Modern nuclear data evaluation increasingly requires not only accurate scattering calculations but also efficient methods for uncertainty quantification and parameter optimization, tasks that benefit from differentiable solvers amenable to gradient-based algorithms. I present a neural network emulator based on bidirectional liquid neural networks that provides a differentiable mapping from the optical potential to scattering wave functions. The key innovation enabling generalization across the parameter space is the use of phase-space coordinates $\rho =kr$ that normalize the oscillation wavelength regardless of projectile energy, allowing a single network to span 1 to 200 MeV. Trained on Numerov solutions for 12 target nuclei (${}^{12}\mathrm{C}$ to ${}^{208}\mathrm{Pb}$), both protons and neutrons, and partial waves up to $l=30$, the network achieves an overall relative error of 0.6%. The predicted wave functions yield accurate $S$-matrix elements and elastic scattering cross sections, reproducing diffraction patterns spanning four orders of magnitude. Importantly, the model generalizes successfully to nuclei not included in training (${}^{24}\mathrm{Mg}, {}^{63}\mathrm{Cu}$, and ${}^{184}\mathrm{W}$) with comparable accuracy, demonstrating that it has captured the smooth functional dependence encoded in the Koning-Delaroche (KD02) parametrization rather than memorizing specific targets. The differentiable nature of the trained model is the key prerequisite for gradient-based optimization of optical model parameters and uncertainty quantification, which I identify as the principal direction for future work.

Close

PrevNext

View figure in article

• Direct reactions
• Elastic scattering reactions
• Nuclear reactions
• Artificial neural networks
• Machine learning models
• Optical, coupled-channel & distorted wave models