[Pw_forum] nspin=1 in NEB calculation

He Jinghui g0800703 at nus.edu.sg
Thu Oct 6 11:27:36 CEST 2011


Dear Guido and others:

Thanks for your kind reply. I have read the manual and still a little 
bit confused. Please point out if i have a wrong understanding.

Does "tot_magnetization" parameter is only for induce starting 
magnetization, which allowed pw.x to determine during SCF cycles, or it 
is used to freeze  the value during whole calculation?

Is this parameter of the same function with:
constrained_magnetization='total'  together with fixed_magnetization(i), 
i=1,3

if i used nspin=0, in output file there is still rho_up and rho_down, 
does this mean the spin-DFT has already been considered even nspin=0?

In my case, I want to perform a cluster calculation with even electrons. 
But the cluster may be at triplet state or singlet one. I just want to 
switch on the spin-DFT, and let the program determine which one is the 
ground state and obtained the stable geometry. So want should I set?

Thank you all so much for your help!

best regards
He jinghui
Department of Chemistry, NUS, singapore

in may


> Dear Jinghui,
>
>>      I was wondering once i set nspin=1, do i have to settot_magnetization?
> do you mean "nspin=2"? This is indeed the option for a spin-polarized
> (collinear) calculation, see Doc/INPUT_PW.txt
>
> In that case, yes, you have to specify the starting magnetization of the
> system (there are a number of ways, again have a look to
> Doc/INPUT_PW.txt). And yes, the default is different from SIESTA where
> -forbid me, I'm not experienced of that code- imposing "spin_polarized
> true" implies by default a ferromagnetic starting point.
>
> HTH,
> Guido
>
>    PS As for NEB, keep in mind that if the magnetization of your system is
> crucially depending on coordinates (as an example, if your transition
> state is magnetic but the initial and final states are not), then you
> might have some difficulty in imposing the true magnetized state for some
> image in the path, because the starting charge density might be
> extrapolated from previous steps where the system is not magnetic. In that
> case, you could either continue the calculation but restarting from atomic
> charges, or replace that charge from one of a separate, controlled SCF
> calculation.
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.democritos.it/pipermail/pw_forum/attachments/20111006/dcdf8c48/attachment.htm 


More information about the Pw_forum mailing list