Abstract
The limits of the nuclear landscape are determined by nuclear bindingenergies. Beyond the proton drip lines, where the separation energy becomesnegative, there is not enough binding energy to prevent protons from escapingthe nucleus. Predicting properties of unstable nuclear states in the vastterritory of proton emitters poses an appreciable challenge for nuclear theoryas it often involves far extrapolations. In addition, significant discrepanciesbetween nuclear models in the proton-rich territory call for quantifiedpredictions. With the help of Bayesian methodology, we mix a family of nuclearmass models corrected with statistical emulators trained on the experimentalmass measurements, in the proton-rich region of the nuclear chart. Separationenergies were computed within nuclear density functional theory using severalSkyrme and Gogny energy density functionals. We also considered masspredictions based on two models used in astrophysical studies. Quantifiedpredictions were obtained for each model using Bayesian Gaussian processestrained on separation-energy residuals and combined via Bayesian modelaveraging. We obtained a good agreement between averaged predictions ofstatistically corrected models and experiment. In particular, we quantifiedmodel results for one- and two-proton separation energies and derivedprobabilities of proton emission. This information enabled us to produce aquantified landscape of proton-rich nuclei. The most promising candidates fortwo-proton decay studies have been identified. The methodology used in thiswork has broad applications to model-based extrapolations of various nuclearobservables. It also provides a reliable uncertainty quantification oftheoretical predictions.