Randomized-NLA methods for fast accurate real-time characterization of tokamak discharges from multiple data-streams

Modern magnetic fusion research involves high-resolution temporal and spatial diagnostics from multiple sensor arrays that generate large data streams well-suited to advanced scalable numerical linear algebra methods. This presentation introduces randomized numerical linear algebra (rNLA) and describes the application of these methods (i) to identify and optimize data reduction methods for real-time discharge control and (ii) to advance our understanding of fundamental behaviors of magnetically-confined plasma.  Randomized methods trade off accuracy for speed and are easily adaptable to distributed computing architectures and streaming data scenarios. We describe our development plans that will incorporate rigorous statistical guarantees, leverage high dimensional statistics, and provide measures of sub-optimality as quantified by tunable parameters. The Columbia University High Beta Tokamak-Extended Pulse (HBT-EP) facility will provide the initial data for algorithm development, and we seek to evaluate the broader use of our algorithms for critical data-intensive control needs in plasma science and related areas.