A Bayesian neural network (BNN) framework is established to probe the nuclear symmetry energy coefficient $C_{\mathrm{sym}}$ through isoscaling analysis of spallation fragments produced in proton-induced reactions on Ca, Ni, and Xe isotopic targets. The reliability of BNN predictions is preliminarily validated against available experimental data and empirical systematics. Systematic analysis reveals that the fragments follow the isoscaling law, and a strong linear correlation is observed between the isoscaling parameter $\alpha$ and the source asymmetry difference $\mathrm{\Delta }\left[{(Z_s/A_s)}^2\right]$ across different systems and fragment groups, supporting statistical equilibrium during fragment formation. Based on this correlation, $C_{\mathrm{sym}}$ is extracted for fragments with charge numbers $Z=4–26$ at specified temperatures. The results indicate a pronounced sensitivity of $C_{\mathrm{sym}}$ to fragment selection, temperature, and reaction system, while also revealing the method's limitations with heavier targets such as Xe. This BNN-based isoscaling methodology offers a viable pathway to constrain the nuclear symmetry energy. Its integration with future high-precision experiments and canonical ensemble treatments will further refine constrains on the nuclear equation of state.

Close

PrevNext

View figure in article

• Low & intermediate energy heavy-ion reactions
• Nuclear fragmentation
• Symmetry energy
• Bayesian networks
• Artificial neural networks
• Bayesian methods