A geometric theory of waves and its applications to plasma physics

Waves play an essential role in many aspects of plasma science, such as plasma manipulation and diagnostics. Due to the complexity of the governing equations, approximate models are often necessary to describe wave dynamics. In this dissertation, waves are treated as geometric objects of a variational theory rather than formal solutions of specific PDEs. This approach simplifies calculations, highlights the underlying wave symmetries, and leads to improved modeling of wave dynamics. This thesis presents two breakthroughs that were obtained in the general theory of waves. The first main contribution is an extension and reformulation of geometrical optics (GO) as a first-principle Lagrangian theory that correctly describes polarization effects, such as polarization precession and the polarization-driven bending of ray trajectories, which appear as leading-order corrections to GO. The theory was applied to several systems of interest, such as relativistic spin-1/2 particles and radio-frequency waves in magnetized plasma. The second main contribution of this thesis is the development of a phase-space method to study basic properties of nonlinear wave--wave interactions. Specifically, I show that waves propagating in modulated media, both classical and quantum, can experience time-averaged refraction caused by effective ponderomotive forces on wave rays. This phenomenon is analogous to the ponderomotive effect encountered by charged particles in high-frequency electromagnetic fields. I also show that phase-space methods can be useful to study problems in the field of wave turbulence, such as the nonlinear interaction of high-frequency waves with large-scale structures. Overall, the results obtained can serve as a basis for future studies on more complex nonlinear wave--wave interactions, such as modulational instabilities in general wave ensembles or wave turbulence.