Thanks for the reply! Great to know that this is on your radar.
My main interest is in exploring numerical methods (and combinations of numerical methods) for solving kinetic equations, so I guess that I would have to build my own Julia type for the variables and the specific sparse matrices for different discretizations of the collision operator (for things like the Boltzmann equation). Given the specific goal of trying out new numerical methods I don’t think I can start at a level higher that this, correct?
I’m looking forward to exploring DifferentialEquations.jl for solving the resulting ODE system. I did see there is basic linear-nonlinear splitting for IMEX methods, but I couldn’t find anything more general that would let me solve different equations in the system using different methods while keeping them coupled. I might not have looked hard enough, though.