Uncertainties with low-resolution nuclear forces

Abstract

Low-resolution nuclear Hamiltonians, obtained from chiral effective field theory (EFT) and softened using renormalization group techniques, have been very successful in nuclear structure theory. The associated EFT truncation uncertainty for these potentials is difficult to quantify. We use singular value decompositions of low-resolution nuclear forces to obtain an operator basis to study Hamiltonian uncertainties for these potentials. We perform Bayesian inference for the singular values and three-body low-energy constants, the free parameters of nuclear Hamiltonians in our framework, using likelihoods based on nucleon-nucleon phase shifts and triton observables to account for the EFT truncation uncertainties in these quantities. Validating our inference, we find good reproduction of input uncertainties for low-energy phase shifts and three-nucleon observables. On the other hand, uncertainties for higher-energy phase shifts are systematically underestimated, which we attribute to limitations of the singular value decomposition and neglected correlations between phase shifts at different energies. We propagate the resulting distribution of Hamiltonians forward to predictions for ground-state properties of $^{24,28}$O and $^{48}$Ca, comparing against other state-of-the-art nuclear structure predictions. Our approach makes it possible to account for EFT uncertainties when using low-resolution potentials, which is important for many ongoing studies in exotic nuclei.