Information and Complex Data Across Physical Systems
JJ Ruby, Lawrence Postdoctoral Fellow, Lawrence Livermore National Laboratory
Friday, December 9, 2022
Abstract: Many modern systems of measurement result in data that is challenging to interpret giving rise to an increased reliance on integrated modelling techniques and purely empirical statistical models, when there is sufficient data. Often the most complete models are sufficiently computationally intensive that they have limited use within an experimental analysis pipeline, outside of heuristic comparison, and have parameters that require adjustment to reproduce measured data. Conversely statistical models have the flexibility to directly infer quantities from data but lack the logical consistency of a physics model and may not provide insight into the mechanisms underlying the observed behavior. This work will present efforts to develop blended physical-statistical models by combining reduced first-principal models with state-of-the-art probabilistic programming languages to extract information from complex measurements. Examples of how these techniques have already been used in both high-energy-density physics and professional baseball will be presented along with the future outlook and directions.
Bio: JJ Ruby received a B.S. in Astrophysics and Planetary Science from Villanova University and is a graduate of the University of Rochester, receiving his PhD from the department of Physics and Astronomy and completing his graduate research at the Laboratory for Laser Energetics performing experiments on the Omega60 Laser and using Bayesian inference techniques to understand the acquired data. He received the Lawrence Postdoctoral Fellowship from Lawrence Livermore National Laboratory to continue working on using modern data science techniques to understand measurements from physical systems. JJ is currently the Lead Innovator for the Houston Astros of Major League Baseball, a role where he uses physics models, Bayesian inference, and in-game measurements to understand the underlying fundamentals of performance and help inform future decision making and player development.