Refining mass formulas for astrophysical applications: a Bayesian neural network approach

Abstract

Exotic nuclei, particularly those near the driplines, are at the core of one of the fundamental questions driving nuclear structure and astrophysics today: what are the limits of nuclear binding? Exotic nuclei play a critical role in both informing theoretical models as well as in our understanding of the origin of the heavy elements. Our purpose is to refine existing mass models through the training of an artificial neural network that will mitigate the large model discrepancies far away from stability. The basic paradigm of our two-pronged approach is an existing mass model that captures as much as possible of the underlying physics followed by the implementation of a Bayesian Neural Network (BNN) refinement to account for the missing physics. Bayesian inference is employed to determine the parameters of the neural network so that model predictions may be accompanied by theoretical uncertainties. Despite the undeniable quality of the mass models adopted in this work, we observe a significant improvement (of about 40%) after the BNN refinement is implemented. Indeed, in the specific case of the Duflo-Zuker mass formula, we find that the rms deviation relative to experiment is reduced from rms =0.503MeV to rms=0.286 MeV. These newly refined mass tables are used to map the neutron drip lines (or rather "drip bands") and to study a few critical r-process nuclei. The BNN approach is highly successful in refining the predictions of existing mass models. In particular, the large discrepancy displayed by the original "bare" models in regions where experimental data is unavailable is considerably quenched after the BNN refinement. This lends credence to our approach and has motivated us to publish refined mass tables that we trust will be helpful for future astrophysical applications.

References in corpus (5)

• Further explorations of Skyrme-Hartree-Fock-Bogoliubov mass formulas. XII: Stiffness and stability of neutron-star matter
• Nuclear charge radii: Density functional theory meets Bayesian neural networks
• Impact of the symmetry energy on the outer crust of non-accreting neutron stars
• Decoding Beta-Decay Systematics: A Global Statistical Model for Beta^- Halflives
• Application Of Support Vector Machines To Global Prediction Of Nuclear Properties

Cited by in corpus (42)

• Machine Learning in Nuclear Physics
• Bayesian approach to model-based extrapolation of nuclear observables
• r-Process Nucleosynthesis: Connecting Rare-Isotope Beam Facilities with the Cosmos
• Neutron drip line in the Ca region from Bayesian model averaging
• Surveying the reach and maturity of machine learning and artificial intelligence in astronomy
• Machine learning the nuclear mass
• Bayesian Evaluation of Incomplete Fission Yields
• Quantified limits of the nuclear landscape
• Nuclear Physics of the Outer Layers of Accreting Neutron Stars
• Taming nuclear complexity with a committee of multilayer neural networks
• Impact of the neutron star crust on the tidal polarizability
• Deep learning: Extrapolation tool for ab initio nuclear theory
• Nuclear masses in extended kernel ridge regression with odd-even effects
• Multi-task learning on nuclear masses and separation energies with the kernel ridge regression
• Beyond the proton drip line: Bayesian analysis of proton-emitting nuclei
• Extrapolation of nuclear structure observables with artificial neural networks
• Validating neural-network refinements of nuclear mass models
• Statistical aspects of nuclear mass models
• Nuclear mass predictions using machine learning models
• Nuclear binding energy predictions using neural networks: Application of the multilayer perceptron
• Electroweak probes of ground state densities
• Impact of statistical uncertainties on the composition of the outer crust of a neutron star
• Trees and Forests in Nuclear Physics
• Markov Chain Monte Carlo Predictions of Neutron-rich Lanthanide Properties as a Probe of $r$-process Dynamics
• Extrapolating from neural network models: a cautionary tale
• Nuclear mass predictions based on convolutional neural network
• Skyrme-Hartree-Fock-Bogoliubov mass models on a 3D mesh. IIb. Fission properties of BSkG2
• Neutron-rich rare isotope production with stable and radioactive beams in the mass range A=40-60 at beam energy around 15 MeV/nucleon
• Application of multilayer perceptron with data augmentation in nuclear physics
• Building Surrogate Models of Nuclear Density Functional Theory with Gaussian Processesand Autoencoders
• Controlling extrapolations of nuclear properties with feature selection
• Calibration of Energy Density Functionals with Deformed Nuclei
• Emergence of low-energy monopole strength in the neutron-rich calcium isotopes
• Statistical learnability of nuclear masses
• Nuclear binding energy predictions based on BP neural network
• Nuclear mass predictions based on deep neural network and finite-range droplet model (2012)
• Shell quenching in nuclear charge radii based on Monte Carlo dropout Bayesian neural network
• Report from the A.I. For Nuclear Physics Workshop
• Bayesian model mixing with multi-reference energy density functional
• FRIB and the GW170817 Kilonova
• Impact of nuclear masses on r-process nucleosynthesis: bulk properties versus shell effects
• Impact of error analysis on the composition the outer crust of a neutron star