The ABINIT Users Mailing List ( CLOSED )

[Matthieu Verstraete <mjv500@york.ac.uk>]

Hello Katalyn,

in principle you're fine. A few things to be careful of:

1) the spin ground state is not unique. The spinat shouldn't affect the 
final result, but it is there to break the natural up/down symmetry.

2) In some systems with several atoms, you can also try the 
antiferromagnetic configuration, which can be more stable. Usually you 
know if you are expecting this, but not always.

3) In more extreme cases (I have not had this often myself) you need a 
larger spinat to get away from the \vec{M}=0 solution. The value seems 
arbitrary but in the numerics it isn't always. Too large a value can also 
make the convergence very difficult. Basically everthing is initialized 
with a magnetized spin density located around each atom. If the final 
magnetized state looks very different from this (localized spins), then 
the convergence will be painful. I don't know if the new iscf values are 
better here too, but in any case they shouldn't be worse.

And so on - it is quite delicate in some cases, and I have seen horrendous 
convergence for complex systems like frustrated Cr.

Have fun

Matthieu

--
================================================================
Dr. Matthieu Verstraete                 mailto:mjv500@york.ac.uk
Dept. of Physics, University of York,     tel: +44 1904 43 22 08
Heslington, YO10 5DD York, United Kingdom fax: +44 1904 43 22 14