Large-scale kinetics modeling of non-LTE plasmas

Abstract

Improvements continue to be made in the modeling capabilities provided by the Los Alamos National Laboratory (LANL) suite of atomic data codes. Recent work includes the addition of parallel rate matrix solvers when obtaining the atomic populations necessary to model non-LTE plasmas and the use of various interpolation techniques to obtain the vast amounts of atomic cross sections and rates that are required to build the data models. These enhancements have enabled the creation of models that contain more than one million relativistic configurations. The ATOMIC code has recently been used to synthesize spectra from such large models with two completely independent, sparse solvers: a serial, iterative, generalized minimum residual algorithm and a parallel, iterative, algebraic multigrid approach. The populations resulting from these two methods can be compared directly in order to determine how sensitive is the solution to the positioning of the usual boundary condition associated with charge neutrality. This sensitivity is an important concern when such large systems of equations are being considered. Additionally, this enhancement in model size can be used to better determine the convergence of calculated spectra with respect to model size. As an application of these capabilities, comparisons with a recent xenon experiment are discussed.