[Wannier] Dissimilar results in band structures for a system obtained from Wannier90

Aron Szabo szaboa at iis.ee.ethz.ch
Mon Jun 29 10:52:12 CEST 2015


Dear Seyed,

I've managed to get nicely converged Wannier functions for phosphorene 
by using sp3 projections on each 4 atoms, therefore including 6 
conduction bands in the calculation too with the disentanglement method. 
The largest spread was 2 A^2. In my experience it should not be much 
larger than 3 A^2 for most semiconductors. And the imaginary part of the 
Hamiltonian should be practically 0 everywhere (or in the order of 
magnitude of 1e-6). I was using 21x21x1 k points. I did not check with 
an even mesh, but sometimes it matters a lot for the wannierization 
whether you use even or odd mesh. I've used VASP, and I'm not an expert 
with espresso, and I did not look into you input and output files. Just 
wanted share my experience with phosphorene.

Best,
Aron

On 06/29/2015 09:38 AM, Seyed Mojtaba Rezaei Sani wrote:
> Dear Nicola,
>
> Thanks. Phosphorene has four P atoms in its unit cell so considering 
> 10 valence bands and therefore 10 WFs, we can't fit a projection of s 
> orbital on every atom and bond to this number.
> But I have found new thing. Instead of increasing the number of 
> kpoints in wannier input, I chose a coarser grid which seems to be 
> fine enough to provide converged WFs with both versions of codes. The 
> resulted band structure completely matches the PWscf one but there is 
> an error in transport section implies that there are 0 unit cells and 
> Wfs inside the principal layer:
>
>  *---------------------------------------------------------------------------*
>  | TRANSPORT                                    |
>  *---------------------------------------------------------------------------*
>
>
>  Calculation of Quantum Conductance and DoS: bulk mode
>
>  Maximum imaginary part of the real-space Hamiltonian: 0.189159
>
>  ------------------------------------------------------------------------------
>      Maximum real part of the real-space Hamiltonian at each lattice point
> --------------------------------------------------------------
>             Lattice point R           Max |H_ij(R)|
>                   -1                     0.000000
>                    0                     7.773991
>                    1                     0.000000
> --------------------------------------------------------------
>
>  Number of unit cells inside the principal layer:     0
>  Number of Wannier Functions inside the principal layer: 0
>
>  Calculating quantum conductance and density of states...
>  ERROR:  IN ZGESV IN tran_transfer, INFO=          -4
>  Exiting.......
>  tran_transfer: problem in ZGESV 1
>
>
>
> in/output files are attached. Any help would be greatly appreciated.
> I have a stupid question too, What are the satisfying values for the 
> imaginary part and spread of WFs?
>
>
>
> On Thu, Jun 25, 2015 at 11:47 PM, Nicola Marzari 
> <nicola.marzari at epfl.ch <mailto:nicola.marzari at epfl.ch>> wrote:
>
>
>
>     Dear Seyed,
>
>
>     any news on this?
>
>                             nicola
>
>
>
>
>     On 15/06/2015 14:17, Nicola Marzari wrote:
>
>
>         Thanks.
>
>         At first sight the two espresso codes give the same
>         eigenvalues in the
>         nscf,
>         but being run with two different parallelization choices, the
>         phases of the wannier functions will all be different, so
>         it's like dealing with two independent calculations. So the
>         question is
>         for Wannier - to see if you have converged to the correct
>         localized state,
>         or not. Usually, if the projections are correct, it shouldn't
>         matter, but
>         maybe in this case they are not correct, and hten the codes
>         randomly
>         converges, or not.
>
>         One of the outputs records a large imaginary part - that's a
>         bad sign.
>         The other has two much less localized wavefunctions - maybe
>         good, maybe
>         bad.
>
>         Can you try with projections that have a s orbital at every
>         atom, and
>         a s orbital at every bond? Those should work, and it would be nice
>         to see that both codes give the same results.
>
>         Otherwise, we'll need to dig deeper.
>
>                      nicola
>
>
>
>
>         On 15/06/2015 13:41, Seyed Mojtaba Rezaei Sani wrote:
>
>             Hi Nicola
>
>             Thanks a lot for your kind response. Surely I did checked
>             outputs before
>             too. There was no considerable difference between them and
>             also there is
>             no unconverged band message. So I can say:
>
>             Re 1) Every thing works perfectly both in 5.1 and 5.1.2
>             versions except
>             the disappointing obtained wannier90 band structure by the
>             later.
>             Re 2) No difference.
>
>             As more details, please find all in/output files for both
>             versions
>             (except UNK ones which are too big) here:
>
>             https://www.dropbox.com/sh/xhd3etb6zfz2e7c/AAClTpe-aGkwXVrzDcIq6L7Aa?dl=0
>
>             Thanks again,
>
>             On Mon, Jun 15, 2015 at 12:23 PM, Nicola Marzari
>             <nicola.marzari at epfl.ch <mailto:nicola.marzari at epfl.ch>
>             <mailto:nicola.marzari at epfl.ch
>             <mailto:nicola.marzari at epfl.ch>>> wrote:
>
>
>
>                 Hi Sani,
>
>
>                 I had mentioned in my email that you should check both
>                 your input and your output files - since the problems
>             seem due
>                 to the change in version of quantum-espresso (I do
>             appreciate you
>                 did check your input files, but I do wonder why you didn't
>                 read the email in full, since it was asking about the
>             output).
>
>                 So - do you see any difference in the output files of
>             quantum
>             espresso?
>
>                  From your plots my first guess would be that you have
>             some states
>             that
>                 are not converged at some k-point.
>
>                 1) can you confirm that in an older version of QE
>             everything
>                 worked perfectly (*which one*?, and in 5.1.2 you have
>             the problems
>                 below?
>
>                 2) do the outputs in the two cases differ - for the QE
>             runs, and
>             for the
>                 wannier runs, when run with *identical* inputs? If so,
>             in what do
>                 they differ.
>
>                 Let us know re 1) and 2) - thanks a lot,
>
>                                          nicola
>
>
>
>                 On 15/06/2015 09:40, Seyed Mojtaba Rezaei Sani wrote:
>
>                     Dear all
>
>                     I am looking forward to receive any kind help.
>
>                     On Wed, Jun 10, 2015 at 2:59 PM, Seyed Mojtaba
>             Rezaei Sani
>                     <s.m.rezaeisani at gmail.com
>             <mailto:s.m.rezaeisani at gmail.com>
>             <mailto:s.m.rezaeisani at gmail.com
>             <mailto:s.m.rezaeisani at gmail.com>>
>                     <mailto:s.m.rezaeisani at gmail.com
>             <mailto:s.m.rezaeisani at gmail.com>
>                     <mailto:s.m.rezaeisani at gmail.com
>             <mailto:s.m.rezaeisani at gmail.com>>>> wrote:
>
>                          Dear Prof. Marzari
>
>                          As you suggested I checked input files and
>             they were exactly
>                          identical. but the difference in result remains.
>                          I tried to make a change in calculations so I
>             increased
>                     number of
>                          kpoints in non scf and wannier calculations from
>                          a mesh of 12*12*1 to 16*16*1. The result was
>             an improvement
>                     in band
>                          structure. you can find the plot attached. But
>                          there are still some inadequacies to the pw
>             band in the
>                     region gamma
>                          to X.
>                          Again I increased the grid to a 24*24*1 mesh
>             which
>                     encountered this
>                          error:
>
>                          Unable to satisfy B1 with any of the first 
>             12 shells
>                            Your cell might be very long, or you may
>             have an
>                     irregular MP grid
>                            Try increasing the parameter search_shells
>             in the win
>                     file (default=12)
>
>                          Setting the parameter search_shells = 60
>             solved the problem
>                     but now
>                          inadequacies are in the region S to Y (the
>             plot is
>             attached).
>                          I don't know if there is any parameter that I
>             could change
>                     to solve
>                          this problem. Please give me any probable hint.
>                          Thanks,
>
>                          On Wed, Jun 10, 2015 at 2:53 PM, Seyed
>             Mojtaba Rezaei Sani
>                          <s.m.rezaeisani at gmail.com
>             <mailto:s.m.rezaeisani at gmail.com>
>             <mailto:s.m.rezaeisani at gmail.com
>             <mailto:s.m.rezaeisani at gmail.com>>
>                     <mailto:s.m.rezaeisani at gmail.com
>             <mailto:s.m.rezaeisani at gmail.com>
>                     <mailto:s.m.rezaeisani at gmail.com
>             <mailto:s.m.rezaeisani at gmail.com>>>> wrote:
>
>                              Dear Prof. Marzari
>
>                              As you suggested I checked input files
>             and they were
>                     exactly
>                              identical. but the difference in result
>             remains.
>                              I tried to make a change in calculations
>             so I increased
>                     number
>                              of kpoints in non scf and wannier
>             calculations from
>                              a mesh of 12*12*1 to 16*16*1. The result
>             was an
>                     improvement in
>                              band structure. you can find the plot
>             attached. But
>                              there are still some inadequacies to the
>             pw band in the
>                     region
>                              gamma to X.
>                              Again I increased the grid to a 24*24*1
>             mesh which
>                     encountered
>                              this error:
>
>                              Unable to satisfy B1 with any of the
>             first  12 shells
>                                Your cell might be very long, or you
>             may have an
>                     irregular MP grid
>                                Try increasing the parameter
>             search_shells in the win
>                     file (default=12)
>
>                              Setting the parameter search_shells = 60
>             solved the
>                     problem but
>                              now inadequacies are in the region S to Y
>             (the plot is
>                     attached).
>                              I don't know if there is any parameter
>             that I could
>                     change to
>                              solve this problem. Please give me any
>             probable hint.
>                              Thanks,
>
>                              On Thu, May 28, 2015 at 12:39 PM, Nicola
>             Marzari
>                              <nicola.marzari at epfl.ch
>             <mailto:nicola.marzari at epfl.ch>
>             <mailto:nicola.marzari at epfl.ch
>             <mailto:nicola.marzari at epfl.ch>>
>                     <mailto:nicola.marzari at epfl.ch
>             <mailto:nicola.marzari at epfl.ch>
>             <mailto:nicola.marzari at epfl.ch
>             <mailto:nicola.marzari at epfl.ch>>>>
>                     wrote:
>
>
>
>                                  Well, the obvious thing to do is
>             making sure your
>                                  inputs are exactly identical, you
>             re-run the
>                     calculations
>                                  to make sure the error shows up
>             again, and then you
>                     start
>                                  doing a diff in the output files of
>                     quantum-espresso, and
>                                  see where there is a difference.
>
>               nicola
>
>
>
>
>                                  On 28/05/2015 09:07, Seyed Mojtaba
>             Rezaei Sani
>             wrote:
>
>                                      Dear Wannier90 users,
>
>                                      I am trying to calculate wannier
>             functions for
>                                      phosphorene using
>                                      wannier90. first I tried to plot
>             band structure
>                     using both
>                                      quantum-espresso and wannier90.
>             Results were in
>                     good
>                                      agreement. Now I
>                                      have upgraded espresso to 5.1.2
>             version but
>                     with the
>                                      same inputs the
>                                      obtained results are not similar.
>                                      All files are attached and here
>             in .win input:
>
>                                      num_bands        = 10
>                                      num_wann         = 10
>                                      num_iter         = 500
>
>                                      Begin Unit_Cell_Cart
>                                      Bohr
>                                            6.282237122  0.000000000
>             0.0000000000
>                                            0.000000000  8.651074073
>             0.0000000000
>                                            0.000000000  0.000000000
>             26.456165734
>                                      End Unit_Cell_Cart
>
>                                      Begin Atoms_Frac
>                                      P1        0.010000000  0.086656382
>             0.152592466
>                                      P1        0.010000000  0.913343530
>             0.001000000
>                                      P2        0.510000000  0.413343618
>             0.152592466
>                                      P2        0.510000000  0.586656470
>             0.001000000
>                                      End Atoms_Frac
>
>                                      Begin Projections
>                                      P1:sp3
>                                      P2:py
>                                      End Projections
>
>                                      bands_plot = true
>
>                                      begin kpoint_path
>                                      G 0.0000 0.0000 0.0000 X 0.5000
>             0.0000 0.0000
>                                      x 0.5000 0.0000 0.0000 S 0.5000
>             0.5000 0.0000
>                                      S 0.5000 0.5000 0.0000 Y 0.0000
>             0.5000 0.0000
>                                      Y 0.0000 0.5000 0.0000 G 0.0000
>             0.0000 0.0000
>                                      end kpoint_path
>
>                                      mp_grid = 12 12 1
>
>                                      begin kpoints
>                                          0.00000000  0.00000000 0.00000000
>                                         .
>                                         .
>                                         .
>                                          0.91666667  0.91666667 0.00000000
>                                      End Kpoints
>
>                                      I would greatly appreciate it if
>             you kindly
>                     give me some
>                                      hints.
>
>                                      Best regards,
>
>                                      --
>                                      Seyed Mojtaba Rezaei Sani
>
>                                      Institute for Research in
>             Fundamental Sciences
>                     (IPM)
>                                      School of Nano-Science
>                                      Shahid Farbin Alley
>                                      Shahid Lavasani st
>                                      P.O. Box 19395-5531
>                                      Tehran, Iran
>                                      Tel: +98 21 2310  (3069)
>
>
>              _______________________________________________
>                                      Wannier mailing list
>             Wannier at quantum-espresso.org
>             <mailto:Wannier at quantum-espresso.org>
>             <mailto:Wannier at quantum-espresso.org
>             <mailto:Wannier at quantum-espresso.org>>
>                                    
>              <mailto:Wannier at quantum-espresso.org
>             <mailto:Wannier at quantum-espresso.org>
>                     <mailto:Wannier at quantum-espresso.org
>             <mailto:Wannier at quantum-espresso.org>>>
>
>             http://mailman.qe-forge.org/cgi-bin/mailman/listinfo/wannier
>
>
>                                  --
>
>
>
>             ----------------------------------------------------------------------
>                                  Prof Nicola Marzari, Chair of Theory and
>             Simulation of
>                                  Materials, EPFL
>                                  Director, National Centre for
>             Competence in
>                     Research NCCR
>                                  MARVEL, EPFL
>
>
>
>
>                              --
>                              Seyed Mojtaba Rezaei Sani
>
>                              Institute for Research in Fundamental
>             Sciences (IPM)
>                              School of Nano-Science
>                              Shahid Farbin Alley
>                              Shahid Lavasani st
>                              P.O. Box 19395-5531
>                              Tehran, Iran
>                              Tel: +98 21 2310  (3069)
>
>
>
>
>                          --
>                          Seyed Mojtaba Rezaei Sani
>
>                          Institute for Research in Fundamental
>             Sciences (IPM)
>                          School of Nano-Science
>                          Shahid Farbin Alley
>                          Shahid Lavasani st
>                          P.O. Box 19395-5531
>                          Tehran, Iran
>                          Tel: +98 21 2310  (3069)
>
>
>
>
>                     --
>                     Seyed Mojtaba Rezaei Sani
>
>                     Institute for Research in Fundamental Sciences (IPM)
>                     School of Nano-Science
>                     Shahid Farbin Alley
>                     Shahid Lavasani st
>                     P.O. Box 19395-5531
>                     Tehran, Iran
>                     Tel: +98 21 2310  (3069)
>
>
>                 --
>
>
>             ----------------------------------------------------------------------
>                 Prof Nicola Marzari, Chair of Theory and Simulation of
>             Materials,
>             EPFL
>                 Director, National Centre for Competence in Research
>             NCCR MARVEL,
>             EPFL
>
>
>
>
>             --
>             Seyed Mojtaba Rezaei Sani
>
>             Institute for Research in Fundamental Sciences (IPM)
>             School of Nano-Science
>             Shahid Farbin Alley
>             Shahid Lavasani st
>             P.O. Box 19395-5531
>             Tehran, Iran
>             Tel: +98 21 2310  (3069)
>
>
>
>     -- 
>
>     ----------------------------------------------------------------------
>     Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL
>     Director, National Centre for Competence in Research NCCR MARVEL, EPFL
>
>
>
>
> -- 
> Seyed Mojtaba Rezaei Sani
>
> Institute for Research in Fundamental Sciences (IPM)
> School of Nano-Science
> Shahid Farbin Alley
> Shahid Lavasani st
> P.O. Box 19395-5531
> Tehran, Iran
> Tel: +98 21 2310  (3069)
>
>
> _______________________________________________
> Wannier mailing list
> Wannier at quantum-espresso.org
> http://mailman.qe-forge.org/cgi-bin/mailman/listinfo/wannier

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/wannier/attachments/20150629/9f9f7f76/attachment.html>


More information about the Wannier mailing list