Opacity Variations for Uncertainty Quantification

Opacity is an important quantity both in its role in mediating radiation transport and as a diagnostic tool for measuring densities, temperatures, and constituents. Understanding our uncertainties in this quantity is a missing piece in high energy density experimental design and interpretation. We take a two-prong approach: (i) data-driven, and (ii) algorithmic. Within the data-driven approach we identify an extensible set of key model variations within LLNL’s Opus opacity code to build an ensemble of reasonable models. As an initial attempt to stand up this new uncertainty quantification capability, we vary the electron collision model, the definition of mean ionization, the gaunt factor, and pressure ionization model for pure carbon. We show that errors are correlated in density- temperature-energy space and quantify the effects of each variation. In addition to the above data-driven approach, we also develop a rigorous ab initio formalism to efficiently account for the effect of quantum fluctuations associated with core ionizations at high temperatures.