about magnetization of magnets

[Ramdane LATEB <ramdane.lateb@xxxxxxxxxxxxxxxxxxx>]

hello femm's members,

My question is about magnetization of magnets, it seems
that the magnetization in femm depends on the number of 
subdivisions in magnets, I mean even if I put a radial 
magnetization in the middle of a span magnet, this seems 
true if the span is very small, but if the span is important 
I have to split the magnet into sub-magnets. What (I think)
I have understood is that it's rather a global magnetization,
more largest the span is, more parallel the magnetization
is, and less largest the span is, more radial the magnetization
is, so to have a radial magnetization I have to split the 
magnet into sub-magnets. I 'm wondering if someone else have 
observed this phenomenon in femm (perhaps it has been discussed 
before, if so , thank you to direct me to the corresponding 
messages). or your own experience, trie with a span magnet 
of 150° (electrical), with out splitting the magnet, and 
look to the back fem (or magnetic flux) in one tooth and 
the cogging torque (no load), and make the same when splitting 
the magnet in four (I mean splitting ,by just adding radial 
segments and adding a radial magnetization in each created 
sub magnet, and use a same mesh to approach the number of 
global nodes), the difference is rather observed in both 
cogging an tooth magnetic flux.

Any advice is welcome, thank you