Reduced-Order Modeling of Backward Wave Oscillator Fields for High-Power Microwave Applications

We present a data-driven, reduced-order model to analyze and predict high-power electric fields within a Backward Wave Oscillator (BWO) using Dynamic Mode Decomposition (DMD). DMD is trained on the high-fidelity data generated from the electromagnetic particle-in-cell (EMPIC) simulation of BWO. We showcase the efficacy of DMD in modeling the time evolution of the BWO fields on a low-dimensional feature space. This application of DMD-driven predictive modeling serves to accelerate the resource-intensive kinetic plasma simulations associated with devices like BWO. Furthermore, it significantly aids in navigating the intricate design processes of such devices by yielding insights into the pivotal spatio-temporal dynamics of the oscillatory field.