SINAPSE neutron-gamma discrimination

Context This dataset contains signals collected with the VENDETA detection array at theLos Alamos National Laboratory (LANL) facility, from the spontaneous fission of $^{240}\text{Pu}$. It is intended to support the development and benchmarking ofneutron-gamma discrimination techniques, with a particular focus on thelow charge (low light output) regime. Experimental details are describedin Syrett et al. (2025). Dataset description The dataset is provided as a single HDF5 file: SINAPSE-neutron_gamma.hdf5.The file contains 4 subsets, spanning different light output and $\beta$ ranges: Subset Light Ouput Range [keVee] beta range description train 200 - 510 0.0 - 0.6 training set val 100 - 200 0.0 - 0.6 validation set test 0 - 200 0.0 - 0.6 test set prompt_gamma 0 - 100 0.8 - 1.2 pure gamma ray set Each subset exposes 3 fields: Field Type Description waveform array of float32 Digitized pulse waveforms, z-score normalized light_output_MeVee float32 Light output in MeVee label int16 Label: 0: $\gamma$, 1: neutron, -1: unidentified Example usage import h5py with h5py.File("SINAPSE-neutron_gamma.hdf5", "r") as file: train_samples = file["dataset/train/samples"][:] train_dataset = { "waveform": train_samples["waveform"], "light_output_MeVee": train_samples["light_output_MeVee"], "label": train_samples["label"], }