I'm currently attempting a 3D FDTD simulation where I'm modeling a plasma as a dielectric tensor based off a frequency dependent Appleton Hartree dispersion relation. The inversion of the dielectric tensor is symmetric positive definite (SPD) when my propagation frequency (f0) is less than the plasma frequency (fpe), when f = fpe the inversion of the dielectric tensor is not possible, and when f < fpe, the inversion of the dielectric matrix is not SPD.
This causes my simulations to blow up in the regions where f