A New Data-Driven Standard Solar Model

Project Overview

Modern high energy density (HED) experiments have succeeded in creating conditions that otherwise exist only in astrophysical objects ranging from planetary and stellar interiors to the envelopes of white dwarf stars. These experiments exhibit physical phenomena that are critical to the behavior of the system, such as electronic screening effects and overlapping atomic orbitals, but cannot otherwise be observed in isolation. Likewise, these effects surely exist within astrophysical bodies, and are likely critical to their structure and evolution, but similar to the HED experiments their effects cannot be studied in isolation in the astrophysical domain.

This project performed early work to build an analysis framework that leveraged relevant data from many different sources to understand physics at extreme conditions. In particular, this included using Inertial Confinement Fusion (ICF) implosions data and HED experimental data from a number of facilities and astrophysical observations all in conjunction to understand the underlying physical mechanisms; in particular, equation of state in hot dense plasmas, nuclear cross-sections in dense plasmas, atomic physics, and thermal conductivity in hot dense plasmas. This framework provides uncertainties for the implicit physical models, where no data exist. The platform is built on Bayesian inference. This is a natural way to marry datasets of varying types and leverage all additional information such as physical laws or scientific intuition in a way that is self-consistent and allows for proper uncertainty quantification. HED experiments were analyzed, and the beginnings of a Bayesian inference model was constructed. To this end a numerical equation of state model was built that can flexibly produce an accurate representation of a dense plasma that is sensitive to the electronic configurations of a given element. In addition, a fully differentiable ray-trace was built that can couple to state-of-the-art Bayesian inference tools in python. Finally, scripts were developed to link these codes with the stellar evolution code MESA with the ultimate goal of performing inference on solar data with the underlying equation of state model.

Mission Impact

This work primarily supports Lawrence Livermore National Laboratory's missions in nuclear deterrence and stockpile stewardship, as well as the Lab`s Core Competency in High Energy Density Science. In particular, the goal of this work is to provide new data streams for constraining physics of interest to the stockpile, especially in regimes that cannot be reached (currently) with terrestrial experiments. This work also supports NASA by making use of stellar observations.

Publications, Presentations, and Patents

John Ruby, "Atomic Energy Levels and Partition Functions for an Empirical Equation of State" (Presentation, Center for Materials at Pressure, University of Rochester. July 2022).

John Ruby,"Inference Techniques for Complex Data in Physics" (Presentation, Los Alamos National Laboratory, Los Alamos, NM, Virtual. August 2022).