The ABINIT Users Mailing List ( CLOSED )

[Xavier Gonze <gonze@pcpm.ucl.ac.be>]

Dear Jorge,

Jorge Iniguez wrote:
> Dear abinitioners,
>
> A couple of weeks ago Gilles Z. mentioned the possibility of doing
> calculations with magnetization constrained to lie along a given spatial
> direction (see emails about Berry curvature & Chern number sent by H. Min
> as well as GZ's response).
>
> I think it is worth to implement such option in abinit. At the very least,
> it would allow us to check whether the spinor calculations are working
> properly.

Yes !
The spinor part of ABINIT was tested already, but I do not think it has
been used for a real scientific work, so I would not consider it as likely 
bug-free !
There are so many issues beyond the first, obvious tests ...

> As you see, I give very similar "seeds" with spinat... but get five
> different answers. The AF modification in spinat5 (which is the way to
> reduce the system energy, since the system ground state is AF) is
> completely missed by the first four runs. (Note that this problem will be
> present even if we do not have spin orbit. An spinor calculation should be
> able to find the ground state (FE or AF) of the system, right?) Regarding
> the first four cases, one could be tempted to think that the energy
> differences involved are so small that the run gets converged before
> actually finding the minimum (in spite of the very small tolerance used).
> However, case 5 involves relatively large energy differences and the
> program fails to go all the way up to an AF ordering. This seems to imply
> something is not working as it should with the spinor calculation? (If
> somebody thinks the above results can be easily explained and there is
> nothing to worry about, please let me know :)

Seems strange indeed ...
> Being able to set the magnetization direction would be helpful to
> understand what is happening here. (Maybe the above result is simply
> indicating that the energy landscape of the system is very complicated? We
> could answer that question if we were able to fix the direction of M.) It
> would also allow to calculate magnetic anisotropies, etc.
>
> I would very much like to work in the implementation of this.

Thanks for this offer !!! ;-)

Best regards,
Xavier