Nuclear mass predictions with anisotropic kernel ridge regression

TL;DR

This paper introduces an anisotropic kernel ridge regression method for nuclear mass predictions, improving accuracy by capturing isotope and isotone correlations without adding new parameters.

Contribution

The paper develops AKRR with a novel anisotropic kernel, enhancing nuclear mass prediction accuracy over traditional KRR without extra input features.

Findings

AKRR outperforms KRR in both interpolation and extrapolation tasks.

The anisotropic kernel reveals correlation patterns among isotopes and isotones.

Significant accuracy improvements are demonstrated in nuclear mass predictions.

Abstract

The anisotropic kernel ridge regression (AKRR) approach in nuclear mass predictions is developed by introducing the anisotropic kernel function into the kernel ridge regression (KRR) approach, without introducing new weight parameter or input in the training. A combination of double two-dimensional Gaussian kernel function is adopted, and the corresponding hyperparameters are optimized carefully by cross-validations. The anisotropic kernel shows cross-shape pattern, which highlights the correlations among the isotopes with the same proton number, and that among the isotones with the same neutron number. Significant improvements are achieved by the AKRR approach in both the interpolation and the extrapolation predictions of nuclear masses comparing with the original KRR approach.