picmi.ElectrostaticSolver's B field solution is wrong when warpx_relativistic=True

[jwestern]

I tested the magnetic field solution on a current loop on a 3D Cartesian grid. The solution is correct when warpx_magnetostatic=True, but is wrong when warpx_relativistic=True. Compared to the latter case, the solution also changes when both warpx_magnetostatic=True and warpx_relativistic=True. All the E fields are fine in all cases.

If more details are needed, or a reproducer is needed, let me know.

Image comparing Bx By Bz for warpx_relativistic=True vs warpx_magnetostatic=True:

Image comparing Bx By Bz for warpx_relativistic=True vs {warpx_relativistic=True & warpx_magnetostatic=True}:

[RemiLehe]

Thanks for raising this issue.

Could you share your WarpX input script for this setup? This would help us reproduce the issue and investigate why the relativistic solver gives incorrect results.