Robust Neutron Lifetime Extraction in UCNτ Measurements using Deep Learning

The neutron lifetime τ_n measurement is a critical parameter used in many physics and cosmology fields. The precise measurement is critical for enhancing and refining various physics models including but not limited to the standard model and the models of the primordial universe, as well as for testing fundamental physics theories and quantum chromodynamic predictions. We present a deep learning approach to extracting the neutron lifetime from the UCNτ experiment performed at the Los Alamos National Laboratory. Specifically, we use a variational autoencoder (VAE) to learn a latent data representation to capture the complex, high dimensional patterns in the measurement data and to extract the lifetime τ_n. We compare the VAE results with those obtained by other analyzers using traditional methods and demonstrate its robustness to systematic uncertainties.