Abstract
The locations of proton-proton collision points in LHC experiments are called primary vertices (PVs). Preliminary results of a hybrid deep learning algorithm for identifying and locating these, targeting the Run 3 incarnation of LHCb, have been described at conferences in 2019 and 2020. In the past year we have made significant progress in a variety of related areas. Using two newer Kernel Density Estimators (KDEs) as input feature sets improves the fidelity of the models, as does using full LHCb simulation rather than the “toy Monte Carlo” originally (and still) used to develop models. We have also built a deep learning model to calculate the KDEs from track information. Connecting a tracks-to-KDE model to a KDE-to-hists model used to find PVs provides a proof-of-concept that a single deep learning model can use track information to find PVs with high efficiency and high fidelity. We have studied a variety of models systematically to understand how variations in their architectures affect performance. While the studies reported here are specific to the LHCb geometry and operating conditions, the results suggest that the same approach could be used by the ATLAS and CMS experiments.
Highlights
The LHCb experiment is currently being upgraded for the planned start of Run 3 of the LHC in 2022
Since we presented results at Connecting-the-Dots 2020, we have tested AllCNN Kernel Density Estimators (KDEs)-tohists models using full LHCb Monte Carlo rather than toy Monte Carlo and find they are (i) more performant without re-training and (ii) even more performant with re-training
If we replace the original KDE with a pair of KDEs, each built from individual track contributions with no pairwise interactions used explicitly, the performance is better again
Summary
The LHCb experiment is currently being upgraded for the planned start of Run 3 of the LHC in 2022. 1 m 390 mrad 70 mrad 15 mrad 66 mm interaction region showing 2xσbeam = ~12.6 cm Figure 1 This diagram illustrates the luminous region of the LHCb experiment and the Vertex Locator (VELO), the high precision silicon pixel detector that surrounds it. As discussed below, using track parameters produced by a proposed LHCb Run 3 Vertex Locator (VELO) tracking algorithm [8] leads to significantly better performance. Modified versions of our original CNN architecture [6] were studied to understand how performance depends on the number of model parameters, the number of network layers, the number of channels per layer, whether batch normalization is used, and the extent to which skip connections are used We refer to these models as members of the AllCNN family. A one-dimensional version of the U-Net architecture [9], originally developed for two-dimensional biomedical images, was tested using toy Monte Carlo and find that it (slightly) outperforms the best of our AllCNN models when both are trained with the same KDE
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have