[nmag-users] Domain wall motion and PBC. Convergence error

[Adrien Pivano]

Dear All,

I would like to compute domain wall motion in a nanowire using periodic boundary conditions, but when I use it I have a convergence problem because I think there is a magnetization contradiction at each sides. In the spin wave example we can see that the right side are the same than the left side. In my example, the right side has a magnetization along [-1,0,0] and in the left side has a magnetization along [1,0,0], with a domain wall in the middle. How I can avoid this convergence error ?

Thanks for your assistance,

Adrien Pivano

[Weiwei Wang]

Hi Adrien,

I think this is a model error: there must be a magnetic defect when you put (-1,0,0) and (1,0,0) together. However, there is no this kind of problem to use periodic boundary conditions if you have two domain walls (i.e., a head-to-head DW and a tail-to-tail DW), but this setting have changed your model, so maybe it's better to recheck why the periodic boundary conditions are needed in your simulation.

Cheers,

Weiwei

[Adrien Pivano]

Hi Weiwei,

Thanks for your help. I need to use the periodic boundary conditions to have a smaller BEM than a normal computation because I'm looking for the Walker threshold and I need a long mesh. So I thought that is possible to reduce the memory with PBC and a small mesh. Apparently  there is no way to avoid this problem cause I think if I have two domain walls and if I use a magnetic field in one direction i should have a 360° domain wall and maybe I couldn’t  see the Walker breakdown.

Thanks a lot for your advice,

Adrien

---

De: "Weiwei Wang" <wangw...@gmail.com>
À: "Adrien Pivano" <adrien...@im2np.fr>
Cc: nmag-...@lists.soton.ac.uk
Envoyé: Lundi 20 Avril 2015 14:23:43
Objet: Re: [nmag-users] Domain wall motion and PBC. Convergence error