Transfer learning as a method to reproduce high-fidelity non-local thermodynamic equilibrium opacities in simulations

Simulations of high-energy density physics often need non-local thermodynamic equilibrium opacity data. These data, however, are expensive to produce at relatively low fidelity. It is even more so high fidelity such that the calculations can contribute 95 % total computation time. This proportion reach large proportions. Neural networks be used replace standard low-fidelity and neural trained reproduce artificial, high-fidelity spectra. In this work, it demonstrated a novel network architecture krypton spectra through transfer learning in simulations. Further, done while achieving relative per cent error peak radiative temperature hohlraum approximately 1 4 19.4 $\times$ speed up.