Speaker
Description
A next-generation neutrinoless double beta decay (0νββ) search in ¹³⁶Xe has the potential to uncover lepton number violation, and determine if neutrinos are their own antiparticle. This rare decay, if discovered, would demonstrate Physics beyond the Standard Model and provide key insights into the evolution of the universe.
Fast machine learning allows real time data processing in hardware, enabling rapid and intelligent triggers for particle physics experiments. At UC San Diego, our group is developing the Ba-tagging technique to increase the sensitivity in future searches in a liquid xenon time projection chamber (TPC) by probing the decay volume for the presence of the ¹³⁶Xe decay daughter, ¹³⁶Ba. We work on a low-latency convolutional neural network capable of identifying candidate 0νββ events using FPGA hardware. By applying to simulated charge and light waveforms, we aim to perform rapid vertex reconstruction, energy estimation, and background discrimination. This enables fast triggering and real time decision making during detector operation.
We evaluated several candidate network architectures for vertex reconstruction accuracy with quantization-aware training and benchmarked their FPGA inference latency across multiple deployment frameworks. The results were used to guide the selection and optimization of the model architecture and deployment approach. A key goal is to achieve inference latency on the order of a few microseconds while maintaining accurate event reconstruction. Preliminary vertex reconstruction results show that the network can recover spatial information from the simulated detector response with millimeter-scale performance in the reconstructed coordinates. The design, implementation, and performance of the fast machine-learning trigger will be presented, together with planned future developments.