A tensor decomposition model for evaluating isotopic yield in neutron-induced fission

TL;DR

This paper introduces a tensor decomposition model (FYTD) for predicting isotopic yields in neutron-induced fission, demonstrating improved accuracy over existing models and capturing multi-dimensional dependencies in fission yield data.

Contribution

The FYTD model applies tensor decomposition to predict fission yields, including isotopic and mass distributions, with validation against experimental and database data, showing superior performance.

Findings

FYTD model predicts 235U and 239Pu fission yields accurately.

Predictions for isotopic yields agree within 1 order of magnitude for 98% cases.

Model captures multi-dimensional dependencies and extends to different neutron energies.

Abstract

After constructing yield tensors with three dimensions for 851 fission products and filling the tensors with the independent yield data from the ENDF/B-VIII.0 database, the tensor decomposition algorithm is applied to predict the independent isotopic yield in fission, which resulting in the Fission Yield Tensor Decomposition (FYTD) model. The fission yields of 235U and 239Pu are set as missing values and then predicted. The predictions for235U fissions by the FYTD model agree with the ENDF/B-VIII.0 data and are better than those by the Talys and BNN+Talys models. Furthermore, we predict not only the mass distribution but also the isotopic yields in the fissions. For fast neutron-induced fission of 239Pu, 98% predictions of the isotopic yields by the FYTD model agree with the ENDF/B-VIII.0 data within 1 order of magnitude. The fission yields of 238Np, 243Am, and 236Np that do not exist in the ENDF/B-VIII.0 database are predicted and compared with those in the JEFF-3.3 database, as well as the experimental data. Good agreement demonstrates the predictive ability of the FYTD model for the target nucleus dependence. The scalability of the yield tensor decomposition model over the incident neutron energy degrees of freedom is examined. After adding a set of 2 MeV neutron-induced 239Pu fission yield data into the yield tensor, the 2 MeV neutron-induced fission yields of 235U and 239Pu are predicted. The comparison with the experimental data shows that the predictions are similar to those by the GEF model in the peak area but more accurate in the valley area. Finally, the yields of the ratio of isomeric states and neutron excess of the products as a function of product charge number are also studied. The FYTD model can capture the multi-dimensional dependence of the fission yield data and make reasonable predictions.