Parallel Implementation of Niched Pareto Genetic Algorithm Code for X-ray Plasma Spectroscopy

X-ray spectroscopic analysis has proved to be a useful technique to determine temperature and density of astrophysical as well as laboratory plasmas [1]. Routinely analysis is performed manually, that is, using an interactive graphical user interface to compare experimental and theoretical spectra calculated using a particular set of plasma model parameters. When a good fit to experimental data is achieved, the parameters used to calculate the synthetic spectrum are considered to be representative of the state of the plasma during the formation of the spectrum. A criterion for measuring good fits is the distance between synthetic and experimental spectra defined in a given metric as well as visual similarity. This procedure is simple and convenient as long as the spectral model is not very complex, does not depend on too many parameters, and is relatively inexpensive (can be performed in real time). For many other cases the availability of a computer-driven automated analysis procedure is important. In this paper we will use a genetic algorithm to estimate plasma temperature and density gradients for high energy density plasmas. Traditional (as described above) spectroscopic analysis has been used to determine averaged or effective temperatures (Te) and densities (Ne). However, the spectroscopic analysis of plasma temperature and density gradients represents a more complicated search problem in parameter space. In this paper we use a genetic algorithm to estimate plasma temperature and density gradients for Inertial Confinement Fusion (ICF) experiments. The goal here is to find temperature and density gradients that simultaneously produce good fits to 1) time-resolved spatially integrated Xray line spectra and 2) X-ray monochromatic emissivity profiles. Spatial emissivity profiles can be extracted from Abel inversion of X-ray monochromatic images provided that the plasma is optically thin and spherically symmetric (see Figure 1). Data: time-resolved spectra Data: time-resolved X-ray monochromatic images I(ν)