Approximating the nuclear binding energy using analytic continued fractions

Area of Science:

• Nuclear Physics
• Computational Physics
• Data Science

Background:

• Understanding nuclear behavior is crucial in nuclear physics.
• Nuclear binding energy (B(A, Z)) is a key property for characterizing nuclides.
• Current data-driven methods have limitations in approximating nuclear binding energy.

Purpose of the Study:

• To introduce a novel data-driven approach, Continued Fraction Regression (cf-r), for analyzing nuclear binding energy.
• To accurately approximate the binding energy of stable and unstable nuclides.
• To assess the model's predictive accuracy and extrapolation capabilities.

Main Methods:

• Utilized Continued Fraction Regression (cf-r) with a tailored loss function.
• Employed analytic continued fractions for approximation.
• Validated the model on experimentally confirmed stable and unstable nuclides.

Main Results:

• Achieved precise predictions for nuclides with residuals smaller than 0.15 MeV.
• Demonstrated accurate approximation of both stable and experimentally confirmed unstable nuclides.
• Showcased robust extrapolation capabilities converging at the nuclear mass limit.

Conclusions:

• Continued Fraction Regression (cf-r) offers a powerful data-driven approach for nuclear binding energy analysis.
• The method provides valuable insights into the limitations of current state-of-the-art techniques.
• The approach has potential applications beyond nuclear physics.