Remedy for non-ellipticity and ill-posedness in the mean-field limit of the 1D Lagrangian zonal-flow model
Determine a mathematically sound remedy for the lack of ellipticity and associated ill-posedness in the mean-field Fokker–Planck limit of the 1D Lagrangian SDE model for zonal flows, namely the equation ∂_t ω + α ∂_x(ω ∂_x ω) + ∂_x(ω β) = κ Δ ω whose effective diffusion term ∂_x((κ − α ω) ∂_x ω) loses ellipticity when ω(x) > κ/α; develop a formulation that remains well-posed while still allowing aggregation phenomena.
References
However, this equation presents problems of well posedness, because the second order differential operator, written on the right-hand-side, takes the form
\partial_x ((\kappa- \alpha\omega)\partial_x\omega)
which is elliptic only when $\omega(x)\leq\kappa/\alpha$ for every $x$. If we start with an initial condition $\omega_0$ satisfying this inequality, one can prove that also the solution does, but this prevents the formation of clusters with increasing strength, the solution will simply decay. If we take an initial condition which violates this inequality, we have a problem of lack of ellipticity.
It is not clear yet how this problem can be remedied.