[FLASH-USERS] Divergence of magnetic field

[Lucas Rubio Fermoso]

Dear all,

I am trying to run a 2D cylindrical simulation with a high uniform 
initial magnetic field in the "Z" direction (~100 T), but I am getting 
very high divergence values that cascade in other errors forward in the 
simulation.

This divergence is generated at the cells in the block boundaries 
perpendicular to the field, while being 0 everywhere else.

Divergence values are created at the first step and they evolve during 
"nrefs" steps, after that the values get frozen for the rest of the 
simulation.

For example: For "nrefs"=2 initial divergence is 0 everywhere, 
divergence changes at steps 1, 2 and 3 and then keeps the same values of 
step 3.

For "nrefs"=4 is the same but evolves up to step 5.

The problem seems related to how paramesh handles fluxes before it 
starts refining.

I have tried to restart the simulation on the initial checkpoint so 
paramesh redistributes blocks on start, but did not work.

Some parameters that may be important:
Grid is Paramesh4
Prolongation methods: injection and balsara, both used with little 
difference (It should be used balsara since the simulation is 
cylindrical)
Solver: Unsplit Staggered Mesh
killdivb = true
flux_correct = true
E_modification = true
energyfix = true

In simulation_initBlock magnetic field face values and magnetic field 
cell values are declared with the same value.

Thanks in advance for your help.

Best regards,
-- 
Lucas Rubio Fermoso
Universidad Politecnica de Madrid