Using dynamical mode decomposition to extract the limit cycle dynamics of modulated turbulence in a plasma simulation

Abstract The novel technique of dynamical mode decomposition (DMD) is applied to the outputs of a numerical simulation of Kelvin–Helmholtz turbulence in a cylindical plasma, so as to capture and quantify the time evolution of the dominant nonlinear structures. Empirically, these structures comprise rotationally symmetric deformations together with spiral patterns, and they are found to be identified as the main modes of the DMD. A new method to calculate the time evolution of DMD mode amplitudes is proposed, based on convolution-type correlation integrals, and then applied to the simulation outputs in a limit cycle regime. The resulting time traces capture the essential physics far better than Fourier techniques applied to the same data.