Multipole boundary data

[mrhardman]

Option to calculate boundary data from a multipole expansion. Effective, fast, and accurate if velocity domain is large enough. Further testing likely required. The option permits the collision operator to run at very large resolutions, by avoiding the need to compute integration weights, but the actual speedup of the simulation at runtime appears to be small for the examples tested. A potentially useful feature for future work.

Now featuring three (enum) options for calculating the boundary data:

• boundary_data_option = "direct_integration". The classic option
• boundary_data_option = "multipole_expansion". Boundary data computed from a direct multipole expansion of the definition of the Rosenbluth potentials, up to 8th order in the expansion.
• boundary_data_option = "delta_f_multipole". A variation on "multipole_expansion" where the multipole is only used for the Non-Maxwellian part of the distribution function, and the exact analytical results are used for the Maxwellian piece. Slightly slower than "multipole_expansion", but likely to perform better on grids with smaller vpa.L and vperp.L when the solution is closer to Maxwellian.

These features are tested in fokker_planck_tests.jl. Further simulations are required to test the performance of these features.