[Wien] Using in1new option

[Stefaan Cottenier]

Hard to give a full answer without knowing the kind of TM and its
position, but inspection of your in1-files shows the following:

* there is probably no compelling reason to use in1new, as the new
global linearization energy of 0.31 Ry is almost identical to the
default one (0.30 Ry)

* Probably atom 2 and 4 are your metal? Both schemes treat it in the
same way, with an s and p LO.

* Ga is treated differently. This is the default scheme:

2   -1.09      0.010 CONT 1
 2    0.30      0.000 CONT 1
 0    0.30      0.000 CONT 1
 1    0.30      0.000 CONT 1

and this is in1new:

0    0.081     0.000 CONT 1
 0   -0.555     0.000 CONT 1
 1    0.259     0.000 CONT 1
 2    0.266     0.000 CONT 1
 2   -0.516     0.000 CONT 1 

The latter has an extra s-LO.

I would check whether the default esepermin and esper0 are correct for
your case. If not, put better values and see whether now you get the
same moments as with the default linearization. If the results are still
different, then the problem is probably not with linearization, but with
 a different starting situation (i.e. you might have stabilized a
high-spin solution in one case, a low-spin solution in the other case --
the total energy will then tell you which of both is the ground state).

Stefaan

 
> We are working on GaN system doped with transition metals. For a 
> particular calculation, using the in1new option gives 
> magnetic moment on the transition metal (TM) site which is about 2
> mu_B 
> higher 
> than what is obtained without the in1new options for TM site in 
> the unit cell. The expected values by us were closer to the one
> obtained 
> using the in1new options.
> The total energies differs by 0.03Ry., and the linearization
> energies in the in1 files are different as indicated in the
> *.in1c files attached.
> 
> I am not sure which results should be trusted because the magnetic
> moment 
> which is of interest to me is so different in the two cases. I am
> sending 
> the two *.in1c files as attachment. It should be add that some other
> 
> calculations on a similar GaN doped system with the  linearization 
> energies generated during the initialization give the
> expected/reported 
> magnetic moments.
> 
> Looking forward to some advice.
> 
> Regards,
> 
> Nandan.
> 
> -- 
>
******************************************************************************
> Nandan Tandon
> Research Student
> Department of Physics,
> University of Pune
> Pune-411007                             Tel.(O)+91-20-25692678 ext
> 426
> MAHARASHTRA, INDIA.                     Fax.(O)+91-20-25691684 
>
******************************************************************************




Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm