New global statistical models of nuclidic (atomic) masses based on multilayered feedforward networks are developed. One goal of such studies is to determine how well the existing data, and only the data, determines the mapping from the proton and neutron numbers to the mass of the nuclear ground state. Another is to provide reliable predictive models that can be used to forecast mass values away from the valley of stability. Our study focuses mainly on the former goal and achieves substantial improvement over previous neural-network models of the mass table by using improved schemes for coding and training. The results suggest that with further development this approach may provide a valuable complement to conventional global models.

Statistical modeling of data sets by neural-network techniques is offered as an alternative to traditional semiempirical approaches to global modeling of nuclear properties. New results are presented to support the position that such novel techniques can rival conventional theory in predictive power, if not in economy of description. Examples include the statistical inference of atomic masses and β-decay halflives based on the information contained in existing databases. Neural network modeling, as well as other statistical strategies based on new algorithms for artificial intelligence, may prove to be a useful asset in the further exploration of nuclear phenomena far from stability.

No abstract

A neural-network model is developed to reproduce the differences between experimental nuclear mass-excess values and the theoretical values given by the Finite Range Droplet Model. The results point to the existence of subtle regularities of nuclear structure not yet contained in the best microscopic/phenomenological models of atomic masses. Combining the FRDM and the neural-network model, we create a hybrid model with improved predictive performance on nuclear-mass systematics and related quantities.