Speaker
Description
Recent decades have witnessed exponential growth in both the quality and volume of experimental nuclear data, driven by advancements in detector technologies and accelerator capabilities. Gamma-ray spectroscopy has particularly benefited from these improvements, with large-scale spectrometers such as GRIFFIN and TIGRESS at TRIUMF enabling collection of increasingly complex, high-dimensional datasets containing hundreds of transitions. Level schemes—the excited-state energies and decay pathways of nuclei—are fundamental to nuclear structure research, yet their construction from spectroscopic data remains a months-to-years manual process of visual pattern recognition, coincidence gating, and iterative refinement. This research reformulates level-scheme construction as a constrained inverse problem, taking γ-ray singles spectra and symmetric coincidence matrices as inputs and recovering directed decay networks.
The approach addresses key challenges inherent to real data: the undirected nature of coincidence measurements, irresolvable doublets, missing weak transitions, and detector artifacts. Building on transition-matrix formalism that analytically relates scheme connectivity to measured intensities, we develop a three-stage pipeline: a probabilistic data layer encoding measurement uncertainties, a learned proposal layer that captures structural priors to constrain combinatorial search, and a physics-enforcing inference layer ensuring energy consistency and intensity-flow conservation.