[Wien] two questions

[Stefaan Cottenier]

> *********************************************************************
> The K-points in BZ are varied in the range of 500 to 2500, in order to 
> find a appropriate
> setting. In my opinion, the total energies tend to be identical with 
> increasing k-pionts.
> However, it is not the case. Are there some errors in my input files?
> ***************************
> BZ   IBZ    Total Energy
> 500  (42)  -53187.869514 Ry <====
> 800  (56)  -53187.869545 Ry
> 1000 (56)  -53187.869236 Ry
> 1500 (108) -53187.882254 Ry <====
> 2000 (135) -53187.865566 Ry
> 2500 (165) -53187.869175 Ry <====
> ***************************

Only the three cases marked by an arrow are sufficiently different in
kmesh to be meaningful. If nothing went wrong with the calculation with
108 points (search for :WAR in the last iteration), then this shows that
you need k-mesh that is larger than what you tried so far in order to
stabilize the second digit of the total energy (see the reg_user section
at www.wien2k.at, textbooks, second document, sec. 6.2 for a discussion
on k-mesh selection).


>   2.  GGA+U scheme are employed to compute the antiferromagnetic 
> Nd2SrMn3O9, in which the moments
> of Nd and Mn sublattce align antiparallel, with the command runsp_lapw. 
> The electronic con-
> figuration of Nd in case.inst is changed to form the antiferromagnetic 
> arrangement of Nd and
> Mn sublattice. The computational details are similar with 1st question, 
> except for case.inst,
> case.indm and case.inorb. The spin-orbit coupling is not included now, 
> in order to check whether
> the magnetic and electronic structures can give me reasonable results 
> without SO coupling.

As you include indm and inorb files: do you use LDA+U? If not, these 
files are not needed.

> case.indm
> *******************************************************************
>   1  3  0                     nmod, natorb, ipr
> PRATT  1.0                    BROYD/PRATT, mixing
>   1 1 3                          iatom nlorb, lorb
>   2 1 2                          iatom nlorb, lorb
>   3 1 2                          iatom nlorb, 
> lorb                            
>   1                              nsic 0..AFM, 1..SIC, 2..HFM
>    0.652 0.00        U J (Ry)   Note: we recommend to use U_eff = U-J 
> and J=0
>    0.463 0.00        U J
>    0.463 0.00        U J
> *********************************************************************

Your last 5 lines are not legal in case.indm. After '3 1 2', there 
should be a line with '0 0', and that's it.

Furthermore, for LDA+U it should be case.indmc, not case.indm.

But most likely you aren't using LDA+U anyway?

>   The total energy fluctuate remarkably even for hundreds of mRy, 
> leading to non-convergence. Are there
> some errors in my input files? Which parameter in my input files should 
> be tuned to improve the convergence?

Convergence with LDA+U might be problematic, that often happens.
Converge with LDA first, then save_lapw. Now converge with -orbc, then
save_lapw. Finally do unconstrained LDA+U (-orb). If it doesn't work
even that way, then consider to crank up the U slowly (first 0.1,
converge, save, then 0.2, etc.)

Stefaan


-- 
==== NEW EMAIL ====

Stefaan Cottenier
Center for Molecular Modeling
Ghent University
Proeftuinstraat 86
BE-9000 Gent
Belgium

http://molmod.Ugent.be
email: Stefaan . Cottenier /at/ UGent . be <===== NEW EMAIL ===