In this work, we study how Gaussian processes can be applied to emulate the output of quantum-mechanical models of nuclear fission. The goal is to emulate the potential energy surface of the fissioning nucleus and use this to calibrate physics parameters. First, we construct an efficient emulator for a 1D projection of the potential energy surface using a Gaussian Stochastic Process (GaSP) model. This model was then used to calibrate the parameters based on experimental data. We then explore modeling the full potential energy surface. This is challenging because there is a large discontinuity in the surface that varies with the input parameters. Additionally, there is some missing data due to some convergence issues with the physics code.
This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344 and was supported by the LLNL-LDRD Program under Project No. 18-ERD-008
IM Release Number: LLNL-ABS-802421
|