how to constrain magnetization in ABINIT?  [SOLVED]

Total energy, DFT+U, BigDFT,...

Moderator: amadon

Post Reply
Posts: 8
Joined: Fri Nov 11, 2016 2:24 am

how to constrain magnetization in ABINIT?

Post by ucsdmavrl » Tue Dec 13, 2016 8:58 am

Dear ABINIT developers and users,

I am wondering how I can successfully constrain the magnetization for a specific atom in a system. I searched the posts in the forum.(viewtopic.php?f=9&t=3337&p=10222&hilit=magconon#p10222)
It is said that we can play with the params of "magcon_lambda" and "magconon". I have a question on the param "magconon". In the doc, it says "The constraint can be either on the direction only (magconon 1) or on the full vector (magconon 2)." I am not clear about what "the direction" and "full vector" is? Could you pls explain a bit more? (say I want to constrain the spin for an atom to be:
0 0 1
how can I set magconon?

Also, when adding these two parameters, the SCF run severely diverged(total energy increase). I play with magcon_lambda from 10 ~ 300 (with step 10), they all still fail to converge, any suggestion on this?


Posts: 271
Joined: Sat Jan 16, 2010 12:05 pm
Location: Dept. of Physics, University J. J. Strossmayer, Osijek, Croatia

Re: how to constrain magnetization in ABINIT?  [SOLVED]

Post by ilukacevic » Wed Dec 14, 2016 8:53 am

Dear ucsdmavrl,

For magconon 1, you constrain only the direction of a vector, while its size will be left open for 'relaxation' during the energy minimization.
For magconon 2, you constrain both the direction and the size of the vector, and the energy is minimized under both of these conditions.

The initial vector is given by spinat variable. So, you just have to add magconon (with the wanted value) and magcon_lambda (for which you have to find the optimal value).

Concerning the SCF issue, maybe Eric or Matthieu can give you the better answer.

Best regards,


Posts: 372
Joined: Tue Apr 19, 2011 11:13 am
Location: University of Liege, Belgium

Re: how to constrain magnetization in ABINIT?

Post by ebousquet » Wed Dec 14, 2016 11:11 pm

Dear Ben,
Are you constraining your magnetization with spin-orbit coupling?
For lambda try smaller values than 10, I got some cases working fine for lambda=1 or so.
Otherwise how bad is the convergence, says on energy residual? It can help to read a wave function obtained without the constraint, but I agree the convergence can be worsened by constraining the magnetization for a reason I do not fully understand and I have to work on it.

Post Reply