New Algorithm Boosts Sensitivity for Detecting Rare Particle Interactions

Scientists are continually refining methods for handling the vast datasets generated by modern particle physics experiments. P. Abratenko and D. Andrade Aldana, working with colleagues from the MicroBooNE Collaboration at Argonne National Laboratory and the University of California, Santa Barbara, demonstrate a novel online data selection algorithm for liquid argon time projection chambers (LArTPCs). This research details a fast technique to identify electron signals from cosmic ray muons within the MicroBooNE detector, utilising only ionization charge information. Representing the first demonstration of online data selection in a LArTPC using real data and charge information exclusively, this work is significant as it offers a crucial proof-of-principle for future large-scale experiments like the Short-Baseline Near Detector and the Deep Underground Neutrino Experiment (DUNE), where efficient data handling will be paramount.

These detectors, crucial for observing particle interactions in detail, produce vast quantities of information requiring efficient processing. The MicroBooNE collaboration successfully demonstrated an online data selection algorithm, capable of filtering data in real-time directly within the detector system. This breakthrough addresses a critical challenge in modern particle physics: managing escalating data rates from increasingly sophisticated experiments. The development centres on identifying electrons originating from cosmic ray muons, high-energy particles from space, within the LArTPC. By focusing on specific signals, the algorithm isolates relevant events as they occur, rather than storing and processing everything. DUNE, designed to unravel the mysteries of neutrinos, will generate even larger datasets, demanding more sophisticated data handling techniques. The MicroBooNE algorithm provides a crucial stepping stone towards achieving the necessary efficiency and sensitivity for these ambitious projects, reducing the volume of data needing storage and analysis, and allowing researchers to focus on the most promising signals. The successful implementation within MicroBooNE’s continuously running supernova data stream demonstrates the algorithm’s practicality and robustness. The system efficiently processes data as it is acquired, enabling immediate identification of interesting events, particularly important for transient phenomena like supernova bursts where timely analysis is essential. The team’s approach streamlines data processing and paves the way for incorporating advanced techniques like artificial intelligence and machine learning to further refine the selection process and enhance the detection of rare signals. MicroBooNE is a liquid argon time projection chamber (LArTPC), a technology creating three-dimensional images of particle interactions by collecting ionization charge from anode wire plane arrays and scintillation light using a dedicated light detection system. The experiment, situated on-surface at Fermilab, collected data via two streams: a beam-based neutrino stream and a continuously running supernova stream, commissioned in November 2017, demanding near 100% data livetime for detecting rare, randomly occurring off-beam physics processes like supernova bursts or proton decay. The detector houses an active mass of 85 metric tons of liquid argon within a 10.4m long, 2.6m wide, and 2.3m high volume. Charged particles traversing the argon create ionization electrons that drift towards three anode wire planes under a uniform electric field, with a maximum drift time of 2.3 milliseconds. These planes, two induction planes (U and V) each containing 2400 wires, and one collection plane (Y) with 3456 wires, all maintain a 3mm wire pitch to precisely detect the ionization trails. The on-surface location exposes the detector to a significant cosmic ray flux of approximately 4kHz, necessitating robust data processing techniques. To demonstrate the algorithm’s efficacy, researchers focused on identifying electrons originating from stopping cosmic ray muons, providing a clear signature within the LArTPC data for validation. The methodology prioritizes utilising ionization charge information exclusively, streamlining the process and reducing computational demands, representing a departure from traditional methods relying on combined charge and light information. The development and validation of the data selection algorithm involved a multi-stage process, beginning with detailed simulations and progressing to analysis of real data collected by the MicroBooNE detector, ensuring accuracy and reliability before implementation in a live data stream. Analysis of cosmic ray muon interactions revealed that approximately 12% of these muons decay within the detector’s active volume, with 8% undergoing nuclear capture and the remaining 80% exiting the active region. These findings are crucial for understanding background processes and refining signal identification techniques. Detailed comparisons between MicroBooNE data and Monte Carlo simulations show strong agreement in trigger primitive distributions, validating the algorithm’s performance. Specifically, the data and simulation exhibit comparable shapes for maximum amplitude of ADC waveforms, integrated ADC charge, and waveform width. Observed discrepancies, such as a broader spread in maximum amplitude in data, are attributed to known detector effects and are being addressed through ongoing calibration procedures. The simulation utilised 25,000 readout windows, corresponding to an exposure of 57 seconds, while the data sample comprised 87,936 drift regions spanning 202 seconds of exposure. Further refinement involved excluding short, incomplete waveforms containing fewer than 20 ADC words, mitigating the effects of buffer overflows, consistent with the detector’s electronics shaping time. The study also examined the fraction of true stopping muons producing Michel electrons, providing a valuable calibration source for low-energy electron identification, with energies below 50 MeV, relevant for future neutrino experiments. Scientists working with the MicroBooNE detector have achieved a crucial advance in handling the deluge of data produced by modern particle physics experiments. Their success in demonstrating real-time data selection within a liquid argon time projection chamber (LArTPC) makes the next generation of experiments genuinely feasible. For years, the ambition to build detectors capable of precisely mapping particle interactions has been hampered by the sheer volume of information generated. The challenge isn’t merely storing the data, but sifting through it quickly enough to identify the rare, fleeting signals revealing new physics. Traditional methods involve processing everything after it’s been recorded, a slow and resource-intensive process. This new algorithm, however, filters data on the fly, discarding uninteresting events before they even reach storage, potentially reducing the load on computing systems and accelerating the pace of discovery. This breakthrough is particularly vital for ambitious projects like DUNE, which will require unprecedented data handling capabilities. While this work focuses on identifying electron signals from cosmic rays, the underlying principle, intelligent, real-time filtering, is broadly applicable, with future developments likely seeing more sophisticated algorithms, perhaps leveraging machine learning, to identify a wider range of signals and further refine the selection process. The field now faces the task of scaling these techniques to even larger detectors and more complex data streams, demanding continued innovation in both hardware and software.