Classical and Bayesian error analysis of the relativistic mean-field model for doubly magic nuclei

Abstract

The information-geometric statistical analysis on the stability of model reductions, reported previously [Imbri\v{s}ak and Nomura, Phys. Rev. C 107, 034304 (2023)] with a focus on the manifold boundary approximation method in the application to the nuclear density-dependent point-coupling model of infinite nuclear matter, is extended to the numerically more challenging case of finite nuclei. A simple procedure is presented for determining the binding energies of doubly magic nuclei within the relativistic mean-field framework using the Woods-Saxon potential. The proposed procedure, employing the Fisher information matrix combined with algorithmic differentiation, is shown to provide reliable estimates of parameter uncertainties of the nuclear energy density functional for finite nuclei, while reducing the time-consuming sampling of the parameter space, which would be required in the numerically more involved Bayesian statistical techniques.