Quantum Circuits Achieve 97.1% Pulse Shape Discrimination for Germanium Detectors with 10 Qubits
Summarize this article with:
Pulse shape discrimination is essential for eliminating unwanted background noise in the search for rare events like neutrinoless double-beta decay and dark matter, particularly when using sensitive germanium detectors. Fabrizio Napolitano from Università di Perugia and INFN, along with colleagues, now presents a novel approach to this challenge, applying quantum machine learning to real experimental data for the first time.
The team develops a hybrid quantum-classical method using variational circuits, which successfully analyses germanium detector waveforms with significantly fewer parameters than existing techniques. This innovative system achieves a remarkably high level of accuracy, matching the performance of current state-of-the-art methods while reducing model complexity by over two orders of magnitude, and opens the possibility of streamlined, efficient data processing in future experiments.
Quantum Machine Learning for Neutrino Detection Scientists are pioneering the application of quantum machine learning (QML) to improve event selection in germanium detectors, a critical step in the VIP experiment’s search for rare phenomena. Researchers explored Variational Quantum Circuits (VQC) as classifiers, demonstrating a complete workflow from data preparation to model evaluation and highlighting the potential of QML for this challenging physics application.
The team utilized a labeled dataset of waveforms from Broad Energy germanium detectors, containing examples of both signal and background events, preparing the data for analysis by extracting relevant features. Performance was evaluated using metrics such as accuracy, precision, and area under the ROC curve, allowing for comparison with classical machine learning algorithms.
Results demonstrate the potential benefits of QML, and the team analyzed the scalability of the approach, assessing its performance with increasing data size and model complexity. This work contributes to the growing field of quantum computing and its application to fundamental physics research.
Quantum Machine Learning for Pulse Shape Discrimination Scientists developed a novel quantum-classical pipeline for pulse shape discrimination, a technique used to reject background noise in rare-event physics experiments utilizing Broad Energy Germanium (BEGe) detectors.
This research pioneers the application of quantum machine learning to real experimental waveforms, moving beyond traditional deep learning approaches. Researchers harnessed a Variational Quantum Circuit (VQC) to directly process 1024-sample waveforms, compressing the input feature space logarithmically by mapping them into a 10-qubit Hilbert space. To establish a performance benchmark, the team referenced a state-of-the-art classical pipeline employing a Denoising Autoencoder (DAE) and Convolutional Neural Network (CNN). In contrast, the quantum pipeline minimizes pre-processing, applying only baseline subtraction and normalization, eliminating the need for a dedicated denoising step. The VQC’s architecture, containing only 302 trainable parameters, represents a significant reduction in complexity compared to the classical CNN. This compact model achieves a remarkable area under the curve of 0. 98 and a global accuracy of 97. 1%, demonstrating that even with current technology, quantum algorithms can match the performance of established classical baselines. This achievement paves the way for future detectors where quantum processing units directly analyze incoming signals, exploiting the exponentially large Hilbert space for enhanced sensitivity.
Quantum Machine Learning Boosts Rare Event Detection Scientists achieved exceptional background rejection in the search for rare events using a novel quantum machine learning approach.
The team developed a 10-qubit Variational Circuit (VQC) that processes germanium detector pulse waveforms directly, achieving an Area Under the Curve (AUC) of 0. 98 in classifying events. This result demonstrates performance comparable to state-of-the-art classical algorithms while dramatically reducing model complexity, with the VQC utilizing only 302 trainable parameters. Experiments revealed a global accuracy of 97. 1% when evaluating the VQC on a test set of 11,377 waveforms. Crucially, the model maintained a high signal efficiency of 98. 7%, correctly identifying the majority of true events, while simultaneously achieving 87. 7% background rejection. The progressive training strategy employed by the scientists demonstrated a clear correlation between the addition of quantum layers and improvements in model performance. Shallow circuits quickly reached approximately 94% accuracy, capturing the gross features of the pulse. As the circuit depth increased, accuracy steadily improved, reaching the final value of 97. 1%, indicating that deeper layers allowed the model to resolve subtle features within the waveforms. Benchmarking against a classical algorithm revealed a model compression factor of approximately 160×, demonstrating the exceptional parametric efficiency of the VQC. The research establishes a pathway towards compact, efficient signal processing for future physics experiments.
Quantum Machine Learning Improves Pulse Discrimination Accuracy This work demonstrates a significant advancement in pulse shape discrimination, a crucial technique for background rejection in rare-event searches using germanium detectors. Researchers achieved high accuracy, a receiver operating characteristic area under the curve of 0. 98 and a global accuracy of 97. 1%, by applying quantum machine learning to experimental pulse waveforms. Notably, this performance matches that of established classical deep learning models, but with a dramatically simplified model requiring only 302 trainable parameters. This achievement bypasses the need for computationally intensive pre-processing steps typically required to manage noise in these systems.
The team successfully addressed the challenge of training deep quantum circuits through a progressive layer-wise training strategy, achieving high signal acceptance, 98. 7%, essential for sensitive rare-event detection. While acknowledging this study represents a proof-of-principle, the results suggest the potential for future detector readout systems where quantum sensors directly interface with quantum processing units. This could enable ultra-low latency, low-power, and high-fidelity event classification for next-generation nuclear and particle physics experiments. Future research will explore extending this approach to reconstruct interaction position and gamma directionality, leveraging subtle correlations often discarded by conventional filters. 👉 More information 🗞 Pulse Shape Discrimination for Germanium Detectors using Variational Quantum Circuits 🧠 ArXiv: https://arxiv.org/abs/2512.08603 Tags: