Abstract

We describe the development of a novel emulator for the fission fragment decay code, $\texttt{CGMF}$, which calculates the emission of prompt neutrons and $\gamma$ rays from fission fragments.  The emulator described in this work uses a combination of a noisy emission model and mixture density network to model the neutron and $\gamma$-ray multiplicities and energies.  In this manuscript, we display the power of this type of emulator for not only modeling average prompt fission observables but also correlations between fission fragment initial conditions and these observables, using both neutron-induced and spontaneous fission reactions. We find that neutron ($\gamma$-ray) multiplicities can be emulated to better than 0.5\% (2\%) across the training and testing data sets, and neutron ($\gamma$-ray) energies can be emulated to better than 1\% (10\%).  While there are improvements to be made in the emulation of $\gamma$-ray properties, our emulator provides a three order of magnitude speed up compared to the original $\texttt{CGMF}$ calculations, paving the way for detailed uncertainty quantification studies. We also develop a new technique for learning the nuclear potential from neutron scattering data. Optical model potentials have long been used to model nuclear reactions and thereby interpret scattering data. Common models have a Woods-Saxon form, where the depth, radius, and diffuseness are parametrized as a function of reaction energy, mass, and charge of the target nucleus. Here, we propose an alternate technique replacing the Woods-Saxon form of the optical potential with a neural network. This network is trained on differential elastic cross section data using differentiable programming. On the differential elastic cross section data originally used to construct the Koning-Delaroche global optical model, our approach reduces the average error by more than a factor of two relative to Koning-Delaroche itself.

Publication Date

7-2026

Document Type

Dissertation

Student Type

Graduate

Degree Name

Mathematical Modeling (Ph.D)

Department, Program, or Center

Mathematics and Statistics, School of

College

College of Science

Advisor

Richard O’Shaughnessy

Advisor/Committee Member

Amy Lovell

Advisor/Committee Member

Arvind Mohan

Campus

RIT – Main Campus

Download

Share

COinS