Abstract
The chart of the nuclides is limited by particle drip lines beyond whichnuclear stability to proton or neutron emission is lost. Predicting the rangeof particle-bound isotopes poses an appreciable challenge for nuclear theory asit involves extreme extrapolations of nuclear masses beyond the regions whereexperimental information is available. Still, quantified extrapolations arecrucial for a variety of applications, including the modeling of stellarnucleosynthesis. We use microscopic nuclear mass models and Bayesianmethodology to provide quantified predictions of proton and neutron separationenergies as well as Bayesian probabilities of existence throughout the nuclearlandscape all the way to the particle drip lines. We apply nuclear densityfunctional theory with several energy density functionals. To account foruncertainties, Bayesian Gaussian processes are trained on the separation-energyresiduals for each individual model, and the resulting predictions are combinedvia Bayesian model averaging. This framework allows to account for systematicand statistical uncertainties and propagate them to extrapolative predictions.We characterize the drip-line regions where the probability that the nucleus isparticle-bound decreases from $1$ to $0$. In these regions, we providequantified predictions for one- and two-nucleon separation energies. Accordingto our Bayesian model averaging analysis, 7759 nuclei with $Z\leq 119$ have aprobability of existence $\geq 0.5$. The extrapolations obtained in this studywill be put through stringent tests when new experimental information on exoticnuclei becomes available. In this respect, the quantified landscape of nuclearexistence obtained in this study should be viewed as a dynamical predictionthat will be fine-tuned when new experimental information and improved globalmass models become available.