Research

We study nuclear reaction processes using a quantum three-body model, treating the projectile as a weakly bound two-body system. Using the Ichimura-Austern-Vincent (IAV) model for inclusive breakup reactions, we have studied fusion and incomplete fusion in reactions induced by deuterons and ⁶Li.

We develop and apply the CDCC method for scattering observables of weakly-bound projectiles. Our code STARS implements the Lagrange-mesh R-matrix method with GPU acceleration. Our reduced-basis emulator achieves 260x speedup, enabling full Bayesian inference of 18-parameter optical potentials.

We pioneer modern ML techniques for nuclear scattering: PINNs with exterior complex scaling for quantum scattering, and bidirectional liquid neural networks (BiCfC) as fast, differentiable surrogates for the radial Schrödinger equation. The BiCfC network enables Fisher information analysis, sensitivity mapping, and gradient-based inverse scattering.

We are developing the Neural Cluster Model (NCM), a variational wave function framework bridging neural network quantum states and nuclear cluster models. The NCM introduces learnable cluster latent variables, allowing cluster structure to emerge from the variational principle. Benchmarks on ⁴He and ⁸Be show excellent agreement with experiment.

We investigate four-body breakup processes. Given the challenges in handling three charged particles within the Faddeev framework, we perform four-body breakup calculations for systems with four charged particles using a DWBA framework, treating the projectile as having a three-body structure.

From two-body alpha decay to three-body two-proton decay and five-body models, we study resonance decay processes. Our inclusive breakup work with the IAV model provides a promising pathway to fully describe surrogate reactions for nuclear astrophysics and stockpile stewardship.