M-PRISM news
Neural Network-Enabled Thermal Feedback Modeling in the Fission Matrix Method
· ANS Annual Conference, Denver
Undergraduate Daris Paine presented this work in the Reactor Physics Method session of the 2026 ANS Annual Conference in Denver. A neural network learns how temperature changes the fission matrix, and its k-eff error falls steadily as the training set grows.
Fig. 4 of the paper: validation error against the number of fission-matrix libraries used for training.
