoing developments towards commissioning.