Consider multigrid techniques for flamelets
Consider the use of a rigorous nonlinear multigrid solver for steady flamelets. Ideally it just flies and lets us use much larger grids on which resolution dependence of the solver smooths out. The two-grid approach currently in the code uses the idea of nonlinear multigrid, but not very well. And yet it works quite well, speeding up calculations where resolution across all mixture fraction space is required.
We may not use a nonlinear multigrid solver for day-to-day options, but having a scalable algorithm for extremely well-refined simulations would be an excellent 'nuclear option.'
The following files do full approximation scheme (FAS) multigrid for a single diffusion equation (yes, it is a linear equation, but this is about testing the nonlinear multigrid framework) and attempts it for the flamelet equations. This flamelet one is not working - noise in species mass fractions that looks like aliasing shows up after the first restriction and prolongation. I don't know why.