[Pw_forum] Fe pseudopotential (pslibrary) test failed

[Andrea Dal Corso]

On Fri, 2012-04-20 at 13:12 -0400, Peng Chen wrote:
> Dear All,
> 
> 
> Recently, I tested Fe.pz-spn-rrkjus.in and Fe.rel-pz-spn-rrkjus.in
> from pslibrary. There are two "problems" in the output:
> 
> 
> 1. the energy in the test with spherical waves is  a little lower than
> that from all electron calculation. Is it reasonable?   
> part of output:
>      n l     nl             e AE (Ry)        e PS (Ry)    De AE-PS
> (Ry)
>      1 0     3S   1( 2.00)       -6.87296       -6.87292
> -0.00004
>      2 1     3P   1( 6.00)       -4.40452       -4.40448
> -0.00004
>      2 0     4S   1( 2.00)       -0.40342       -0.40341
> -0.00001
>      3 2     3D   1( 6.00)       -0.56965       -0.56961
> -0.00003
>      3 1     4P   1( 0.00)       -0.10550       -0.10550
> -0.00000
> 
> 
>  Cutoff (Ry) :  110.0
>                            N = 1       N = 2       N = 3
>      E(L=0) =        -6.8963 Ry   -0.4136 Ry   -0.0008 Ry
>      E(L=1) =        -4.4514 Ry   -0.1311 Ry    0.0203 Ry
>      E(L=2) =        -0.5736 Ry    0.0340 Ry    0.0747 Ry
> 
> 
> Here -6.8963 < -6.87296, -4.4514 < -4.40452.
> 
> 
> 2. For some test configurations, the difference of PS and AE are very
> large. And some configuration tests even failed and give error  "from
> chargeps : error #         1      negative rho".
> 
> 
>      n l     nl             e AE (Ry)        e PS (Ry)    De AE-PS
> (Ry)
>      1 0     3S   1( 2.00)       -9.60696      -12.19076
>  2.58380
>      2 1     3P   1( 6.00)       -7.12183       -7.12183
>  0.00000
>      2 0     4S   1( 0.00)       -2.30301       -2.30301
>  0.00000
>      3 2     3D   1( 5.00)       -3.19807       -0.29435
> -2.90371
>      3 1     4P   1( 0.00)       -1.76774       -1.76774
>  0.00000
>  
> The inputs for Fe pseudopotential and my test are shown below, I am
> not sure if there is something wrong with my test input. Please let me
> know if it is not correct.
> 
Thank you for reporting this. Actually the Fe PP was tested for
configurations with 6 or 7 d electrons. Configurations with less
electrons in the d shell are not converging. This seems to be due to the
3p core radius. Decreasing it to 1.0 or 0.9 seems to work better
although the wavefunction cut-off increases, and the PP must be tested
again. Note however that for any PP, if your test configuration is too
far in energy from the configuration used to generate the PP, the test
will fail.


> 
> Fe.pz-spn-rrkjus.in:
> 
> 
> &input
>    title='Fe',
>    zed=26.,
>    rel=1,
>    config='[Ar] 4s2.0 3d6.0 4p0',
>    iswitch=3,
>    dft='LDA'
>  /
>  &inputp
>    pseudotype=3,
>    file_pseudopw='Fe.pz-spn-rrkjus.UPF',
>    author='ADC',
>    lloc=-2,
>    rcloc=1.8,
>    which_augfun='PSQ',
>    rmatch_augfun=0.68,
>    nlcc=.true.,
>    new_core_ps=.true.,
>    rcore=0.6,
>    tm=.true.
>  /
> 6
> 3S  1  0  2.00  0.00  1.10  1.30  0.0
> 4S  2  0  2.00  0.00  0.80  1.40  0.0
> 3P  2  1  6.00  0.00  1.00  1.30  0.0
> 4P  3  1  0.00  0.00  1.00  1.60  0.0
> 3D  3  2  6.00  0.00  1.40  2.00  0.0
> 3D  3  2  0.00 -0.40  1.40  2.00  0.0
> 
> 
> Fe.testpz-spn-rrkjus.in:
> 
> 
>  &input
>         iswitch=2,
> 
> 
>         rel=1,
>         zed=26.0,
>         config="[Ne] 3s2 3p6 4s2 3d6 4p0"
>         dft='LDA',
>  /
>  &test
>         file_pseudo="Fe.pz-spn-rrkjus.UPF",
>         nconf=8,
>         configts(1)="3s2 3p6 4s2 3d6 4p0",
>         configts(2)="3s2 3p6 4s2 3d5 4p0",
>         configts(3)="3s2 3p6 4s2 3d4 4p0",
>         configts(4)="3s2 3p6 4s2 3d3 4p0",
>         configts(5)="3s2 3p6 4s1 3d6 4p0",
>         configts(6)="3s2 3p6 4s0 3d6 4p0",
>         configts(7)="3s2 3p6 4s0 3d5 4p0",
>         configts(8)="3s2 3p6 4s0 3d4 4p0",
>        ecutmin=30,ecutmax=110,decut=20
>  /
> 
> 
> P.S. I didn't find Tb and Dy  pseudopotential  for Quantum Espresso,
> please kindly let me know if someone can provide or know  where to
> find them. I actually tried to create them using ld1.x, but the result
> is very bad and it is frustrating. 
> 
> 
We are still trying to tune the spd PPs. Those for f electron are not
yet available, at least within pslibrary.

HTH,

Andrea


> -- 
>   Best Regards.
>         Peng  
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum