Online Electron Reconstruction at CLAS12

TL;DR

The paper tackles real-time electron reconstruction at CLAS12 to enhance online monitoring and triggering. It introduces a two-stage ML pipeline: first, fast online track-to-calorimeter association via multilayer perceptrons, and second, an electron classifier that fuses signals from the DC, HTCC, and ECAL using 51 input features across multiple layers, achieving efficiency near 100% and purity above 75% at data rates that meet the DAQ. Validation against offline electron identification shows that the online approach can provide reconstructed electrons for immediate analysis and can serve as a data-reduction filter, with retraining strategies further improving performance in challenging regions. The method has implications for triggerless streaming readout and online event selection, and the authors discuss integration, monitoring, continual learning, and potential extensions to hadron identification for broader applicability in current and future experiments such as ePIC, SoLID, and MOLLER.

Abstract

Online reconstruction is key for monitoring purposes and real time analysis in High Energy and Nuclear Physics experiments. A necessary component of reconstruction algorithms is particle identification that combines information left by a particle passing through several detector components to identify the particle's type. Of particular interest to electroproduction Nuclear Physics experiments such as CLAS12 is electron identification which is used to trigger data recording. A machine learning approach was developed for CLAS12 to reconstruct and identify electrons by combining raw signals at the data acquisition level from several detector components. This approach achieves an electron identification purity above 75% whilst retaining an efficiency close to 100%. The machine learning tools are capable of running at high rates exceeding the data acquisition rates and will allow electron reconstruction in real time. This work enhances online analyses and monitoring and can contribute to improved triggering at CLAS12. This machine learning driven approach will also be crucial for experiments aiming to transition to streaming readout operations where online reconstruction will be a key component of the data taking paradigm.

Figures (19)

• Figure 1: The CLAS12 Detector, with highlighted detector subsystems. Image taken from C12Overview.
• Figure 2: A schematic of the CLAS12 FD drift chambers (left) and one of its superlayers (right). Images taken from C12DC
• Figure 3: The number of photo-electrons recorded in the HTCC as a function of momentum for positively charged particles. Images taken from C12DC
• Figure 4: Schematic representation of the calorimeters' three views. Image taken from C12ECAL.
• Figure 5: The electron sampling fraction as a function of momentum. Images taken from C12ECAL