[Pw_forum] Generation of H and O GIPAW PP

[Davide Ceresoli]

Dear Jérôme,
     the two output files you attached do not show large differences,
The core contribution is only for oxygen, in both cases.

What can lead to wrong/inaccurate results is the linear dependence
between GIPAW projectors (in your "not-good" calculation). Moreover,
I strongly advise to generate always two projectors per angular
momentum in the &test section.

I've never tried to generate hydrogen with 'p' locality. Maybe someone
else can comment. In any case, I'm attaching original the input files
for the hydrogen and oxygen pseudopotentials. You can try to add to them.

Best wishes,
     Davide


On 08/28/2012 02:37 PM, jérôme cuny wrote:
> Dear quantum-espresso users,
>
> I would like to share with you a problem I have for which I hope you could help me.
>
> I am trying to generate H and O norm-conserving PP to calculate the NMR
> properties of water molecules using the GIPAW module.
> I am using the 4.3.2 version of both quantum-espresso and gipaw (I join to this
> e-mail the make.sys file I used for the compilation).
>
> My problem is that, even if the generation of the PPs seems fine, their use by
> the gipaw module leads to bad results.
> I consider as good results the results I obtain using the H and O PP provided in
> the example folder of the gipaw module (see file water-nmr_1.log).
>
> Using the input file H.pbe-tm_good.in for the PP generation of H, the result I
> obtain for one water molecule in a box is not too bad (see water-nmr_1_good.log)
>
> However, my problem appear when I want to use as local part of the PP, the p
> orbital (see H.pbe-tm_bad.in and water-nmr_1_bad.log). Strangely, when I do
> this, I always obtain a contribution to the total chemical shit coming from some
> core orbitals! The same problem occur when I generate the O PP (in that case I
> use as input, the data I found in a presentation on GIPAW, see fig
> O_PP_Slide.tiff) as the core contribution become really different from the one
> of the reference calculation.
>
> So, does somebody see what I do wrong in my PP generation? Or, of course, is
> there any other source of error that could lead to this problem?
>
> I have checked the previous postings on the subject (GIPAW+PP) and I have not
> find one that answered my question. If I missed the good one, I am sorry.
>
> Best Regards,
>
> Jérôme Cuny
> ETH-Zurich/USI-Campus
>
> --------------------------------------------------------------------
> Jérôme Cuny
> Department of Chemistry and Applied Biochemistry
> ETH-Zurich
> USI-Campus, Via Giuseppe Buffi 13
> Computational Sciences/Parrinello
> 6900 Lugano, Switzerland
> Tel : +41 (0) 58666-4802       Fax : +41 (0) 58666-4817
> --------------------------------------------------------------------
>
>

-------------- next part --------------
 &input
        title = 'H'
	prefix = 'h'
        zed = 1
        dft = 'PBE'
	rel = 0
        iswitch = 3
        beta = 0.2
        xmin = -8.0, dx = 0.01
        nld = 3, rlderiv = 2.2, eminld  = -5.0, emaxld  = 5.0, deld = 0.005
 /
2
1S  1  0  0.5  1
2S  2  0  0.0  1
 &inputp
	pseudotype = 1
   	lloc = 0
   	tm = .true.
   	file_pseudopw = 'H.pbe-tm-new-gipaw.UPF'
        lgipaw_reconstruction = .true.
        upf_v1_format = .true.
        author = 'D.C.'
 /
1
1S  1  0  0.5  0.00  0.90  0.90
 &test
   ecutmin = 60
   ecutmax = 120
   decut   = 10
 /
2
1S  1  0  0.50  0.00  0.90  0.90
2S  2  0  0.00  0.00  0.90  0.90

-------------- next part --------------
 &input
        title = 'O'
	prefix = 'o'
        zed = 8
        dft = 'PBE'
	rel = 1
        iswitch = 3
        beta = 0.2
        xmin = -8.0, dx = 0.01
        nld = 3, rlderiv = 2.0, eminld  = -5.0, emaxld  = 5.0, deld = 0.005
 /
5
1S  1  0  2.0  1
2S  2  0  2.0  1
2P  2  1  4.0  1
3S  3  0  0.0  1
3P  3  1 -1.0  1
 &inputp
	pseudotype = 1
   	lloc = 1
   	tm = .true.
   	file_pseudopw = 'O.pbe-tm-new-gipaw.UPF'
        lgipaw_reconstruction = .true.
        upf_v1_format = .true.
        author = 'D.C.'
 /
2
2S  1  0  2.0  0.00  1.40  1.40
2P  2  1  4.0  0.00  1.40  1.40
 &test
   ecutmin = 60
   ecutmax = 120
   decut   = 10
 /
4
2S  1  0  2.00  0.00  1.40  1.40
3S  2  0  0.00  0.00  1.40  1.40
2P  2  1  4.00  0.00  1.40  1.40
3P  3  1  0.00 -0.10  1.40  1.40