Using Physics Guided Deep Learning to Investigate Performance and Variability in Inertial Confinement Fusion Experiments

On December 5^th, 2022 the National Ignition Facility (NIF) demonstrated fusion ignition in the laboratory with the inertial confinement fusion (ICF) experiment N221204. Fusion ignition marks one of science’s greatest achievements, that has been decades in the making, with contributions from several different fields of expertise, including machine and deep learning.

Here we present our findings using physics guided deep learning models with experimental data, recorded from ICF experiments at the NIF, to predict multiple ICF performance metrics and investigate variability in shot performance, as shots start to approach the ignition boundary. We highlight features that significantly impact shot performance and affect variability in shot metrics. We investigate the performance and variability space to help research designs that lead to repeatable high performing shots.