iljah / pamhd Goto Github PK

Currently magnetic fields on cell faces are ad-hoc mix of several eigen::vector3d and dedicated type (Face_Type,

Solar wind carries/induces ultra low frequency waves in magnetosphere, try simulating 3d test with cells that are e.g. 2-8x narrower towards sun before adaptive mesh refinement with https://github.com/fmihpc/dccrg/blob/master/dccrg_stretched_cartesian_geometry.hpp
This should increase frequencies that can get from solar wind boundary to magnetosphere with e.g. 4-64x fewer cells on dayside than with cubic cells.

Why each background face B is defined as a 3D vector, shouldn't it be a scalar because only a component perpendicular to the face is stored?

2d blast wave test with runtime adaptive mesh refinement (https://github.com/iljah/pamhd/blob/amr/tests/mhd/config_files/magnetohydrodynamic/blast_wave/2d/x1/rusanov_staggered_amr5.json) creates large divergence of magnetic field in some cells, this should be fixed.

It would be useful to name PAMHD output files such that they are correctly ordered in time in Linux, VisIt, etc. Now the name includes simulation time in scientific notation, which is somewhat cumbersome to deal with.

I suggest replacing the time parameter in the file name with an integer giving the number of time steps since the start. With appropriate amount of padding zeros of course.

For example:

mhd_staggered_7.205e+03_s.dc
->
mhd_staggered_00000698.dc

Pressure has bit of sawtooth in dynamic amr blast wave test if cell size changes too close to B pressure front. Perhaps better pressure adjustment when overriding MHD flux solution with staggered B solver?..

Small pressure/B1 anomaly forms in solar wind before magnetosphere in 3d test. Might be related: #7

muParser requires compilation but ExprTk doesn't, perhaps it would be better for handling math expressions?
http://www.partow.net/programming/exprtk/index.html