Automated Design of HED Experiments on the Z Machine: a Metrics-Based Approach

We present a numerical optimization framework for automated design of high energy density (HED) materials science experiments on the Z Machine at Sandia National Laboratories. High energy density dynamic compression experiments in the 100s of GPa regime can only be performed on a limited number of experimental facilities, often requiring bespoke designs tailored to the individual objectives of the experiment. Through rigorous definition of traditional experimental objectives (peak pressure, shockless compression paths, etc.) we have developed a metric-based optimization method that computationally designs the experiment end-to-end, starting with the driver (Z Machine) input conditions. The system also allows for the incorporation of metrics less intuitively obvious to a human designer that can increase experimental objective success within known shot-to-shot driver variability. Initial results using this method will be presented, along with a discussion of advanced metrics under investigation, including the ability to optimize the difference between the observable response of various material models to design experiments to discriminate between theories.