Machine learning in nuclear physics at low and intermediate energies

Machine learning(ML)is becoming a new paradigm for scientific research in various research fields due to its exciting and powerful capability of modeling tools used for big-data processing tasks.In this review,we first briefly introduce the different methodologies used in ML algorithms and techniques.As a snapshot of many applications by ML,some selected applications are presented,especially for low-and intermediate-energy nuclear physics,which include topics on theoretical applications in nuclear structure,nuclear reactions,properties of nuclear matter,and experimental applications in event identification/reconstruction,complex system control,and firmware performance.Finally,we present a summary and outlook on the possible directions of ML use in low-intermediate energy nuclear physics and possible improvements in ML algorithms.

Influence of nuclear physics inputs and astrophysical conditions on r-process

The rapid neutron-capture process(r-process) is one of the main mechanisms to explain the origin of heavy elements in the universe. Although the past decades have seen great progress in understanding this process, the related nuclear physics inputs to r-process models include significant uncertainty. In this study, ten nuclear mass models, including macroscopic, macroscopicmicroscopic, and microscopic models, are used to calculate the β-decay rates and neutron-capture rates of the neutron-rich isotopes for the r-process simulations occurring in three classes of astrophysical conditions. The final r-process abundances include uncertainties introduced by the nuclear mass model mainly through the variation of neutron-capture rates, whereas the uncertainties of β-decay rates make a relatively small contribution. The uncertainties in different astrophysical scenarios are also investigated,and are found to be connected to the diverse groups of nuclei produced during nucleosynthesis.

High precision nuclear mass predictions towards a hundred kilo-electron-volt accuracy

Mass is a fundamental property and an important fingerprint of atomic nucleus.It provides an extremely useful test ground for nuclear models and is crucial to understand energy generation in stars as well as the heavy elements synthesized in stellar explosions.Nuclear physicists have been attempting at developing a precise,reliable,and predictive nuclear model that is suitable for the whole nuclear chart,while this still remains a great challenge even in recent days.Here we employ the Fourier spectral analysis to examine the deviations of nuclear mass predictions to the experimental data and to present a novel way for accurate nuclear mass predictions.In this analysis,we map the mass deviations from the space of nucleon number to its conjugate space of frequency,and are able to pin down the main contributions to the model deficiencies.By using the radial basis function approach we can further isolate and quantify the sources.Taking a pedagogical mass model as an example,we examine explicitly the correlation between nuclear effective interactions and the distributions of mass deviations in the frequency domain.The method presented in this work,therefore,opens up a new way for improving the nuclear mass predictions towards a hundred kilo-electron-volt accuracy,which is argued to be the chaos-related limit for the nuclear mass predictions.

Magnetic moment predictions of odd-A nuclei with the Bayesian neural network approach

The Bayesian neural network approach has been employed to improve the nuclear magnetic moment predictions of odd-A nuclei.The Schmidt magnetic moment obtained from the extreme single-particle shell model makes large root-mean-square(rms)deviations from data,i.e.,0.949μN and 1.272μN for odd-neutron nuclei and odd-proton nuclei,respectively.By including the dependence of the nuclear spin and Schmidt magnetic moment,the machine-learning approach precisely describes the magnetic moments of odd-A uclei with rms deviations of 0.036μN for odd-neutron nuclei and 0.061μN for odd-proton nuclei.Furthermore,the evolution of magnetic moments along isotopic chains,including the staggering and sudden jump trend,which are difficult to describe using nuclear models,have been well reproduced by the Bayesian neural network(BNN)approach.The magnetic moments of doubly closed-shell±1 nuclei,for example,isoscalar and isovector magnetic moments,have been well studied and compared with the corresponding non-relativistic and relativistic calculations.