Pseudo-Finslerian spacetimes and multi-refringence

It is reasonably well-known that birefringent crystal optics can to some extent be described by the use of pseudo-Finslerian spacetimes (an extension of pseudo-Riemannian spacetime). What is less commonly appreciated is that there are two separate and quite disjoint pseudo-Finsler structures for the two photon polarizations, and further, that there are separate tangent-space pseudo-Finsler structures defined by the group velocity and co-tangent-space pseudo-co-Finsler structures defined by the phase velocity. The inter-connections between these four separate pseudo-Finsler structures are rather subtle. One particular source of technical difficulty is the fact that because physicists need to use pseudo-Finsler structures to describe propagation of signals, there will be nonzero null vectors in both the tangent and cotangent spaces -- this causes significant problems in that many of the mathematical results normally obtained for "usual'' (Euclidean signature) Finsler structures either do not apply, or require significant modifications to their formulation. We shall first provide a few basic definitions, and then present a tutorial outline of the relevant physics of birefringent optics, explicitly demonstrating the interpretation in terms of pseudo-Finsler spacetimes. We shall then discuss the tricky issues that arise when trying to inter-relate the various pseudo-Finsler structures we encounter, and finish by connecting these technical questions (which of course arise in what is a physically a very well-understood situation) back to ongoing research into possible extensions and modifications of general relativity.