[gpaw-users] vdW-DF2

Duy Le ttduyle at gmail.com
Tue Oct 5 16:47:46 CEST 2010


I don't know if anyone has any trick for the problem. I am surrendering to
it.
--------------------------------------------------
Duy Le
PhD Student
Department of Physics
University of Central Florida.

"Men don't need hand to do things"


On Wed, Sep 29, 2010 at 5:41 PM, Duy Le <ttduyle at gmail.com> wrote:

> Hi,
> Sure. I am sending a tar file, which contains input file, and 3 output
> files corresponding to vdW-DF, revPBE, vdW-DF2.
> Thank you for your help.
> Best,
> Link to the file: http://physics.ucf.edu/~dle/vdW-DF2.problem.tar
>
> --------------------------------------------------
> Duy Le
> PhD Student
> Department of Physics
> University of Central Florida.
>
> "Men don't need hand to do things"
>
>
> On Wed, Sep 29, 2010 at 5:26 PM, Duy Le <ttduyle at gmail.com> wrote:
>
>> Hi,
>> Sure. I am sending a tar file, which contains input file, and 3 output
>> files corresponding to vdW-DF, revPBE, vdW-DF2.
>> Thank you for your help.
>> Best,
>>
>> --------------------------------------------------
>> Duy Le
>> PhD Student
>> Department of Physics
>> University of Central Florida.
>>
>> "Men don't need hand to do things"
>>
>>
>> On Wed, Sep 29, 2010 at 3:13 AM, Jess Wellendorff Pedersen <
>> printerfejl at gmail.com> wrote:
>>
>>> Hi Duy,
>>>
>>> could you send us a s ample of the text output?
>>>
>>> / Jess
>>>
>>> Duy Le wrote:
>>>
>>>> Dear all,
>>>>
>>>> Did any one try this system? Whatever I did- such as: change mixer
>>>> parameter, eigeinsolver- I could not get the scf converged. If anyone has
>>>> any tip, please let me know.
>>>>
>>>> Best regards,
>>>> Duy
>>>> --------------------------------------------------
>>>> Duy Le
>>>> PhD Student
>>>> Department of Physics
>>>> University of Central Florida.
>>>>
>>>> "Men don't need hand to do things"
>>>>
>>>>
>>>> On Thu, Sep 23, 2010 at 6:41 PM, Duy Le <ttduyle at gmail.com <mailto:
>>>> ttduyle at gmail.com>> wrote:
>>>>
>>>>    Dear all,
>>>>    Sorry for my belated response. I was trying do it myself first but
>>>>    unfortunately I could not get any of them converged.
>>>>    Below are the input (input.py) and structure (MOL.xsf). With this
>>>>    input and structure, I have no problem with vdW-DF1 or revPBE or any
>>>>    other DFT functional but I can not get vdW-DF2 converged. The "Error
>>>>    in Wave Functions" was oscillating and it could not be smaller than
>>>>    1.0e-6.
>>>>
>>>>    Best regards,
>>>>    Duy
>>>>
>>>>    Here are the inputs:
>>>>    ~~~~ input.py
>>>>    ##/usr/bin/env python
>>>>    from ase import *
>>>>    from ase.lattice.surface import *
>>>>    from gpaw import GPAW
>>>>    from gpaw.utilities.tools import cutoff2gridspacing,
>>>>    gridspacing2cutoff
>>>>    from gpaw.vdw import FFTVDWFunctional
>>>>    from gpaw.poisson import PoissonSolver
>>>>    vdw = FFTVDWFunctional(nspins=1,exchange='X_PW86-None',Zab=-1.887)
>>>>    system=read('MOL.xsf')
>>>>    d=0
>>>>    output='MOL.'+str(d)+'.txt'
>>>>    tmpfile='MOL.'+str(d)+'.gpw'
>>>>    xsf='MOL.'+str(d)+'.xsf'
>>>>
>>>>    k=1
>>>>    ngx=96
>>>>    ngy=96
>>>>    ngz=112
>>>>
>>>>
>>>>    calc=GPAW(kpts=(k,k,1),
>>>>             xc=vdw,verbose=1,
>>>>             gpts=(ngx,ngy,ngz),
>>>>             convergence={'energy':0.0001,
>>>>                          'density':1.e-4,
>>>>    #                       'eigenstates': 1.0e-6, # default 1.0e-9
>>>>                          'band':'occupied'},
>>>>    #          eigensolver='cg',
>>>>    #          poissonsolver=PoissonSolver(nn=4, eps=1e-7),
>>>>             txt=output)
>>>>    system.set_calculator(calc)
>>>>    Etot = system.get_potential_energy()
>>>>
>>>>
>>>>    ~~~~ MOL.xsf
>>>>    ANIMSTEPS 1
>>>>    CRYSTAL
>>>>    PRIMVEC 1
>>>>     17.14730299493188 0.00000000000000 0.00000000000000
>>>>     0.00000000000000 17.32500000000000 0.00000000000000
>>>>     0.00000000000000 0.00000000000000 19.80000000000000
>>>>    PRIMCOORD 1
>>>>     15 1
>>>>     6     7.88929861131941     8.74279897333303     9.23519036609335
>>>>     6     9.29407452321463     8.66235256964292     9.22660802911678
>>>>     6     9.25953637200172     6.44054447324515     9.21345896750818
>>>>     6     7.19487577477512     7.50640945431773     9.23391747470668
>>>>     6     8.57448589108522    10.77139928082912     9.22321194485284
>>>>     7    10.03927789166028     7.54045396872909     9.21601308563562
>>>>     7     7.91048599589602     6.35523280726322     9.21957979402641
>>>>     7     7.45472256395586    10.06949990739891     9.23111842694947
>>>>     7     9.71946649561474     9.98338351912395     9.22313940828820
>>>>     7     5.83477936812194     7.43368881826066     9.24868794967438
>>>>     1     8.63391314599401    11.85291790879339     9.20920053359968
>>>>     1    10.68059547151664    10.30415787931885     9.19583460569330
>>>>     1     9.78701750033542     5.48653725614175     9.20204409378673
>>>>     1     5.38210093732843     6.53173729472345     9.20183809503345
>>>>     1     5.27897101044724     8.27578470491815     9.20575836661537
>>>>
>>>>    --------------------------------------------------
>>>>    Duy Le
>>>>    PhD Student
>>>>    Department of Physics
>>>>    University of Central Florida.
>>>>
>>>>    "Men don't need hand to do things"
>>>>
>>>>
>>>>
>>>>    On Fri, Sep 17, 2010 at 7:18 AM, Jess Wellendorff Pedersen
>>>>    <printerfejl at gmail.com <mailto:printerfejl at gmail.com>> wrote:
>>>>    > Hi Duy,
>>>>    >
>>>>    > The present GPAW approach to vdW-DF-2 calculations uses revPBE
>>>>    setups. I
>>>>    > have not myself experienced convergence problems with vdW-DF-2
>>>>    this way.
>>>>    > Convergence for the 148 G2-2 molecules is pretty much the same
>>>>    as for
>>>>    > vdW-DF-1, in my experience. Please provide a bit more info on your
>>>>    > calculations, especially which kind of molecules.
>>>>    >
>>>>    > /Jess
>>>>    >
>>>>    > Duy Le wrote:
>>>>    >>
>>>>    >> Dear all,
>>>>    >>
>>>>    >> I noticed that it is hard to get a SCF converged with vdW-DF2 for
>>>>    >> isolated molecules. In fact I could not get any of them
>>>>    converged. I
>>>>    >> have not seen this problem with older version of vdW-DF.
>>>>    >>
>>>>    >> There is no such problem in the case of a system containing
>>>>    molecules
>>>>    >> on metal surfaces.
>>>>    >>
>>>>    >> Best regards,
>>>>    >> Duy
>>>>    >>
>>>>    >> ********** example of input ****************
>>>>    >> #!/usr/bin/env python
>>>>    >> from ase import *
>>>>    >> from ase.io <http://ase.io> import *
>>>>
>>>>    >> from ase.lattice.surface import *
>>>>    >> from gpaw import GPAW, Mixer, FermiDirac
>>>>    >> from gpaw.utilities.tools import cutoff2gridspacing,
>>>>    gridspacing2cutoff
>>>>    >> from gpaw.vdw import FFTVDWFunctional
>>>>    >> ##vdw = FFTVDWFunctional(nspins=1,Nalpha=20, lambd=1.2,rcut=125.0,
>>>>    >> Nr=2048,size=None)
>>>>    >> vdw = FFTVDWFunctional(nspins=2,exchange='X_PW86-None',Zab=-1.887)
>>>>    >> system = read('POSCAR')
>>>>    >> system.set_pbc((1,1,1))
>>>>    >> output='output.txt'
>>>>    >> k=1
>>>>    >> ngx=152
>>>>    >> ngy=100
>>>>    >> ngz=180
>>>>    >> print ngx,ngy,ngz
>>>>    >> mm = [0] * len(system)
>>>>    >> mm[192-162]=-1.
>>>>    >> mm[195-162]=1.
>>>>    >> mm[196-162]=1.
>>>>    >> system.set_initial_magnetic_moments(mm)
>>>>    >>
>>>>    >> calc=GPAW(kpts=(k,k,1),
>>>>    >> #          occupations=FermiDirac(0.1),
>>>>    >> #          mixer=Mixer(beta=0.05, nmaxold=3, weight=50.0),
>>>>    >>          xc=vdw,spinpol=True,
>>>>    >>          gpts=(ngx,ngy,ngz),
>>>>    >>          convergence={'energy':0.0001,
>>>>    >>                       'density':1.e-4,
>>>>    >>                       'band':'occupied'},
>>>>    >>          txt=output)
>>>>    >> system.set_calculator(calc)
>>>>    >> GGA_energy = system.get_potential_energy()
>>>>    >>
>>>>    >>
>>>>    >>
>>>>    >> --------------------------------------------------
>>>>    >> Duy Le
>>>>    >> PhD Student
>>>>    >> Department of Physics
>>>>    >> University of Central Florida.
>>>>    >>
>>>>    >> "Men don't need hand to do things"
>>>>    >>
>>>>    >>
>>>>    >>
>>>>    >> On Mon, Aug 23, 2010 at 10:20 AM, Duy Le <ttduyle at gmail.com
>>>>    <mailto:ttduyle at gmail.com>> wrote:
>>>>    >>
>>>>    >>>
>>>>    >>> This's great news! I am also waiting for this.
>>>>    >>> Thanks.
>>>>    >>> --------------------------------------------------
>>>>    >>> Duy Le
>>>>    >>> PhD Student
>>>>    >>> Department of Physics
>>>>    >>> University of Central Florida.
>>>>    >>>
>>>>    >>> "Men don't need hand to do things"
>>>>    >>>
>>>>    >>>
>>>>    >>>
>>>>    >>> On Mon, Aug 23, 2010 at 5:32 AM, Jess Wellendorff Pedersen
>>>>    >>> <printerfejl at gmail.com <mailto:printerfejl at gmail.com>> wrote:
>>>>    >>>
>>>>    >>>>
>>>>    >>>> You're right, as compared to vdW-DF-1, the vdW-DF-2 is
>>>>    supposed to give
>>>>    >>>> better vdW energetics as well as better vdW binding distances.
>>>>    >>>>
>>>>    >>>> I will commit the code-changes within a few hours, drop you a
>>>>    mail when
>>>>    >>>> it's there. The changes are only to the Python-code
>>>>    (gpaw/vdw.py) so no
>>>>    >>>> recompilation of C/Fortran-code is needed for that particular
>>>>    update in
>>>>    >>>> itself.
>>>>    >>>>
>>>>    >>>> - Jess
>>>>    >>>>
>>>>    >>>> Michael F. Peintinger wrote:
>>>>    >>>>
>>>>    >>>>>
>>>>    >>>>> I would like to use it. I calculate Molecular Crystals and thus
>>>>    >>>>> van-der-Waals binding may be important. As I try to find
>>>>    structures I
>>>>    >>>>> would like to try it.
>>>>    >>>>> At the summer school there was a talk by Bengt where he
>>>>    showed, that
>>>>    >>>>> the
>>>>    >>>>> second version is a significant improvement!
>>>>    >>>>>
>>>>    >>>>> Regards
>>>>    >>>>>
>>>>    >>>>> Michael
>>>>    >>>>>
>>>>    >>>>>
>>>>    >>>>> On Mon, 2010-08-23 at 11:19 +0200, Jess Wellendorff Pedersen
>>>>    wrote:
>>>>    >>>>>
>>>>    >>>>>
>>>>    >>>>>>
>>>>    >>>>>> Hi Michael,
>>>>    >>>>>>
>>>>    >>>>>> no, it is not. However, if there is a general need for it,
>>>>    I can
>>>>    >>>>>> commit
>>>>    >>>>>> my personal modification of the vdW-DF code to the GPAW
>>>>    svn, such that
>>>>    >>>>>> one can use vdW-DF-2 like this:
>>>>    >>>>>> *****************
>>>>    >>>>>> from gpaw.vdw import FFTVDWFunctional
>>>>    >>>>>> vdw =
>>>>    FFTVDWFunctional(nspins=2,exchange='X_PW86-None',Zab=-1.887)
>>>>    >>>>>> calc = GPAW(......, xc=vdw,.......)
>>>>    >>>>>> *******************
>>>>    >>>>>> All one would then need to remember is to set the number of
>>>>    spins
>>>>    >>>>>> correctly, i.r. nspins=1 for spin-paired systems, and
>>>>    nspins=2 for
>>>>    >>>>>> spin-polarized systems.
>>>>    >>>>>>
>>>>    >>>>>> Regards,
>>>>    >>>>>> Jess Wellendorff
>>>>    >>>>>>
>>>>    >>>>>> Michael F. Peintinger wrote:
>>>>    >>>>>>
>>>>    >>>>>>
>>>>    >>>>>>>
>>>>    >>>>>>> Hi all,
>>>>    >>>>>>>
>>>>    >>>>>>> is the functional vdW-DF2 already in the svn?
>>>>    >>>>>>>
>>>>    >>>>>>> Best regards,
>>>>    >>>>>>>
>>>>    >>>>>>>
>>>>    >>>>>>> Michael
>>>>    >>>>>>>
>>>>    >>>>>>>
>>>>    >>>>>>>
>>>>    >>>>>>>
>>>>    >>>>>
>>>>    >>>>>
>>>>    >>>>
>>>>    >>>> _______________________________________________
>>>>    >>>> gpaw-users mailing list
>>>>    >>>> gpaw-users at listserv.fysik.dtu.dk
>>>>    <mailto:gpaw-users at listserv.fysik.dtu.dk>
>>>>
>>>>    >>>> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>>>>    >>>>
>>>>    >>>>
>>>>    >
>>>>
>>>>
>>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20101005/c5a784ef/attachment-0001.html 


More information about the gpaw-users mailing list