[Thermo_pw-forum] nonstoichiometric compounds
Andrea Dal Corso
dalcorso at sissa.it
Mon Oct 17 17:45:29 CEST 2016
On Sun, 2016-10-16 at 16:41 +0200, Gunnar Palsson wrote:
> Dear Andrea,
>
> I seem to be stuck with this problem. I tried setting ESPRESSO_TMPDIR instead of specifying outdir in the scf and ph files but I still get the same error.
> Any advice would be most helpful? I was also able to run a manual phonon run and plot the dispersion and the densites of states, which look fine. As a workaround, is it possible
> for thermo_pw to calculate the thermodynamic quantities from the dynamic matrices generated from ph.x directly without having run thermo_pw.x?
>
Dear Gunnar,
Thank you for reporting the problem.
I have analyzed your input file and the problem is that the automatic
path generated by thermo_pw is not very clever when some parts of the
path are much shorter than others as in your case. The code should be
able to deal with this, but presently it is not.
There are two solutions to your problem:
Possibility one: give explicitely the path where to compute the phonon
dispersions in the input of thermo_pw. The same path that works in QE
should work also here.
Possibility number two. A quick and dirty fix for your case:
Change the routine thermo_pw/src/sym_band_sub.f90 at line 291 there is
factor_dx = MAX(5.0_DP, 2.0_DP * sizeb, 2.0_DP * sizec, &
2.0_DP/sizeb, 2.0_DP/sizec)
change the 5.0_DP to 6.0_DP.
The same change should be done in thermo_pw/src/plotband_sub.f90
at line 462. Then recompile.
Thermo_pw can use the dynamical matrices calculated in a previous run,
or using the phonon code. You need to use the two input variables
after_disp=.TRUE.
and fildyn='name of the dynamical matrix used in the ph.x input'
in the thermo_pw input.
Note however that you need to make the change written above or the code
will stop at the same point even if you read the dynamical matrices from
file.
HTH,
Andrea
> Best,
> Gunnar
>
> > On 11 Oct 2016, at 00:08, Gunnar Palsson <gunnar.karl at gmail.com> wrote:
> >
> > Andrea,
> >
> > I compiled qe-6.0 and thermo_pw 0.6.0, deleted the output folder, and repeated the calculation. I still get the same error. Then I tried calculating pure vanadium with as similar
> > settings as possible which went fine.
> >
> > Attached are the input and output files.
> >
> >
> >
> > Best,
> > Gunnar
> > <ph_control><thermo_control><v2h.scf.in><v2h.scf.out>
> >> On 10 Oct 2016, at 11:03, Andrea Dal Corso <dalcorso at sissa.it> wrote:
> >>
> >> It is a strange error. It seems that your working directory was not
> >> empty. The output has no information.
> >> I would try with the last version of thermo_pw and with a clean working
> >> directory. If the problem is still there I will check what is happening.
> >>
> >> Andrea
> >>
> >>
> >> On Mon, 2016-10-10 at 10:51 +0200, Gunnar Palsson wrote:
> >>> Andrea,
> >>>
> >>> Thank you very much for explaining things very clearly! I was able to create an input file based on the space group and Wyckoff symbols. It now works nicely to relax the structure and calculate the elastic constants. However when I did a phonon calculation I got a message that I do not know how to interpret:
> >>>
> >>> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
> >>> Error in routine plotband_sub (1):
> >>> unexpected change of symmetry
> >>> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
> >>>
> >>> Any idea of how to proceed from here?
> >>>
> >>>
> >>> Best regards,
> >>> Gunnar
> >>>
> >>>
> >>>> On 8 Oct 2016, at 16:02, Andrea Dal Corso <dalcorso at sissa.it> wrote:
> >>>>
> >>>> This is not a problem of thermo_pw. If you use the same input with pw.x you will have only four symmetry operations as well. The problem is that using ibrav=6 with 24 atoms the
> >>>> code finds that it is a supercell and neglects all symmetries with fractional translations.
> >>>> Although slow however this cell is correct and should provide reasonable results.
> >>>> thermo_pw just tells you that you are not exploiting correctly the symmetry.
> >>>> If you use ibrav=7 without converting the coordinates to the centered cell, the coordinates are
> >>>> wrong because they are in the crystal basis of the conventional cell and the atoms should be 12.
> >>>> cif2qe should now have an option that allow you to obtain the crystal coordinates for the centered cell.
> >>>>
> >>>> The alternative is to use the QE ability to deal directly with the space group.
> >>>> You have to specify space_group=141 and origin_choice=2 in the system namelist, to use
> >>>> nat=2 because there are only two inequivalent atoms and specify the atomic coordinates as crystal_sg,
> >>>> giving
> >>>>
> >>>> V 16h 0.49964 0.26712
> >>>> H 8c
> >>>>
> >>>> extracted from your cif file. With this input QE finds a cell with ibrav=7,
> >>>> and 12 atoms and thermo_pw recognizes correctly the space group 141 and finds a point group D_4h
> >>>> compatible with the centered tetragonal lattice.
> >>>>
> >>>> HTH,
> >>>>
> >>>> Andrea
> >>>>
> >>>>
> >>>>
> >>>>
> >>>> Quoting Gunnar Palsson <gunnar.karl at gmail.com>:
> >>>>
> >>>>> Dear Andrea,
> >>>>>
> >>>>>
> >>>>> I have obtained a cif file for the stoichiometric compound V2H and I have run cif2cell (1.2.10) to generate a pw input file. Since the cell is body centered tetragonal I changed the Bravais lattice to 7. When running thermo_pw to calculate the phonon dispersion, the program mentions that there is an incompatibility of the point group with the Bravais lattice.
> >>>>>
> >>>>> However if I run the cif2cell with the following option: cif2cell v2h.cif -p pwscf --no-reduce --print-symmetry-operations I get 32 symmetry operations.
> >>>>>
> >>>>> The output of thermo_pw (0.5.0) running on QE 5.4.0 is:
> >>>>>
> >>>>> C_i (-1) is incompatible with the centered tetragonal Bravais lattice
> >>>>> It is compatible with the
> >>>>> triclinic Bravais lattice; ibrav= 14
> >>>>>
> >>>>> You might want to change the Bravais lattice or to
> >>>>>
> >>>>> understand why the symmetries are wrong beforecontinuing
> >>>>> The point group or the Laue class are not used to reduce the number of
> >>>>> computed tensor components
> >>>>> Found symmetry operation: I + ( 0.5000 -0.5000 -0.5000)
> >>>>> This is a supercell, fractional translations are disabled
> >>>>>
> >>>>> FFT mesh: ( 50, 50, 50 )
> >>>>>
> >>>>> 2 Sym. Ops., with inversion, found
> >>>>>
> >>>>>
> >>>>> s frac. trans.
> >>>>>
> >>>>> isym = 1 identity
> >>>>>
> >>>>> cryst. s( 1) = ( 1 0 0 )
> >>>>> ( 0 1 0 )
> >>>>> ( 0 0 1 )
> >>>>>
> >>>>> cart. s( 1) = ( 1.0000000 0.0000000 0.0000000 )
> >>>>> ( 0.0000000 1.0000000 0.0000000 )
> >>>>> ( 0.0000000 0.0000000 1.0000000 )
> >>>>>
> >>>>>
> >>>>> isym = 2 inversion
> >>>>>
> >>>>> cryst. s( 2) = ( -1 0 0 )
> >>>>> ( 0 -1 0 )
> >>>>> ( 0 0 -1 )
> >>>>>
> >>>>>
> >>>>> cart. s( 2) = ( -1.0000000 0.0000000 0.0000000 )
> >>>>> ( 0.0000000 -1.0000000 0.0000000 )
> >>>>> ( 0.0000000 0.0000000 -1.0000000 )
> >>>>>
> >>>>>
> >>>>> point group C_i (-1)
> >>>>> there are 2 classes
> >>>>> the character table:
> >>>>>
> >>>>> E i
> >>>>> A_g 1.00 1.00
> >>>>> A_u 1.00 -1.00
> >>>>>
> >>>>>
> >>>>> Changing the Bravais lattice to 6 yields a different response:
> >>>>>
> >>>>> C_2h (2/m) is incompatible with the tetragonal Bravais lattice
> >>>>> It is compatible with the
> >>>>> monoclinic (c unique) Bravais lattice; ibrav= 12
> >>>>> base centered monoclinic (c unique) Bravais lattice; ibrav= 13
> >>>>> monoclinic (b unique) Bravais lattice; ibrav= -12
> >>>>> base centered monoclinic (b unique) Bravais lattice; ibrav= -13
> >>>>>
> >>>>> You might want to change the Bravais lattice or to
> >>>>>
> >>>>> understand why the symmetries are wrong beforecontinuing
> >>>>> The point group or the Laue class are not used to reduce the number of
> >>>>> computed tensor components
> >>>>> Found symmetry operation: I + ( -0.5000 -0.5000 -0.5000)
> >>>>> This is a supercell, fractional translations are disabled
> >>>>>
> >>>>> FFT mesh: ( 60, 60, 64 )
> >>>>>
> >>>>> 4 Sym. Ops., with inversion, found
> >>>>>
> >>>>>
> >>>>> s frac. trans.
> >>>>>
> >>>>> isym = 1 identity
> >>>>>
> >>>>> cryst. s( 1) = ( 1 0 0 )
> >>>>> ( 0 1 0 )
> >>>>> ( 0 0 1 )
> >>>>>
> >>>>> cart. s( 1) = ( 1.0000000 0.0000000 0.0000000 )
> >>>>> ( 0.0000000 1.0000000 0.0000000 )
> >>>>> ( 0.0000000 0.0000000 1.0000000 )
> >>>>>
> >>>>>
> >>>>> isym = 2 180 deg rotation - cart. axis [1,0,0]
> >>>>>
> >>>>> cryst. s( 2) = ( 1 0 0 )
> >>>>> ( 0 -1 0 )
> >>>>> ( 0 0 -1 )
> >>>>>
> >>>>> cart. s( 2) = ( 1.0000000 0.0000000 0.0000000 )
> >>>>>
> >>>>> isym = 3 inversion
> >>>>>
> >>>>> cryst. s( 3) = ( -1 0 0 )
> >>>>> ( 0 -1 0 )
> >>>>> ( 0 0 -1 )
> >>>>>
> >>>>> cart. s( 3) = ( -1.0000000 0.0000000 0.0000000 )
> >>>>> ( 0.0000000 -1.0000000 0.0000000 )
> >>>>> ( 0.0000000 0.0000000 -1.0000000 )
> >>>>>
> >>>>>
> >>>>> isym = 4 inv. 180 deg rotation - cart. axis [1,0,0]
> >>>>>
> >>>>> cryst. s( 4) = ( -1 0 0 )
> >>>>> ( 0 1 0 )
> >>>>> ( 0 0 1 )
> >>>>>
> >>>>> cart. s( 4) = ( -1.0000000 0.0000000 0.0000000 )
> >>>>> ( 0.0000000 1.0000000 0.0000000 )
> >>>>> ( 0.0000000 0.0000000 1.0000000 )
> >>>>>
> >>>>>
> >>>>> point group C_2h (2/m)
> >>>>> there are 4 classes
> >>>>> the character table:
> >>>>>
> >>>>> I have spent a lot of time trying to figure this out but I am unable. Any advice would be most welcome,
> >>>>>
> >>>>> Best regards,
> >>>>> Gunnar
> >>>>>
> >>>>>
> >>>>>> On 16 Aug 2016, at 12:05, Andrea Dal Corso <dalcorso at sissa.it> wrote:
> >>>>>>
> >>>>>> On Sat, 2016-08-13 at 13:16 +0200, Gunnar Palsson wrote:
> >>>>>>> Dear Andrea,
> >>>>>>>
> >>>>>>> First I would like to tell you that I find your thermo_pw package extremely useful in many ways. I was inspired by your article J. Phys. Condensed Matter 26, 335401 (2014) and have been practising using your PS library v1.0 and the PAW potentials, for pure vanadium and niobium to calculate the thermoelastic properties. I have been thinking for quite a while about calculating the thermophysical properties and changes in free energy of metal hydrides, specifically vanadium hydride, in an attempt to determine the phase boundaries. I am also interested in the changes in the elastic constants with temperature and hydrogen. The question I have is whether one can use thermo_pw to calculate the Helmholtz free energy in the quasi harmonic approximation as a function of temperature for a supercell of vanadium with few hydrogen atoms. The size of the supercell determines what part of the phase diagram one can access in terms of concentration. So for 16 vanadium atoms, one can go from
> >>>>>> 1/
> >>>>>>> 16 to 16/16 in H/V concentration for example. It is my understanding that this requires ibrav=0, which may not be fully implemented? If not it would still be valuable to look into the stoichiometric hydride, V2H, and VH. The V2H belongs to space group 141. Finally, I have some neutron scattering data on hydrogen containing vanadium, and it would be extremely valuable to use the output of thermo_pw to calculate various neutron scattering cross sections and compare with experiment, which I can probably do.
> >>>>>>>
> >>>>>> Using ibrav=0 thermo_pw cannot use symmetry to reduce the number of
> >>>>>> tensor components that it calculates. Moreover it does not know the
> >>>>>> Bravais lattice number and presently it cannot calculate it, so it
> >>>>>> cannot find the automatic path, you have to provide it from input.
> >>>>>> I think it might also be possible to use ibrav=0 for phonon calculations
> >>>>>> and for the calculation of harmonic thermal properties, but all the
> >>>>>> rest, such as the anharmonic properties is most probably uncorrect. In
> >>>>>> any case I have not worked on this option so far, so I think it is very
> >>>>>> probable that there are problems.
> >>>>>>
> >>>>>> Andrea
> >>>>>>
> >>>>>>> Best regards,
> >>>>>>> Dr. Gunnar Palsson,
> >>>>>>> Researcher,
> >>>>>>> Uppsala University.
> >>>>>>> _______________________________________________
> >>>>>>> Thermo_pw-forum mailing list
> >>>>>>> Thermo_pw-forum at qeforge.qe-forge.org
> >>>>>>> http://qeforge.qe-forge.org/mailman/listinfo/thermo_pw-forum
> >>>>>>
> >>>>>> --
> >>>>>> Andrea Dal Corso Tel. 0039-040-3787428
> >>>>>> SISSA, Via Bonomea 265 Fax. 0039-040-3787249
> >>>>>> I-34136 Trieste (Italy) e-mail: dalcorso at sissa.it
> >>>>>>
> >>>>>> _______________________________________________
> >>>>>> Thermo_pw-forum mailing list
> >>>>>> Thermo_pw-forum at qeforge.qe-forge.org
> >>>>>> http://qeforge.qe-forge.org/mailman/listinfo/thermo_pw-forum
> >>>>
> >>>>
> >>>>
> >>>
> >>
> >> --
> >> Andrea Dal Corso Tel. 0039-040-3787428
> >> SISSA, Via Bonomea 265 Fax. 0039-040-3787249
> >> I-34136 Trieste (Italy) e-mail: dalcorso at sissa.it
> >>
> >
>
--
Andrea Dal Corso Tel. 0039-040-3787428
SISSA, Via Bonomea 265 Fax. 0039-040-3787249
I-34136 Trieste (Italy) e-mail: dalcorso at sissa.it
More information about the Thermo_pw-forum
mailing list