Abstract: In cancer patients undergoing proton therapy, a very intense secondary radiation is produced during the treatment, which lasts around one minute. About one billion prompt gamma-rays are emitted per second, and their detection with fast scintillation detectors is useful for monitoring a correct beam delivery. To cope with the expected count rate and pile-up, as well as the scarce statistics due to the short treatment duration, we developed an eidetic data acquisition system capable of continuously digitizing the detector signal with a high sampling rate and without any dead time. By streaming the fully unprocessed waveforms to the computer, complex pile-up decomposition algorithms can be applied and optimized offline. We describe the data acquisition architecture and the multiple experimental tests designed to verify the sustained data throughput speed and the absence of dead time. While the system is tailored for the proton therapy environment, the methodology can be deployed in any other field requiring the recording of raw waveforms at high sampling rates with zero dead time.

Abstract: The dark matter sector remains completely unknown. It is therefore crucial to keep an open mind regarding its nature and possible interactions. Focusing on the case of Weakly Interacting Massive Particles, in this work we make this general philosophy more concrete by applying modern machine learning techniques to dark matter direct detection. We do this by encoding and decoding the graphical representation of background events in the XENONnT experiment with a convolutional variational autoencoder. We describe a methodology that utilizes the `anomaly score' derived from the reconstruction loss of the convolutional variational autoencoder as well as a pre-trained standard convolutional neural network, in a semi-supervised fashion. Indeed, we observe that optimum results are obtained only when both unsupervised and supervised anomaly scores are considered together. A data set that has a higher proportion of anomaly score is deemed anomalous and deserves further investigation. Contrary to classical analyses, in principle all information about the events is used, preventing unnecessary information loss. Lastly, we demonstrate the reach of learning-focused anomaly detection in this context by comparing results with classical inference, observing that, if tuned properly, these techniques have the potential to outperform likelihood-based methods.