Using data science methods to efficiently and effectively plan HED experiments and analyze data

High Energy Density (HED) science is the study of the behavior of material under extreme conditions of temperature and pressure (above 1 Mbar). Understanding the growth and properties of hydrodynamic instabilities and the transition into turbulence in high energy density regimes is important in many HED processes and will also yield insights into other areas of science where hydrodynamic instabilities occur.

Unlike “classical” fluids experiments, examining hydrodynamic instabilities in an HED regime is a little more difficult. HED experiments are expensive and often performed at large oversubscribed facilities. Additionally, there are many limitations when it comes to the available diagnostics. Further, these problems contain complex physics: hydrodynamics flows, radiation, conduction, and magnetic fields to name a few. Such complexity means that modeling them can become prohibitively expensive. Improving the models and conducting accurate simulations is of great importance due to these physical and financial limitations. 

A big question that exists within the context of improving simulations is: how do we create a general model that can both accurately capture the underlying physics and be applicable to more than one problem? In an attempt to answer this question, we incorporate experimental design and uncertainty quantification techniques to connect experimental data with simulations and take a statistical approach in order to efficiently plan experiments so that the data obtained can be analyzed to yield valid and objective conclusions.