Controlling extrapolations of nuclear properties with feature selection

Abstract

Predictions of nuclear properties far from measured data are inherently imprecise because of uncertainties in our knowledge of nuclear forces and in our treatment of quantum many-body effects in strongly-interacting systems. While the model bias can be directly calculated when experimental data is available, only an estimate can be made in the absence of such measurements. Current approaches to compute the estimated bias quickly lose predictive power when input variables such as proton or neutron number are extrapolated, resulting in uncontrolled uncertainties in applications such as nucleosynthesis simulations. In this letter, we present a novel technique to identify the input variables of machine learning algorithms that can provide robust estimates of model bias. Our process is based on selecting input variables, or features, based on their probability distribution functions across the entire nuclear chart. We illustrate our approach on the problem of quantifying the model bias in nuclear binding energies calculated with Density Functional Theory (DFT). We show that feature selection can systematically improve theoretical predictions without increasing uncertainties.

arXiv commentManuscript: 6 pages. Supplemental material: 3 pages

References in corpus (29)

• Nuclear ground-state masses and deformations: FRDM(2012)
• Quantum Monte Carlo methods for nuclear physics
• Origin of the Heaviest Elements: the Rapid Neutron-Capture Process
• Coupled-cluster computations of atomic nuclei
• Nuclear Energy Density Optimization
• Nuclear effective field theory: status and perspectives
• The In-Medium Similarity Renormalization Group: A Novel Ab Initio Method for Nuclei
• Multi-system Bayesian constraints on the transport coefficients of QCD matter
• Non-Empirical Interactions for the Nuclear Shell Model: An Update
• Nuclear Mass Predictions for the Crustal Composition of Neutron Stars: A Bayesian Neural Network Approach
• Bayesian approach to model-based extrapolation of nuclear observables
• Neutron drip line in the Ca region from Bayesian model averaging
• Predictions of nuclear $\beta$-decay half-lives with machine learning and their impacts on $r$ process
• Bayesian Evaluation of Incomplete Fission Yields
• Theory of Nuclear Fission
• Quantified limits of the nuclear landscape
• Refining mass formulas for astrophysical applications: a Bayesian neural network approach
• Taming nuclear complexity with a committee of multilayer neural networks
• Chiral EFT based nuclear forces: Achievements and challenges
• Axially deformed solution of the Skyrme-Hartree-Fock-Bogolyubov equations using the transformed harmonic oscillator basis (III) hfbtho (v3.00): a new version of the program
• Nuclear Theory and Science of the Facility for Rare Isotope Beams
• Machine learning the deuteron
• Extrapolation of nuclear structure observables with artificial neural networks
• Validating neural-network refinements of nuclear mass models
• Error Analysis in Nuclear Density Functional Theory
• Statistical aspects of nuclear mass models
• Microscopically-based energy density functionals for nuclei using the density matrix expansion: Full optimization and validation
• Trees and Forests in Nuclear Physics
• Extrapolating from neural network models: a cautionary tale

Cited by in corpus (5)

• Impact of ground-state properties and collective excitations on the Skyrme ansatz: a Bayesian study
• Multipole responses in fissioning nuclei and their uncertainties
• Quantifying uncertainty in machine learning on nuclear binding energy
• Bayesian model mixing with multi-reference energy density functional
• Numerical Convergence of Electromagnetic Responses with the Finite-Amplitude Method