Physics-constrained data-driven methods in MHD

Accurate and efficient plasma models are essential to understand and control experimental devices. Data-driven techniques recently developed in fluid dynamics can be leveraged to develop interpretable reduced-order models of plasmas that strike a balance between accuracy and efficiency. The dynamic mode decomposition, POD-Galerkin methods, and other reduced order models are applied to experimental and simulation data, and suggest possible uses in real-time control and modeling for fusion devices.