Relationship between B and H fields in cylindrical magnetization

[VinnyCee]

So I've been simulating a really simple geometry using ANSYS Maxwell. It is a cylinder only and I am looking at the $\overrightarrow{B}$ and $\overrightarrow{H}$ fields in order to see their relationship between them when the material is magnetized in the circumferential direction. I used a cylindrical coordinate system to describe the magnetization. So the direction of magnetization is in the $\phi$ direction.

$\overrightarrow{B}$ using a BH-curve to define the magnetic material:




$\overrightarrow{H}$ using a BH-curve to define the magnetic material:




$\overrightarrow{B}$ using a scalar to define the magnetic material:




$\overrightarrow{H}$ using a scalar to define the magnetic material:


Notice that both BH-curve defined and scalar defined magnetic materials exhibit a kind of random/noise $\overrightarrow{H}$. What is causing this phenomenon? Isn't $\overrightarrow{H}$ supposed to be in same direction as $\overrightarrow{B}$ external to the material and in the opposite direction interior?

 

[marcusl]

Possibly its an artifact of the finite-sized and coarse meshing, and of finite-precision arithmetic effects (number of digits, roundoff error, etc). The results are wrong in other ways as well; for instance, there should be no fields outside the cylinder.