The ABINIT Users Mailing List ( CLOSED )

[Masayoshi Mikami <mmikami@yk.rim.or.jp>]

Dear Aldo,

As mentioned by Andrew, it appears important to use 
semi-core electrons for both V and Y. 

Your attachment (Thanks !) reads: 
- pspatm: opening atomic psp file    ./39y.pspnc
  Troullier-Martins psp for element  Y         Thu Oct 27 17:39:06 EDT 1994
  39.00000   3.00000    940714                znucl, zion, pspdat 

- pspatm: opening atomic psp file    ./23v.pspnc
  Troullier-Martins psp for element  V         Thu Oct 27 17:33:59 EDT 1994
  23.00000   5.00000    940714                znucl, zion, pspdat

These pseudopotentials do not contain the semicore states.
Such pseudopotential may not produce plausible results, 
as explained ~ABINIT/Infos/Psp_infos/psp1.data.
(I also mentioned something related to the semi-core states:
 http://www.abinit.org/wws/arc/forum/2004-01/msg00036.html )

Although there are few pseudopotentials with semi-core states 
for Y and V on the ABINIT WEB, you can still find such PPs
from HGH PP page: (http://www.abinit.org/ABINIT/Psps/HGH/hgh.html).
V for 23v.13.hgh, Y for 39v.11.hgh.(click the green-highlighted area)
Rather high ecut may be needed, due to the semi-core states
that are rather "localized".

In any case, if you take PPs from the ABINIT web, please notice
the message "Use at own risk ! Pseudopotentials should always 
be tested in well-known situations, before using them for
predictions ..." 
If the HGH PPs still do not work, you will have to make your own 
PPs for your case, with PP-making codes(e.g. FHI98PP, OPIUM, ...).

In passing, I noticed that you adopted ecuts as
   ecut1 30
   ecut2 40
   ecut3 50
   ecut4 60
   ecut5 70
Please note that the default unit for ecut is Hartree, not Ry.
So your ecut should be enough for Troullier-Martins Oxygen PP.
(If you try Y/V pseudopotentials with semi-core states,
 you will have to check the ecut, probably higher than 
 30 Hartree (60 Ry). So this dataset may work for such tests.)

Good continuation,
Masayoshi

From: Aldo Humberto Romero <ahromero@bethe.fis.puc.cl>
Subject: Re: [abinit-forum] Pseudos  for YVO4
Date: Fri, 19 Mar 2004 11:14:03 -0400 (CLT)
Message-ID: <Pine.LNX.4.21.0403191043130.18807-200000@bethe.fis.puc.cl>

> 
> 
> Dear Masayoshi and Ab init users,
> 
> 
> let me then complete my question, here I am including the input file
> I have used to optimized the cell parameters as function of tue
> cutoff energy. 
> 
> With respect to the pseudos, I have used the pseudos found at the
> Abinit web page.... I would like also to point out that
> as I increases the cutoff the agreement with respct to
> the experimental lattice parameters gets worst (at 30ry is 5% at 60 ry
> oiis 8%!) ....here is the report  in the output:
> 
> - pspatm: opening atomic psp file    ./39y.pspnc
>   Troullier-Martins psp for element  Y         Thu Oct 27 17:39:06 EDT
> 1994
>   39.00000   3.00000    940714                znucl, zion, pspdat
>     1    1    2    0      2001   0.00000
> pspcod,pspxc,lmax,lloc,mmax,r2well
>     0   3.763   7.439    0   2.8904620        l,e99.0,e99.9,nproj,rcpsp
>    0.00000000   0.00000000   0.00000000   0.00000000     rms, ekb1, ekb2,
> epsatm
>     1   1.340   3.336    1   3.5304324        l,e99.0,e99.9,nproj,rcpsp
>    0.00000000   0.00000000   0.00000000   0.00000000     rms, ekb1, ekb2,
> epsatm
>     2   7.352   9.722    1   2.8904620        l,e99.0,e99.9,nproj,rcpsp
>    0.00000000   0.00000000   0.00000000   0.00000000     rms, ekb1, ekb2,
> epsatm
>     3.77869318742782    0.03790408035416    1.78466965569179
> rchrg,fchrg,qchrg
> 
> pspatm: epsatm=   84.09799435
>          --- l  ekb(1:nproj) -->
>              1    0.397139
>              2   -2.848780
>  pspatm: atomic psp has been read  and splines computed
>  
> - pspatm: opening atomic psp file    ./23v.pspnc
>   Troullier-Martins psp for element  V         Thu Oct 27 17:33:59 EDT
> 1994
>   23.00000   5.00000    940714                znucl, zion, pspdat
>     1    1    2    0      2001   0.00000
> pspcod,pspxc,lmax,lloc,mmax,r2well
>     0   6.704  10.456    0   2.1748425        l,e99.0,e99.9,nproj,rcpsp
>    0.00000000   0.00000000   0.00000000   0.00000000     rms, ekb1, ekb2,
> epsatm
>     1   3.062   5.016    1   2.6897970        l,e99.0,e99.9,nproj,rcpsp
>    0.00000000   0.00000000   0.00000000   0.00000000     rms, ekb1, ekb2,
> epsatm
>     2  21.816  28.020    1   1.9926227        l,e99.0,e99.9,nproj,rcpsp
>    0.00000000   0.00000000   0.00000000   0.00000000     rms, ekb1, ekb2,
> epsatm
>     2.54061776212796    0.14301071947837    2.04660280594678
> rchrg,fchrg,qchrg
>  pspatm: epsatm=   66.12637516
>          --- l  ekb(1:nproj) -->
>              1    1.496370
>              2   -9.013216
>  pspatm: atomic psp has been read  and splines computed
>  
> - pspatm: opening atomic psp file    ./8o.pspnc
>   Troullier-Martins psp for element  O         Thu Oct 27 17:29:57 EDT
> 1994
>    8.00000   6.00000    940714                znucl, zion, pspdat
>     1    1    1    1      2001   0.00000
> pspcod,pspxc,lmax,lloc,mmax,r2well
>     0   5.480  16.893    1   1.4482335        l,e99.0,e99.9,nproj,rcpsp
>    0.00000000   0.00000000   0.00000000   0.00000000     rms, ekb1, ekb2,
> epsatm
>     1  20.911  28.075    0   1.4482335        l,e99.0,e99.9,nproj,rcpsp
>    0.00000000   0.00000000   0.00000000   0.00000000     rms, ekb1, ekb2,
> epsatm
>     0.56990156784787    3.95561236318184    0.63894027514378
> rchrg,fchrg,qchrg
>  pspatm: epsatm=    1.57752239
>          --- l  ekb(1:nproj) -->
>              0    5.670783
>  pspatm: atomic psp has been read  and splines computed
> ---
> 
> 
> 
> 
> I didn't check over any otehr possible compound, mainly because many of
> the
> examples I have found for Vanadium are oxides, which the electronic
> correlation is important and the agrement could be bad... I just went
> straigth to the system I am interested of...
> 
> 
> I have not been able to use OPIUM pseudos due to an erro in my compilation
> in my
> linux system and I have not been able to solve it.
> 
> Thansk again and I hope this will clarify some of my problems..
> 
> Best regards
> 
> -aldo.